Oniguruma 6.9.4 fixes several CVEs.
- GD:
. Fixed bug #78849 (GD build broken with -D SIGNED_COMPARE_SLOW). (cmb)
+- MBString:
+ . Upgraded bundled Oniguruma to 6.9.4. (cmb)
+
- OPcache:
. Fixed potential ASLR related invalid opline handler issues. (cmb)
. Fixed $x = (bool)$x; with opcache (should emit undeclared variable notice).
History
+2019/11/29: Version 6.9.4
+
+2019/11/22: Release Candidate 3 for Version 6.9.4
+
+2019/11/20: fix a problem found by libFuzzer test
+2019/11/14: Release Candidate 2 for Version 6.9.4
+2019/11/12: fix integer overflow by nested quantifier
+2019/11/11: fix CVE-2019-19012: Integer overflow related to reg->dmax in search_in_range()
+2019/11/07: fix CVE-2019-19203: heap-buffer-overflow in gb18030_mbc_enc_len()
+2019/11/06: fix CVE-2019-19204: heap-buffer-overflow in fetch_interval_quantifier()
+2019/11/06: add HAVE_INTTYPES_H into config.h.windows.in and config.h.win{32,64}
+2019/11/06: add HAVE_STDINT_H into config.h.win{32,64}
+2019/11/05: Release Candidate 1 for Version 6.9.4
+2019/10/31: Update Unicode Emoji version to 12.1 (Nothing data changed)
+2019/10/29: implement USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR configuration
+2019/10/18: re-implement case fold conversion
+2019/10/04: fix #156: Heap buffer overflow in match_at() with case-insensitive match
+2019/09/30: NEW API: add onig_regset_replace()
+2019/09/30: change Unicode VERSION value format
+2019/09/20: NEW API: add regset functions
+2019/09/20: add data ensure check before peek string value in OP_PUSH_IF_PEEK_NEXT
+2019/09/20: fix loose code in encode-harness.c
+2019/08/13: fix heap-buffer-overflow
+2019/08/13: Add a macro to disable direct threading in the match engine (PR#149)
+
2019/08/06: Version 6.9.3 (secirity fix release)
-2019/07/30: add ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE
+2019/07/30: add ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC
2019/07/29: add STK_PREC_READ_START/END stack type
2019/07/29: Fix #147: Stack Exhaustion Problem caused by some parsing functions
2019/07/11: add a dictionary file for libfuzzer
* doc/SYNTAX.md: contributed by seanofw
+Version 6.9.4
+-------------
+
+* NEW API: RegSet (set of regexes)
+* Fixed CVE-2019-19012
+* Fixed CVE-2019-19203 (Does not affect UTF-8, UTF-16 and UTF-32 encodings)
+* Fixed CVE-2019-19204 (Affects only PosixBasic, Emacs and Grep syntaxes)
+* Fixed CVE-2019-19246
+* Fixed some problems (found by libFuzzer test)
+
+
Version 6.9.3 (security fix release)
------------------------------------
* Fixed CVE-2019-13224
* Fixed CVE-2019-13225
-* Fixed many problems (found by libfuzzer programs)
+* Fixed CVE-2019-16163
+* Fixed many problems (found by libFuzzer test)
Version 6.9.2 (Reiwa)
---------------------
* add doc/SYNTAX.md
+* Direct threaded code (for GCC and Clang)
* Update Unicode version 12.1.0
* NEW: Unicode Text Segment mode option (?y{g}) (?y{w}) (*original)
- g: Extended Grapheme Cluster mode / w: Word mode
-
- (Unicode Standard Annex #29 [http://unicode.org/reports/tr29/])
-
Version 6.9.1
-------------
* NEW: \O (true anychar)
* NEW: if-then-else (?(...)...\|...)
* NEW: Backreference validity checker (?(xxx)) (*original)
-* NEW: Absent repeater (?~absent) \[is equal to (?\~\|absent|\O*)]
+* NEW: Absent repeater (?~absent) \[is equal to (?\~\|(?:absent)|\O*)]
* NEW: Absent expression (?~|absent|expr) (*original)
* NEW: Absent stopper (?~|absent) (*original)
|File |Description |
|:---------------------|:-----------------------------------------|
+|sample/callout.c |example of callouts |
+|sample/count.c |example of built-in callout *COUNT |
+|sample/echo.c |example of user defined callouts of name |
+|sample/encode.c |example of some encodings |
+|sample/listcap.c |example of the capture history |
+|sample/names.c |example of the named group callback |
+|sample/posix.c |POSIX API sample |
+|sample/regset.c |example of using RegSet API |
+|sample/scan.c |example of using onig_scan() |
|sample/simple.c |example of the minimum (Oniguruma API) |
-|sample/names.c |example of the named group callback. |
-|sample/encode.c |example of some encodings. |
-|sample/listcap.c |example of the capture history. |
-|sample/posix.c |POSIX API sample. |
-|sample/scan.c |example of using onig_scan(). |
-|sample/sql.c |example of the variable meta characters. |
-|sample/user_property.c|example of user defined Unicode property. |
-|sample/callout.c |example of callouts. |
+|sample/sql.c |example of the variable meta characters |
+|sample/user_property.c|example of user defined Unicode property |
Test Programs
-Oniguruma API Version 6.9.3 2019/07/06
+Oniguruma API Version 6.9.4 2019/09/30
#include <oniguruma.h>
# int onig_new_deluxe(regex_t** reg, const UChar* pattern, const UChar* pattern_end,
OnigCompileInfo* ci, OnigErrorInfo* einfo)
- This function is deprecate, and it does not allow the case where
+ This function is deprecated, and it does not allow the case where
the encoding of pattern and target is different.
Create a regex object.
normal return: match position offset (i.e. p - str >= 0)
not found: ONIG_MISMATCH (< 0)
+ error: error code (< 0)
arguments
1 reg: regex object
Do not pass invalid byte string in the regex character encoding.
normal return: match length (>= 0)
- not match: ONIG_MISMATCH ( < 0)
+ not match: ONIG_MISMATCH (< 0)
+ error: error code (< 0)
arguments
1 reg: regex object
7 callback_arg: optional argument passed to callback
+# int onig_regset_new(OnigRegSet** rset, int n, regex_t* regs[])
+
+ Create a regset object.
+ All regex objects must have the same character encoding.
+ All regex objects are prohibited from having the ONIG_OPTION_FIND_LONGEST option.
+
+ arguments
+ 1 rset: return address of regset object
+ 2 n: number of regex in regs
+ 3 regs: array of regex
+
+ normal return: ONIG_NORMAL
+
+
+# int onig_regset_add(OnigRegSet* set, regex_t* reg)
+
+ Add a regex into regset.
+ The regex object must have the same character encoding with the regset.
+ The regex object is prohibited from having the ONIG_OPTION_FIND_LONGEST option.
+
+ arguments
+ 1 set: regset object
+ 2 reg: regex object
+
+ normal return: ONIG_NORMAL
+
+
+# int onig_regset_replace(OnigRegSet* set, int at, regex_t* reg)
+
+ Replace a regex in regset with another one.
+ If the reg argument value is NULL, then remove at-th regex. (and indexes of other regexes are changed)
+
+ arguments
+ 1 set: regset object
+ 2 at: index of regex (zero origin)
+ 3 reg: regex object
+
+ normal return: ONIG_NORMAL
+
+
+# void onig_regset_free(OnigRegSet* set)
+
+ Free memory used by regset object and regex objects in the regset.
+ If the same regex object is registered twice, the situation becomes destructive.
+
+ arguments
+ 1 set: regset object
+
+
+# int onig_regset_number_of_regex(OnigRegSet* set)
+
+ Returns number of regex objects in the regset.
+
+ arguments
+ 1 set: regset object
+
+
+# regex_t* onig_regset_get_regex(OnigRegSet* set, int at)
+
+ Returns the regex object corresponding to the at-th regex.
+
+ arguments
+ 1 set: regset object
+ 2 at: index of regex array (zero origin)
+
+
+# OnigRegion* onig_regset_get_region(OnigRegSet* set, int at)
+
+ Returns the region object corresponding to the at-th regex.
+
+ arguments
+ 1 set: regset object
+ 2 at: index of regex array (zero origin)
+
+
+# int onig_regset_search(OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos)
+
+ Perform a search with regset.
+
+ return value:
+ normal return: index of match regex (zero origin)
+ not found: ONIG_MISMATCH (< 0)
+ error: error code (< 0)
+
+ arguments
+ 1 set: regset object
+ 2 str: target string
+ 3 end: terminate address of target string
+ 4 start: search start address of target string
+ 5 range: search terminate address of target string
+ 6 lead: outer loop element
+ ONIG_REGSET_POSITION_LEAD (returns most left position)
+ ONIG_REGSET_REGEX_LEAD (returns most left position)
+ ONIG_REGSET_PRIORITY_TO_REGEX_ORDER (returns first match regex)
+ 7 option: search time option
+ ONIG_OPTION_NOTBOL string head(str) isn't considered as begin of line
+ ONIG_OPTION_NOTEOL string end (end) isn't considered as end of line
+ 8 rmatch_pos: return address of match position (match_address - str)
+
+ * ONIG_REGSET_POSITION_LEAD and ONIG_REGSET_REGEX_LEAD return the same result.
+ These differences only appear in search time.
+ In most cases, ONIG_REGSET_POSITION_LEAD seems to be faster.
+
+
+# int onig_regset_search_with_param(OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos)
+
+ Perform a search with regset and match-params.
+
+ return value:
+ normal return: index of match regex (zero origin)
+ not found: ONIG_MISMATCH (< 0)
+ error: error code (< 0)
+
+ arguments
+ 1 set: regset object
+ 2 str: target string
+ 3 end: terminate address of target string
+ 4 start: search start address of target string
+ 5 range: search terminate address of target string
+ 6 lead: outer loop element
+ ONIG_REGSET_POSITION_LEAD (returns most left position)
+ ONIG_REGSET_REGEX_LEAD (returns most left position)
+ ONIG_REGSET_PRIORITY_TO_REGEX_ORDER (returns first match regex)
+ 7 option: search time option
+ ONIG_OPTION_NOTBOL string head(str) isn't considered as begin of line
+ ONIG_OPTION_NOTEOL string end (end) isn't considered as end of line
+ 8 mps: array of match-params
+ 9 rmatch_pos: return address of match position (match_address - str)
+
+
# OnigRegion* onig_region_new(void)
Create a region.
-鬼車インターフェース Version 6.9.3 2019/07/06
+鬼車インターフェース Version 6.9.4 2019/09/30
#include <oniguruma.h>
7 callback_arg: コールバック関数に渡される付加引数値
+# int onig_regset_new(OnigRegSet** rset, int n, regex_t* regs[])
+
+ regsetオブジェクトを生成する。
+ 全ての正規表現オブジェクトは、同じ文字エンコーディングでなければならない。
+ 全ての正規表現オブジェクトは、ONIG_OPTION_FIND_LONGESTオプションでコンパイルされていてはならない。
+
+ 引数
+ 1 rset: regsetオブジェクトを返すためのアドレス
+ 2 n: 正規表現の個数
+ 3 regs: 正規表現オブジェクトの配列
+
+ 正常終了戻り値: ONIG_NORMAL
+
+
+# int onig_regset_add(OnigRegSet* set, regex_t* reg)
+
+ regsetオブジェクトに正規表現を追加する。
+ 正規表現オブジェクトは、regsetと同じ文字エンコーディングでなければならない。
+ 正規表現オブジェクトは、ONIG_OPTION_FIND_LONGESTオプションでコンパイルされていてはならない。
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 reg: 正規表現オブジェクト
+
+ 正常終了戻り値: ONIG_NORMAL
+
+
+# int onig_regset_replace(OnigRegSet* set, int at, regex_t* reg)
+
+ regsetの中の一個の正規表現オブジェクトを別のものに変更する。
+ 若しreg引数の値がNULLであれば、at番目の正規表現オブジェクトを外す。(そして、以降の正規表現オブジェクトのインデックスは変化する)
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 at: 変更する場所のインデックス
+ 2 reg: 正規表現オブジェクト
+
+ 正常終了戻り値: ONIG_NORMAL
+
+
+# void onig_regset_free(OnigRegSet* set)
+
+ regsetオブジェクトとその中の正規表現オブジェクトの使用メモリを開放する。
+ 若し、同一の正規表現オブジェクトを重複して登録していれば、破壊的な状況になる。
+
+ 引数
+ 1 set: regsetオブジェクト
+
+
+# int onig_regset_number_of_regex(OnigRegSet* set)
+
+ regsetの中の正規表現オブジェクトの個数を返す。
+
+ 引数
+ 1 set: regsetオブジェクト
+
+
+# regex_t* onig_regset_get_regex(OnigRegSet* set, int at)
+
+ regsetのat番目の正規表現を返す。
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 at: 正規表現オブジェクトのインデックス (ゼロ開始)
+
+
+# OnigRegion* onig_regset_get_region(OnigRegSet* set, int at)
+
+ regsetのat番目の正規表現に対応する領域を返す。
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 at: 正規表現オブジェクトのインデックス (ゼロ開始)
+
+
+# int onig_regset_search(OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos)
+
+ regsetによる検索を実行する。
+
+ 戻り値:
+ 検索成功: マッチした正規表現オブジェクトのインデックス (ゼロ開始)
+ 検索失敗: ONIG_MISMATCH (< 0)
+ エラー: エラーコード (< 0)
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 str: 検索対象文字列
+ 3 end: 検索対象文字列の終端アドレス
+ 4 start: 検索対象文字列の検索先頭位置アドレス
+ 5 range: 検索対象文字列の検索終了位置アドレス
+ (start <= 探索される文字列 < range)
+ 6 lead: 外側のループ要素
+ ONIG_REGSET_POSITION_LEAD (最左位置でマッチした結果を返す)
+ ONIG_REGSET_REGEX_LEAD (最左位置でマッチした結果を返す)
+ ONIG_REGSET_PRIORITY_TO_REGEX_ORDER (最初にマッチした正規表現の結果を返す)
+ 7 option: 検索時オプション
+ ONIG_OPTION_NOTBOL 文字列の先頭(str)を行頭と看做さない
+ ONIG_OPTION_NOTEOL 文字列の終端(end)を行末と看做さない
+ 8 rmatch_pos: マッチした位置を返すためのアドレス (match_address - str)
+
+ * ONIG_REGSET_POSITION_LEADとONIG_REGSET_REGEX_LEADは同じ結果を返す。
+ これらの違いは検索時間にしか現れない。
+ ほとんどの場合、ONIG_REGSET_POSITION_LEADのほうが速いと思われる。
+
+
+# int onig_regset_search_with_param(OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos)
+
+ regsetとOnigMatchParamオブジェクトによる検索を実行する。
+
+ 戻り値:
+ 検索成功: マッチした正規表現オブジェクトのインデックス (ゼロ開始)
+ 検索失敗: ONIG_MISMATCH (< 0)
+ エラー: エラーコード (< 0)
+
+ 引数
+ 1 set: regsetオブジェクト
+ 2 str: 検索対象文字列
+ 3 end: 検索対象文字列の終端アドレス
+ 4 start: 検索対象文字列の検索先頭位置アドレス
+ 5 range: 検索対象文字列の検索終了位置アドレス
+ (start <= 探索される文字列 < range)
+ 6 lead: 外側のループ要素
+ ONIG_REGSET_POSITION_LEAD (最左位置でマッチした結果を返す)
+ ONIG_REGSET_REGEX_LEAD (最左位置でマッチした結果を返す)
+ ONIG_REGSET_PRIORITY_TO_REGEX_ORDER (最初にマッチした正規表現の結果を返す)
+ 7 option: 検索時オプション
+ ONIG_OPTION_NOTBOL 文字列の先頭(str)を行頭と看做さない
+ ONIG_OPTION_NOTEOL 文字列の終端(end)を行末と看做さない
+ 8 mps: OnigMatchParamオブジェクトの配列
+ 9 rmatch_pos: マッチした位置を返すためのアドレス (match_address - str)
+
+
# OnigRegion* onig_region_new(void)
マッチ領域情報(region)を作成する。
-Oniguruma Regular Expressions Version 6.9.2 2019/03/29
+Oniguruma Regular Expressions Version 6.9.4 2019/10/31
syntax: ONIG_SYNTAX_ONIGURUMA (default)
In negative look-behind, capturing group isn't allowed,
but non-capturing group (?:) is allowed.
+ * In look-behind and negative look-behind, support for
+ ignore-case option is limited. Only supports conversion
+ between single characters. (Does not support conversion
+ of multiple characters in Unicode)
+
(?>subexp) atomic group
no backtracks in subexp.
This works like .* (more precisely \O*), but it is
limited by the range that does not include the string
match with <absent>.
- This is a written abbreviation of (?~|absent|\O*).
+ This is a written abbreviation of (?~|(?:absent)|\O*).
\O* is used as a repeater.
(?~|absent|exp) Absent expression (* original)
-鬼車 正規表現 Version 6.9.2 2019/03/29
+鬼車 正規表現 Version 6.9.4 2019/10/31
使用文法: ONIG_SYNTAX_ONIGURUMA (既定値)
\f 改頁 (0x0C)
\a 鐘 (0x07)
\e 退避修飾 (0x1B)
- \nnn 八進数表現 符号化バイト値(の一部)
+ \nnn 八進数表現 符号化バイト値
\o{17777777777} 拡張八進数表現 コードポイント値
\uHHHH 拡張十六進数表現 コードポイント値
- \xHH 十六進数表現 符号化バイト値(の一部)
+ \xHH 十六進数表現 符号化バイト値
\x{7HHHHHHH} 拡張十六進数表現 コードポイント値
\cx 制御文字表現 コードポイント値
\C-x 制御文字表現 コードポイント値
否定戻り読みでは、捕獲式集合は許されないが、
非捕獲式集合は許される。
+ * 戻り読み、否定戻り読みの中では、ignore-caseオプションの
+ 対応が制限される。一文字と一文字の間の変換しか対応しない。
+ (Unicodeでの複数文字の変換に対応しない)
+
(?>式) 原子的式集合
式全体を通過したとき、式の中での後退再試行を行なわない
<不在機能群>
- (?~不在式) 不在繰り返し (*原案 田中哲)
- これは .*(より正確には\O*)のように動作するが、<不在式>に
+ (?~不在) 不在繰り返し (*原案 田中哲)
+ これは .*(より正確には\O*)のように動作するが、<不在>に
適合する文字列を含まない範囲に制限される。
- これは(?~|不在式|\O*)の省略表記である。
+ これは(?~|(?:不在)|\O*)の省略表記である。
- (?~|不在式|式) 不在式 (* 原作)
- これは<式>のように動作するが、<不在式>に適合する文字列を
+ (?~|不在|式) 不在式 (* 原作)
+ これは<式>のように動作するが、<不在>に適合する文字列を
含まない範囲に制限される。
例 (?~|345|\d*) "12345678" ==> "12", "1", ""
- (?~|不在式) 不在停止 (* 原作)
+ (?~|不在) 不在停止 (* 原作)
この演算子を通過した後は、対象文字列の適合範囲が
- <不在式>に適合する文字列を含まない範囲に制限される。
+ <不在>に適合する文字列を含まない範囲に制限される。
(?~|) 範囲消去
不在停止の効果を消して、それ以前の状態にする。
# Oniguruma syntax (operator) configuration
-_Documented for Oniguruma 6.9.2 (2019/03/28)_
+_Documented for Oniguruma 6.9.3 (2019/08/08)_
----------
If this flag is set, Oniguruma will warn about nested repeat operators those have no meaning, like `(?:a*)+`.
If this flag is clear, Oniguruma will allow the nested repeat operators without warning about them.
+### 26. ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC (allow [a-\x{7fffffff}])
+
+_Set in: Oniguruma_
+
+If this flag is set, then invalid code points at the end of range in character class are allowed.
+
### 31. ONIG_SYN_CONTEXT_INDEP_ANCHORS
_Set in: PosixExtended, GnuRegex, Java, Perl, Perl_NG, Ruby, Oniguruma_
| 23 | `ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC` | - | Yes | - | - | Yes | Yes | Yes | Yes | Yes | Yes |
| 24 | `ONIG_SYN_WARN_CC_OP_NOT_ESCAPED` | - | - | - | - | - | - | - | - | Yes | Yes |
| 25 | `ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT` | - | - | - | - | - | - | - | - | Yes | Yes |
+| 26 | `ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC` | - | - | - | - | - | - | - | - | - | Yes |
| 31 | `ONIG_SYN_CONTEXT_INDEP_ANCHORS` | - | Yes | - | - | Yes | Yes | Yes | Yes | Yes | Yes |
-Unicode Properties (from Unicode Version: 12.1.0)
+Unicode Properties (Unicode Version: 12.1.0, Emoji: 12.1)
15: ASCII_Hex_Digit
16: Adlam
ascii.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
big5.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return EncLen_BIG5[*p];
}
+static int
+big5_code_to_mbclen(OnigCodePoint code)
+{
+ if ((code & (~0xffff)) != 0) return ONIGERR_INVALID_CODE_POINT_VALUE;
+ if ((code & 0xff00) != 0) return 2;
+ if (EncLen_BIG5[(int )(code & 0xff)] == 1) return 1;
+
+ return ONIGERR_INVALID_CODE_POINT_VALUE;
+}
+
static int
is_valid_mbc_string(const UChar* p, const UChar* end)
{
pp, end, lower);
}
-#if 0
-static int
-big5_is_mbc_ambiguous(OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- return onigenc_mbn_is_mbc_ambiguous(ONIG_ENCODING_BIG5, flag, pp, end);
-}
-#endif
-
static int
big5_is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
1, /* min enc length */
onigenc_is_mbc_newline_0x0a,
big5_mbc_to_code,
- onigenc_mb2_code_to_mbclen,
+ big5_code_to_mbclen,
big5_code_to_mbc,
big5_mbc_case_fold,
onigenc_ascii_apply_all_case_fold,
-#define STDC_HEADERS 1\r
-#define HAVE_SYS_TYPES_H 1\r
-#define HAVE_SYS_STAT_H 1\r
-#define HAVE_STDLIB_H 1\r
-#define HAVE_STRING_H 1\r
-#define HAVE_MEMORY_H 1\r
-#define HAVE_FLOAT_H 1\r
-#define HAVE_OFF_T 1\r
-#define SIZEOF_INT 4\r
-#define SIZEOF_SHORT 2\r
-#define SIZEOF_LONG 4\r
-#define SIZEOF_LONG_LONG 8\r
-#define SIZEOF___INT64 8\r
-#define SIZEOF_OFF_T 4\r
-#define SIZEOF_VOIDP 4\r
-#define SIZEOF_FLOAT 4\r
-#define SIZEOF_DOUBLE 8\r
-#define SIZEOF_SIZE_T 4\r
-#define HAVE_PROTOTYPES 1\r
-#define TOKEN_PASTE(x,y) x##y\r
-#define HAVE_STDARG_PROTOTYPES 1\r
-#ifndef NORETURN\r
-#if _MSC_VER > 1100\r
-#define NORETURN(x) __declspec(noreturn) x\r
-#else\r
-#define NORETURN(x) x\r
-#endif\r
-#endif\r
-#define HAVE_DECL_SYS_NERR 1\r
-#define STDC_HEADERS 1\r
-#define HAVE_STDLIB_H 1\r
-#define HAVE_STRING_H 1\r
-#define HAVE_LIMITS_H 1\r
-#define HAVE_FCNTL_H 1\r
-#define HAVE_SYS_UTIME_H 1\r
-#define HAVE_MEMORY_H 1\r
-#define uid_t int\r
-#define gid_t int\r
-#define GETGROUPS_T int\r
-#define HAVE_ALLOCA 1\r
-#define HAVE_DUP2 1\r
-#define HAVE_MEMCMP 1\r
-#define HAVE_MEMMOVE 1\r
-#define HAVE_MKDIR 1\r
-#define HAVE_STRCASECMP 1\r
-#define HAVE_STRNCASECMP 1\r
-#define HAVE_STRERROR 1\r
-#define HAVE_STRFTIME 1\r
-#define HAVE_STRCHR 1\r
-#define HAVE_STRSTR 1\r
-#define HAVE_STRTOD 1\r
-#define HAVE_STRTOL 1\r
-#define HAVE_STRTOUL 1\r
-#define HAVE_FLOCK 1\r
-#define HAVE_VSNPRINTF 1\r
-#define HAVE_FINITE 1\r
-#define HAVE_FMOD 1\r
-#define HAVE_FREXP 1\r
-#define HAVE_HYPOT 1\r
-#define HAVE_MODF 1\r
-#define HAVE_WAITPID 1\r
-#define HAVE_CHSIZE 1\r
-#define HAVE_TIMES 1\r
-#define HAVE__SETJMP 1\r
-#define HAVE_TELLDIR 1\r
-#define HAVE_SEEKDIR 1\r
-#define HAVE_MKTIME 1\r
-#define HAVE_COSH 1\r
-#define HAVE_SINH 1\r
-#define HAVE_TANH 1\r
-#define HAVE_EXECVE 1\r
-#define HAVE_TZNAME 1\r
-#define HAVE_DAYLIGHT 1\r
-#define SETPGRP_VOID 1\r
-#define inline __inline\r
-#define NEED_IO_SEEK_BETWEEN_RW 1\r
-#define RSHIFT(x,y) ((x)>>(int)y)\r
-#define FILE_COUNT _cnt\r
-#define FILE_READPTR _ptr\r
-#define DEFAULT_KCODE KCODE_NONE\r
-#define DLEXT ".so"\r
-#define DLEXT2 ".dll"\r
+#if defined(__MINGW32__) || _MSC_VER >= 1600
+#define HAVE_STDINT_H 1
+#endif
+#if defined(__MINGW32__) || _MSC_VER >= 1800
+#define HAVE_INTTYPES_H 1
+#endif
+#define HAVE_SYS_TYPES_H 1
+#define HAVE_SYS_STAT_H 1
+#define HAVE_MEMORY_H 1
+#define HAVE_OFF_T 1
+#define SIZEOF_INT 4
+#define SIZEOF_LONG 4
+#define SIZEOF_LONG_LONG 8
+#define SIZEOF___INT64 8
+#define SIZEOF_OFF_T 4
+#define SIZEOF_VOIDP 4
+#define SIZEOF_FLOAT 4
+#define SIZEOF_DOUBLE 8
+#define SIZEOF_SIZE_T 4
+#define TOKEN_PASTE(x,y) x##y
+#ifndef NORETURN
+#if _MSC_VER > 1100
+#define NORETURN(x) __declspec(noreturn) x
+#else
+#define NORETURN(x) x
+#endif
+#endif
+#define HAVE_DECL_SYS_NERR 1
+#define HAVE_FCNTL_H 1
+#define HAVE_SYS_UTIME_H 1
+#define HAVE_MEMORY_H 1
+#define uid_t int
+#define gid_t int
+#define GETGROUPS_T int
+#define HAVE_ALLOCA 1
+#define HAVE_DUP2 1
+#define HAVE_MKDIR 1
+#define HAVE_FLOCK 1
+#define HAVE_FINITE 1
+#define HAVE_HYPOT 1
+#define HAVE_WAITPID 1
+#define HAVE_CHSIZE 1
+#define HAVE_TIMES 1
+#define HAVE_TELLDIR 1
+#define HAVE_SEEKDIR 1
+#define HAVE_EXECVE 1
+#define HAVE_DAYLIGHT 1
+#define SETPGRP_VOID 1
+#define inline __inline
+#define NEED_IO_SEEK_BETWEEN_RW 1
+#define RSHIFT(x,y) ((x)>>(int)y)
+#define FILE_COUNT _cnt
+#define FILE_READPTR _ptr
+#define DEFAULT_KCODE KCODE_NONE
+#define DLEXT ".so"
+#define DLEXT2 ".dll"
-#define STDC_HEADERS 1\r
-#define HAVE_SYS_TYPES_H 1\r
-#define HAVE_SYS_STAT_H 1\r
-#define HAVE_STDLIB_H 1\r
-#define HAVE_STRING_H 1\r
-#define HAVE_MEMORY_H 1\r
-#define HAVE_FLOAT_H 1\r
-#define HAVE_OFF_T 1\r
-#define SIZEOF_INT 4\r
-#define SIZEOF_SHORT 2\r
-#define SIZEOF_LONG 4\r
-#define SIZEOF_LONG_LONG 8\r
-#define SIZEOF___INT64 8\r
-#define SIZEOF_OFF_T 4\r
-#define SIZEOF_VOIDP 8\r
-#define SIZEOF_FLOAT 4\r
-#define SIZEOF_DOUBLE 8\r
-#define SIZEOF_SIZE_T 8\r
-#define HAVE_PROTOTYPES 1\r
-#define TOKEN_PASTE(x,y) x##y\r
-#define HAVE_STDARG_PROTOTYPES 1\r
-#ifndef NORETURN\r
-#if _MSC_VER > 1100\r
-#define NORETURN(x) __declspec(noreturn) x\r
-#else\r
-#define NORETURN(x) x\r
-#endif\r
-#endif\r
-#define HAVE_DECL_SYS_NERR 1\r
-#define STDC_HEADERS 1\r
-#define HAVE_STDLIB_H 1\r
-#define HAVE_STRING_H 1\r
-#define HAVE_LIMITS_H 1\r
-#define HAVE_FCNTL_H 1\r
-#define HAVE_SYS_UTIME_H 1\r
-#define HAVE_MEMORY_H 1\r
-#define uid_t int\r
-#define gid_t int\r
-#define GETGROUPS_T int\r
-#define HAVE_ALLOCA 1\r
-#define HAVE_DUP2 1\r
-#define HAVE_MEMCMP 1\r
-#define HAVE_MEMMOVE 1\r
-#define HAVE_MKDIR 1\r
-#define HAVE_STRCASECMP 1\r
-#define HAVE_STRNCASECMP 1\r
-#define HAVE_STRERROR 1\r
-#define HAVE_STRFTIME 1\r
-#define HAVE_STRCHR 1\r
-#define HAVE_STRSTR 1\r
-#define HAVE_STRTOD 1\r
-#define HAVE_STRTOL 1\r
-#define HAVE_STRTOUL 1\r
-#define HAVE_FLOCK 1\r
-#define HAVE_VSNPRINTF 1\r
-#define HAVE_FINITE 1\r
-#define HAVE_FMOD 1\r
-#define HAVE_FREXP 1\r
-#define HAVE_HYPOT 1\r
-#define HAVE_MODF 1\r
-#define HAVE_WAITPID 1\r
-#define HAVE_CHSIZE 1\r
-#define HAVE_TIMES 1\r
-#define HAVE__SETJMP 1\r
-#define HAVE_TELLDIR 1\r
-#define HAVE_SEEKDIR 1\r
-#define HAVE_MKTIME 1\r
-#define HAVE_COSH 1\r
-#define HAVE_SINH 1\r
-#define HAVE_TANH 1\r
-#define HAVE_EXECVE 1\r
-#define HAVE_TZNAME 1\r
-#define HAVE_DAYLIGHT 1\r
-#define SETPGRP_VOID 1\r
-#define inline __inline\r
-#define NEED_IO_SEEK_BETWEEN_RW 1\r
-#define RSHIFT(x,y) ((x)>>(int)y)\r
-#define FILE_COUNT _cnt\r
-#define FILE_READPTR _ptr\r
-#define DEFAULT_KCODE KCODE_NONE\r
-#define DLEXT ".so"\r
-#define DLEXT2 ".dll"\r
+#if defined(__MINGW32__) || _MSC_VER >= 1600
+#define HAVE_STDINT_H 1
+#endif
+#if defined(__MINGW32__) || _MSC_VER >= 1800
+#define HAVE_INTTYPES_H 1
+#endif
+#define HAVE_SYS_TYPES_H 1
+#define HAVE_SYS_STAT_H 1
+#define HAVE_MEMORY_H 1
+#define HAVE_OFF_T 1
+#define SIZEOF_INT 4
+#define SIZEOF_LONG 4
+#define SIZEOF_LONG_LONG 8
+#define SIZEOF___INT64 8
+#define SIZEOF_OFF_T 4
+#define SIZEOF_VOIDP 8
+#define SIZEOF_FLOAT 4
+#define SIZEOF_DOUBLE 8
+#define SIZEOF_SIZE_T 8
+#define TOKEN_PASTE(x,y) x##y
+#ifndef NORETURN
+#if _MSC_VER > 1100
+#define NORETURN(x) __declspec(noreturn) x
+#else
+#define NORETURN(x) x
+#endif
+#endif
+#define HAVE_DECL_SYS_NERR 1
+#define HAVE_FCNTL_H 1
+#define HAVE_SYS_UTIME_H 1
+#define HAVE_MEMORY_H 1
+#define uid_t int
+#define gid_t int
+#define GETGROUPS_T int
+#define HAVE_ALLOCA 1
+#define HAVE_DUP2 1
+#define HAVE_MKDIR 1
+#define HAVE_FLOCK 1
+#define HAVE_FINITE 1
+#define HAVE_HYPOT 1
+#define HAVE_WAITPID 1
+#define HAVE_CHSIZE 1
+#define HAVE_TIMES 1
+#define HAVE_TELLDIR 1
+#define HAVE_SEEKDIR 1
+#define HAVE_EXECVE 1
+#define HAVE_DAYLIGHT 1
+#define SETPGRP_VOID 1
+#define inline __inline
+#define NEED_IO_SEEK_BETWEEN_RW 1
+#define RSHIFT(x,y) ((x)>>(int)y)
+#define FILE_COUNT _cnt
+#define FILE_READPTR _ptr
+#define DEFAULT_KCODE KCODE_NONE
+#define DLEXT ".so"
+#define DLEXT2 ".dll"
cp1251.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2006-2018 Byte <byte AT mail DOT kna DOT ru>
- * K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2006-2019 Byte <byte AT mail DOT kna DOT ru>
+ * K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
euc_jp.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return ONIGERR_INVALID_CODE_POINT_VALUE;
}
-#if 0
-static int
-code_to_mbc_first(OnigCodePoint code)
-{
- int first;
-
- if ((code & 0xff0000) != 0) {
- first = (code >> 16) & 0xff;
- }
- else if ((code & 0xff00) != 0) {
- first = (code >> 8) & 0xff;
- }
- else {
- return (int )code;
- }
- return first;
-}
-#endif
-
static int
code_to_mbc(OnigCodePoint code, UChar *buf)
{
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -pt -T -L ANSI-C -N onigenc_euc_jp_lookup_property_name --output-file gperf1.tmp euc_jp_prop.gperf */
+/* Command-line: gperf -pt -T -L ANSI-C -N onigenc_euc_jp_lookup_property_name --output-file gperf1.tmp euc_jp_prop.gperf */
/* Computed positions: -k'1,3' */
#if !((' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
euc_kr.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return EncLen_EUCKR[*p];
}
+static int
+euckr_code_to_mbclen(OnigCodePoint code)
+{
+ if ((code & (~0xffff)) != 0) return ONIGERR_INVALID_CODE_POINT_VALUE;
+ if ((code & 0xff00) != 0) return 2;
+ if (EncLen_EUCKR[(int )(code & 0xff)] == 1) return 1;
+
+ return ONIGERR_INVALID_CODE_POINT_VALUE;
+}
+
static int
is_valid_mbc_string(const UChar* p, const UChar* end)
{
pp, end, lower);
}
-#if 0
-static int
-euckr_is_mbc_ambiguous(OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- return onigenc_mbn_is_mbc_ambiguous(ONIG_ENCODING_EUC_KR, flag, pp, end);
-}
-#endif
-
static int
euckr_is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
1, /* min enc length */
onigenc_is_mbc_newline_0x0a,
euckr_mbc_to_code,
- onigenc_mb2_code_to_mbclen,
+ euckr_code_to_mbclen,
euckr_code_to_mbc,
euckr_mbc_case_fold,
onigenc_ascii_apply_all_case_fold,
1, /* min enc length */
onigenc_is_mbc_newline_0x0a,
euckr_mbc_to_code,
- onigenc_mb2_code_to_mbclen,
+ euckr_code_to_mbclen,
euckr_code_to_mbc,
euckr_mbc_case_fold,
onigenc_ascii_apply_all_case_fold,
euc_tw.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return EncLen_EUCTW[*p];
}
+static int
+euctw_code_to_mbclen(OnigCodePoint code)
+{
+ if ((code & 0xff000000) != 0) return 4;
+ else if ((code & 0xff0000) != 0) return ONIGERR_INVALID_CODE_POINT_VALUE;
+ else if ((code & 0xff00) != 0) return 2;
+ else {
+ if (EncLen_EUCTW[(int )(code & 0xff)] == 1)
+ return 1;
+
+ return ONIGERR_INVALID_CODE_POINT_VALUE;
+ }
+}
+
static int
is_valid_mbc_string(const UChar* p, const UChar* end)
{
1, /* min enc length */
onigenc_is_mbc_newline_0x0a,
euctw_mbc_to_code,
- onigenc_mb4_code_to_mbclen,
+ euctw_code_to_mbclen,
euctw_code_to_mbc,
euctw_mbc_case_fold,
onigenc_ascii_apply_all_case_fold,
**********************************************************************/
/*-
* Copyright (c) 2005-2019 KUBO Takehiro <kubo AT jiubao DOT org>
- * K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#if 1
#define DEBUG_GB18030(arg)
#else
+#include <stdio.h>
#define DEBUG_GB18030(arg) printf arg
#endif
return 2;
}
+static int
+gb18030_code_to_mbclen(OnigCodePoint code)
+{
+ if ((code & 0xff000000) != 0) return 4;
+ else if ((code & 0xff0000) != 0) return ONIGERR_INVALID_CODE_POINT_VALUE;
+ else if ((code & 0xff00) != 0) return 2;
+ else {
+ if (GB18030_MAP[(int )(code & 0xff)] == CM)
+ return ONIGERR_INVALID_CODE_POINT_VALUE;
+
+ return 1;
+ }
+}
+
static int
is_valid_mbc_string(const UChar* p, const UChar* end)
{
pp, end, lower);
}
-#if 0
-static int
-gb18030_is_mbc_ambiguous(OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- return onigenc_mbn_is_mbc_ambiguous(ONIG_ENCODING_GB18030, flag, pp, end);
-}
-#endif
-
static int
gb18030_is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
1, /* min enc length */
onigenc_is_mbc_newline_0x0a,
gb18030_mbc_to_code,
- onigenc_mb4_code_to_mbclen,
+ gb18030_code_to_mbclen,
gb18030_code_to_mbc,
gb18030_mbc_case_fold,
onigenc_ascii_apply_all_case_fold,
iso8859_1.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_1_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xdf, 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (*p >= 0xaa && *p <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_10.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_10_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_11.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
iso8859_13.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_13_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xdf, 0xb5 are lower case letter, but can't convert. */
- if (*p == 0xb5)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_14.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1; /* return byte length of converted char to lower */
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_14_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_15.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1; /* return byte length of converted char to lower */
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_15_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xdf etc.. are lower case letter, but can't convert. */
- if (*p == 0xaa || *p == 0xb5 || *p == 0xba)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_16.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1; /* return byte length of converted char to lower */
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_16_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_2.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1; /* return byte length of converted char to lower */
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_2_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static const OnigPairCaseFoldCodes CaseFoldMap[] = {
{ 0xa1, 0xb1 },
{ 0xa3, 0xb3 },
iso8859_3.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_3_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (*p == 0xb5)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_4.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1; /* return byte length of converted char to lower */
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_4_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- if (*p == 0xa2)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_5.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- (*pp)++;
- v = (EncISO_8859_5_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_6.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
iso8859_7.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- (*pp)++;
- v = (EncISO_8859_7_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- if (*p == 0xc0 || *p == 0xe0)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
iso8859_8.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
iso8859_9.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- (*pp)++;
- return TRUE;
- }
-
- (*pp)++;
- v = (EncISO_8859_9_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xdf etc.. are lower case letter, but can't convert. */
- if (*p >= 0xaa && *p <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
-
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
koi8.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-koi8_is_mbc_ambiguous(OnigAmbigType flag, const OnigUChar** pp, const OnigUChar* end)
-{
- const OnigUChar* p = *pp;
-
- (*pp)++;
- if (((flag & ONIGENC_CASE_FOLD_ASCII_CASE) != 0 &&
- ONIGENC_IS_MBC_ASCII(p)) ||
- ((flag & ONIGENC_CASE_FOLD_NONASCII_CASE) != 0 &&
- !ONIGENC_IS_MBC_ASCII(p))) {
- int v = (EncKOI8_CtypeTable[*p] &
- (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- return (v != 0 ? TRUE : FALSE);
- }
- return FALSE;
-}
-#endif
-
static int
koi8_is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
koi8_r.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 1;
}
-#if 0
-static int
-koi8_r_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- int v;
- const UChar* p = *pp;
-
- (*pp)++;
- v = (EncKOI8_R_CtypeTable[*p] & (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- return (v != 0 ? TRUE : FALSE);
-}
-#endif
-
static int
koi8_r_is_code_ctype(OnigCodePoint code, unsigned int ctype)
{
mktable.c
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
onig_init.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2016-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2016-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
oniggnu.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2005 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
onigposix.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#endif
#endif
+#ifndef ONIG_STATIC
#ifndef ONIG_EXTERN
#if defined(_WIN32) && !defined(__GNUC__)
#if defined(ONIGURUMA_EXPORT)
#ifndef ONIG_EXTERN
#define ONIG_EXTERN extern
#endif
+#else
+#define ONIG_EXTERN extern
+#endif
#ifndef ONIGURUMA_H
typedef unsigned int OnigOptionType;
oniguruma.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#define ONIGURUMA
#define ONIGURUMA_VERSION_MAJOR 6
#define ONIGURUMA_VERSION_MINOR 9
-#define ONIGURUMA_VERSION_TEENY 3
+#define ONIGURUMA_VERSION_TEENY 4
-#define ONIGURUMA_VERSION_INT 60903
+#define ONIGURUMA_VERSION_INT 60904
#ifndef P_
#if defined(__STDC__) || defined(_WIN32)
typedef OnigRegexType regex_t;
#endif
+struct OnigRegSetStruct;
+typedef struct OnigRegSetStruct OnigRegSet;
+
+typedef enum {
+ ONIG_REGSET_POSITION_LEAD = 0,
+ ONIG_REGSET_REGEX_LEAD = 1,
+ ONIG_REGSET_PRIORITY_TO_REGEX_ORDER = 2
+} OnigRegSetLead;
typedef struct {
int num_of_elements;
int onig_match P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option));
ONIG_EXTERN
int onig_match_with_param P_((OnigRegex, const OnigUChar* str, const OnigUChar* end, const OnigUChar* at, OnigRegion* region, OnigOptionType option, OnigMatchParam* mp));
+
+ONIG_EXTERN
+int onig_regset_new P_((OnigRegSet** rset, int n, regex_t* regs[]));
+ONIG_EXTERN
+int onig_regset_add P_((OnigRegSet* set, regex_t* reg));
+ONIG_EXTERN
+int onig_regset_replace P_((OnigRegSet* set, int at, regex_t* reg));
+ONIG_EXTERN
+void onig_regset_free P_((OnigRegSet* set));
+ONIG_EXTERN
+int onig_regset_number_of_regex P_((OnigRegSet* set));
+ONIG_EXTERN
+regex_t* onig_regset_get_regex P_((OnigRegSet* set, int at));
+ONIG_EXTERN
+OnigRegion* onig_regset_get_region P_((OnigRegSet* set, int at));
+ONIG_EXTERN
+int onig_regset_search P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos));
+ONIG_EXTERN
+int onig_regset_search_with_param P_((OnigRegSet* set, const OnigUChar* str, const OnigUChar* end, const OnigUChar* start, const OnigUChar* range, OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos));
+
ONIG_EXTERN
OnigRegion* onig_region_new P_((void));
ONIG_EXTERN
regcomp.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#endif
switch (opcode) {
- case OP_EXACTMBN:
+ case OP_STR_MBN:
if (! is_in_string_pool(reg, op->exact_len_n.s))
xfree(op->exact_len_n.s);
break;
- case OP_EXACTN: case OP_EXACTMB2N: case OP_EXACTMB3N: case OP_EXACTN_IC:
+ case OP_STR_N: case OP_STR_MB2N: case OP_STR_MB3N: case OP_STR_N_IC:
if (! is_in_string_pool(reg, op->exact_n.s))
xfree(op->exact_n.s);
break;
- case OP_EXACT1: case OP_EXACT2: case OP_EXACT3: case OP_EXACT4:
- case OP_EXACT5: case OP_EXACTMB2N1: case OP_EXACTMB2N2:
- case OP_EXACTMB2N3: case OP_EXACT1_IC:
+ case OP_STR_1: case OP_STR_2: case OP_STR_3: case OP_STR_4:
+ case OP_STR_5: case OP_STR_MB2N1: case OP_STR_MB2N2:
+ case OP_STR_MB2N3: case OP_STR_1_IC:
break;
case OP_CCLASS_NOT: case OP_CCLASS:
#endif
switch (opcode) {
- case OP_EXACTMBN:
+ case OP_STR_MBN:
total += op->exact_len_n.len * op->exact_len_n.n;
break;
- case OP_EXACTN:
- case OP_EXACTN_IC:
+ case OP_STR_N:
+ case OP_STR_N_IC:
total += op->exact_n.n;
break;
- case OP_EXACTMB2N:
+ case OP_STR_MB2N:
total += op->exact_n.n * 2;
break;
- case OP_EXACTMB3N:
+ case OP_STR_MB3N:
total += op->exact_n.n * 3;
break;
#endif
switch (opcode) {
- case OP_EXACTMBN:
+ case OP_STR_MBN:
len = op->exact_len_n.len * op->exact_len_n.n;
xmemcpy(curr, op->exact_len_n.s, len);
xfree(op->exact_len_n.s);
op->exact_len_n.s = curr;
curr += len;
break;
- case OP_EXACTN:
- case OP_EXACTN_IC:
+ case OP_STR_N:
+ case OP_STR_N_IC:
len = op->exact_n.n;
copy:
xmemcpy(curr, op->exact_n.s, len);
op->exact_n.s = curr;
curr += len;
break;
- case OP_EXACTMB2N:
+ case OP_STR_MB2N:
len = op->exact_n.n * 2;
goto copy;
break;
- case OP_EXACTMB3N:
+ case OP_STR_MB3N:
len = op->exact_n.n * 3;
goto copy;
break;
static void
-swap_node(Node* a, Node* b)
+node_swap(Node* a, Node* b)
{
Node c;
}
}
+static int
+node_list_len(Node* list)
+{
+ int len;
+
+ len = 1;
+ while (IS_NOT_NULL(NODE_CDR(list))) {
+ list = NODE_CDR(list);
+ len++;
+ }
+
+ return len;
+}
+
+static Node*
+node_list_add(Node* list, Node* x)
+{
+ Node *n;
+
+ n = onig_node_new_list(x, NULL);
+ if (IS_NULL(n)) return NULL_NODE;
+
+ if (IS_NOT_NULL(list)) {
+ while (IS_NOT_NULL(NODE_CDR(list)))
+ list = NODE_CDR(list);
+
+ NODE_CDR(list) = n;
+ }
+
+ return n;
+}
+
+static int
+node_str_node_cat(Node* node, Node* add)
+{
+ int r;
+
+ if (STR_(node)->flag != STR_(add)->flag)
+ return ONIGERR_TYPE_BUG;
+
+ r = onig_node_str_cat(node, STR_(add)->s, STR_(add)->end);
+ if (r != 0) return r;
+
+ if (NODE_STRING_IS_CASE_FOLD_MATCH(node))
+ STR_(node)->case_min_len += STR_(add)->case_min_len;
+
+ return 0;
+}
+
+static int
+node_str_cat_case_fold(Node* node, const UChar* s, const UChar* end, int case_min_len)
+{
+ int r;
+
+ if (! NODE_STRING_IS_CASE_FOLD_MATCH(node))
+ return ONIGERR_TYPE_BUG;
+
+ r = onig_node_str_cat(node, s, end);
+ if (r != 0) return r;
+
+ STR_(node)->case_min_len += case_min_len;
+ return 0;
+}
+
+static void
+node_conv_to_str_node(Node* node, int flag)
+{
+ NODE_SET_TYPE(node, NODE_STRING);
+ STR_(node)->flag = flag;
+ STR_(node)->s = STR_(node)->buf;
+ STR_(node)->end = STR_(node)->buf;
+ STR_(node)->capacity = 0;
+ STR_(node)->case_min_len = 0;
+}
+
static OnigLen
distance_add(OnigLen d1, OnigLen d2)
{
static int compile_tree(Node* node, regex_t* reg, ScanEnv* env);
-#define IS_NEED_STR_LEN_OP_EXACT(op) \
- ((op) == OP_EXACTN || (op) == OP_EXACTMB2N ||\
- (op) == OP_EXACTMB3N || (op) == OP_EXACTMBN || (op) == OP_EXACTN_IC)
+#define IS_NEED_STR_LEN_OP(op) \
+ ((op) == OP_STR_N || (op) == OP_STR_MB2N ||\
+ (op) == OP_STR_MB3N || (op) == OP_STR_MBN || (op) == OP_STR_N_IC)
static int
-select_str_opcode(int mb_len, int str_len, int ignore_case)
+select_str_opcode(int mb_len, int str_len)
{
int op;
- if (ignore_case) {
+ switch (mb_len) {
+ case 1:
switch (str_len) {
- case 1: op = OP_EXACT1_IC; break;
- default: op = OP_EXACTN_IC; break;
+ case 1: op = OP_STR_1; break;
+ case 2: op = OP_STR_2; break;
+ case 3: op = OP_STR_3; break;
+ case 4: op = OP_STR_4; break;
+ case 5: op = OP_STR_5; break;
+ default: op = OP_STR_N; break;
}
- }
- else {
- switch (mb_len) {
- case 1:
- switch (str_len) {
- case 1: op = OP_EXACT1; break;
- case 2: op = OP_EXACT2; break;
- case 3: op = OP_EXACT3; break;
- case 4: op = OP_EXACT4; break;
- case 5: op = OP_EXACT5; break;
- default: op = OP_EXACTN; break;
- }
- break;
+ break;
- case 2:
- switch (str_len) {
- case 1: op = OP_EXACTMB2N1; break;
- case 2: op = OP_EXACTMB2N2; break;
- case 3: op = OP_EXACTMB2N3; break;
- default: op = OP_EXACTMB2N; break;
- }
- break;
+ case 2:
+ switch (str_len) {
+ case 1: op = OP_STR_MB2N1; break;
+ case 2: op = OP_STR_MB2N2; break;
+ case 3: op = OP_STR_MB2N3; break;
+ default: op = OP_STR_MB2N; break;
+ }
+ break;
- case 3:
- op = OP_EXACTMB3N;
- break;
+ case 3:
+ op = OP_STR_MB3N;
+ break;
- default:
- op = OP_EXACTMBN;
- break;
- }
+ default:
+ op = OP_STR_MBN;
+ break;
}
+
return op;
}
}
static int
-compile_tree_empty_check(Node* node, regex_t* reg, int emptiness, ScanEnv* env)
+compile_quant_body_with_empty_check(QuantNode* qn, regex_t* reg, ScanEnv* env)
{
int r;
- int saved_num_null_check = reg->num_null_check;
+ int saved_num_empty_check;
+ int emptiness;
+ Node* body;
+
+ body = NODE_BODY((Node* )qn);
+ emptiness = qn->emptiness;
+ saved_num_empty_check = reg->num_empty_check;
if (emptiness != BODY_IS_NOT_EMPTY) {
r = add_op(reg, OP_EMPTY_CHECK_START);
if (r != 0) return r;
- COP(reg)->empty_check_start.mem = reg->num_null_check; /* NULL CHECK ID */
- reg->num_null_check++;
+ COP(reg)->empty_check_start.mem = reg->num_empty_check; /* NULL CHECK ID */
+ reg->num_empty_check++;
}
- r = compile_tree(node, reg, env);
+ r = compile_tree(body, reg, env);
if (r != 0) return r;
if (emptiness != BODY_IS_NOT_EMPTY) {
if (emptiness == BODY_IS_EMPTY_POSSIBILITY)
r = add_op(reg, OP_EMPTY_CHECK_END);
- else if (emptiness == BODY_IS_EMPTY_POSSIBILITY_MEM)
- r = add_op(reg, OP_EMPTY_CHECK_END_MEMST);
+ else if (emptiness == BODY_IS_EMPTY_POSSIBILITY_MEM) {
+ if (NODE_IS_EMPTY_STATUS_CHECK(qn) != 0)
+ r = add_op(reg, OP_EMPTY_CHECK_END_MEMST);
+ else
+ r = add_op(reg, OP_EMPTY_CHECK_END);
+ }
+#ifdef USE_CALL
else if (emptiness == BODY_IS_EMPTY_POSSIBILITY_REC)
r = add_op(reg, OP_EMPTY_CHECK_END_MEMST_PUSH);
+#endif
if (r != 0) return r;
- COP(reg)->empty_check_end.mem = saved_num_null_check; /* NULL CHECK ID */
+ COP(reg)->empty_check_end.mem = saved_num_empty_check; /* NULL CHECK ID */
}
return r;
}
static int
add_compile_string_length(UChar* s ARG_UNUSED, int mb_len, int str_len,
- regex_t* reg ARG_UNUSED, int ignore_case)
+ regex_t* reg ARG_UNUSED)
{
return 1;
}
static int
-add_compile_string(UChar* s, int mb_len, int str_len,
- regex_t* reg, int ignore_case)
+add_compile_string(UChar* s, int mb_len, int str_len, regex_t* reg)
{
int op;
int r;
UChar* p;
UChar* end;
- op = select_str_opcode(mb_len, str_len, ignore_case);
+ op = select_str_opcode(mb_len, str_len);
r = add_op(reg, op);
if (r != 0) return r;
byte_len = mb_len * str_len;
end = s + byte_len;
- if (op == OP_EXACTMBN) {
+ if (op == OP_STR_MBN) {
p = onigenc_strdup(reg->enc, s, end);
CHECK_NULL_RETURN_MEMERR(p);
COP(reg)->exact_len_n.n = str_len;
COP(reg)->exact_len_n.s = p;
}
- else if (IS_NEED_STR_LEN_OP_EXACT(op)) {
+ else if (IS_NEED_STR_LEN_OP(op)) {
p = onigenc_strdup(reg->enc, s, end);
CHECK_NULL_RETURN_MEMERR(p);
- if (op == OP_EXACTN_IC)
+ if (op == OP_STR_N_IC)
COP(reg)->exact_n.n = byte_len;
else
COP(reg)->exact_n.n = str_len;
static int
compile_length_string_node(Node* node, regex_t* reg)
{
- int rlen, r, len, prev_len, slen, ambig;
+ int rlen, r, len, prev_len, slen;
UChar *p, *prev;
StrNode* sn;
OnigEncoding enc = reg->enc;
if (sn->end <= sn->s)
return 0;
- ambig = NODE_STRING_IS_AMBIG(node);
+ if (NODE_STRING_IS_CASE_FOLD_MATCH(node) != 0) return 1;
p = prev = sn->s;
- SAFE_ENC_LEN(enc, p, sn->end, prev_len);
+ prev_len = enclen(enc, p);
p += prev_len;
slen = 1;
rlen = 0;
for (; p < sn->end; ) {
- SAFE_ENC_LEN(enc, p, sn->end, len);
+ len = enclen(enc, p);
if (len == prev_len) {
slen++;
}
else {
- r = add_compile_string_length(prev, prev_len, slen, reg, ambig);
+ r = add_compile_string_length(prev, prev_len, slen, reg);
rlen += r;
prev = p;
slen = 1;
p += len;
}
- r = add_compile_string_length(prev, prev_len, slen, reg, ambig);
+ r = add_compile_string_length(prev, prev_len, slen, reg);
rlen += r;
return rlen;
}
static int
-compile_length_string_raw_node(StrNode* sn, regex_t* reg)
+compile_length_string_crude_node(StrNode* sn, regex_t* reg)
{
if (sn->end <= sn->s)
return 0;
return add_compile_string_length(sn->s, 1 /* sb */, (int )(sn->end - sn->s),
- reg, 0);
+ reg);
+}
+
+static int
+compile_ambig_string_node(Node* node, regex_t* reg)
+{
+ int r;
+ int len;
+ int byte_len;
+ UChar* p;
+ StrNode* sn;
+ OnigEncoding enc = reg->enc;
+
+ sn = STR_(node);
+ len = enclen(enc, sn->s);
+ byte_len = (int )(sn->end - sn->s);
+ if (len == byte_len) {
+ r = add_op(reg, OP_STR_1_IC);
+ if (r != 0) return r;
+
+ xmemset(COP(reg)->exact.s, 0, sizeof(COP(reg)->exact.s));
+ xmemcpy(COP(reg)->exact.s, sn->s, (size_t )byte_len);
+ }
+ else {
+ r = add_op(reg, OP_STR_N_IC);
+ if (r != 0) return r;
+
+ p = onigenc_strdup(enc, sn->s, sn->end);
+ CHECK_NULL_RETURN_MEMERR(p);
+
+ COP(reg)->exact_n.s = p;
+ COP(reg)->exact_n.n = byte_len;
+ }
+
+ return 0;
}
static int
compile_string_node(Node* node, regex_t* reg)
{
- int r, len, prev_len, slen, ambig;
+ int r, len, prev_len, slen;
UChar *p, *prev, *end;
StrNode* sn;
OnigEncoding enc = reg->enc;
return 0;
end = sn->end;
- ambig = NODE_STRING_IS_AMBIG(node);
+ if (NODE_STRING_IS_CASE_FOLD_MATCH(node) != 0) {
+ return compile_ambig_string_node(node, reg);
+ }
p = prev = sn->s;
- SAFE_ENC_LEN(enc, p, end, prev_len);
+ prev_len = enclen(enc, p);
p += prev_len;
slen = 1;
for (; p < end; ) {
- SAFE_ENC_LEN(enc, p, end, len);
+ len = enclen(enc, p);
if (len == prev_len) {
slen++;
}
else {
- r = add_compile_string(prev, prev_len, slen, reg, ambig);
+ r = add_compile_string(prev, prev_len, slen, reg);
if (r != 0) return r;
prev = p;
p += len;
}
- return add_compile_string(prev, prev_len, slen, reg, ambig);
+ return add_compile_string(prev, prev_len, slen, reg);
}
static int
-compile_string_raw_node(StrNode* sn, regex_t* reg)
+compile_string_crude_node(StrNode* sn, regex_t* reg)
{
if (sn->end <= sn->s)
return 0;
- return add_compile_string(sn->s, 1 /* sb */, (int )(sn->end - sn->s), reg, 0);
+ return add_compile_string(sn->s, 1 /* sb */, (int )(sn->end - sn->s), reg);
}
static void*
return 0;
}
+static void
+set_addr_in_repeat_range(regex_t* reg)
+{
+ int i;
+
+ for (i = 0; i < reg->num_repeat; i++) {
+ RepeatRange* p = reg->repeat_range + i;
+ int offset = p->u.offset;
+ p->u.pcode = reg->ops + offset;
+ }
+}
+
static int
-entry_repeat_range(regex_t* reg, int id, int lower, int upper)
+entry_repeat_range(regex_t* reg, int id, int lower, int upper, int ops_index)
{
#define REPEAT_RANGE_ALLOC 4
- OnigRepeatRange* p;
+ RepeatRange* p;
if (reg->repeat_range_alloc == 0) {
- p = (OnigRepeatRange* )xmalloc(sizeof(OnigRepeatRange) * REPEAT_RANGE_ALLOC);
+ p = (RepeatRange* )xmalloc(sizeof(RepeatRange) * REPEAT_RANGE_ALLOC);
CHECK_NULL_RETURN_MEMERR(p);
reg->repeat_range = p;
reg->repeat_range_alloc = REPEAT_RANGE_ALLOC;
else if (reg->repeat_range_alloc <= id) {
int n;
n = reg->repeat_range_alloc + REPEAT_RANGE_ALLOC;
- p = (OnigRepeatRange* )xrealloc(reg->repeat_range, sizeof(OnigRepeatRange) * n);
+ p = (RepeatRange* )xrealloc(reg->repeat_range, sizeof(RepeatRange) * n);
CHECK_NULL_RETURN_MEMERR(p);
reg->repeat_range = p;
reg->repeat_range_alloc = n;
p = reg->repeat_range;
}
- p[id].lower = lower;
- p[id].upper = (IS_INFINITE_REPEAT(upper) ? 0x7fffffff : upper);
+ p[id].lower = lower;
+ p[id].upper = (IS_INFINITE_REPEAT(upper) ? 0x7fffffff : upper);
+ p[id].u.offset = ops_index;
return 0;
}
if (r != 0) return r;
COP(reg)->repeat.id = num_repeat;
- COP(reg)->repeat.addr = SIZE_INC_OP + target_len + SIZE_OP_REPEAT_INC;
+ COP(reg)->repeat.addr = SIZE_INC + target_len + OPSIZE_REPEAT_INC;
- r = entry_repeat_range(reg, num_repeat, qn->lower, qn->upper);
+ r = entry_repeat_range(reg, num_repeat, qn->lower, qn->upper,
+ COP_CURR_OFFSET(reg) + OPSIZE_REPEAT);
if (r != 0) return r;
- r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, emptiness, env);
+ r = compile_quant_body_with_empty_check(qn, reg, env);
if (r != 0) return r;
- if (
-#ifdef USE_CALL
- NODE_IS_IN_MULTI_ENTRY(qn) ||
-#endif
- NODE_IS_IN_REAL_REPEAT(qn)) {
- r = add_op(reg, qn->greedy ? OP_REPEAT_INC_SG : OP_REPEAT_INC_NG_SG);
- }
- else {
- r = add_op(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG);
- }
+ r = add_op(reg, qn->greedy ? OP_REPEAT_INC : OP_REPEAT_INC_NG);
if (r != 0) return r;
COP(reg)->repeat_inc.id = num_repeat;
if (qn->lower <= 1 ||
int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0) {
if (IS_NOT_NULL(qn->next_head_exact))
- return SIZE_OP_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower;
+ return OPSIZE_ANYCHAR_STAR_PEEK_NEXT + tlen * qn->lower;
else
- return SIZE_OP_ANYCHAR_STAR + tlen * qn->lower;
+ return OPSIZE_ANYCHAR_STAR + tlen * qn->lower;
}
}
mod_tlen = tlen;
if (emptiness != BODY_IS_NOT_EMPTY)
- mod_tlen += SIZE_OP_EMPTY_CHECK_START + SIZE_OP_EMPTY_CHECK_END;
+ mod_tlen += OPSIZE_EMPTY_CHECK_START + OPSIZE_EMPTY_CHECK_END;
if (infinite &&
(qn->lower <= 1 ||
int_multiply_cmp(tlen, qn->lower, QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {
if (qn->lower == 1 && tlen > QUANTIFIER_EXPAND_LIMIT_SIZE) {
- len = SIZE_OP_JUMP;
+ len = OPSIZE_JUMP;
}
else {
len = tlen * qn->lower;
if (qn->greedy) {
#ifdef USE_OP_PUSH_OR_JUMP_EXACT
if (IS_NOT_NULL(qn->head_exact))
- len += SIZE_OP_PUSH_OR_JUMP_EXACT1 + mod_tlen + SIZE_OP_JUMP;
+ len += OPSIZE_PUSH_OR_JUMP_EXACT1 + mod_tlen + OPSIZE_JUMP;
else
#endif
if (IS_NOT_NULL(qn->next_head_exact))
- len += SIZE_OP_PUSH_IF_PEEK_NEXT + mod_tlen + SIZE_OP_JUMP;
+ len += OPSIZE_PUSH_IF_PEEK_NEXT + mod_tlen + OPSIZE_JUMP;
else
- len += SIZE_OP_PUSH + mod_tlen + SIZE_OP_JUMP;
+ len += OPSIZE_PUSH + mod_tlen + OPSIZE_JUMP;
}
else
- len += SIZE_OP_JUMP + mod_tlen + SIZE_OP_PUSH;
+ len += OPSIZE_JUMP + mod_tlen + OPSIZE_PUSH;
}
else if (qn->upper == 0) {
- if (qn->is_refered != 0) { /* /(?<n>..){0}/ */
- len = SIZE_OP_JUMP + tlen;
+ if (qn->include_referred != 0) { /* /(?<n>..){0}/ */
+ len = OPSIZE_JUMP + tlen;
}
else
len = 0;
}
else if (!infinite && qn->greedy &&
(qn->upper == 1 ||
- int_multiply_cmp(tlen + SIZE_OP_PUSH, qn->upper,
+ int_multiply_cmp(tlen + OPSIZE_PUSH, qn->upper,
QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {
len = tlen * qn->lower;
- len += (SIZE_OP_PUSH + tlen) * (qn->upper - qn->lower);
+ len += (OPSIZE_PUSH + tlen) * (qn->upper - qn->lower);
}
else if (!qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */
- len = SIZE_OP_PUSH + SIZE_OP_JUMP + tlen;
+ len = OPSIZE_PUSH + OPSIZE_JUMP + tlen;
}
else {
- len = SIZE_OP_REPEAT_INC + mod_tlen + SIZE_OP_REPEAT;
+ len = OPSIZE_REPEAT_INC + mod_tlen + OPSIZE_REPEAT;
}
return len;
mod_tlen = tlen;
if (emptiness != BODY_IS_NOT_EMPTY)
- mod_tlen += SIZE_OP_EMPTY_CHECK_START + SIZE_OP_EMPTY_CHECK_END;
+ mod_tlen += OPSIZE_EMPTY_CHECK_START + OPSIZE_EMPTY_CHECK_END;
if (infinite &&
(qn->lower <= 1 ||
if (qn->greedy) {
#ifdef USE_OP_PUSH_OR_JUMP_EXACT
if (IS_NOT_NULL(qn->head_exact))
- COP(reg)->jump.addr = SIZE_OP_PUSH_OR_JUMP_EXACT1 + SIZE_INC_OP;
+ COP(reg)->jump.addr = OPSIZE_PUSH_OR_JUMP_EXACT1 + SIZE_INC;
else
#endif
if (IS_NOT_NULL(qn->next_head_exact))
- COP(reg)->jump.addr = SIZE_OP_PUSH_IF_PEEK_NEXT + SIZE_INC_OP;
+ COP(reg)->jump.addr = OPSIZE_PUSH_IF_PEEK_NEXT + SIZE_INC;
else
- COP(reg)->jump.addr = SIZE_OP_PUSH + SIZE_INC_OP;
+ COP(reg)->jump.addr = OPSIZE_PUSH + SIZE_INC;
}
else {
- COP(reg)->jump.addr = SIZE_OP_JUMP + SIZE_INC_OP;
+ COP(reg)->jump.addr = OPSIZE_JUMP + SIZE_INC;
}
}
else {
if (IS_NOT_NULL(qn->head_exact)) {
r = add_op(reg, OP_PUSH_OR_JUMP_EXACT1);
if (r != 0) return r;
- COP(reg)->push_or_jump_exact1.addr = SIZE_INC_OP + mod_tlen + SIZE_OP_JUMP;
+ COP(reg)->push_or_jump_exact1.addr = SIZE_INC + mod_tlen + OPSIZE_JUMP;
COP(reg)->push_or_jump_exact1.c = STR_(qn->head_exact)->s[0];
- r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, emptiness, env);
+ r = compile_quant_body_with_empty_check(qn, reg, env);
if (r != 0) return r;
- addr = -(mod_tlen + (int )SIZE_OP_PUSH_OR_JUMP_EXACT1);
+ addr = -(mod_tlen + (int )OPSIZE_PUSH_OR_JUMP_EXACT1);
}
else
#endif
if (IS_NOT_NULL(qn->next_head_exact)) {
r = add_op(reg, OP_PUSH_IF_PEEK_NEXT);
if (r != 0) return r;
- COP(reg)->push_if_peek_next.addr = SIZE_INC_OP + mod_tlen + SIZE_OP_JUMP;
+ COP(reg)->push_if_peek_next.addr = SIZE_INC + mod_tlen + OPSIZE_JUMP;
COP(reg)->push_if_peek_next.c = STR_(qn->next_head_exact)->s[0];
- r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, emptiness, env);
+ r = compile_quant_body_with_empty_check(qn, reg, env);
if (r != 0) return r;
- addr = -(mod_tlen + (int )SIZE_OP_PUSH_IF_PEEK_NEXT);
+ addr = -(mod_tlen + (int )OPSIZE_PUSH_IF_PEEK_NEXT);
}
else {
r = add_op(reg, OP_PUSH);
if (r != 0) return r;
- COP(reg)->push.addr = SIZE_INC_OP + mod_tlen + SIZE_OP_JUMP;
+ COP(reg)->push.addr = SIZE_INC + mod_tlen + OPSIZE_JUMP;
- r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, emptiness, env);
+ r = compile_quant_body_with_empty_check(qn, reg, env);
if (r != 0) return r;
- addr = -(mod_tlen + (int )SIZE_OP_PUSH);
+ addr = -(mod_tlen + (int )OPSIZE_PUSH);
}
r = add_op(reg, OP_JUMP);
else {
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = mod_tlen + SIZE_INC_OP;
+ COP(reg)->jump.addr = mod_tlen + SIZE_INC;
- r = compile_tree_empty_check(NODE_QUANT_BODY(qn), reg, emptiness, env);
+ r = compile_quant_body_with_empty_check(qn, reg, env);
if (r != 0) return r;
r = add_op(reg, OP_PUSH);
}
}
else if (qn->upper == 0) {
- if (qn->is_refered != 0) { /* /(?<n>..){0}/ */
+ if (qn->include_referred != 0) { /* /(?<n>..){0}/ */
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = tlen + SIZE_INC_OP;
+ COP(reg)->jump.addr = tlen + SIZE_INC;
r = compile_tree(NODE_QUANT_BODY(qn), reg, env);
}
}
else if (! infinite && qn->greedy &&
(qn->upper == 1 ||
- int_multiply_cmp(tlen + SIZE_OP_PUSH, qn->upper,
+ int_multiply_cmp(tlen + OPSIZE_PUSH, qn->upper,
QUANTIFIER_EXPAND_LIMIT_SIZE) <= 0)) {
int n = qn->upper - qn->lower;
if (r != 0) return r;
for (i = 0; i < n; i++) {
- int v = onig_positive_int_multiply(n - i, tlen + SIZE_OP_PUSH);
+ int v = onig_positive_int_multiply(n - i, tlen + OPSIZE_PUSH);
if (v < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
r = add_op(reg, OP_PUSH);
else if (! qn->greedy && qn->upper == 1 && qn->lower == 0) { /* '??' */
r = add_op(reg, OP_PUSH);
if (r != 0) return r;
- COP(reg)->push.addr = SIZE_INC_OP + SIZE_OP_JUMP;
+ COP(reg)->push.addr = SIZE_INC + OPSIZE_JUMP;
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = tlen + SIZE_INC_OP;
+ COP(reg)->jump.addr = tlen + SIZE_INC;
r = compile_tree(NODE_QUANT_BODY(qn), reg, env);
}
#ifdef USE_CALL
if (node->m.regnum == 0 && NODE_IS_CALLED(node)) {
- len = tlen + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;
+ len = tlen + OPSIZE_CALL + OPSIZE_JUMP + OPSIZE_RETURN;
return len;
}
if (NODE_IS_CALLED(node)) {
- len = SIZE_OP_MEMORY_START_PUSH + tlen
- + SIZE_OP_CALL + SIZE_OP_JUMP + SIZE_OP_RETURN;
- if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))
+ len = OPSIZE_MEM_START_PUSH + tlen
+ + OPSIZE_CALL + OPSIZE_JUMP + OPSIZE_RETURN;
+ if (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum))
len += (NODE_IS_RECURSION(node)
- ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);
+ ? OPSIZE_MEM_END_PUSH_REC : OPSIZE_MEM_END_PUSH);
else
len += (NODE_IS_RECURSION(node)
- ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);
+ ? OPSIZE_MEM_END_REC : OPSIZE_MEM_END);
}
else if (NODE_IS_RECURSION(node)) {
- len = SIZE_OP_MEMORY_START_PUSH;
- len += tlen + (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum)
- ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_REC);
+ len = OPSIZE_MEM_START_PUSH;
+ len += tlen + (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum)
+ ? OPSIZE_MEM_END_PUSH_REC : OPSIZE_MEM_END_REC);
}
else
#endif
{
- if (MEM_STATUS_AT0(reg->bt_mem_start, node->m.regnum))
- len = SIZE_OP_MEMORY_START_PUSH;
+ if (MEM_STATUS_AT0(reg->push_mem_start, node->m.regnum))
+ len = OPSIZE_MEM_START_PUSH;
else
- len = SIZE_OP_MEMORY_START;
+ len = OPSIZE_MEM_START;
- len += tlen + (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum)
- ? SIZE_OP_MEMORY_END_PUSH : SIZE_OP_MEMORY_END);
+ len += tlen + (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum)
+ ? OPSIZE_MEM_END_PUSH : OPSIZE_MEM_END);
}
break;
v = onig_positive_int_multiply(qn->lower, tlen);
if (v < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
- len = v + SIZE_OP_PUSH + tlen + SIZE_OP_POP_OUT + SIZE_OP_JUMP;
+ len = v + OPSIZE_PUSH + tlen + OPSIZE_POP_OUT + OPSIZE_JUMP;
}
else {
- len = SIZE_OP_ATOMIC_START + tlen + SIZE_OP_ATOMIC_END;
+ len = OPSIZE_ATOMIC_START + tlen + OPSIZE_ATOMIC_END;
}
break;
len = compile_length_tree(cond, reg);
if (len < 0) return len;
- len += SIZE_OP_PUSH;
- len += SIZE_OP_ATOMIC_START + SIZE_OP_ATOMIC_END;
+ len += OPSIZE_PUSH;
+ len += OPSIZE_ATOMIC_START + OPSIZE_ATOMIC_END;
if (IS_NOT_NULL(Then)) {
tlen = compile_length_tree(Then, reg);
len += tlen;
}
- len += SIZE_OP_JUMP + SIZE_OP_ATOMIC_END;
+ len += OPSIZE_JUMP + OPSIZE_ATOMIC_END;
if (IS_NOT_NULL(Else)) {
tlen = compile_length_tree(Else, reg);
compile_bag_memory_node(BagNode* node, regex_t* reg, ScanEnv* env)
{
int r;
- int len;
#ifdef USE_CALL
if (NODE_IS_CALLED(node)) {
+ int len;
+
r = add_op(reg, OP_CALL);
if (r != 0) return r;
- node->m.called_addr = COP_CURR_OFFSET(reg) + 1 + SIZE_OP_JUMP;
+ node->m.called_addr = COP_CURR_OFFSET(reg) + 1 + OPSIZE_JUMP;
NODE_STATUS_ADD(node, ADDR_FIXED);
COP(reg)->call.addr = (int )node->m.called_addr;
if (node->m.regnum == 0) {
len = compile_length_tree(NODE_BAG_BODY(node), reg);
- len += SIZE_OP_RETURN;
+ len += OPSIZE_RETURN;
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = len + SIZE_INC_OP;
+ COP(reg)->jump.addr = len + SIZE_INC;
r = compile_tree(NODE_BAG_BODY(node), reg, env);
if (r != 0) return r;
}
else {
len = compile_length_tree(NODE_BAG_BODY(node), reg);
- len += (SIZE_OP_MEMORY_START_PUSH + SIZE_OP_RETURN);
- if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))
+ len += (OPSIZE_MEM_START_PUSH + OPSIZE_RETURN);
+ if (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum))
len += (NODE_IS_RECURSION(node)
- ? SIZE_OP_MEMORY_END_PUSH_REC : SIZE_OP_MEMORY_END_PUSH);
+ ? OPSIZE_MEM_END_PUSH_REC : OPSIZE_MEM_END_PUSH);
else
- len += (NODE_IS_RECURSION(node)
- ? SIZE_OP_MEMORY_END_REC : SIZE_OP_MEMORY_END);
+ len += (NODE_IS_RECURSION(node) ? OPSIZE_MEM_END_REC : OPSIZE_MEM_END);
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = len + SIZE_INC_OP;
+ COP(reg)->jump.addr = len + SIZE_INC;
}
}
#endif
- if (MEM_STATUS_AT0(reg->bt_mem_start, node->m.regnum))
- r = add_op(reg, OP_MEMORY_START_PUSH);
+ if (MEM_STATUS_AT0(reg->push_mem_start, node->m.regnum))
+ r = add_op(reg, OP_MEM_START_PUSH);
else
- r = add_op(reg, OP_MEMORY_START);
+ r = add_op(reg, OP_MEM_START);
if (r != 0) return r;
COP(reg)->memory_start.num = node->m.regnum;
if (r != 0) return r;
#ifdef USE_CALL
- if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))
+ if (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum))
r = add_op(reg, (NODE_IS_RECURSION(node)
- ? OP_MEMORY_END_PUSH_REC : OP_MEMORY_END_PUSH));
+ ? OP_MEM_END_PUSH_REC : OP_MEM_END_PUSH));
else
- r = add_op(reg, (NODE_IS_RECURSION(node) ? OP_MEMORY_END_REC : OP_MEMORY_END));
+ r = add_op(reg, (NODE_IS_RECURSION(node) ? OP_MEM_END_REC : OP_MEM_END));
if (r != 0) return r;
COP(reg)->memory_end.num = node->m.regnum;
r = add_op(reg, OP_RETURN);
}
#else
- if (MEM_STATUS_AT0(reg->bt_mem_end, node->m.regnum))
- r = add_op(reg, OP_MEMORY_END_PUSH);
+ if (MEM_STATUS_AT0(reg->push_mem_end, node->m.regnum))
+ r = add_op(reg, OP_MEM_END_PUSH);
else
- r = add_op(reg, OP_MEMORY_END);
+ r = add_op(reg, OP_MEM_END);
if (r != 0) return r;
COP(reg)->memory_end.num = node->m.regnum;
#endif
r = add_op(reg, OP_PUSH);
if (r != 0) return r;
- COP(reg)->push.addr = SIZE_INC_OP + len + SIZE_OP_POP_OUT + SIZE_OP_JUMP;
+ COP(reg)->push.addr = SIZE_INC + len + OPSIZE_POP_OUT + OPSIZE_JUMP;
r = compile_tree(NODE_QUANT_BODY(qn), reg, env);
if (r != 0) return r;
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = -((int )SIZE_OP_PUSH + len + (int )SIZE_OP_POP_OUT);
+ COP(reg)->jump.addr = -((int )OPSIZE_PUSH + len + (int )OPSIZE_POP_OUT);
}
else {
r = add_op(reg, OP_ATOMIC_START);
else
then_len = 0;
- jump_len = cond_len + then_len + SIZE_OP_ATOMIC_END + SIZE_OP_JUMP;
+ jump_len = cond_len + then_len + OPSIZE_ATOMIC_END + OPSIZE_JUMP;
r = add_op(reg, OP_PUSH);
if (r != 0) return r;
- COP(reg)->push.addr = SIZE_INC_OP + jump_len;
+ COP(reg)->push.addr = SIZE_INC + jump_len;
r = compile_tree(cond, reg, env);
if (r != 0) return r;
r = add_op(reg, OP_JUMP);
if (r != 0) return r;
- COP(reg)->jump.addr = SIZE_OP_ATOMIC_END + else_len + SIZE_INC_OP;
+ COP(reg)->jump.addr = OPSIZE_ATOMIC_END + else_len + SIZE_INC;
r = add_op(reg, OP_ATOMIC_END);
if (r != 0) return r;
switch (node->type) {
case ANCR_PREC_READ:
- len = SIZE_OP_PREC_READ_START + tlen + SIZE_OP_PREC_READ_END;
+ len = OPSIZE_PREC_READ_START + tlen + OPSIZE_PREC_READ_END;
break;
case ANCR_PREC_READ_NOT:
- len = SIZE_OP_PREC_READ_NOT_START + tlen + SIZE_OP_PREC_READ_NOT_END;
+ len = OPSIZE_PREC_READ_NOT_START + tlen + OPSIZE_PREC_READ_NOT_END;
break;
case ANCR_LOOK_BEHIND:
- len = SIZE_OP_LOOK_BEHIND + tlen;
+ len = OPSIZE_LOOK_BEHIND + tlen;
break;
case ANCR_LOOK_BEHIND_NOT:
- len = SIZE_OP_LOOK_BEHIND_NOT_START + tlen + SIZE_OP_LOOK_BEHIND_NOT_END;
+ len = OPSIZE_LOOK_BEHIND_NOT_START + tlen + OPSIZE_LOOK_BEHIND_NOT_END;
break;
case ANCR_WORD_BOUNDARY:
case ANCR_WORD_BEGIN:
case ANCR_WORD_END:
#endif
- len = SIZE_OP_WORD_BOUNDARY;
+ len = OPSIZE_WORD_BOUNDARY;
break;
case ANCR_TEXT_SEGMENT_BOUNDARY:
r = add_op(reg, OP_PREC_READ_NOT_START);
if (r != 0) return r;
- COP(reg)->prec_read_not_start.addr = SIZE_INC_OP + len + SIZE_OP_PREC_READ_NOT_END;
+ COP(reg)->prec_read_not_start.addr = SIZE_INC + len + OPSIZE_PREC_READ_NOT_END;
r = compile_tree(NODE_ANCHOR_BODY(node), reg, env);
if (r != 0) return r;
r = add_op(reg, OP_PREC_READ_NOT_END);
len = compile_length_tree(NODE_ANCHOR_BODY(node), reg);
r = add_op(reg, OP_LOOK_BEHIND_NOT_START);
if (r != 0) return r;
- COP(reg)->look_behind_not_start.addr = SIZE_INC_OP + len + SIZE_OP_LOOK_BEHIND_NOT_END;
+ COP(reg)->look_behind_not_start.addr = SIZE_INC + len + OPSIZE_LOOK_BEHIND_NOT_END;
if (node->char_len < 0) {
r = get_char_len_node(NODE_ANCHOR_BODY(node), reg, &n);
switch (node->type) {
case GIMMICK_FAIL:
- len = SIZE_OP_FAIL;
+ len = OPSIZE_FAIL;
break;
case GIMMICK_SAVE:
- len = SIZE_OP_PUSH_SAVE_VAL;
+ len = OPSIZE_PUSH_SAVE_VAL;
break;
case GIMMICK_UPDATE_VAR:
- len = SIZE_OP_UPDATE_VAR;
+ len = OPSIZE_UPDATE_VAR;
break;
#ifdef USE_CALLOUT
case GIMMICK_CALLOUT:
switch (node->detail_type) {
case ONIG_CALLOUT_OF_CONTENTS:
- len = SIZE_OP_CALLOUT_CONTENTS;
+ len = OPSIZE_CALLOUT_CONTENTS;
break;
case ONIG_CALLOUT_OF_NAME:
- len = SIZE_OP_CALLOUT_NAME;
+ len = OPSIZE_CALLOUT_NAME;
break;
default:
r += compile_length_tree(NODE_CAR(node), reg);
n++;
} while (IS_NOT_NULL(node = NODE_CDR(node)));
- r += (SIZE_OP_PUSH + SIZE_OP_JUMP) * (n - 1);
+ r += (OPSIZE_PUSH + OPSIZE_JUMP) * (n - 1);
}
break;
case NODE_STRING:
- if (NODE_STRING_IS_RAW(node))
- r = compile_length_string_raw_node(STR_(node), reg);
+ if (NODE_STRING_IS_CRUDE(node))
+ r = compile_length_string_crude_node(STR_(node), reg);
else
r = compile_length_string_node(node, reg);
break;
break;
case NODE_BACKREF:
- r = SIZE_OP_BACKREF;
+ r = OPSIZE_BACKREF;
break;
#ifdef USE_CALL
case NODE_CALL:
- r = SIZE_OP_CALL;
+ r = OPSIZE_CALL;
break;
#endif
do {
len += compile_length_tree(NODE_CAR(x), reg);
if (IS_NOT_NULL(NODE_CDR(x))) {
- len += SIZE_OP_PUSH + SIZE_OP_JUMP;
+ len += OPSIZE_PUSH + OPSIZE_JUMP;
}
} while (IS_NOT_NULL(x = NODE_CDR(x)));
pos = COP_CURR_OFFSET(reg) + 1 + len; /* goal position */
enum OpCode push = NODE_IS_SUPER(node) ? OP_PUSH_SUPER : OP_PUSH;
r = add_op(reg, push);
if (r != 0) break;
- COP(reg)->push.addr = SIZE_INC_OP + len + SIZE_OP_JUMP;
+ COP(reg)->push.addr = SIZE_INC + len + OPSIZE_JUMP;
}
r = compile_tree(NODE_CAR(node), reg, env);
if (r != 0) break;
break;
case NODE_STRING:
- if (NODE_STRING_IS_RAW(node))
- r = compile_string_raw_node(STR_(node), reg);
+ if (NODE_STRING_IS_CRUDE(node))
+ r = compile_string_crude_node(STR_(node), reg);
else
r = compile_string_node(node, reg);
break;
Node** ptarget = &(NODE_BODY(node));
Node* old = *ptarget;
r = noname_disable_map(ptarget, map, counter);
+ if (r != 0) return r;
if (*ptarget != old && NODE_TYPE(*ptarget) == NODE_QUANT) {
- onig_reduce_nested_quantifier(node, *ptarget);
+ r = onig_reduce_nested_quantifier(node);
}
}
break;
}
}
- loc = env->capture_history;
- MEM_STATUS_CLEAR(env->capture_history);
+ loc = env->cap_history;
+ MEM_STATUS_CLEAR(env->cap_history);
for (i = 1; i <= ONIG_MAX_CAPTURE_HISTORY_GROUP; i++) {
if (MEM_STATUS_AT(loc, i)) {
- MEM_STATUS_ON_SIMPLE(env->capture_history, map[i].new_val);
+ MEM_STATUS_ON_SIMPLE(env->cap_history, map[i].new_val);
}
}
len = NODE_STRING_LEN(x);
if (len > NODE_STRING_LEN(y)) len = NODE_STRING_LEN(y);
- if (NODE_STRING_IS_AMBIG(x) || NODE_STRING_IS_AMBIG(y)) {
+ if (NODE_STRING_IS_CASE_FOLD_MATCH(x) || NODE_STRING_IS_CASE_FOLD_MATCH(y)) {
/* tiny version */
return 0;
}
break;
if (exact == 0 ||
- ! IS_IGNORECASE(reg->options) || NODE_STRING_IS_RAW(node)) {
+ ! IS_IGNORECASE(reg->options) || NODE_STRING_IS_CRUDE(node)) {
n = node;
}
}
if (NODE_IS_RECURSION(node)) break;
backs = BACKREFS_P(br);
- len = tree_min_len(mem_env[backs[0]].node, env);
+ len = tree_min_len(mem_env[backs[0]].mem_node, env);
for (i = 1; i < br->back_num; i++) {
- tmin = tree_min_len(mem_env[backs[i]].node, env);
+ tmin = tree_min_len(mem_env[backs[i]].mem_node, env);
if (len > tmin) len = tmin;
}
}
}
backs = BACKREFS_P(br);
for (i = 0; i < br->back_num; i++) {
- tmax = tree_max_len(mem_env[backs[i]].node, env);
+ tmax = tree_max_len(mem_env[backs[i]].mem_node, env);
if (len < tmax) len = tmax;
}
}
if (backs[i] > env->num_mem)
return ONIGERR_INVALID_BACKREF;
- NODE_STATUS_ADD(mem_env[backs[i]].node, BACKREF);
+ NODE_STATUS_ADD(mem_env[backs[i]].mem_node, BACKREF);
}
r = 0;
}
return r;
}
+static int
+set_empty_repeat_node_trav(Node* node, Node* empty, ScanEnv* env)
+{
+ int r;
+
+ switch (NODE_TYPE(node)) {
+ case NODE_LIST:
+ case NODE_ALT:
+ do {
+ r = set_empty_repeat_node_trav(NODE_CAR(node), empty, env);
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
+ break;
+
+ case NODE_ANCHOR:
+ {
+ AnchorNode* an = ANCHOR_(node);
+
+ if (! ANCHOR_HAS_BODY(an)) {
+ r = 0;
+ break;
+ }
+
+ switch (an->type) {
+ case ANCR_PREC_READ:
+ case ANCR_LOOK_BEHIND:
+ empty = NULL_NODE;
+ break;
+ default:
+ break;
+ }
+ r = set_empty_repeat_node_trav(NODE_BODY(node), empty, env);
+ }
+ break;
+
+ case NODE_QUANT:
+ {
+ QuantNode* qn = QUANT_(node);
+
+ if (qn->emptiness != BODY_IS_NOT_EMPTY) empty = node;
+ r = set_empty_repeat_node_trav(NODE_BODY(node), empty, env);
+ }
+ break;
+
+ case NODE_BAG:
+ if (IS_NOT_NULL(NODE_BODY(node))) {
+ r = set_empty_repeat_node_trav(NODE_BODY(node), empty, env);
+ if (r != 0) return r;
+ }
+ {
+ BagNode* en = BAG_(node);
+
+ if (en->type == BAG_MEMORY) {
+ if (NODE_IS_BACKREF(node)) {
+ if (IS_NOT_NULL(empty))
+ SCANENV_MEMENV(env)[en->m.regnum].empty_repeat_node = empty;
+ }
+ }
+ else if (en->type == BAG_IF_ELSE) {
+ if (IS_NOT_NULL(en->te.Then)) {
+ r = set_empty_repeat_node_trav(en->te.Then, empty, env);
+ if (r != 0) return r;
+ }
+ if (IS_NOT_NULL(en->te.Else)) {
+ r = set_empty_repeat_node_trav(en->te.Else, empty, env);
+ }
+ }
+ }
+ break;
+
+ default:
+ r = 0;
+ break;
+ }
+
+ return r;
+}
+
+static int
+is_ancestor_node(Node* node, Node* me)
+{
+ Node* parent;
+
+ while ((parent = NODE_PARENT(me)) != NULL_NODE) {
+ if (parent == node) return 1;
+ me = parent;
+ }
+ return 0;
+}
+
+static void
+set_empty_status_check_trav(Node* node, ScanEnv* env)
+{
+ switch (NODE_TYPE(node)) {
+ case NODE_LIST:
+ case NODE_ALT:
+ do {
+ set_empty_status_check_trav(NODE_CAR(node), env);
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));
+ break;
+
+ case NODE_ANCHOR:
+ {
+ AnchorNode* an = ANCHOR_(node);
+
+ if (! ANCHOR_HAS_BODY(an)) break;
+ set_empty_status_check_trav(NODE_BODY(node), env);
+ }
+ break;
+
+ case NODE_QUANT:
+ set_empty_status_check_trav(NODE_BODY(node), env);
+ break;
+
+ case NODE_BAG:
+ if (IS_NOT_NULL(NODE_BODY(node)))
+ set_empty_status_check_trav(NODE_BODY(node), env);
+ {
+ BagNode* en = BAG_(node);
+
+ if (en->type == BAG_IF_ELSE) {
+ if (IS_NOT_NULL(en->te.Then)) {
+ set_empty_status_check_trav(en->te.Then, env);
+ }
+ if (IS_NOT_NULL(en->te.Else)) {
+ set_empty_status_check_trav(en->te.Else, env);
+ }
+ }
+ }
+ break;
+
+ case NODE_BACKREF:
+ {
+ int i;
+ int* backs;
+ MemEnv* mem_env = SCANENV_MEMENV(env);
+ BackRefNode* br = BACKREF_(node);
+ backs = BACKREFS_P(br);
+ for (i = 0; i < br->back_num; i++) {
+ Node* ernode = mem_env[backs[i]].empty_repeat_node;
+ if (IS_NOT_NULL(ernode)) {
+ if (! is_ancestor_node(ernode, node)) {
+ MEM_STATUS_LIMIT_ON(env->reg->empty_status_mem, backs[i]);
+ NODE_STATUS_ADD(ernode, EMPTY_STATUS_CHECK);
+ NODE_STATUS_ADD(mem_env[backs[i]].mem_node, EMPTY_STATUS_CHECK);
+ }
+ }
+ }
+ }
+ break;
+
+ default:
+ break;
+ }
+}
+
+static void
+set_parent_node_trav(Node* node, Node* parent)
+{
+ NODE_PARENT(node) = parent;
+
+ switch (NODE_TYPE(node)) {
+ case NODE_LIST:
+ case NODE_ALT:
+ do {
+ set_parent_node_trav(NODE_CAR(node), node);
+ } while (IS_NOT_NULL(node = NODE_CDR(node)));
+ break;
+
+ case NODE_ANCHOR:
+ if (! ANCHOR_HAS_BODY(ANCHOR_(node))) break;
+ set_parent_node_trav(NODE_BODY(node), node);
+ break;
+
+ case NODE_QUANT:
+ set_parent_node_trav(NODE_BODY(node), node);
+ break;
+
+ case NODE_BAG:
+ if (IS_NOT_NULL(NODE_BODY(node)))
+ set_parent_node_trav(NODE_BODY(node), node);
+ {
+ BagNode* en = BAG_(node);
+
+ if (en->type == BAG_IF_ELSE) {
+ if (IS_NOT_NULL(en->te.Then))
+ set_parent_node_trav(en->te.Then, node);
+ if (IS_NOT_NULL(en->te.Else)) {
+ set_parent_node_trav(en->te.Else, node);
+ }
+ }
+ }
+ break;
+
+ default:
+ break;
+ }
+}
+
#ifdef USE_CALL
if ((eret & RECURSION_MUST) == 0)
r &= ~RECURSION_MUST;
}
+ else {
+ r &= ~RECURSION_MUST;
+ }
}
else {
r = infinite_recursive_call_check(NODE_BODY(node), env, head);
r = recursive_call_check_trav(NODE_BODY(node), env, state);
if (QUANT_(node)->upper == 0) {
if (r == FOUND_CALLED_NODE)
- QUANT_(node)->is_refered = 1;
+ QUANT_(node)->include_referred = 1;
}
break;
if (! NODE_IS_RECURSION(node)) {
NODE_STATUS_ADD(node, MARK1);
r = recursive_call_check(NODE_BODY(node));
- if (r != 0)
+ if (r != 0) {
NODE_STATUS_ADD(node, RECURSION);
+ MEM_STATUS_ON(env->backtrack_mem, en->m.regnum);
+ }
NODE_STATUS_REMOVE(node, MARK1);
}
- if (NODE_IS_CALLED(node))
- r = FOUND_CALLED_NODE;
- }
- }
+ if (NODE_IS_CALLED(node))
+ r = FOUND_CALLED_NODE;
+ }
+ }
+
+ state1 = state;
+ if (NODE_IS_RECURSION(node))
+ state1 |= IN_RECURSION;
+
+ ret = recursive_call_check_trav(NODE_BODY(node), env, state1);
+ if (ret == FOUND_CALLED_NODE)
+ r = FOUND_CALLED_NODE;
+
+ if (en->type == BAG_IF_ELSE) {
+ if (IS_NOT_NULL(en->te.Then)) {
+ ret = recursive_call_check_trav(en->te.Then, env, state1);
+ if (ret == FOUND_CALLED_NODE)
+ r = FOUND_CALLED_NODE;
+ }
+ if (IS_NOT_NULL(en->te.Else)) {
+ ret = recursive_call_check_trav(en->te.Else, env, state1);
+ if (ret == FOUND_CALLED_NODE)
+ r = FOUND_CALLED_NODE;
+ }
+ }
+ }
+ break;
+
+ default:
+ break;
+ }
+
+ return r;
+}
+
+#endif
+
+static void
+remove_from_list(Node* prev, Node* a)
+{
+ if (NODE_CDR(prev) != a) return ;
+
+ NODE_CDR(prev) = NODE_CDR(a);
+ NODE_CDR(a) = NULL_NODE;
+}
+
+static int
+reduce_string_list(Node* node)
+{
+ int r = 0;
+
+ switch (NODE_TYPE(node)) {
+ case NODE_LIST:
+ {
+ Node* prev;
+ Node* curr;
+ Node* prev_node;
+ Node* next_node;
+
+ prev = NULL_NODE;
+ do {
+ next_node = NODE_CDR(node);
+ curr = NODE_CAR(node);
+ if (NODE_TYPE(curr) == NODE_STRING) {
+ if (IS_NULL(prev) || STR_(curr)->flag != STR_(prev)->flag) {
+ prev = curr;
+ prev_node = node;
+ }
+ else {
+ r = node_str_node_cat(prev, curr);
+ if (r != 0) return r;
+ remove_from_list(prev_node, node);
+ onig_node_free(node);
+ }
+ }
+ else {
+ prev = NULL_NODE;
+ prev_node = node;
+ }
+
+ node = next_node;
+ } while (r == 0 && IS_NOT_NULL(node));
+ }
+ break;
+
+ case NODE_ALT:
+ do {
+ r = reduce_string_list(NODE_CAR(node));
+ } while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
+ break;
+
+ case NODE_ANCHOR:
+ if (IS_NULL(NODE_BODY(node)))
+ break;
+ /* fall */
+ case NODE_QUANT:
+ r = reduce_string_list(NODE_BODY(node));
+ break;
- state1 = state;
- if (NODE_IS_RECURSION(node))
- state1 |= IN_RECURSION;
+ case NODE_BAG:
+ {
+ BagNode* en = BAG_(node);
- ret = recursive_call_check_trav(NODE_BODY(node), env, state1);
- if (ret == FOUND_CALLED_NODE)
- r = FOUND_CALLED_NODE;
+ r = reduce_string_list(NODE_BODY(node));
+ if (r != 0) return r;
if (en->type == BAG_IF_ELSE) {
if (IS_NOT_NULL(en->te.Then)) {
- ret = recursive_call_check_trav(en->te.Then, env, state1);
- if (ret == FOUND_CALLED_NODE)
- r = FOUND_CALLED_NODE;
+ r = reduce_string_list(en->te.Then);
+ if (r != 0) return r;
}
if (IS_NOT_NULL(en->te.Else)) {
- ret = recursive_call_check_trav(en->te.Else, env, state1);
- if (ret == FOUND_CALLED_NODE)
- r = FOUND_CALLED_NODE;
+ r = reduce_string_list(en->te.Else);
+ if (r != 0) return r;
}
}
}
return r;
}
-#endif
#define IN_ALT (1<<0)
#define IN_NOT (1<<1)
head = NODE_ANCHOR_BODY(an);
np = NODE_CAR(head);
- swap_node(node, head);
+ node_swap(node, head);
NODE_CAR(node) = head;
NODE_BODY(head) = np;
}
static int
-setup_look_behind(Node* node, regex_t* reg, ScanEnv* env)
+tune_look_behind(Node* node, regex_t* reg, ScanEnv* env)
{
int r, len;
AnchorNode* an = ANCHOR_(node);
}
static int
-next_setup(Node* node, Node* next_node, regex_t* reg)
+tune_next(Node* node, Node* next_node, regex_t* reg)
{
NodeType type;
Node* en = onig_node_new_bag(BAG_STOP_BACKTRACK);
CHECK_NULL_RETURN_MEMERR(en);
NODE_STATUS_ADD(en, STRICT_REAL_REPEAT);
- swap_node(node, en);
+ node_swap(node, en);
NODE_BODY(node) = en;
}
}
static int
-update_string_node_case_fold(regex_t* reg, Node *node)
+is_all_code_len_1_items(int n, OnigCaseFoldCodeItem items[])
{
- UChar *p, *end, buf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
+ int i;
+
+ for (i = 0; i < n; i++) {
+ OnigCaseFoldCodeItem* item = items + i;
+ if (item->code_len != 1) return 0;
+ }
+
+ return 1;
+}
+
+static int
+get_min_max_byte_len_case_fold_items(int n, OnigCaseFoldCodeItem items[], int* rmin, int* rmax)
+{
+ int i, len, minlen, maxlen;
+
+ minlen = INT_MAX;
+ maxlen = 0;
+ for (i = 0; i < n; i++) {
+ OnigCaseFoldCodeItem* item = items + i;
+
+ len = item->byte_len;
+ if (len < minlen) minlen = len;
+ if (len > maxlen) maxlen = len;
+ }
+
+ *rmin = minlen;
+ *rmax = maxlen;
+ return 0;
+}
+
+static int
+conv_string_case_fold(OnigEncoding enc, OnigCaseFoldType case_fold_flag,
+ UChar* s, UChar* end, UChar** rs, UChar** rend, int* rcase_min_len)
+{
+ UChar *p, buf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
UChar *sbuf, *ebuf, *sp;
- int r, i, len, sbuf_size;
- StrNode* sn = STR_(node);
+ int i, n, len, sbuf_size;
- end = sn->end;
- sbuf_size = (int )(end - sn->s) * 2;
+ *rs = NULL;
+ sbuf_size = (int )(end - s) * 2;
sbuf = (UChar* )xmalloc(sbuf_size);
CHECK_NULL_RETURN_MEMERR(sbuf);
ebuf = sbuf + sbuf_size;
+ n = 0;
sp = sbuf;
- p = sn->s;
+ p = s;
while (p < end) {
- len = ONIGENC_MBC_CASE_FOLD(reg->enc, reg->case_fold_flag, &p, end, buf);
+ len = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, buf);
for (i = 0; i < len; i++) {
if (sp >= ebuf) {
sbuf = (UChar* )xrealloc(sbuf, sbuf_size * 2);
*sp++ = buf[i];
}
+ n++;
}
- r = onig_node_str_set(node, sbuf, sp);
- if (r != 0) {
- xfree(sbuf);
- return r;
- }
-
- xfree(sbuf);
+ *rs = sbuf;
+ *rend = sp;
+ *rcase_min_len = n;
return 0;
}
static int
-expand_case_fold_make_rem_string(Node** rnode, UChar *s, UChar *end, regex_t* reg)
+make_code_list_to_string(Node** rnode, OnigEncoding enc,
+ int n, OnigCodePoint codes[])
{
- int r;
- Node *node;
+ int r, i, len;
+ Node* node;
+ UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
- node = onig_node_new_str(s, end);
- if (IS_NULL(node)) return ONIGERR_MEMORY;
+ *rnode = NULL_NODE;
+ node = onig_node_new_str(NULL, NULL);
+ CHECK_NULL_RETURN_MEMERR(node);
- r = update_string_node_case_fold(reg, node);
- if (r != 0) {
- onig_node_free(node);
- return r;
+ for (i = 0; i < n; i++) {
+ len = ONIGENC_CODE_TO_MBC(enc, codes[i], buf);
+ if (len < 0) {
+ r = len;
+ goto err;
+ }
+
+ r = onig_node_str_cat(node, buf, buf + len);
+ if (r != 0) goto err;
}
- NODE_STRING_SET_AMBIG(node);
- NODE_STRING_SET_DONT_GET_OPT_INFO(node);
*rnode = node;
return 0;
+
+ err:
+ onig_node_free(node);
+ return r;
}
static int
-expand_case_fold_string_alt(int item_num, OnigCaseFoldCodeItem items[], UChar *p,
- int slen, UChar *end, regex_t* reg, Node **rnode)
+unravel_cf_node_add(Node** rlist, Node* add)
{
- int r, i, j;
- int len;
- int varlen;
- Node *anode, *var_anode, *snode, *xnode, *an;
- UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
-
- *rnode = var_anode = NULL_NODE;
+ Node *list;
- varlen = 0;
- for (i = 0; i < item_num; i++) {
- if (items[i].byte_len != slen) {
- varlen = 1;
- break;
- }
+ list = *rlist;
+ if (IS_NULL(list)) {
+ list = onig_node_new_list(add, NULL);
+ CHECK_NULL_RETURN_MEMERR(list);
+ *rlist = list;
}
+ else {
+ Node* r = node_list_add(list, add);
+ CHECK_NULL_RETURN_MEMERR(r);
+ }
+
+ return 0;
+}
- if (varlen != 0) {
- *rnode = var_anode = onig_node_new_alt(NULL_NODE, NULL_NODE);
- if (IS_NULL(var_anode)) return ONIGERR_MEMORY;
+static int
+unravel_cf_string_add(Node** rlist, Node** rsn, UChar* s, UChar* end,
+ unsigned int flag, int case_min_len)
+{
+ int r;
+ Node *sn, *list;
- xnode = onig_node_new_list(NULL, NULL);
- if (IS_NULL(xnode)) goto mem_err;
- NODE_CAR(var_anode) = xnode;
+ list = *rlist;
+ sn = *rsn;
- anode = onig_node_new_alt(NULL_NODE, NULL_NODE);
- if (IS_NULL(anode)) goto mem_err;
- NODE_CAR(xnode) = anode;
+ if (IS_NOT_NULL(sn) && STR_(sn)->flag == flag) {
+ if (NODE_STRING_IS_CASE_FOLD_MATCH(sn))
+ r = node_str_cat_case_fold(sn, s, end, case_min_len);
+ else
+ r = onig_node_str_cat(sn, s, end);
}
else {
- *rnode = anode = onig_node_new_alt(NULL_NODE, NULL_NODE);
- if (IS_NULL(anode)) return ONIGERR_MEMORY;
+ sn = onig_node_new_str(s, end);
+ CHECK_NULL_RETURN_MEMERR(sn);
+
+ STR_(sn)->flag = flag;
+ STR_(sn)->case_min_len = case_min_len;
+ r = unravel_cf_node_add(&list, sn);
}
- snode = onig_node_new_str(p, p + slen);
- if (IS_NULL(snode)) goto mem_err;
+ if (r == 0) {
+ *rlist = list;
+ *rsn = sn;
+ }
+ return r;
+}
- NODE_CAR(anode) = snode;
+static int
+unravel_cf_string_fold_add(Node** rlist, Node** rsn, OnigEncoding enc,
+ OnigCaseFoldType case_fold_flag, UChar* s, UChar* end)
+{
+ int r;
+ int case_min_len;
+ UChar *rs, *rend;
- for (i = 0; i < item_num; i++) {
- snode = onig_node_new_str(NULL, NULL);
- if (IS_NULL(snode)) goto mem_err;
+ r = conv_string_case_fold(enc, case_fold_flag, s, end,
+ &rs, &rend, &case_min_len);
+ if (r != 0) return r;
- for (j = 0; j < items[i].code_len; j++) {
- len = ONIGENC_CODE_TO_MBC(reg->enc, items[i].code[j], buf);
- if (len < 0) {
- r = len;
- goto mem_err2;
- }
+ r = unravel_cf_string_add(rlist, rsn, rs, rend,
+ NODE_STRING_CASE_FOLD_MATCH, case_min_len);
+ xfree(rs);
- r = onig_node_str_cat(snode, buf, buf + len);
- if (r != 0) goto mem_err2;
- }
+ return r;
+}
- an = onig_node_new_alt(NULL_NODE, NULL_NODE);
- if (IS_NULL(an)) {
- goto mem_err2;
- }
+static int
+unravel_cf_string_alt_or_cc_add(Node** rlist, int n,
+ OnigCaseFoldCodeItem items[], int byte_len, OnigEncoding enc,
+ OnigCaseFoldType case_fold_flag, UChar* s, UChar* end)
+{
+ int r, i;
+ Node* node;
- if (items[i].byte_len != slen && IS_NOT_NULL(var_anode)) {
- Node *rem;
- UChar *q = p + items[i].byte_len;
+ if (is_all_code_len_1_items(n, items)) {
+ OnigCodePoint codes[14];/* least ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM + 1 */
- if (q < end) {
- r = expand_case_fold_make_rem_string(&rem, q, end, reg);
- if (r != 0) {
- onig_node_free(an);
- goto mem_err2;
- }
+ codes[0] = ONIGENC_MBC_TO_CODE(enc, s, end);
+ for (i = 0; i < n; i++) {
+ OnigCaseFoldCodeItem* item = items + i;
+ codes[i+1] = item->code[0];
+ }
+ r = onig_new_cclass_with_code_list(&node, enc, n + 1, codes);
+ if (r != 0) return r;
+ }
+ else {
+ Node *snode, *alt, *curr;
- xnode = onig_node_list_add(NULL_NODE, snode);
- if (IS_NULL(xnode)) {
- onig_node_free(an);
- onig_node_free(rem);
- goto mem_err2;
- }
- if (IS_NULL(onig_node_list_add(xnode, rem))) {
- onig_node_free(an);
- onig_node_free(xnode);
- onig_node_free(rem);
- goto mem_err;
- }
+ snode = onig_node_new_str(s, end);
+ CHECK_NULL_RETURN_MEMERR(snode);
+ node = curr = onig_node_new_alt(snode, NULL_NODE);
+ if (IS_NULL(curr)) {
+ onig_node_free(snode);
+ return ONIGERR_MEMORY;
+ }
- NODE_CAR(an) = xnode;
+ r = 0;
+ for (i = 0; i < n; i++) {
+ OnigCaseFoldCodeItem* item = items + i;
+ r = make_code_list_to_string(&snode, enc, item->code_len, item->code);
+ if (r != 0) {
+ onig_node_free(node);
+ return r;
}
- else {
- NODE_CAR(an) = snode;
+
+ alt = onig_node_new_alt(snode, NULL_NODE);
+ if (IS_NULL(alt)) {
+ onig_node_free(snode);
+ onig_node_free(node);
+ return ONIGERR_MEMORY;
}
- NODE_CDR(var_anode) = an;
- var_anode = an;
- }
- else {
- NODE_CAR(an) = snode;
- NODE_CDR(anode) = an;
- anode = an;
+ NODE_CDR(curr) = alt;
+ curr = alt;
}
}
- return varlen;
-
- mem_err2:
- onig_node_free(snode);
-
- mem_err:
- onig_node_free(*rnode);
-
- return ONIGERR_MEMORY;
+ r = unravel_cf_node_add(rlist, node);
+ if (r != 0) onig_node_free(node);
+ return r;
}
static int
-is_good_case_fold_items_for_search(OnigEncoding enc, int slen,
- int n, OnigCaseFoldCodeItem items[])
+unravel_cf_look_behind_add(Node** rlist, Node** rsn,
+ int n, OnigCaseFoldCodeItem items[], OnigEncoding enc,
+ UChar* s, int one_len)
{
- int i, len;
- UChar buf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
+ int r, i, found;
+ found = 0;
for (i = 0; i < n; i++) {
OnigCaseFoldCodeItem* item = items + i;
+ if (item->byte_len == one_len) {
+ if (item->code_len == 1) {
+ found = 1;
+ }
+ }
+ }
- if (item->code_len != 1) return 0;
- if (item->byte_len != slen) return 0;
- len = ONIGENC_CODE_TO_MBC(enc, item->code[0], buf);
- if (len != slen) return 0;
+ if (found == 0) {
+ r = unravel_cf_string_add(rlist, rsn, s, s + one_len, 0 /* flag */, 0);
}
+ else {
+ Node* node;
+ OnigCodePoint codes[14];/* least ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM + 1 */
- return 1;
-}
+ found = 0;
+ codes[found++] = ONIGENC_MBC_TO_CODE(enc, s, s + one_len);
+ for (i = 0; i < n; i++) {
+ OnigCaseFoldCodeItem* item = items + i;
+ if (item->byte_len == one_len) {
+ if (item->code_len == 1) {
+ codes[found++] = item->code[0];
+ }
+ }
+ }
+ r = onig_new_cclass_with_code_list(&node, enc, found, codes);
+ if (r != 0) return r;
+
+ r = unravel_cf_node_add(rlist, node);
+ if (r != 0) onig_node_free(node);
+
+ *rsn = NULL_NODE;
+ }
-#define THRESHOLD_CASE_FOLD_ALT_FOR_EXPANSION 8
+ return r;
+}
static int
-expand_case_fold_string(Node* node, regex_t* reg, int state)
-{
- int r, n, len, alt_num;
- int fold_len;
- int prev_is_ambig, prev_is_good, is_good, is_in_look_behind;
- UChar *start, *end, *p;
- UChar* foldp;
- Node *top_root, *root, *snode, *prev_node;
+unravel_case_fold_string(Node* node, regex_t* reg, int state)
+{
+ int r, n, one_len, min_len, max_len, in_look_behind;
+ UChar *start, *end, *p, *q;
+ StrNode* snode;
+ Node *sn, *list;
+ OnigEncoding enc;
OnigCaseFoldCodeItem items[ONIGENC_GET_CASE_FOLD_CODES_MAX_NUM];
- UChar buf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
- StrNode* sn;
- if (NODE_STRING_IS_AMBIG(node)) return 0;
+ if (NODE_STRING_IS_CASE_EXPANDED(node)) return 0;
- sn = STR_(node);
+ snode = STR_(node);
- start = sn->s;
- end = sn->end;
+ start = snode->s;
+ end = snode->end;
if (start >= end) return 0;
- is_in_look_behind = (state & IN_LOOK_BEHIND) != 0;
+ in_look_behind = (state & IN_LOOK_BEHIND) != 0;
+ enc = reg->enc;
- r = 0;
- top_root = root = prev_node = snode = NULL_NODE;
- alt_num = 1;
+ list = sn = NULL_NODE;
p = start;
while (p < end) {
- n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(reg->enc, reg->case_fold_flag,
- p, end, items);
+ n = ONIGENC_GET_CASE_FOLD_CODES_BY_STR(enc, reg->case_fold_flag, p, end,
+ items);
if (n < 0) {
r = n;
goto err;
}
- SAFE_ENC_LEN(reg->enc, p, end, len);
- is_good = is_good_case_fold_items_for_search(reg->enc, len, n, items);
-
- if (is_in_look_behind ||
- (IS_NOT_NULL(snode) ||
- (is_good
- /* expand single char case: ex. /(?i:a)/ */
- && !(p == start && p + len >= end)))) {
- if (IS_NULL(snode)) {
- if (IS_NULL(root) && IS_NOT_NULL(prev_node)) {
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);
- if (IS_NULL(root)) {
- onig_node_free(prev_node);
- goto mem_err;
- }
- }
-
- prev_node = snode = onig_node_new_str(NULL, NULL);
- if (IS_NULL(snode)) goto mem_err;
- if (IS_NOT_NULL(root)) {
- if (IS_NULL(onig_node_list_add(root, snode))) {
- onig_node_free(snode);
- goto mem_err;
- }
- }
-
- prev_is_ambig = -1; /* -1: new */
- prev_is_good = 0; /* escape compiler warning */
- }
- else {
- prev_is_ambig = NODE_STRING_IS_AMBIG(snode);
- prev_is_good = NODE_STRING_IS_GOOD_AMBIG(snode);
- }
-
- if (n != 0) {
- foldp = p;
- fold_len = ONIGENC_MBC_CASE_FOLD(reg->enc, reg->case_fold_flag,
- &foldp, end, buf);
- foldp = buf;
- }
- else {
- foldp = p; fold_len = len;
- }
-
- if ((prev_is_ambig == 0 && n != 0) ||
- (prev_is_ambig > 0 && (n == 0 || prev_is_good != is_good))) {
- if (IS_NULL(root) /* && IS_NOT_NULL(prev_node) */) {
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);
- if (IS_NULL(root)) {
- onig_node_free(prev_node);
- goto mem_err;
- }
- }
-
- prev_node = snode = onig_node_new_str(foldp, foldp + fold_len);
- if (IS_NULL(snode)) goto mem_err;
- if (IS_NULL(onig_node_list_add(root, snode))) {
- onig_node_free(snode);
- goto mem_err;
- }
- }
- else {
- r = onig_node_str_cat(snode, foldp, foldp + fold_len);
- if (r != 0) goto err;
- }
-
- if (n != 0) NODE_STRING_SET_AMBIG(snode);
- if (is_good != 0) NODE_STRING_SET_GOOD_AMBIG(snode);
+ one_len = enclen(enc, p);
+ if (n == 0) {
+ q = p + one_len;
+ r = unravel_cf_string_add(&list, &sn, p, q, 0 /* flag */, 0);
+ if (r != 0) goto err;
}
else {
- alt_num *= (n + 1);
- if (alt_num > THRESHOLD_CASE_FOLD_ALT_FOR_EXPANSION) break;
-
- if (IS_NULL(root) && IS_NOT_NULL(prev_node)) {
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);
- if (IS_NULL(root)) {
- onig_node_free(prev_node);
- goto mem_err;
- }
+ if (in_look_behind != 0) {
+ q = p + one_len;
+ r = unravel_cf_look_behind_add(&list, &sn, n, items, enc, p, one_len);
+ if (r != 0) goto err;
}
-
- r = expand_case_fold_string_alt(n, items, p, len, end, reg, &prev_node);
- if (r < 0) goto mem_err;
- if (r == 1) {
- if (IS_NULL(root)) {
- top_root = prev_node;
+ else {
+ get_min_max_byte_len_case_fold_items(n, items, &min_len, &max_len);
+ q = p + max_len;
+ if (one_len == max_len && min_len == max_len) {
+ r = unravel_cf_string_alt_or_cc_add(&list, n, items, max_len, enc,
+ reg->case_fold_flag, p, q);
+ if (r != 0) goto err;
+ sn = NULL_NODE;
}
else {
- if (IS_NULL(onig_node_list_add(root, prev_node))) {
- onig_node_free(prev_node);
- goto mem_err;
- }
- }
-
- root = NODE_CAR(prev_node);
- }
- else { /* r == 0 */
- if (IS_NOT_NULL(root)) {
- if (IS_NULL(onig_node_list_add(root, prev_node))) {
- onig_node_free(prev_node);
- goto mem_err;
- }
+ r = unravel_cf_string_fold_add(&list, &sn, enc, reg->case_fold_flag,
+ p, q);
+ if (r != 0) goto err;
}
}
-
- snode = NULL_NODE;
}
- p += len;
+ p = q;
}
- if (p < end) {
- Node *srem;
-
- r = expand_case_fold_make_rem_string(&srem, p, end, reg);
- if (r != 0) goto mem_err;
-
- if (IS_NOT_NULL(prev_node) && IS_NULL(root)) {
- top_root = root = onig_node_list_add(NULL_NODE, prev_node);
- if (IS_NULL(root)) {
- onig_node_free(srem);
- onig_node_free(prev_node);
- goto mem_err;
- }
- }
-
- if (IS_NULL(root)) {
- prev_node = srem;
+ if (IS_NOT_NULL(list)) {
+ if (node_list_len(list) == 1) {
+ node_swap(node, NODE_CAR(list));
}
else {
- if (IS_NULL(onig_node_list_add(root, srem))) {
- onig_node_free(srem);
- goto mem_err;
- }
+ node_swap(node, list);
}
+ onig_node_free(list);
+ }
+ else {
+ node_swap(node, sn);
+ onig_node_free(sn);
}
-
- /* ending */
- top_root = (IS_NOT_NULL(top_root) ? top_root : prev_node);
- swap_node(node, top_root);
- onig_node_free(top_root);
return 0;
- mem_err:
- r = ONIGERR_MEMORY;
-
err:
- onig_node_free(top_root);
+ if (IS_NOT_NULL(list))
+ onig_node_free(list);
+ else if (IS_NOT_NULL(sn))
+ onig_node_free(sn);
+
return r;
}
__inline
#endif
static int
-setup_call_node_call(CallNode* cn, ScanEnv* env, int state)
+tune_call_node_call(CallNode* cn, ScanEnv* env, int state)
{
MemEnv* mem_env = SCANENV_MEMENV(env);
}
set_call_attr:
- NODE_CALL_BODY(cn) = mem_env[cn->group_num].node;
+ NODE_CALL_BODY(cn) = mem_env[cn->group_num].mem_node;
if (IS_NULL(NODE_CALL_BODY(cn))) {
onig_scan_env_set_error_string(env, ONIGERR_UNDEFINED_NAME_REFERENCE,
cn->name, cn->name_end);
}
static void
-setup_call2_call(Node* node)
+tune_call2_call(Node* node)
{
switch (NODE_TYPE(node)) {
case NODE_LIST:
case NODE_ALT:
do {
- setup_call2_call(NODE_CAR(node));
+ tune_call2_call(NODE_CAR(node));
} while (IS_NOT_NULL(node = NODE_CDR(node)));
break;
case NODE_QUANT:
- setup_call2_call(NODE_BODY(node));
+ tune_call2_call(NODE_BODY(node));
break;
case NODE_ANCHOR:
if (ANCHOR_HAS_BODY(ANCHOR_(node)))
- setup_call2_call(NODE_BODY(node));
+ tune_call2_call(NODE_BODY(node));
break;
case NODE_BAG:
if (en->type == BAG_MEMORY) {
if (! NODE_IS_MARK1(node)) {
NODE_STATUS_ADD(node, MARK1);
- setup_call2_call(NODE_BODY(node));
+ tune_call2_call(NODE_BODY(node));
NODE_STATUS_REMOVE(node, MARK1);
}
}
else if (en->type == BAG_IF_ELSE) {
- setup_call2_call(NODE_BODY(node));
+ tune_call2_call(NODE_BODY(node));
if (IS_NOT_NULL(en->te.Then))
- setup_call2_call(en->te.Then);
+ tune_call2_call(en->te.Then);
if (IS_NOT_NULL(en->te.Else))
- setup_call2_call(en->te.Else);
+ tune_call2_call(en->te.Else);
}
else {
- setup_call2_call(NODE_BODY(node));
+ tune_call2_call(NODE_BODY(node));
}
}
break;
NODE_STATUS_ADD(called, CALLED);
BAG_(called)->m.entry_count++;
- setup_call2_call(called);
+ tune_call2_call(called);
}
NODE_STATUS_REMOVE(node, MARK1);
}
}
static int
-setup_call(Node* node, ScanEnv* env, int state)
+tune_call(Node* node, ScanEnv* env, int state)
{
int r;
case NODE_LIST:
case NODE_ALT:
do {
- r = setup_call(NODE_CAR(node), env, state);
+ r = tune_call(NODE_CAR(node), env, state);
} while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
break;
if (QUANT_(node)->upper == 0)
state |= IN_ZERO_REPEAT;
- r = setup_call(NODE_BODY(node), env, state);
+ r = tune_call(NODE_BODY(node), env, state);
break;
case NODE_ANCHOR:
if (ANCHOR_HAS_BODY(ANCHOR_(node)))
- r = setup_call(NODE_BODY(node), env, state);
+ r = tune_call(NODE_BODY(node), env, state);
else
r = 0;
break;
NODE_STATUS_ADD(node, IN_ZERO_REPEAT);
BAG_(node)->m.entry_count--;
}
- r = setup_call(NODE_BODY(node), env, state);
+ r = tune_call(NODE_BODY(node), env, state);
}
else if (en->type == BAG_IF_ELSE) {
- r = setup_call(NODE_BODY(node), env, state);
+ r = tune_call(NODE_BODY(node), env, state);
if (r != 0) return r;
if (IS_NOT_NULL(en->te.Then)) {
- r = setup_call(en->te.Then, env, state);
+ r = tune_call(en->te.Then, env, state);
if (r != 0) return r;
}
if (IS_NOT_NULL(en->te.Else))
- r = setup_call(en->te.Else, env, state);
+ r = tune_call(en->te.Else, env, state);
}
else
- r = setup_call(NODE_BODY(node), env, state);
+ r = tune_call(NODE_BODY(node), env, state);
}
break;
CALL_(node)->entry_count--;
}
- r = setup_call_node_call(CALL_(node), env, state);
+ r = tune_call_node_call(CALL_(node), env, state);
break;
default:
}
static int
-setup_call2(Node* node)
+tune_call2(Node* node)
{
int r = 0;
case NODE_LIST:
case NODE_ALT:
do {
- r = setup_call2(NODE_CAR(node));
+ r = tune_call2(NODE_CAR(node));
} while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
break;
case NODE_QUANT:
if (QUANT_(node)->upper != 0)
- r = setup_call2(NODE_BODY(node));
+ r = tune_call2(NODE_BODY(node));
break;
case NODE_ANCHOR:
if (ANCHOR_HAS_BODY(ANCHOR_(node)))
- r = setup_call2(NODE_BODY(node));
+ r = tune_call2(NODE_BODY(node));
break;
case NODE_BAG:
if (! NODE_IS_IN_ZERO_REPEAT(node))
- r = setup_call2(NODE_BODY(node));
+ r = tune_call2(NODE_BODY(node));
{
BagNode* en = BAG_(node);
if (r != 0) return r;
if (en->type == BAG_IF_ELSE) {
if (IS_NOT_NULL(en->te.Then)) {
- r = setup_call2(en->te.Then);
+ r = tune_call2(en->te.Then);
if (r != 0) return r;
}
if (IS_NOT_NULL(en->te.Else))
- r = setup_call2(en->te.Else);
+ r = tune_call2(en->te.Else);
}
}
break;
case NODE_CALL:
if (! NODE_IS_IN_ZERO_REPEAT(node)) {
- setup_call2_call(node);
+ tune_call2_call(node);
}
break;
static void
-setup_called_state_call(Node* node, int state)
+tune_called_state_call(Node* node, int state)
{
switch (NODE_TYPE(node)) {
case NODE_ALT:
/* fall */
case NODE_LIST:
do {
- setup_called_state_call(NODE_CAR(node), state);
+ tune_called_state_call(NODE_CAR(node), state);
} while (IS_NOT_NULL(node = NODE_CDR(node)));
break;
if (qn->lower != qn->upper)
state |= IN_VAR_REPEAT;
- setup_called_state_call(NODE_QUANT_BODY(qn), state);
+ tune_called_state_call(NODE_QUANT_BODY(qn), state);
}
break;
/* fall */
case ANCR_PREC_READ:
case ANCR_LOOK_BEHIND:
- setup_called_state_call(NODE_ANCHOR_BODY(an), state);
+ tune_called_state_call(NODE_ANCHOR_BODY(an), state);
break;
default:
break;
if (NODE_IS_MARK1(node)) {
if ((~en->m.called_state & state) != 0) {
en->m.called_state |= state;
- setup_called_state_call(NODE_BODY(node), state);
+ tune_called_state_call(NODE_BODY(node), state);
}
}
else {
NODE_STATUS_ADD(node, MARK1);
en->m.called_state |= state;
- setup_called_state_call(NODE_BODY(node), state);
+ tune_called_state_call(NODE_BODY(node), state);
NODE_STATUS_REMOVE(node, MARK1);
}
}
else if (en->type == BAG_IF_ELSE) {
+ state |= IN_ALT;
+ tune_called_state_call(NODE_BODY(node), state);
if (IS_NOT_NULL(en->te.Then)) {
- setup_called_state_call(en->te.Then, state);
+ tune_called_state_call(en->te.Then, state);
}
if (IS_NOT_NULL(en->te.Else))
- setup_called_state_call(en->te.Else, state);
+ tune_called_state_call(en->te.Else, state);
}
else {
- setup_called_state_call(NODE_BODY(node), state);
+ tune_called_state_call(NODE_BODY(node), state);
}
}
break;
case NODE_CALL:
- setup_called_state_call(NODE_BODY(node), state);
+ tune_called_state_call(NODE_BODY(node), state);
break;
default:
}
static void
-setup_called_state(Node* node, int state)
+tune_called_state(Node* node, int state)
{
switch (NODE_TYPE(node)) {
case NODE_ALT:
/* fall */
case NODE_LIST:
do {
- setup_called_state(NODE_CAR(node), state);
+ tune_called_state(NODE_CAR(node), state);
} while (IS_NOT_NULL(node = NODE_CDR(node)));
break;
#ifdef USE_CALL
case NODE_CALL:
- setup_called_state_call(node, state);
+ tune_called_state_call(node, state);
break;
#endif
/* fall */
case BAG_OPTION:
case BAG_STOP_BACKTRACK:
- setup_called_state(NODE_BODY(node), state);
+ tune_called_state(NODE_BODY(node), state);
break;
case BAG_IF_ELSE:
- setup_called_state(NODE_BODY(node), state);
+ state |= IN_ALT;
+ tune_called_state(NODE_BODY(node), state);
if (IS_NOT_NULL(en->te.Then))
- setup_called_state(en->te.Then, state);
+ tune_called_state(en->te.Then, state);
if (IS_NOT_NULL(en->te.Else))
- setup_called_state(en->te.Else, state);
+ tune_called_state(en->te.Else, state);
break;
}
}
if (qn->lower != qn->upper)
state |= IN_VAR_REPEAT;
- setup_called_state(NODE_QUANT_BODY(qn), state);
+ tune_called_state(NODE_QUANT_BODY(qn), state);
}
break;
/* fall */
case ANCR_PREC_READ:
case ANCR_LOOK_BEHIND:
- setup_called_state(NODE_ANCHOR_BODY(an), state);
+ tune_called_state(NODE_ANCHOR_BODY(an), state);
break;
default:
break;
#endif /* USE_CALL */
-static int setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env);
+static int tune_tree(Node* node, regex_t* reg, int state, ScanEnv* env);
#ifdef __GNUC__
__inline
#endif
static int
-setup_anchor(Node* node, regex_t* reg, int state, ScanEnv* env)
+tune_anchor(Node* node, regex_t* reg, int state, ScanEnv* env)
{
/* allowed node types in look-behind */
#define ALLOWED_TYPE_IN_LB \
switch (an->type) {
case ANCR_PREC_READ:
- r = setup_tree(NODE_ANCHOR_BODY(an), reg, state, env);
+ r = tune_tree(NODE_ANCHOR_BODY(an), reg, state, env);
break;
case ANCR_PREC_READ_NOT:
- r = setup_tree(NODE_ANCHOR_BODY(an), reg, (state | IN_NOT), env);
+ r = tune_tree(NODE_ANCHOR_BODY(an), reg, (state | IN_NOT), env);
break;
case ANCR_LOOK_BEHIND:
ALLOWED_BAG_IN_LB, ALLOWED_ANCHOR_IN_LB);
if (r < 0) return r;
if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;
- r = setup_tree(NODE_ANCHOR_BODY(an), reg, (state|IN_LOOK_BEHIND), env);
+ r = tune_tree(NODE_ANCHOR_BODY(an), reg, (state|IN_LOOK_BEHIND), env);
if (r != 0) return r;
- r = setup_look_behind(node, reg, env);
+ r = tune_look_behind(node, reg, env);
}
break;
ALLOWED_BAG_IN_LB_NOT, ALLOWED_ANCHOR_IN_LB_NOT);
if (r < 0) return r;
if (r > 0) return ONIGERR_INVALID_LOOK_BEHIND_PATTERN;
- r = setup_tree(NODE_ANCHOR_BODY(an), reg, (state|IN_NOT|IN_LOOK_BEHIND),
- env);
+ r = tune_tree(NODE_ANCHOR_BODY(an), reg, (state|IN_NOT|IN_LOOK_BEHIND),
+ env);
if (r != 0) return r;
- r = setup_look_behind(node, reg, env);
+ r = tune_look_behind(node, reg, env);
}
break;
__inline
#endif
static int
-setup_quant(Node* node, regex_t* reg, int state, ScanEnv* env)
+tune_quant(Node* node, regex_t* reg, int state, ScanEnv* env)
{
int r;
OnigLen d;
if (d == 0) {
#ifdef USE_STUBBORN_CHECK_CAPTURES_IN_EMPTY_REPEAT
qn->emptiness = quantifiers_memory_node_info(body);
- if (qn->emptiness == BODY_IS_EMPTY_POSSIBILITY_REC) {
- if (NODE_TYPE(body) == NODE_BAG &&
- BAG_(body)->type == BAG_MEMORY) {
- MEM_STATUS_ON(env->bt_mem_end, BAG_(body)->m.regnum);
- }
- }
#else
qn->emptiness = BODY_IS_EMPTY_POSSIBILITY;
#endif
if (qn->lower != qn->upper)
state |= IN_VAR_REPEAT;
- r = setup_tree(body, reg, state, env);
+ r = tune_tree(body, reg, state, env);
if (r != 0) return r;
/* expand string */
if (!IS_INFINITE_REPEAT(qn->lower) && qn->lower == qn->upper &&
qn->lower > 1 && qn->lower <= EXPAND_STRING_MAX_LENGTH) {
int len = NODE_STRING_LEN(body);
- StrNode* sn = STR_(body);
if (len * qn->lower <= EXPAND_STRING_MAX_LENGTH) {
int i, n = qn->lower;
- onig_node_conv_to_str_node(node, STR_(body)->flag);
+ node_conv_to_str_node(node, STR_(body)->flag);
for (i = 0; i < n; i++) {
- r = onig_node_str_cat(node, sn->s, sn->end);
+ r = node_str_node_cat(node, body);
if (r != 0) return r;
}
onig_node_free(body);
return r;
}
-/* setup_tree does the following work.
+/* tune_tree does the following work.
1. check empty loop. (set qn->emptiness)
2. expand ignore-case in char class.
3. set memory status bit flags. (reg->mem_stats)
6. expand repeated string.
*/
static int
-setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env)
+tune_tree(Node* node, regex_t* reg, int state, ScanEnv* env)
{
int r = 0;
{
Node* prev = NULL_NODE;
do {
- r = setup_tree(NODE_CAR(node), reg, state, env);
+ r = tune_tree(NODE_CAR(node), reg, state, env);
if (IS_NOT_NULL(prev) && r == 0) {
- r = next_setup(prev, NODE_CAR(node), reg);
+ r = tune_next(prev, NODE_CAR(node), reg);
}
prev = NODE_CAR(node);
} while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
case NODE_ALT:
do {
- r = setup_tree(NODE_CAR(node), reg, (state | IN_ALT), env);
+ r = tune_tree(NODE_CAR(node), reg, (state | IN_ALT), env);
} while (r == 0 && IS_NOT_NULL(node = NODE_CDR(node)));
break;
case NODE_STRING:
- if (IS_IGNORECASE(reg->options) && !NODE_STRING_IS_RAW(node)) {
- r = expand_case_fold_string(node, reg, state);
+ if (IS_IGNORECASE(reg->options) && !NODE_STRING_IS_CRUDE(node)) {
+ r = unravel_case_fold_string(node, reg, state);
}
break;
for (i = 0; i < br->back_num; i++) {
if (p[i] > env->num_mem) return ONIGERR_INVALID_BACKREF;
MEM_STATUS_ON(env->backrefed_mem, p[i]);
- MEM_STATUS_ON(env->bt_mem_start, p[i]);
+#if 0
#ifdef USE_BACKREF_WITH_LEVEL
if (NODE_IS_NEST_LEVEL(node)) {
- MEM_STATUS_ON(env->bt_mem_end, p[i]);
+ MEM_STATUS_ON(env->backtrack_mem, p[i]);
}
+#endif
+#else
+ /* More precisely, it should be checked whether alt/repeat exists before
+ the subject capture node, and then this backreference position
+ exists before (or in) the capture node. */
+ MEM_STATUS_ON(env->backtrack_mem, p[i]);
#endif
}
}
{
OnigOptionType options = reg->options;
reg->options = BAG_(node)->o.options;
- r = setup_tree(NODE_BODY(node), reg, state, env);
+ r = tune_tree(NODE_BODY(node), reg, state, env);
reg->options = options;
}
break;
if ((state & (IN_ALT | IN_NOT | IN_VAR_REPEAT | IN_MULTI_ENTRY)) != 0
|| NODE_IS_RECURSION(node)) {
- MEM_STATUS_ON(env->bt_mem_start, en->m.regnum);
+ MEM_STATUS_ON(env->backtrack_mem, en->m.regnum);
}
- r = setup_tree(NODE_BODY(node), reg, state, env);
+ r = tune_tree(NODE_BODY(node), reg, state, env);
break;
case BAG_STOP_BACKTRACK:
{
Node* target = NODE_BODY(node);
- r = setup_tree(target, reg, state, env);
+ r = tune_tree(target, reg, state, env);
if (NODE_TYPE(target) == NODE_QUANT) {
QuantNode* tqn = QUANT_(target);
if (IS_INFINITE_REPEAT(tqn->upper) && tqn->lower <= 1 &&
break;
case BAG_IF_ELSE:
- r = setup_tree(NODE_BODY(node), reg, (state | IN_ALT), env);
+ r = tune_tree(NODE_BODY(node), reg, (state | IN_ALT), env);
if (r != 0) return r;
if (IS_NOT_NULL(en->te.Then)) {
- r = setup_tree(en->te.Then, reg, (state | IN_ALT), env);
+ r = tune_tree(en->te.Then, reg, (state | IN_ALT), env);
if (r != 0) return r;
}
if (IS_NOT_NULL(en->te.Else))
- r = setup_tree(en->te.Else, reg, (state | IN_ALT), env);
+ r = tune_tree(en->te.Else, reg, (state | IN_ALT), env);
break;
}
}
break;
case NODE_QUANT:
- r = setup_quant(node, reg, state, env);
+ r = tune_quant(node, reg, state, env);
break;
case NODE_ANCHOR:
- r = setup_anchor(node, reg, state, env);
+ r = tune_anchor(node, reg, state, env);
break;
#ifdef USE_CALL
} MinMax;
typedef struct {
- MinMax mmd;
+ MinMax mm;
OnigEncoding enc;
OnigOptionType options;
OnigCaseFoldType case_fold_flag;
} OptAnc;
typedef struct {
- MinMax mmd; /* position */
+ MinMax mm; /* position */
OptAnc anc;
int reach_end;
int case_fold;
- int good_case_fold;
int len;
UChar s[OPT_EXACT_MAXLEN];
} OptStr;
typedef struct {
- MinMax mmd; /* position */
+ MinMax mm; /* position */
OptAnc anc;
int value; /* weighted value */
UChar map[CHAR_MAP_SIZE];
static void
clear_opt_exact(OptStr* e)
{
- clear_mml(&e->mmd);
+ clear_mml(&e->mm);
clear_opt_anc_info(&e->anc);
e->reach_end = 0;
e->case_fold = 0;
- e->good_case_fold = 0;
e->len = 0;
e->s[0] = '\0';
}
to->case_fold = 1;
}
- else {
- if (to->good_case_fold != 0) {
- if (add->good_case_fold == 0) return 0;
- }
- }
}
r = 0;
p = add->s;
end = p + add->len;
for (i = to->len; p < end; ) {
- SAFE_ENC_LEN(enc, p, end, len);
+ len = enclen(enc, p);
if (i + len > OPT_EXACT_MAXLEN) {
r = 1; /* 1:full */
break;
UChar *p;
for (i = to->len, p = s; p < end && i < OPT_EXACT_MAXLEN; ) {
- SAFE_ENC_LEN(enc, p, end, len);
+ len = enclen(enc, p);
if (i + len > OPT_EXACT_MAXLEN) break;
for (j = 0; j < len && p < end; j++)
to->s[i++] = *p++;
return ;
}
- if (! is_equal_mml(&to->mmd, &add->mmd)) {
+ if (! is_equal_mml(&to->mm, &add->mm)) {
clear_opt_exact(to);
return ;
}
to->len = i;
if (add->case_fold != 0)
to->case_fold = 1;
- if (add->good_case_fold == 0)
- to->good_case_fold = 0;
alt_merge_opt_anc_info(&to->anc, &add->anc);
if (! to->reach_end) to->anc.right = 0;
if (now->case_fold == 0) vn *= 2;
if (alt->case_fold == 0) va *= 2;
- if (now->good_case_fold != 0) vn *= 4;
- if (alt->good_case_fold != 0) va *= 4;
-
- if (comp_distance_value(&now->mmd, &alt->mmd, vn, va) > 0)
+ if (comp_distance_value(&now->mm, &alt->mm, vn, va) > 0)
copy_opt_exact(now, alt);
}
vn = z / now->value;
va = z / alt->value;
- if (comp_distance_value(&now->mmd, &alt->mmd, vn, va) > 0)
+ if (comp_distance_value(&now->mm, &alt->mm, vn, va) > 0)
copy_opt_map(now, alt);
}
if (m->value <= 0) return -1;
if (e->case_fold != 0) {
- if (e->good_case_fold != 0)
- case_value = 2;
- else
- case_value = 1;
+ case_value = 1;
}
else
case_value = 3;
ae = COMP_EM_BASE * e->len * case_value;
am = COMP_EM_BASE * 5 * 2 / m->value;
- return comp_distance_value(&e->mmd, &m->mmd, ae, am);
+ return comp_distance_value(&e->mm, &m->mm, ae, am);
}
static void
{
int i, val;
- /* if (! is_equal_mml(&to->mmd, &add->mmd)) return ; */
+ /* if (! is_equal_mml(&to->mm, &add->mm)) return ; */
if (to->value == 0) return ;
- if (add->value == 0 || to->mmd.max < add->mmd.min) {
+ if (add->value == 0 || to->mm.max < add->mm.min) {
clear_opt_map(to);
return ;
}
- alt_merge_mml(&to->mmd, &add->mmd);
+ alt_merge_mml(&to->mm, &add->mm);
val = 0;
for (i = 0; i < CHAR_MAP_SIZE; i++) {
static void
set_bound_node_opt_info(OptNode* opt, MinMax* plen)
{
- copy_mml(&(opt->sb.mmd), plen);
- copy_mml(&(opt->spr.mmd), plen);
- copy_mml(&(opt->map.mmd), plen);
+ copy_mml(&(opt->sb.mm), plen);
+ copy_mml(&(opt->spr.mm), plen);
+ copy_mml(&(opt->map.mm), plen);
}
static void
}
if (add->map.value > 0 && to->len.max == 0) {
- if (add->map.mmd.max == 0)
+ if (add->map.mm.max == 0)
add->map.anc.left |= to->anc.left;
}
if (to->spr.len > 0) {
if (add->len.max > 0) {
- if (to->spr.len > (int )add->len.max)
- to->spr.len = add->len.max;
-
- if (to->spr.mmd.max == 0)
+ if (to->spr.mm.max == 0)
select_opt_exact(enc, &to->sb, &to->spr);
else
select_opt_exact(enc, &to->sm, &to->spr);
r = 0;
enc = env->enc;
clear_node_opt_info(opt);
- set_bound_node_opt_info(opt, &env->mmd);
+ set_bound_node_opt_info(opt, &env->mm);
switch (NODE_TYPE(node)) {
case NODE_LIST:
do {
r = optimize_nodes(NODE_CAR(nd), &xo, &nenv);
if (r == 0) {
- add_mml(&nenv.mmd, &xo.len);
+ add_mml(&nenv.mm, &xo.len);
concat_left_node_opt_info(enc, opt, &xo);
}
} while (r == 0 && IS_NOT_NULL(nd = NODE_CDR(nd)));
{
StrNode* sn = STR_(node);
int slen = (int )(sn->end - sn->s);
- /* int is_raw = NODE_STRING_IS_RAW(node); */
- if (! NODE_STRING_IS_AMBIG(node)) {
+ if (! NODE_STRING_IS_CASE_FOLD_MATCH(node)) {
concat_opt_exact_str(&opt->sb, sn->s, sn->end, enc);
if (slen > 0) {
add_char_opt_map(&opt->map, *(sn->s), enc);
set_mml(&opt->len, slen, slen);
}
else {
- int max;
+ int max, min;
- if (NODE_STRING_IS_DONT_GET_OPT_INFO(node)) {
- int n = onigenc_strlen(enc, sn->s, sn->end);
- max = ONIGENC_MBC_MAXLEN_DIST(enc) * n;
- }
- else {
- concat_opt_exact_str(&opt->sb, sn->s, sn->end, enc);
- opt->sb.case_fold = 1;
- if (NODE_STRING_IS_GOOD_AMBIG(node))
- opt->sb.good_case_fold = 1;
-
- if (slen > 0) {
- r = add_char_amb_opt_map(&opt->map, sn->s, sn->end,
- enc, env->case_fold_flag);
- if (r != 0) break;
- }
+ concat_opt_exact_str(&opt->sb, sn->s, sn->end, enc);
+ opt->sb.case_fold = 1;
- max = slen;
+ if (slen > 0) {
+ r = add_char_amb_opt_map(&opt->map, sn->s, sn->end,
+ enc, env->case_fold_flag);
+ if (r != 0) break;
}
- set_mml(&opt->len, slen, max);
+ max = slen;
+ min = sn->case_min_len * ONIGENC_MBC_MINLEN(enc);
+ set_mml(&opt->len, min, max);
}
}
break;
int z;
CClassNode* cc = CCLASS_(node);
- /* no need to check ignore case. (set in setup_tree()) */
+ /* no need to check ignore case. (set in tune_tree()) */
if (IS_NOT_NULL(cc->mbuf) || IS_NCCLASS_NOT(cc)) {
OnigLen min = ONIGENC_MBC_MINLEN(enc);
break;
}
backs = BACKREFS_P(br);
- min = tree_min_len(mem_env[backs[0]].node, env->scan_env);
- max = tree_max_len(mem_env[backs[0]].node, env->scan_env);
+ min = tree_min_len(mem_env[backs[0]].mem_node, env->scan_env);
+ max = tree_max_len(mem_env[backs[0]].mem_node, env->scan_env);
for (i = 1; i < br->back_num; i++) {
- tmin = tree_min_len(mem_env[backs[i]].node, env->scan_env);
- tmax = tree_max_len(mem_env[backs[i]].node, env->scan_env);
+ tmin = tree_min_len(mem_env[backs[i]].mem_node, env->scan_env);
+ tmax = tree_max_len(mem_env[backs[i]].mem_node, env->scan_env);
if (min > tmin) min = tmin;
if (max < tmax) max = tmax;
}
}
if (IS_INFINITE_REPEAT(qn->upper)) {
- if (env->mmd.max == 0 &&
+ if (env->mm.max == 0 &&
NODE_IS_ANYCHAR(NODE_BODY(node)) && qn->greedy != 0) {
if (IS_MULTILINE(CTYPE_OPTION(NODE_QUANT_BODY(qn), env)))
add_opt_anc_info(&opt->anc, ANCR_ANYCHAR_INF_ML);
copy_opt_env(&nenv, env);
r = optimize_nodes(NODE_BAG_BODY(en), &xo, &nenv);
if (r == 0) {
- add_mml(&nenv.mmd, &xo.len);
+ add_mml(&nenv.mm, &xo.len);
concat_left_node_opt_info(enc, opt, &xo);
if (IS_NOT_NULL(en->te.Then)) {
r = optimize_nodes(en->te.Then, &xo, &nenv);
if (e->case_fold) {
reg->optimize = OPTIMIZE_STR_CASE_FOLD;
- if (e->good_case_fold != 0) {
- if (e->len >= 2) {
- r = set_sunday_quick_search_or_bmh_skip_table(reg, 1,
- reg->exact, reg->exact_end,
- reg->map, &(reg->map_offset));
- if (r != 0) return r;
- reg->optimize = OPTIMIZE_STR_CASE_FOLD_FAST;
- }
- }
}
else {
int allow_reverse;
}
}
- reg->dmin = e->mmd.min;
- reg->dmax = e->mmd.max;
+ reg->dist_min = e->mm.min;
+ reg->dist_max = e->mm.max;
+
+ if (reg->dist_min != INFINITE_LEN) {
+ int n;
+ if (e->case_fold != 0)
+ n = 1;
+ else
+ n = (int )(reg->exact_end - reg->exact);
- if (reg->dmin != INFINITE_LEN) {
- reg->threshold_len = reg->dmin + (int )(reg->exact_end - reg->exact);
+ reg->threshold_len = reg->dist_min + n;
}
return 0;
reg->map[i] = m->map[i];
reg->optimize = OPTIMIZE_MAP;
- reg->dmin = m->mmd.min;
- reg->dmax = m->mmd.max;
+ reg->dist_min = m->mm.min;
+ reg->dist_max = m->mm.max;
- if (reg->dmin != INFINITE_LEN) {
- reg->threshold_len = reg->dmin + 1;
+ if (reg->dist_min != INFINITE_LEN) {
+ reg->threshold_len = reg->dist_min + 1;
}
}
env.options = reg->options;
env.case_fold_flag = reg->case_fold_flag;
env.scan_env = scan_env;
- clear_mml(&env.mmd);
+ clear_mml(&env.mm);
r = optimize_nodes(node, &opt, &env);
if (r != 0) return r;
ANCR_PREC_READ_NOT);
if (reg->anchor & (ANCR_END_BUF | ANCR_SEMI_END_BUF)) {
- reg->anchor_dmin = opt.len.min;
- reg->anchor_dmax = opt.len.max;
+ reg->anc_dist_min = opt.len.min;
+ reg->anc_dist_max = opt.len.max;
}
if (opt.sb.len > 0 || opt.sm.len > 0) {
{
reg->optimize = OPTIMIZE_NONE;
reg->anchor = 0;
- reg->anchor_dmin = 0;
- reg->anchor_dmax = 0;
+ reg->anc_dist_min = 0;
+ reg->anc_dist_max = 0;
reg->sub_anchor = 0;
reg->exact_end = (UChar* )NULL;
reg->map_offset = 0;
{
static const char* on[] = { "NONE", "STR",
"STR_FAST", "STR_FAST_STEP_FORWARD",
- "STR_CASE_FOLD_FAST", "STR_CASE_FOLD", "MAP" };
+ "STR_CASE_FOLD", "MAP" };
fprintf(f, "optimize: %s\n", on[reg->optimize]);
fprintf(f, " anchor: "); print_anchor(f, reg->anchor);
if ((reg->anchor & ANCR_END_BUF_MASK) != 0)
- print_distance_range(f, reg->anchor_dmin, reg->anchor_dmax);
+ print_distance_range(f, reg->anc_dist_min, reg->anc_dist_max);
fprintf(f, "\n");
if (reg->optimize) {
Node* root;
ScanEnv scan_env;
#ifdef USE_CALL
- UnsetAddrList uslist;
+ UnsetAddrList uslist = {0};
#endif
root = 0;
reg->string_pool_end = 0;
reg->num_mem = 0;
reg->num_repeat = 0;
- reg->num_null_check = 0;
+ reg->num_empty_check = 0;
reg->repeat_range_alloc = 0;
- reg->repeat_range = (OnigRepeatRange* )NULL;
+ reg->repeat_range = (RepeatRange* )NULL;
+ reg->empty_status_mem = 0;
r = onig_parse_tree(&root, pattern, pattern_end, reg, &scan_env);
if (r != 0) goto err;
+ r = reduce_string_list(root);
+ if (r != 0) goto err;
+
/* mixed use named group and no-named group */
if (scan_env.num_named > 0 &&
IS_SYNTAX_BV(scan_env.syntax, ONIG_SYN_CAPTURE_ONLY_NAMED_GROUP) &&
r = unset_addr_list_init(&uslist, scan_env.num_call);
if (r != 0) goto err;
scan_env.unset_addr_list = &uslist;
- r = setup_call(root, &scan_env, 0);
+ r = tune_call(root, &scan_env, 0);
if (r != 0) goto err_unset;
- r = setup_call2(root);
+ r = tune_call2(root);
if (r != 0) goto err_unset;
r = recursive_call_check_trav(root, &scan_env, 0);
if (r < 0) goto err_unset;
r = infinite_recursive_call_check_trav(root, &scan_env);
if (r != 0) goto err_unset;
- setup_called_state(root, 0);
+ tune_called_state(root, 0);
}
reg->num_call = scan_env.num_call;
#endif
- r = setup_tree(root, reg, 0, &scan_env);
+#ifdef ONIG_DEBUG_PARSE
+ fprintf(stderr, "MAX PARSE DEPTH: %d\n", scan_env.max_parse_depth);
+ fprintf(stderr, "TREE (parsed)\n");
+ print_tree(stderr, root);
+ fprintf(stderr, "\n");
+#endif
+
+ r = tune_tree(root, reg, 0, &scan_env);
if (r != 0) goto err_unset;
+ if (scan_env.backref_num != 0) {
+ set_parent_node_trav(root, NULL_NODE);
+ r = set_empty_repeat_node_trav(root, NULL_NODE, &scan_env);
+ if (r != 0) goto err_unset;
+ set_empty_status_check_trav(root, &scan_env);
+ }
+
#ifdef ONIG_DEBUG_PARSE
+ fprintf(stderr, "TREE (after tune)\n");
print_tree(stderr, root);
+ fprintf(stderr, "\n");
#endif
- reg->capture_history = scan_env.capture_history;
- reg->bt_mem_start = scan_env.bt_mem_start;
- reg->bt_mem_start |= reg->capture_history;
- if (IS_FIND_CONDITION(reg->options))
- MEM_STATUS_ON_ALL(reg->bt_mem_end);
+ reg->capture_history = scan_env.cap_history;
+ reg->push_mem_start = scan_env.backtrack_mem | scan_env.cap_history;
+
+#ifdef USE_CALLOUT
+ if (IS_NOT_NULL(reg->extp) && reg->extp->callout_num != 0) {
+ reg->push_mem_end = reg->push_mem_start;
+ }
else {
- reg->bt_mem_end = scan_env.bt_mem_end;
- reg->bt_mem_end |= reg->capture_history;
+ if (MEM_STATUS_IS_ALL_ON(reg->push_mem_start))
+ reg->push_mem_end = scan_env.backrefed_mem | scan_env.cap_history;
+ else
+ reg->push_mem_end = reg->push_mem_start &
+ (scan_env.backrefed_mem | scan_env.cap_history);
}
- reg->bt_mem_start |= reg->bt_mem_end;
+#else
+ if (MEM_STATUS_IS_ALL_ON(reg->push_mem_start))
+ reg->push_mem_end = scan_env.backrefed_mem | scan_env.cap_history;
+ else
+ reg->push_mem_end = reg->push_mem_start &
+ (scan_env.backrefed_mem | scan_env.cap_history);
+#endif
clear_optimize_info(reg);
#ifndef ONIG_DONT_OPTIMIZE
}
#endif
- if ((reg->num_repeat != 0) || (reg->bt_mem_end != 0)
+ set_addr_in_repeat_range(reg);
+
+ if ((reg->push_mem_end != 0)
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+ || (reg->num_repeat != 0)
+ || (reg->num_empty_check != 0)
+#endif
#ifdef USE_CALLOUT
|| (IS_NOT_NULL(reg->extp) && reg->extp->callout_num != 0)
#endif
)
reg->stack_pop_level = STACK_POP_LEVEL_ALL;
else {
- if (reg->bt_mem_start != 0)
+ if (reg->push_mem_start != 0)
reg->stack_pop_level = STACK_POP_LEVEL_MEM_START;
else
reg->stack_pop_level = STACK_POP_LEVEL_FREE;
if (IS_NULL(*reg)) return ONIGERR_MEMORY;
r = onig_reg_init(*reg, option, ONIGENC_CASE_FOLD_DEFAULT, enc, syntax);
- if (r != 0) goto err;
+ if (r != 0) {
+ xfree(*reg);
+ *reg = NULL;
+ return r;
+ }
r = onig_compile(*reg, pattern, pattern_end, einfo);
if (r != 0) {
- err:
onig_free(*reg);
*reg = NULL;
}
#ifdef ONIG_DEBUG_PARSE
+#ifdef USE_CALL
static void
p_string(FILE* f, int len, UChar* s)
{
fputs(":", f);
while (len-- > 0) { fputc(*s++, f); }
}
+#endif
static void
Indent(FILE* f, int indent)
Indent(f, indent);
if (IS_NULL(node)) {
fprintf(f, "ERROR: null node!!!\n");
- exit (0);
+ exit(0);
}
type = NODE_TYPE(node);
case NODE_STRING:
{
+ char* str;
char* mode;
- char* dont;
- char* good;
- if (NODE_STRING_IS_RAW(node))
- mode = "-raw";
- else if (NODE_STRING_IS_AMBIG(node))
- mode = "-ambig";
+ if (NODE_STRING_IS_CRUDE(node))
+ mode = "-crude";
+ else if (NODE_STRING_IS_CASE_FOLD_MATCH(node))
+ mode = "-case_fold_match";
else
mode = "";
- if (NODE_STRING_IS_GOOD_AMBIG(node))
- good = "-good";
- else
- good = "";
-
- if (NODE_STRING_IS_DONT_GET_OPT_INFO(node))
- dont = " (dont-opt)";
+ if (STR_(node)->s == STR_(node)->end)
+ str = "empty-string";
else
- dont = "";
+ str = "string";
- fprintf(f, "<string%s%s%s:%p>", mode, good, dont, node);
+ fprintf(f, "<%s%s:%p>", str, mode, node);
for (p = STR_(node)->s; p < STR_(node)->end; p++) {
if (*p >= 0x20 && *p < 0x7f)
fputc(*p, f);
case NODE_BAG:
fprintf(f, "<bag:%p> ", node);
+ if (BAG_(node)->type == BAG_IF_ELSE) {
+ Node* Then;
+ Node* Else;
+ BagNode* bn;
+
+ bn = BAG_(node);
+ fprintf(f, "if-else\n");
+ print_indent_tree(f, NODE_BODY(node), indent + add);
+
+ Then = bn->te.Then;
+ Else = bn->te.Else;
+ if (IS_NULL(Then)) {
+ Indent(f, indent + add);
+ fprintf(f, "THEN empty\n");
+ }
+ else
+ print_indent_tree(f, Then, indent + add);
+
+ if (IS_NULL(Else)) {
+ Indent(f, indent + add);
+ fprintf(f, "ELSE empty\n");
+ }
+ else
+ print_indent_tree(f, Else, indent + add);
+
+ break;
+ }
+
switch (BAG_(node)->type) {
case BAG_OPTION:
fprintf(f, "option:%d", BAG_(node)->o.options);
case BAG_STOP_BACKTRACK:
fprintf(f, "stop-bt");
break;
- case BAG_IF_ELSE:
- fprintf(f, "if-else");
+ default:
break;
}
fprintf(f, "\n");
regenc.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
p += enclen(enc, p);
}
else {
- if (prev) *prev = (const UChar* )NULL; /* Sorry */
+ if (prev)
+ *prev = onigenc_get_prev_char_head(enc, start, p);
}
return p;
}
return (UChar* )s;
}
-#if 0
-extern int
-onigenc_mbc_enc_len_end(OnigEncoding enc, const UChar* p, const UChar* end)
-{
- int len;
- int n;
-
- len = ONIGENC_MBC_ENC_LEN(enc, p);
- n = (int )(end - p);
-
- return (n < len ? n : len);
-}
-#endif
-
extern UChar*
onigenc_step(OnigEncoding enc, const UChar* p, const UChar* end, int n)
{
return 1; /* return byte length of converted char to lower */
}
-#if 0
-extern int
-onigenc_ascii_is_mbc_ambiguous(OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- const UChar* p = *pp;
-
- (*pp)++;
- return ONIGENC_IS_ASCII_CODE_CASE_AMBIG(*p);
-}
-#endif
-
extern int
onigenc_single_byte_mbc_enc_len(const UChar* p ARG_UNUSED)
{
}
}
-#if 0
-extern int
-onigenc_mbn_is_mbc_ambiguous(OnigEncoding enc, OnigCaseFoldType flag,
- const UChar** pp, const UChar* end)
-{
- const UChar* p = *pp;
-
- if (ONIGENC_IS_MBC_ASCII(p)) {
- (*pp)++;
- return ONIGENC_IS_ASCII_CODE_CASE_AMBIG(*p);
- }
-
- (*pp) += enclen(enc, p);
- return FALSE;
-}
-#endif
-
-extern int
-onigenc_mb2_code_to_mbclen(OnigCodePoint code)
-{
- if ((code & (~0xffff)) != 0) return ONIGERR_INVALID_CODE_POINT_VALUE;
-
- if ((code & 0xff00) != 0) return 2;
- else return 1;
-}
-
-extern int
-onigenc_mb4_code_to_mbclen(OnigCodePoint code)
-{
- if ((code & 0xff000000) != 0) return 4;
- else if ((code & 0xff0000) != 0) return 3;
- else if ((code & 0xff00) != 0) return 2;
- else return 1;
-}
-
extern int
onigenc_mb2_code_to_mbc(OnigEncoding enc, OnigCodePoint code, UChar *buf)
{
regenc.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
/* methods for multi byte encoding */
extern OnigCodePoint onigenc_mbn_mbc_to_code P_((OnigEncoding enc, const UChar* p, const UChar* end));
extern int onigenc_mbn_mbc_case_fold P_((OnigEncoding enc, OnigCaseFoldType flag, const UChar** p, const UChar* end, UChar* lower));
-extern int onigenc_mb2_code_to_mbclen P_((OnigCodePoint code));
extern int onigenc_mb2_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, UChar *buf));
extern int onigenc_minimum_property_name_to_ctype P_((OnigEncoding enc, UChar* p, UChar* end));
extern int onigenc_unicode_property_name_to_ctype P_((OnigEncoding enc, UChar* p, UChar* end));
extern int onigenc_is_mbc_word_ascii P_((OnigEncoding enc, UChar* s, const UChar* end));
extern int onigenc_mb2_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, unsigned int ctype));
-extern int onigenc_mb4_code_to_mbclen P_((OnigCodePoint code));
extern int onigenc_mb4_code_to_mbc P_((OnigEncoding enc, OnigCodePoint code, UChar *buf));
extern int onigenc_mb4_is_code_ctype P_((OnigEncoding enc, OnigCodePoint code, unsigned int ctype));
extern struct PropertyNameCtype* onigenc_euc_jp_lookup_property_name P_((register const char *str, register size_t len));
regerror.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regexec.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#define CHECK_INTERRUPT_IN_MATCH
+#define STACK_MEM_START(reg, i) \
+ (MEM_STATUS_AT((reg)->push_mem_start, (i)) != 0 ? \
+ STACK_AT(mem_start_stk[i])->u.mem.pstr : (UChar* )((void* )(mem_start_stk[i])))
+
+#define STACK_MEM_END(reg, i) \
+ (MEM_STATUS_AT((reg)->push_mem_end, (i)) != 0 ? \
+ STACK_AT(mem_end_stk[i])->u.mem.pstr : (UChar* )((void* )(mem_end_stk[i])))
+
+static int forward_search(regex_t* reg, const UChar* str, const UChar* end, UChar* start, UChar* range, UChar** low, UChar** high, UChar** low_prev);
+
+static int
+search_in_range(regex_t* reg, const UChar* str, const UChar* end, const UChar* start, const UChar* range, /* match range */ const UChar* data_range, /* subject string range */ OnigRegion* region, OnigOptionType option, OnigMatchParam* mp);
+
+
#ifdef USE_CALLOUT
typedef struct {
int last_match_at_call_counter;
} MatchArg;
-#ifdef ONIG_DEBUG
+#if defined(ONIG_DEBUG_COMPILE) || defined(ONIG_DEBUG_MATCH)
/* arguments type */
typedef enum {
} OpInfoType;
static OpInfoType OpInfo[] = {
- { OP_FINISH, "finish" },
- { OP_END, "end" },
- { OP_EXACT1, "exact1" },
- { OP_EXACT2, "exact2" },
- { OP_EXACT3, "exact3" },
- { OP_EXACT4, "exact4" },
- { OP_EXACT5, "exact5" },
- { OP_EXACTN, "exactn" },
- { OP_EXACTMB2N1, "exactmb2-n1" },
- { OP_EXACTMB2N2, "exactmb2-n2" },
- { OP_EXACTMB2N3, "exactmb2-n3" },
- { OP_EXACTMB2N, "exactmb2-n" },
- { OP_EXACTMB3N, "exactmb3n" },
- { OP_EXACTMBN, "exactmbn" },
- { OP_EXACT1_IC, "exact1-ic" },
- { OP_EXACTN_IC, "exactn-ic" },
- { OP_CCLASS, "cclass" },
- { OP_CCLASS_MB, "cclass-mb" },
- { OP_CCLASS_MIX, "cclass-mix" },
- { OP_CCLASS_NOT, "cclass-not" },
- { OP_CCLASS_MB_NOT, "cclass-mb-not" },
- { OP_CCLASS_MIX_NOT, "cclass-mix-not" },
- { OP_ANYCHAR, "anychar" },
- { OP_ANYCHAR_ML, "anychar-ml" },
- { OP_ANYCHAR_STAR, "anychar*" },
- { OP_ANYCHAR_ML_STAR, "anychar-ml*" },
- { OP_ANYCHAR_STAR_PEEK_NEXT, "anychar*-peek-next" },
- { OP_ANYCHAR_ML_STAR_PEEK_NEXT, "anychar-ml*-peek-next" },
- { OP_WORD, "word" },
- { OP_WORD_ASCII, "word-ascii" },
- { OP_NO_WORD, "not-word" },
- { OP_NO_WORD_ASCII, "not-word-ascii" },
- { OP_WORD_BOUNDARY, "word-boundary" },
- { OP_NO_WORD_BOUNDARY, "not-word-boundary" },
- { OP_WORD_BEGIN, "word-begin" },
- { OP_WORD_END, "word-end" },
- { OP_TEXT_SEGMENT_BOUNDARY, "text-segment-boundary" },
- { OP_BEGIN_BUF, "begin-buf" },
- { OP_END_BUF, "end-buf" },
- { OP_BEGIN_LINE, "begin-line" },
- { OP_END_LINE, "end-line" },
- { OP_SEMI_END_BUF, "semi-end-buf" },
- { OP_BEGIN_POSITION, "begin-position" },
- { OP_BACKREF1, "backref1" },
- { OP_BACKREF2, "backref2" },
- { OP_BACKREF_N, "backref-n" },
- { OP_BACKREF_N_IC, "backref-n-ic" },
- { OP_BACKREF_MULTI, "backref_multi" },
- { OP_BACKREF_MULTI_IC, "backref_multi-ic" },
- { OP_BACKREF_WITH_LEVEL, "backref_with_level" },
- { OP_BACKREF_WITH_LEVEL_IC, "backref_with_level-c" },
- { OP_BACKREF_CHECK, "backref_check" },
- { OP_BACKREF_CHECK_WITH_LEVEL, "backref_check_with_level" },
- { OP_MEMORY_START_PUSH, "mem-start-push" },
- { OP_MEMORY_START, "mem-start" },
- { OP_MEMORY_END_PUSH, "mem-end-push" },
- { OP_MEMORY_END_PUSH_REC, "mem-end-push-rec" },
- { OP_MEMORY_END, "mem-end" },
- { OP_MEMORY_END_REC, "mem-end-rec" },
- { OP_FAIL, "fail" },
- { OP_JUMP, "jump" },
- { OP_PUSH, "push" },
- { OP_PUSH_SUPER, "push-super" },
- { OP_POP_OUT, "pop-out" },
+ { OP_FINISH, "finish"},
+ { OP_END, "end"},
+ { OP_STR_1, "str_1"},
+ { OP_STR_2, "str_2"},
+ { OP_STR_3, "str_3"},
+ { OP_STR_4, "str_4"},
+ { OP_STR_5, "str_5"},
+ { OP_STR_N, "str_n"},
+ { OP_STR_MB2N1, "str_mb2-n1"},
+ { OP_STR_MB2N2, "str_mb2-n2"},
+ { OP_STR_MB2N3, "str_mb2-n3"},
+ { OP_STR_MB2N, "str_mb2-n"},
+ { OP_STR_MB3N, "str_mb3n"},
+ { OP_STR_MBN, "str_mbn"},
+ { OP_STR_1_IC, "str_1-ic"},
+ { OP_STR_N_IC, "str_n-ic"},
+ { OP_CCLASS, "cclass"},
+ { OP_CCLASS_MB, "cclass-mb"},
+ { OP_CCLASS_MIX, "cclass-mix"},
+ { OP_CCLASS_NOT, "cclass-not"},
+ { OP_CCLASS_MB_NOT, "cclass-mb-not"},
+ { OP_CCLASS_MIX_NOT, "cclass-mix-not"},
+ { OP_ANYCHAR, "anychar"},
+ { OP_ANYCHAR_ML, "anychar-ml"},
+ { OP_ANYCHAR_STAR, "anychar*"},
+ { OP_ANYCHAR_ML_STAR, "anychar-ml*"},
+ { OP_ANYCHAR_STAR_PEEK_NEXT, "anychar*-peek-next"},
+ { OP_ANYCHAR_ML_STAR_PEEK_NEXT, "anychar-ml*-peek-next"},
+ { OP_WORD, "word"},
+ { OP_WORD_ASCII, "word-ascii"},
+ { OP_NO_WORD, "not-word"},
+ { OP_NO_WORD_ASCII, "not-word-ascii"},
+ { OP_WORD_BOUNDARY, "word-boundary"},
+ { OP_NO_WORD_BOUNDARY, "not-word-boundary"},
+ { OP_WORD_BEGIN, "word-begin"},
+ { OP_WORD_END, "word-end"},
+ { OP_TEXT_SEGMENT_BOUNDARY, "text-segment-boundary"},
+ { OP_BEGIN_BUF, "begin-buf"},
+ { OP_END_BUF, "end-buf"},
+ { OP_BEGIN_LINE, "begin-line"},
+ { OP_END_LINE, "end-line"},
+ { OP_SEMI_END_BUF, "semi-end-buf"},
+ { OP_BEGIN_POSITION, "begin-position"},
+ { OP_BACKREF1, "backref1"},
+ { OP_BACKREF2, "backref2"},
+ { OP_BACKREF_N, "backref-n"},
+ { OP_BACKREF_N_IC, "backref-n-ic"},
+ { OP_BACKREF_MULTI, "backref_multi"},
+ { OP_BACKREF_MULTI_IC, "backref_multi-ic"},
+ { OP_BACKREF_WITH_LEVEL, "backref_with_level"},
+ { OP_BACKREF_WITH_LEVEL_IC, "backref_with_level-c"},
+ { OP_BACKREF_CHECK, "backref_check"},
+ { OP_BACKREF_CHECK_WITH_LEVEL, "backref_check_with_level"},
+ { OP_MEM_START_PUSH, "mem-start-push"},
+ { OP_MEM_START, "mem-start"},
+ { OP_MEM_END_PUSH, "mem-end-push"},
+#ifdef USE_CALL
+ { OP_MEM_END_PUSH_REC, "mem-end-push-rec"},
+#endif
+ { OP_MEM_END, "mem-end"},
+#ifdef USE_CALL
+ { OP_MEM_END_REC, "mem-end-rec"},
+#endif
+ { OP_FAIL, "fail"},
+ { OP_JUMP, "jump"},
+ { OP_PUSH, "push"},
+ { OP_PUSH_SUPER, "push-super"},
+ { OP_POP_OUT, "pop-out"},
#ifdef USE_OP_PUSH_OR_JUMP_EXACT
- { OP_PUSH_OR_JUMP_EXACT1, "push-or-jump-e1" },
+ { OP_PUSH_OR_JUMP_EXACT1, "push-or-jump-e1"},
+#endif
+ { OP_PUSH_IF_PEEK_NEXT, "push-if-peek-next"},
+ { OP_REPEAT, "repeat"},
+ { OP_REPEAT_NG, "repeat-ng"},
+ { OP_REPEAT_INC, "repeat-inc"},
+ { OP_REPEAT_INC_NG, "repeat-inc-ng"},
+ { OP_EMPTY_CHECK_START, "empty-check-start"},
+ { OP_EMPTY_CHECK_END, "empty-check-end"},
+ { OP_EMPTY_CHECK_END_MEMST, "empty-check-end-memst"},
+#ifdef USE_CALL
+ { OP_EMPTY_CHECK_END_MEMST_PUSH,"empty-check-end-memst-push"},
+#endif
+ { OP_PREC_READ_START, "push-pos"},
+ { OP_PREC_READ_END, "pop-pos"},
+ { OP_PREC_READ_NOT_START, "prec-read-not-start"},
+ { OP_PREC_READ_NOT_END, "prec-read-not-end"},
+ { OP_ATOMIC_START, "atomic-start"},
+ { OP_ATOMIC_END, "atomic-end"},
+ { OP_LOOK_BEHIND, "look-behind"},
+ { OP_LOOK_BEHIND_NOT_START, "look-behind-not-start"},
+ { OP_LOOK_BEHIND_NOT_END, "look-behind-not-end"},
+ { OP_PUSH_SAVE_VAL, "push-save-val"},
+ { OP_UPDATE_VAR, "update-var"},
+#ifdef USE_CALL
+ { OP_CALL, "call"},
+ { OP_RETURN, "return"},
#endif
- { OP_PUSH_IF_PEEK_NEXT, "push-if-peek-next" },
- { OP_REPEAT, "repeat" },
- { OP_REPEAT_NG, "repeat-ng" },
- { OP_REPEAT_INC, "repeat-inc" },
- { OP_REPEAT_INC_NG, "repeat-inc-ng" },
- { OP_REPEAT_INC_SG, "repeat-inc-sg" },
- { OP_REPEAT_INC_NG_SG, "repeat-inc-ng-sg" },
- { OP_EMPTY_CHECK_START, "empty-check-start" },
- { OP_EMPTY_CHECK_END, "empty-check-end" },
- { OP_EMPTY_CHECK_END_MEMST, "empty-check-end-memst" },
- { OP_EMPTY_CHECK_END_MEMST_PUSH,"empty-check-end-memst-push" },
- { OP_PREC_READ_START, "push-pos" },
- { OP_PREC_READ_END, "pop-pos" },
- { OP_PREC_READ_NOT_START, "prec-read-not-start" },
- { OP_PREC_READ_NOT_END, "prec-read-not-end" },
- { OP_ATOMIC_START, "atomic-start" },
- { OP_ATOMIC_END, "atomic-end" },
- { OP_LOOK_BEHIND, "look-behind" },
- { OP_LOOK_BEHIND_NOT_START, "look-behind-not-start" },
- { OP_LOOK_BEHIND_NOT_END, "look-behind-not-end" },
- { OP_CALL, "call" },
- { OP_RETURN, "return" },
- { OP_PUSH_SAVE_VAL, "push-save-val" },
- { OP_UPDATE_VAR, "update-var" },
#ifdef USE_CALLOUT
- { OP_CALLOUT_CONTENTS, "callout-contents" },
- { OP_CALLOUT_NAME, "callout-name" },
+ { OP_CALLOUT_CONTENTS, "callout-contents"},
+ { OP_CALLOUT_NAME, "callout-name"},
#endif
- { -1, "" }
+ { -1, ""}
};
static char*
fprintf(f, "%s", op2name(opcode));
switch (opcode) {
- case OP_EXACT1:
+ case OP_STR_1:
p_string(f, 1, p->exact.s); break;
- case OP_EXACT2:
+ case OP_STR_2:
p_string(f, 2, p->exact.s); break;
- case OP_EXACT3:
+ case OP_STR_3:
p_string(f, 3, p->exact.s); break;
- case OP_EXACT4:
+ case OP_STR_4:
p_string(f, 4, p->exact.s); break;
- case OP_EXACT5:
+ case OP_STR_5:
p_string(f, 5, p->exact.s); break;
- case OP_EXACTN:
+ case OP_STR_N:
len = p->exact_n.n;
p_string(f, len, p->exact_n.s); break;
- case OP_EXACTMB2N1:
+ case OP_STR_MB2N1:
p_string(f, 2, p->exact.s); break;
- case OP_EXACTMB2N2:
+ case OP_STR_MB2N2:
p_string(f, 4, p->exact.s); break;
- case OP_EXACTMB2N3:
+ case OP_STR_MB2N3:
p_string(f, 3, p->exact.s); break;
- case OP_EXACTMB2N:
+ case OP_STR_MB2N:
len = p->exact_n.n;
p_len_string(f, len, 2, p->exact_n.s); break;
- case OP_EXACTMB3N:
+ case OP_STR_MB3N:
len = p->exact_n.n;
p_len_string(f, len, 3, p->exact_n.s); break;
- case OP_EXACTMBN:
+ case OP_STR_MBN:
{
int mb_len;
while (n-- > 0) { fputc(*q++, f); }
}
break;
- case OP_EXACT1_IC:
+ case OP_STR_1_IC:
len = enclen(enc, p->exact.s);
p_string(f, len, p->exact.s);
break;
- case OP_EXACTN_IC:
+ case OP_STR_N_IC:
len = p->exact_n.n;
p_len_string(f, len, 1, p->exact_n.s);
break;
case OP_CCLASS_MB_NOT:
{
OnigCodePoint ncode;
- OnigCodePoint* codes;
+ OnigCodePoint* codes;
codes = (OnigCodePoint* )p->cclass_mb.mb;
GET_CODE_POINT(ncode, codes);
codes++;
GET_CODE_POINT(code, codes);
- fprintf(f, ":%u:%u", code, ncode);
+ fprintf(f, ":%d:0x%x", ncode, code);
}
break;
case OP_CCLASS_MIX:
}
break;
- case OP_MEMORY_START:
- case OP_MEMORY_START_PUSH:
+ case OP_MEM_START:
+ case OP_MEM_START_PUSH:
mem = p->memory_start.num;
fprintf(f, ":%d", mem);
break;
- case OP_MEMORY_END_PUSH:
- case OP_MEMORY_END_PUSH_REC:
- case OP_MEMORY_END:
- case OP_MEMORY_END_REC:
+
+ case OP_MEM_END:
+ case OP_MEM_END_PUSH:
+#ifdef USE_CALL
+ case OP_MEM_END_REC:
+ case OP_MEM_END_PUSH_REC:
+#endif
mem = p->memory_end.num;
fprintf(f, ":%d", mem);
break;
case OP_REPEAT_INC:
case OP_REPEAT_INC_NG:
- case OP_REPEAT_INC_SG:
- case OP_REPEAT_INC_NG_SG:
mem = p->repeat.id;
fprintf(f, ":%d", mem);
break;
break;
case OP_EMPTY_CHECK_END:
case OP_EMPTY_CHECK_END_MEMST:
+#ifdef USE_CALL
case OP_EMPTY_CHECK_END_MEMST_PUSH:
+#endif
mem = p->empty_check_end.mem;
fprintf(f, ":%d", mem);
break;
p_rel_addr(f, addr, p, start);
break;
+#ifdef USE_CALL
case OP_CALL:
addr = p->call.addr;
fprintf(f, ":{/%d}", addr);
break;
+#endif
case OP_PUSH_SAVE_VAL:
{
case OP_ATOMIC_START:
case OP_ATOMIC_END:
case OP_LOOK_BEHIND_NOT_END:
+#ifdef USE_CALL
case OP_RETURN:
+#endif
break;
default:
break;
}
}
-#endif /* ONIG_DEBUG */
+#endif /* defined(ONIG_DEBUG_COMPILE) || defined(ONIG_DEBUG_MATCH) */
#ifdef ONIG_DEBUG_COMPILE
extern void
Operation* start = reg->ops;
Operation* end = reg->ops + reg->ops_used;
- fprintf(f, "bt_mem_start: 0x%x, bt_mem_end: 0x%x\n",
- reg->bt_mem_start, reg->bt_mem_end);
+ fprintf(f, "push_mem_start: 0x%x, push_mem_end: 0x%x\n",
+ reg->push_mem_start, reg->push_mem_end);
fprintf(f, "code-length: %d\n", reg->ops_used);
bp = start;
result = ONIGERR_INVALID_ARGUMENT;\
}\
best_len = result;\
- goto finish;\
+ goto match_at_end;\
break;\
}\
} while(0)
/* handled by normal-POP */
#define STK_MEM_START 0x0010
#define STK_MEM_END 0x8030
-#define STK_REPEAT_INC 0x0050
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+#define STK_REPEAT_INC (0x0040 | STK_MASK_POP_HANDLED)
+#else
+#define STK_REPEAT_INC 0x0040
+#endif
#ifdef USE_CALLOUT
#define STK_CALLOUT 0x0070
#endif
/* avoided by normal-POP */
#define STK_VOID 0x0000 /* for fill a blank */
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+#define STK_EMPTY_CHECK_START (0x3000 | STK_MASK_POP_HANDLED)
+#else
#define STK_EMPTY_CHECK_START 0x3000
+#endif
#define STK_EMPTY_CHECK_END 0x5000 /* for recursive call */
#define STK_MEM_END_MARK 0x8100
#define STK_TO_VOID_START 0x1200 /* mark for "(?>...)" */
-#define STK_REPEAT 0x0300
+/* #define STK_REPEAT 0x0300 */
#define STK_CALL_FRAME 0x0400
#define STK_RETURN 0x0500
#define STK_SAVE_VAL 0x0600
UChar* pstr_prev; /* previous char position of pstr */
} state;
struct {
- int count; /* for OP_REPEAT_INC, OP_REPEAT_INC_NG */
- Operation* pcode; /* byte code position (head of repeated target) */
- } repeat;
- struct {
- StackIndex si; /* index of stack */
+ int count;
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+ StackIndex prev_index; /* index of stack */
+#endif
} repeat_inc;
struct {
UChar *pstr; /* start/end position */
StackIndex prev_end; /* prev. info (for backtrack "(...)*" ) */
} mem;
struct {
- UChar *pstr; /* start position */
+ UChar *pstr; /* start position */
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+ StackIndex prev_index; /* index of stack */
+#endif
} empty_check;
#ifdef USE_CALL
struct {
#endif
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+
+#define PTR_NUM_SIZE(reg) ((reg)->num_repeat + (reg)->num_empty_check + ((reg)->num_mem + 1) * 2)
+#define UPDATE_FOR_STACK_REALLOC do{\
+ repeat_stk = (StackIndex* )alloc_base;\
+ empty_check_stk = (StackIndex* )(repeat_stk + reg->num_repeat);\
+ mem_start_stk = (StackIndex* )(empty_check_stk + reg->num_empty_check);\
+ mem_end_stk = mem_start_stk + num_mem + 1;\
+} while(0)
+
+#define SAVE_REPEAT_STK_VAR(sid) stk->u.repeat_inc.prev_index = repeat_stk[sid]
+#define LOAD_TO_REPEAT_STK_VAR(sid) repeat_stk[sid] = GET_STACK_INDEX(stk)
+#define POP_REPEAT_INC else if (stk->type == STK_REPEAT_INC) {repeat_stk[stk->zid] = stk->u.repeat_inc.prev_index;}
+
+#define SAVE_EMPTY_CHECK_STK_VAR(sid) stk->u.empty_check.prev_index = empty_check_stk[sid]
+#define LOAD_TO_EMPTY_CHECK_STK_VAR(sid) empty_check_stk[sid] = GET_STACK_INDEX(stk)
+#define POP_EMPTY_CHECK_START else if (stk->type == STK_EMPTY_CHECK_START) {empty_check_stk[stk->zid] = stk->u.empty_check.prev_index;}
+
+#else
+
+#define PTR_NUM_SIZE(reg) (((reg)->num_mem + 1) * 2)
+#define UPDATE_FOR_STACK_REALLOC do{\
+ mem_start_stk = (StackIndex* )alloc_base;\
+ mem_end_stk = mem_start_stk + num_mem + 1;\
+} while(0)
+
+#define SAVE_REPEAT_STK_VAR(sid)
+#define LOAD_TO_REPEAT_STK_VAR(sid)
+#define POP_REPEAT_INC
+
+#define SAVE_EMPTY_CHECK_STK_VAR(sid)
+#define LOAD_TO_EMPTY_CHECK_STK_VAR(sid)
+#define POP_EMPTY_CHECK_START
+
+#endif /* USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR */
#ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
-#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mp) do { \
+#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
(msa).stack_p = (void* )0;\
(msa).options = (arg_option);\
(msa).region = (arg_region);\
(msa).start = (arg_start);\
- (msa).match_stack_limit = (mp)->match_stack_limit;\
- (msa).retry_limit_in_match = (mp)->retry_limit_in_match;\
- (msa).mp = mp;\
+ (msa).match_stack_limit = (mpv)->match_stack_limit;\
+ (msa).retry_limit_in_match = (mpv)->retry_limit_in_match;\
+ (msa).mp = mpv;\
(msa).best_len = ONIG_MISMATCH;\
- (msa).ptr_num = (reg)->num_repeat + ((reg)->num_mem + 1) * 2; \
+ (msa).ptr_num = PTR_NUM_SIZE(reg);\
} while(0)
#else
-#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mp) do { \
+#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
(msa).stack_p = (void* )0;\
(msa).options = (arg_option);\
(msa).region = (arg_region);\
(msa).start = (arg_start);\
- (msa).match_stack_limit = (mp)->match_stack_limit;\
- (msa).retry_limit_in_match = (mp)->retry_limit_in_match;\
- (msa).mp = mp;\
- (msa).ptr_num = (reg)->num_repeat + ((reg)->num_mem + 1) * 2; \
+ (msa).match_stack_limit = (mpv)->match_stack_limit;\
+ (msa).retry_limit_in_match = (mpv)->retry_limit_in_match;\
+ (msa).mp = mpv;\
+ (msa).ptr_num = PTR_NUM_SIZE(reg);\
} while(0)
#endif
};\
} while(0)
-#define UPDATE_FOR_STACK_REALLOC do{\
- repeat_stk = (StackIndex* )alloc_base;\
- mem_start_stk = (StackIndex* )(repeat_stk + reg->num_repeat);\
- mem_end_stk = mem_start_stk + num_mem + 1;\
-} while(0)
-
static unsigned int MatchStackLimit = DEFAULT_MATCH_STACK_LIMIT_SIZE;
extern unsigned int
static unsigned long RetryLimitInMatch = DEFAULT_RETRY_LIMIT_IN_MATCH;
#define CHECK_RETRY_LIMIT_IN_MATCH do {\
- if (retry_in_match_counter++ > retry_limit_in_match) goto retry_limit_in_match_over;\
+ if (retry_in_match_counter++ > retry_limit_in_match) {\
+ MATCH_AT_ERROR_RETURN(ONIGERR_RETRY_LIMIT_IN_MATCH_OVER);\
+ }\
} while (0)
#else
#define STACK_PUSH_ALT_LOOK_BEHIND_NOT(pat,s,sprev) \
STACK_PUSH(STK_ALT_LOOK_BEHIND_NOT,pat,s,sprev)
+#if 0
#define STACK_PUSH_REPEAT(sid, pat) do {\
STACK_ENSURE(1);\
stk->type = STK_REPEAT;\
stk->zid = (sid);\
- stk->u.repeat.pcode = (pat);\
- stk->u.repeat.count = 0;\
+ stk->u.repeat.pcode = (pat);\
STACK_INC;\
} while(0)
+#endif
-#define STACK_PUSH_REPEAT_INC(sindex) do {\
+#define STACK_PUSH_REPEAT_INC(sid, ct) do {\
STACK_ENSURE(1);\
stk->type = STK_REPEAT_INC;\
- stk->u.repeat_inc.si = (sindex);\
+ stk->zid = (sid);\
+ stk->u.repeat_inc.count = (ct);\
+ SAVE_REPEAT_STK_VAR(sid);\
+ LOAD_TO_REPEAT_STK_VAR(sid);\
STACK_INC;\
} while(0)
stk->type = STK_EMPTY_CHECK_START;\
stk->zid = (cnum);\
stk->u.empty_check.pstr = (s);\
+ SAVE_EMPTY_CHECK_STK_VAR(cnum);\
+ LOAD_TO_EMPTY_CHECK_STK_VAR(cnum);\
STACK_INC;\
} while(0)
#define STACK_BASE_CHECK(p, at) \
if ((p) < stk_base) {\
fprintf(stderr, "at %s\n", at);\
- goto stack_error;\
+ MATCH_AT_ERROR_RETURN(ONIGERR_STACK_BUG);\
}
#else
#define STACK_BASE_CHECK(p, at)
mem_start_stk[stk->zid] = stk->u.mem.prev_start;\
mem_end_stk[stk->zid] = stk->u.mem.prev_end;\
}\
- else if (stk->type == STK_REPEAT_INC) {\
- STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
- }\
else if (stk->type == STK_MEM_END) {\
mem_start_stk[stk->zid] = stk->u.mem.prev_start;\
mem_end_stk[stk->zid] = stk->u.mem.prev_end;\
}\
+ POP_REPEAT_INC \
+ POP_EMPTY_CHECK_START \
POP_CALLOUT_CASE\
}\
}\
mem_start_stk[stk->zid] = stk->u.mem.prev_start;\
mem_end_stk[stk->zid] = stk->u.mem.prev_end;\
}\
- else if (stk->type == STK_REPEAT_INC) {\
- STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
- }\
else if (stk->type == STK_MEM_END) {\
mem_start_stk[stk->zid] = stk->u.mem.prev_start;\
mem_end_stk[stk->zid] = stk->u.mem.prev_end;\
}\
+ POP_REPEAT_INC \
+ POP_EMPTY_CHECK_START \
/* Don't call callout here because negation of total success by (?!..) (?<!..) */\
}\
}\
}\
} while(0)
-#define STACK_EMPTY_CHECK(isnull,sid,s) do {\
- StackType* k = stk;\
+
+#define EMPTY_CHECK_START_SEARCH(sid, k) do {\
+ k = stk;\
while (1) {\
k--;\
- STACK_BASE_CHECK(k, "STACK_EMPTY_CHECK"); \
+ STACK_BASE_CHECK(k, "EMPTY_CHECK_START_SEARCH"); \
if (k->type == STK_EMPTY_CHECK_START) {\
- if (k->zid == (sid)) {\
- (isnull) = (k->u.empty_check.pstr == (s));\
- break;\
- }\
+ if (k->zid == (sid)) break;\
}\
}\
} while(0)
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+
+#define GET_EMPTY_CHECK_START(sid, k) do {\
+ if (reg->num_call == 0) {\
+ k = STACK_AT(empty_check_stk[sid]);\
+ }\
+ else {\
+ EMPTY_CHECK_START_SEARCH(sid, k);\
+ }\
+} while(0)
+#else
+
+#define GET_EMPTY_CHECK_START(sid, k) EMPTY_CHECK_START_SEARCH(sid, k)
+
+#endif
+
+
+#define STACK_EMPTY_CHECK(isnull, sid, s) do {\
+ StackType* k;\
+ GET_EMPTY_CHECK_START(sid, k);\
+ (isnull) = (k->u.empty_check.pstr == (s));\
+} while(0)
+
#define STACK_MEM_START_GET_PREV_END_ADDR(k /* STK_MEM_START*/, reg, addr) do {\
if (k->u.mem.prev_end == INVALID_STACK_INDEX) {\
(addr) = 0;\
}\
else {\
- if (MEM_STATUS_AT((reg)->bt_mem_end, k->zid))\
+ if (MEM_STATUS_AT((reg)->push_mem_end, k->zid))\
(addr) = STACK_AT(k->u.mem.prev_end)->u.mem.pstr;\
else\
(addr) = (UChar* )k->u.mem.prev_end;\
} while (0)
#ifdef USE_STUBBORN_CHECK_CAPTURES_IN_EMPTY_REPEAT
-#define STACK_EMPTY_CHECK_MEM(isnull,sid,s,reg) do {\
- StackType* k = stk;\
- while (1) {\
- k--;\
- STACK_BASE_CHECK(k, "STACK_EMPTY_CHECK_MEM"); \
- if (k->type == STK_EMPTY_CHECK_START) {\
- if (k->zid == (sid)) {\
- if (k->u.empty_check.pstr != (s)) {\
- (isnull) = 0;\
- break;\
+#define STACK_EMPTY_CHECK_MEM(isnull, sid, s, reg) do {\
+ StackType* k;\
+ GET_EMPTY_CHECK_START(sid, k);\
+ if (k->u.empty_check.pstr != (s)) {\
+ (isnull) = 0;\
+ }\
+ else {\
+ UChar* endp;\
+ (isnull) = 1;\
+ while (k < stk) {\
+ if (k->type == STK_MEM_START &&\
+ MEM_STATUS_LIMIT_AT((reg)->empty_status_mem, k->zid)) {\
+ STACK_MEM_START_GET_PREV_END_ADDR(k, reg, endp);\
+ if (endp == 0) {\
+ (isnull) = 0; break;\
}\
- else {\
- UChar* endp;\
- int level = 0;\
- (isnull) = 1;\
- while (k < stk) {\
- if (k->type == STK_MEM_START && level == 0) {\
- STACK_MEM_START_GET_PREV_END_ADDR(k, reg, endp);\
- if (endp == 0) {\
- (isnull) = 0; break;\
- }\
- else if (STACK_AT(k->u.mem.prev_start)->u.mem.pstr != endp) {\
- (isnull) = 0; break;\
- }\
- else if (endp != s) {\
- (isnull) = -1; /* empty, but position changed */ \
- }\
- }\
- else if (k->type == STK_PREC_READ_START) {\
- level++;\
- }\
- else if (k->type == STK_PREC_READ_END) {\
- level--;\
- }\
- k++;\
- }\
- break;\
+ else if (STACK_AT(k->u.mem.prev_start)->u.mem.pstr != endp) {\
+ (isnull) = 0; break;\
+ }\
+ else if (endp != s) {\
+ (isnull) = -1; /* empty, but position changed */ \
}\
}\
+ k++;\
}\
}\
} while(0)
}\
else {\
UChar* endp;\
- int prec_level = 0;\
(isnull) = 1;\
while (k < stk) {\
if (k->type == STK_MEM_START) {\
- if (level == 0 && prec_level == 0) {\
+ if (level == 0 && \
+ MEM_STATUS_LIMIT_AT((reg)->empty_status_mem, k->zid) !=0) {\
STACK_MEM_START_GET_PREV_END_ADDR(k, reg, endp);\
if (endp == 0) {\
(isnull) = 0; break;\
else if (k->type == STK_EMPTY_CHECK_END) {\
if (k->zid == (sid)) level--;\
}\
- else if (k->type == STK_PREC_READ_START) {\
- prec_level++;\
- }\
- else if (k->type == STK_PREC_READ_END) {\
- prec_level--;\
- }\
k++;\
}\
break;\
} while(0)
#endif /* USE_STUBBORN_CHECK_CAPTURES_IN_EMPTY_REPEAT */
-#define STACK_GET_REPEAT(sid, k) do {\
- int level = 0;\
- k = stk;\
+#define STACK_GET_REPEAT_COUNT_SEARCH(sid, c) do {\
+ StackType* k = stk;\
while (1) {\
- k--;\
- STACK_BASE_CHECK(k, "STACK_GET_REPEAT"); \
- if (k->type == STK_REPEAT) {\
- if (level == 0) {\
- if (k->zid == (sid)) {\
- break;\
+ (k)--;\
+ STACK_BASE_CHECK(k, "STACK_GET_REPEAT_COUNT_SEARCH");\
+ if ((k)->type == STK_REPEAT_INC) {\
+ if ((k)->zid == (sid)) {\
+ (c) = (k)->u.repeat_inc.count;\
+ break;\
+ }\
+ }\
+ else if ((k)->type == STK_RETURN) {\
+ int level = -1;\
+ while (1) {\
+ (k)--;\
+ if ((k)->type == STK_CALL_FRAME) {\
+ level++;\
+ if (level == 0) break;\
}\
+ else if ((k)->type == STK_RETURN) level--;\
}\
}\
- else if (k->type == STK_CALL_FRAME) level--;\
- else if (k->type == STK_RETURN) level++;\
}\
} while(0)
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+
+#define STACK_GET_REPEAT_COUNT(sid, c) do {\
+ if (reg->num_call == 0) {\
+ (c) = (STACK_AT(repeat_stk[sid]))->u.repeat_inc.count;\
+ }\
+ else {\
+ STACK_GET_REPEAT_COUNT_SEARCH(sid, c);\
+ }\
+} while(0)
+#else
+#define STACK_GET_REPEAT_COUNT(sid, c) STACK_GET_REPEAT_COUNT_SEARCH(sid, c)
+#endif
+
#define STACK_RETURN(addr) do {\
int level = 0;\
StackType* k = stk;\
#define MATCH_DEBUG_OUT(offset)
#endif
+#define MATCH_AT_ERROR_RETURN(err_code) best_len = err_code; goto match_at_end
+
/* match data(str - end) from position (sstart). */
/* if sstart == str then set sprev to NULL. */
static const void *opcode_to_label[] = {
&&L_FINISH,
&&L_END,
- &&L_EXACT1,
- &&L_EXACT2,
- &&L_EXACT3,
- &&L_EXACT4,
- &&L_EXACT5,
- &&L_EXACTN,
- &&L_EXACTMB2N1,
- &&L_EXACTMB2N2,
- &&L_EXACTMB2N3,
- &&L_EXACTMB2N,
- &&L_EXACTMB3N,
- &&L_EXACTMBN,
- &&L_EXACT1_IC,
- &&L_EXACTN_IC,
+ &&L_STR_1,
+ &&L_STR_2,
+ &&L_STR_3,
+ &&L_STR_4,
+ &&L_STR_5,
+ &&L_STR_N,
+ &&L_STR_MB2N1,
+ &&L_STR_MB2N2,
+ &&L_STR_MB2N3,
+ &&L_STR_MB2N,
+ &&L_STR_MB3N,
+ &&L_STR_MBN,
+ &&L_STR_1_IC,
+ &&L_STR_N_IC,
&&L_CCLASS,
&&L_CCLASS_MB,
&&L_CCLASS_MIX,
&&L_BACKREF_WITH_LEVEL_IC,
&&L_BACKREF_CHECK,
&&L_BACKREF_CHECK_WITH_LEVEL,
- &&L_MEMORY_START,
- &&L_MEMORY_START_PUSH,
- &&L_MEMORY_END_PUSH,
- &&L_MEMORY_END_PUSH_REC,
- &&L_MEMORY_END,
- &&L_MEMORY_END_REC,
+ &&L_MEM_START,
+ &&L_MEM_START_PUSH,
+ &&L_MEM_END_PUSH,
+#ifdef USE_CALL
+ &&L_MEM_END_PUSH_REC,
+#endif
+ &&L_MEM_END,
+#ifdef USE_CALL
+ &&L_MEM_END_REC,
+#endif
&&L_FAIL,
&&L_JUMP,
&&L_PUSH,
&&L_REPEAT_NG,
&&L_REPEAT_INC,
&&L_REPEAT_INC_NG,
- &&L_REPEAT_INC_SG,
- &&L_REPEAT_INC_NG_SG,
&&L_EMPTY_CHECK_START,
&&L_EMPTY_CHECK_END,
&&L_EMPTY_CHECK_END_MEMST,
+#ifdef USE_CALL
&&L_EMPTY_CHECK_END_MEMST_PUSH,
+#endif
&&L_PREC_READ_START,
&&L_PREC_READ_END,
&&L_PREC_READ_NOT_START,
&&L_LOOK_BEHIND,
&&L_LOOK_BEHIND_NOT_START,
&&L_LOOK_BEHIND_NOT_END,
- &&L_CALL,
- &&L_RETURN,
&&L_PUSH_SAVE_VAL,
&&L_UPDATE_VAR,
+#ifdef USE_CALL
+ &&L_CALL,
+ &&L_RETURN,
+#endif
#ifdef USE_CALLOUT
&&L_CALLOUT_CONTENTS,
&&L_CALLOUT_NAME,
char *alloc_base;
StackType *stk_base, *stk, *stk_end;
StackType *stkp; /* used as any purpose. */
- StackIndex si;
- StackIndex *repeat_stk;
StackIndex *mem_start_stk, *mem_end_stk;
UChar* keep;
+
+#ifdef USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR
+ StackIndex *repeat_stk;
+ StackIndex *empty_check_stk;
+#endif
#ifdef USE_RETRY_LIMIT_IN_MATCH
unsigned long retry_limit_in_match;
unsigned long retry_in_match_counter;
#endif
-
#ifdef USE_CALLOUT
int of;
#endif
rmt[0].rm_eo = (regoff_t )(s - str);
for (i = 1; i <= num_mem; i++) {
if (mem_end_stk[i] != INVALID_STACK_INDEX) {
- if (MEM_STATUS_AT(reg->bt_mem_start, i))
- rmt[i].rm_so = (regoff_t )(STACK_AT(mem_start_stk[i])->u.mem.pstr - str);
- else
- rmt[i].rm_so = (regoff_t )((UChar* )((void* )(mem_start_stk[i])) - str);
-
- rmt[i].rm_eo = (regoff_t )((MEM_STATUS_AT(reg->bt_mem_end, i)
- ? STACK_AT(mem_end_stk[i])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[i]))
- - str);
+ rmt[i].rm_so = (regoff_t )(STACK_MEM_START(reg, i) - str);
+ rmt[i].rm_eo = (regoff_t )(STACK_MEM_END(reg, i) - str);
}
else {
rmt[i].rm_so = rmt[i].rm_eo = ONIG_REGION_NOTPOS;
region->end[0] = (int )(s - str);
for (i = 1; i <= num_mem; i++) {
if (mem_end_stk[i] != INVALID_STACK_INDEX) {
- if (MEM_STATUS_AT(reg->bt_mem_start, i))
- region->beg[i] = (int )(STACK_AT(mem_start_stk[i])->u.mem.pstr - str);
- else
- region->beg[i] = (int )((UChar* )((void* )mem_start_stk[i]) - str);
-
- region->end[i] = (int )((MEM_STATUS_AT(reg->bt_mem_end, i)
- ? STACK_AT(mem_end_stk[i])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[i])) - str);
+ region->beg[i] = (int )(STACK_MEM_START(reg, i) - str);
+ region->end[i] = (int )(STACK_MEM_END(reg, i) - str);
}
else {
region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
stkp = stk_base;
r = make_capture_history_tree(region->history_root, &stkp,
stk, (UChar* )str, reg);
- if (r < 0) {
- best_len = r; /* error code */
- goto finish;
- }
+ if (r < 0) MATCH_AT_ERROR_RETURN(r);
}
#endif /* USE_CAPTURE_HISTORY */
#ifdef USE_POSIX_API_REGION_OPTION
}
/* default behavior: return first-matching result. */
- goto finish;
+ goto match_at_end;
- CASE_OP(EXACT1)
+ CASE_OP(STR_1)
DATA_ENSURE(1);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
NEXT_OUT;
- CASE_OP(EXACT1_IC)
+ CASE_OP(STR_1_IC)
{
int len;
UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
INC_OP;
NEXT_OUT;
- CASE_OP(EXACT2)
+ CASE_OP(STR_2)
DATA_ENSURE(2);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACT3)
+ CASE_OP(STR_3)
DATA_ENSURE(3);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACT4)
+ CASE_OP(STR_4)
DATA_ENSURE(4);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACT5)
+ CASE_OP(STR_5)
DATA_ENSURE(5);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTN)
+ CASE_OP(STR_N)
tlen = p->exact_n.n;
DATA_ENSURE(tlen);
ps = p->exact_n.s;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTN_IC)
+ CASE_OP(STR_N_IC)
{
int len;
UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTMB2N1)
+ CASE_OP(STR_MB2N1)
DATA_ENSURE(2);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
NEXT_OUT;
- CASE_OP(EXACTMB2N2)
+ CASE_OP(STR_MB2N2)
DATA_ENSURE(4);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTMB2N3)
+ CASE_OP(STR_MB2N3)
DATA_ENSURE(6);
ps = p->exact.s;
if (*ps != *s) goto fail;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTMB2N)
+ CASE_OP(STR_MB2N)
tlen = p->exact_n.n;
DATA_ENSURE(tlen * 2);
ps = p->exact_n.s;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTMB3N)
+ CASE_OP(STR_MB3N)
tlen = p->exact_n.n;
DATA_ENSURE(tlen * 3);
ps = p->exact_n.s;
INC_OP;
JUMP_OUT;
- CASE_OP(EXACTMBN)
+ CASE_OP(STR_MBN)
tlen = p->exact_len_n.len; /* mb byte len */
tlen2 = p->exact_len_n.n; /* number of chars */
tlen2 *= tlen;
UChar *ss;
int mb_len;
- DATA_ENSURE(1);
mb_len = enclen(encode, s);
DATA_ENSURE(mb_len);
ss = s;
break;
#endif
default:
- goto bytecode_error;
+ MATCH_AT_ERROR_RETURN(ONIGERR_UNDEFINED_BYTECODE);
break;
}
INC_OP;
JUMP_OUT;
- CASE_OP(MEMORY_START_PUSH)
+ CASE_OP(MEM_START_PUSH)
mem = p->memory_start.num;
STACK_PUSH_MEM_START(mem, s);
INC_OP;
JUMP_OUT;
- CASE_OP(MEMORY_START)
+ CASE_OP(MEM_START)
mem = p->memory_start.num;
mem_start_stk[mem] = (StackIndex )((void* )s);
INC_OP;
JUMP_OUT;
- CASE_OP(MEMORY_END_PUSH)
+ CASE_OP(MEM_END_PUSH)
mem = p->memory_end.num;
STACK_PUSH_MEM_END(mem, s);
INC_OP;
JUMP_OUT;
- CASE_OP(MEMORY_END)
+ CASE_OP(MEM_END)
mem = p->memory_end.num;
mem_end_stk[mem] = (StackIndex )((void* )s);
INC_OP;
JUMP_OUT;
#ifdef USE_CALL
- CASE_OP(MEMORY_END_PUSH_REC)
- mem = p->memory_end.num;
- STACK_GET_MEM_START(mem, stkp); /* should be before push mem-end. */
- si = GET_STACK_INDEX(stkp);
- STACK_PUSH_MEM_END(mem, s);
- mem_start_stk[mem] = si;
- INC_OP;
- JUMP_OUT;
+ CASE_OP(MEM_END_PUSH_REC)
+ {
+ StackIndex si;
+
+ mem = p->memory_end.num;
+ STACK_GET_MEM_START(mem, stkp); /* should be before push mem-end. */
+ si = GET_STACK_INDEX(stkp);
+ STACK_PUSH_MEM_END(mem, s);
+ mem_start_stk[mem] = si;
+ INC_OP;
+ JUMP_OUT;
+ }
- CASE_OP(MEMORY_END_REC)
+ CASE_OP(MEM_END_REC)
mem = p->memory_end.num;
mem_end_stk[mem] = (StackIndex )((void* )s);
STACK_GET_MEM_START(mem, stkp);
- if (MEM_STATUS_AT(reg->bt_mem_start, mem))
+ if (MEM_STATUS_AT(reg->push_mem_start, mem))
mem_start_stk[mem] = GET_STACK_INDEX(stkp);
else
mem_start_stk[mem] = (StackIndex )((void* )stkp->u.mem.pstr);
if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
- if (MEM_STATUS_AT(reg->bt_mem_start, mem))
- pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
- else
- pstart = (UChar* )((void* )mem_start_stk[mem]);
-
- pend = (MEM_STATUS_AT(reg->bt_mem_end, mem)
- ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[mem]));
+ pstart = STACK_MEM_START(reg, mem);
+ pend = STACK_MEM_END(reg, mem);
n = (int )(pend - pstart);
if (n != 0) {
DATA_ENSURE(n);
if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
- if (MEM_STATUS_AT(reg->bt_mem_start, mem))
- pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
- else
- pstart = (UChar* )((void* )mem_start_stk[mem]);
-
- pend = (MEM_STATUS_AT(reg->bt_mem_end, mem)
- ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[mem]));
+ pstart = STACK_MEM_START(reg, mem);
+ pend = STACK_MEM_END(reg, mem);
n = (int )(pend - pstart);
if (n != 0) {
DATA_ENSURE(n);
if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
- if (MEM_STATUS_AT(reg->bt_mem_start, mem))
- pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
- else
- pstart = (UChar* )((void* )mem_start_stk[mem]);
-
- pend = (MEM_STATUS_AT(reg->bt_mem_end, mem)
- ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[mem]));
+ pstart = STACK_MEM_START(reg, mem);
+ pend = STACK_MEM_END(reg, mem);
n = (int )(pend - pstart);
if (n != 0) {
DATA_ENSURE(n);
if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
- if (MEM_STATUS_AT(reg->bt_mem_start, mem))
- pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
- else
- pstart = (UChar* )((void* )mem_start_stk[mem]);
-
- pend = (MEM_STATUS_AT(reg->bt_mem_end, mem)
- ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
- : (UChar* )((void* )mem_end_stk[mem]));
+ pstart = STACK_MEM_START(reg, mem);
+ pend = STACK_MEM_END(reg, mem);
n = (int )(pend - pstart);
if (n != 0) {
DATA_ENSURE(n);
case OP_PUSH:
case OP_REPEAT_INC:
case OP_REPEAT_INC_NG:
- case OP_REPEAT_INC_SG:
- case OP_REPEAT_INC_NG_SG:
INC_OP;
break;
default:
- goto unexpected_bytecode_error;
+ MATCH_AT_ERROR_RETURN(ONIGERR_UNEXPECTED_BYTECODE);
break;
}
#else
addr = p->push_if_peek_next.addr;
c = p->push_if_peek_next.c;
- if (c == *s) {
+ if (DATA_ENSURE_CHECK1 && c == *s) {
STACK_PUSH_ALT(p + addr, s, sprev);
INC_OP;
JUMP_OUT;
mem = p->repeat.id; /* mem: OP_REPEAT ID */
addr = p->repeat.addr;
- STACK_ENSURE(1);
- repeat_stk[mem] = GET_STACK_INDEX(stk);
- STACK_PUSH_REPEAT(mem, p + 1);
-
+ STACK_PUSH_REPEAT_INC(mem, 0);
if (reg->repeat_range[mem].lower == 0) {
STACK_PUSH_ALT(p + addr, s, sprev);
}
mem = p->repeat.id; /* mem: OP_REPEAT ID */
addr = p->repeat.addr;
- STACK_ENSURE(1);
- repeat_stk[mem] = GET_STACK_INDEX(stk);
- STACK_PUSH_REPEAT(mem, p + 1);
-
+ STACK_PUSH_REPEAT_INC(mem, 0);
if (reg->repeat_range[mem].lower == 0) {
STACK_PUSH_ALT(p + 1, s, sprev);
p += addr;
CASE_OP(REPEAT_INC)
mem = p->repeat_inc.id; /* mem: OP_REPEAT ID */
- si = repeat_stk[mem];
- stkp = STACK_AT(si);
-
- repeat_inc:
- stkp->u.repeat.count++;
- if (stkp->u.repeat.count >= reg->repeat_range[mem].upper) {
+ STACK_GET_REPEAT_COUNT(mem, n);
+ n++;
+ if (n >= reg->repeat_range[mem].upper) {
/* end of repeat. Nothing to do. */
INC_OP;
}
- else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
+ else if (n >= reg->repeat_range[mem].lower) {
INC_OP;
STACK_PUSH_ALT(p, s, sprev);
- p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */
+ p = reg->repeat_range[mem].u.pcode;
}
else {
- p = stkp->u.repeat.pcode;
+ p = reg->repeat_range[mem].u.pcode;
}
- STACK_PUSH_REPEAT_INC(si);
+ STACK_PUSH_REPEAT_INC(mem, n);
CHECK_INTERRUPT_JUMP_OUT;
- CASE_OP(REPEAT_INC_SG)
- mem = p->repeat_inc.id; /* mem: OP_REPEAT ID */
- STACK_GET_REPEAT(mem, stkp);
- si = GET_STACK_INDEX(stkp);
- goto repeat_inc;
-
CASE_OP(REPEAT_INC_NG)
mem = p->repeat_inc.id; /* mem: OP_REPEAT ID */
- si = repeat_stk[mem];
- stkp = STACK_AT(si);
-
- repeat_inc_ng:
- stkp->u.repeat.count++;
- if (stkp->u.repeat.count < reg->repeat_range[mem].upper) {
- if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
- Operation* pcode = stkp->u.repeat.pcode;
-
- STACK_PUSH_REPEAT_INC(si);
- STACK_PUSH_ALT(pcode, s, sprev);
+ STACK_GET_REPEAT_COUNT(mem, n);
+ n++;
+ STACK_PUSH_REPEAT_INC(mem, n);
+ if (n == reg->repeat_range[mem].upper) {
+ INC_OP;
+ }
+ else {
+ if (n >= reg->repeat_range[mem].lower) {
+ STACK_PUSH_ALT(reg->repeat_range[mem].u.pcode, s, sprev);
INC_OP;
}
else {
- p = stkp->u.repeat.pcode;
- STACK_PUSH_REPEAT_INC(si);
+ p = reg->repeat_range[mem].u.pcode;
}
}
- else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) {
- STACK_PUSH_REPEAT_INC(si);
- INC_OP;
- }
CHECK_INTERRUPT_JUMP_OUT;
- CASE_OP(REPEAT_INC_NG_SG)
- mem = p->repeat_inc.id; /* mem: OP_REPEAT ID */
- STACK_GET_REPEAT(mem, stkp);
- si = GET_STACK_INDEX(stkp);
- goto repeat_inc_ng;
-
CASE_OP(PREC_READ_START)
STACK_PUSH_PREC_READ_START(s, sprev);
INC_OP;
OnigCalloutFunc func;
OnigCalloutArgs args;
- of = ONIG_CALLOUT_OF_NAME;
- name_id = p->callout_name.id;
- mem = p->callout_name.num;
+ of = ONIG_CALLOUT_OF_NAME;
+ mem = p->callout_name.num;
callout_common_entry:
e = onig_reg_callout_list_at(reg, mem);
in = e->in;
if (of == ONIG_CALLOUT_OF_NAME) {
+ name_id = p->callout_name.id;
func = onig_get_callout_start_func(reg, mem);
}
else {
call_result = ONIGERR_INVALID_ARGUMENT;
}
best_len = call_result;
- goto finish;
+ goto match_at_end;
break;
}
}
#endif
CASE_OP(FINISH)
- goto finish;
+ goto match_at_end;
#ifdef ONIG_DEBUG_STATISTICS
fail:
JUMP_OUT;
DEFAULT_OP
- goto bytecode_error;
+ MATCH_AT_ERROR_RETURN(ONIGERR_UNDEFINED_BYTECODE);
} BYTECODE_INTERPRETER_END;
- finish:
+ match_at_end:
STACK_SAVE;
return best_len;
+}
-#ifdef ONIG_DEBUG
- stack_error:
- STACK_SAVE;
- return ONIGERR_STACK_BUG;
-#endif
+typedef struct {
+ regex_t* reg;
+ OnigRegion* region;
+} RR;
+
+struct OnigRegSetStruct {
+ RR* rs;
+ int n;
+ int alloc;
+ OnigEncoding enc;
+ int anchor; /* BEGIN_BUF, BEGIN_POS, (SEMI_)END_BUF */
+ OnigLen anc_dmin; /* (SEMI_)END_BUF anchor distance */
+ OnigLen anc_dmax; /* (SEMI_)END_BUF anchor distance */
+ int all_low_high;
+ int anychar_inf;
+};
- bytecode_error:
- STACK_SAVE;
- return ONIGERR_UNDEFINED_BYTECODE;
+enum SearchRangeStatus {
+ SRS_DEAD = 0,
+ SRS_LOW_HIGH = 1,
+ SRS_ALL_RANGE = 2
+};
-#if defined(ONIG_DEBUG) && !defined(USE_DIRECT_THREADED_CODE)
- unexpected_bytecode_error:
- STACK_SAVE;
- return ONIGERR_UNEXPECTED_BYTECODE;
-#endif
+typedef struct {
+ int state; /* value of enum SearchRangeStatus */
+ UChar* low;
+ UChar* high;
+ UChar* low_prev;
+ UChar* sch_range;
+} SearchRange;
+
+#define REGSET_MATCH_AND_RETURN_CHECK(upper_range) \
+ r = match_at(reg, str, end, (upper_range), s, prev, msas + i); \
+ if (r != ONIG_MISMATCH) {\
+ if (r >= 0) {\
+ goto match;\
+ }\
+ else goto finish; /* error */ \
+ }
-#ifdef USE_RETRY_LIMIT_IN_MATCH
- retry_limit_in_match_over:
- STACK_SAVE;
- return ONIGERR_RETRY_LIMIT_IN_MATCH_OVER;
-#endif
-}
+static inline int
+regset_search_body_position_lead(OnigRegSet* set,
+ const UChar* str, const UChar* end,
+ const UChar* start, const UChar* range, /* match start range */
+ const UChar* orig_range, /* data range */
+ OnigOptionType option, MatchArg* msas, int* rmatch_pos)
+{
+ int r, n, i;
+ UChar *s, *prev;
+ UChar *low, *high, *low_prev;
+ UChar* sch_range;
+ regex_t* reg;
+ OnigEncoding enc;
+ SearchRange* sr;
+ n = set->n;
+ enc = set->enc;
-static UChar*
-slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
- const UChar* text, const UChar* text_end, UChar* text_range)
-{
- UChar *t, *p, *s, *end;
+ s = (UChar* )start;
+ if (s > str)
+ prev = onigenc_get_prev_char_head(enc, str, s);
+ else
+ prev = (UChar* )NULL;
- end = (UChar* )text_end;
- end -= target_end - target - 1;
- if (end > text_range)
- end = text_range;
+ sr = (SearchRange* )xmalloc(sizeof(*sr) * n);
+ CHECK_NULL_RETURN_MEMERR(sr);
- s = (UChar* )text;
+ for (i = 0; i < n; i++) {
+ reg = set->rs[i].reg;
- while (s < end) {
- if (*s == *target) {
- p = s + 1;
- t = target + 1;
- while (t < target_end) {
- if (*t != *p++)
- break;
- t++;
+ sr[i].state = SRS_DEAD;
+ if (reg->optimize != OPTIMIZE_NONE) {
+ if (reg->dist_max != INFINITE_LEN) {
+ if (end - range > reg->dist_max)
+ sch_range = (UChar* )range + reg->dist_max;
+ else
+ sch_range = (UChar* )end;
+
+ if (forward_search(reg, str, end, s, sch_range, &low, &high, &low_prev)) {
+ sr[i].state = SRS_LOW_HIGH;
+ sr[i].low = low;
+ sr[i].high = high;
+ sr[i].low_prev = low_prev;
+ sr[i].sch_range = sch_range;
+ }
+ }
+ else {
+ sch_range = (UChar* )end;
+ if (forward_search(reg, str, end, s, sch_range,
+ &low, &high, (UChar** )NULL)) {
+ goto total_active;
+ }
}
- if (t == target_end)
- return s;
}
- s += enclen(enc, s);
+ else {
+ total_active:
+ sr[i].state = SRS_ALL_RANGE;
+ sr[i].low = s;
+ sr[i].high = (UChar* )range;
+ sr[i].low_prev = prev;
+ }
}
- return (UChar* )NULL;
-}
+#define ACTIVATE_ALL_LOW_HIGH_SEARCH_THRESHOLD_LEN 500
-static int
-str_lower_case_match(OnigEncoding enc, int case_fold_flag,
- const UChar* t, const UChar* tend,
- const UChar* p, const UChar* end)
-{
- int lowlen;
- UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
+ if (set->all_low_high != 0
+ && range - start > ACTIVATE_ALL_LOW_HIGH_SEARCH_THRESHOLD_LEN) {
+ do {
+ int try_count = 0;
+ for (i = 0; i < n; i++) {
+ if (sr[i].state == SRS_DEAD) continue;
+
+ if (s < sr[i].low) continue;
+ if (s >= sr[i].high) {
+ if (forward_search(set->rs[i].reg, str, end, s, sr[i].sch_range,
+ &low, &high, &low_prev) != 0) {
+ sr[i].low = low;
+ sr[i].high = high;
+ sr[i].low_prev = low_prev;
+ if (s < low) continue;
+ }
+ else {
+ sr[i].state = SRS_DEAD;
+ continue;
+ }
+ }
- while (t < tend) {
- lowlen = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, lowbuf);
- q = lowbuf;
- while (lowlen > 0) {
- if (t >= tend) return 0;
- if (*t++ != *q++) return 0;
- lowlen--;
- }
- }
+ reg = set->rs[i].reg;
+ REGSET_MATCH_AND_RETURN_CHECK(orig_range);
+ try_count++;
+ } /* for (i) */
- return 1;
-}
+ if (s >= range) break;
-static UChar*
-slow_search_ic(OnigEncoding enc, int case_fold_flag,
- UChar* target, UChar* target_end,
- const UChar* text, const UChar* text_end, UChar* text_range)
+ if (try_count == 0) {
+ low = (UChar* )range;
+ for (i = 0; i < n; i++) {
+ if (sr[i].state == SRS_LOW_HIGH && low > sr[i].low) {
+ low = sr[i].low;
+ low_prev = sr[i].low_prev;
+ }
+ }
+ if (low == range) break;
+
+ s = low;
+ prev = low_prev;
+ }
+ else {
+ prev = s;
+ s += enclen(enc, s);
+ }
+ } while (1);
+ }
+ else {
+ int prev_is_newline = 1;
+ do {
+ for (i = 0; i < n; i++) {
+ if (sr[i].state == SRS_DEAD) continue;
+ if (sr[i].state == SRS_LOW_HIGH) {
+ if (s < sr[i].low) continue;
+ if (s >= sr[i].high) {
+ if (forward_search(set->rs[i].reg, str, end, s, sr[i].sch_range,
+ &low, &high, &low_prev) != 0) {
+ sr[i].low = low;
+ sr[i].high = high;
+ /* sr[i].low_prev = low_prev; */
+ if (s < low) continue;
+ }
+ else {
+ sr[i].state = SRS_DEAD;
+ continue;
+ }
+ }
+ }
+
+ reg = set->rs[i].reg;
+ if ((reg->anchor & ANCR_ANYCHAR_INF) == 0 || prev_is_newline != 0) {
+ REGSET_MATCH_AND_RETURN_CHECK(orig_range);
+ }
+ }
+
+ if (s >= range) break;
+
+ if (set->anychar_inf != 0)
+ prev_is_newline = ONIGENC_IS_MBC_NEWLINE(set->enc, s, end);
+
+ prev = s;
+ s += enclen(enc, s);
+ } while (1);
+ }
+
+ xfree(sr);
+ return ONIG_MISMATCH;
+
+ finish:
+ xfree(sr);
+ return r;
+
+ match:
+ xfree(sr);
+ *rmatch_pos = (int )(s - str);
+ return i;
+}
+
+static inline int
+regset_search_body_regex_lead(OnigRegSet* set,
+ const UChar* str, const UChar* end,
+ const UChar* start, const UChar* orig_range, OnigRegSetLead lead,
+ OnigOptionType option, OnigMatchParam* mps[], int* rmatch_pos)
+{
+ int r;
+ int i;
+ int n;
+ int match_index;
+ const UChar* ep;
+ regex_t* reg;
+ OnigRegion* region;
+
+ n = set->n;
+
+ match_index = ONIG_MISMATCH;
+ ep = orig_range;
+ for (i = 0; i < n; i++) {
+ reg = set->rs[i].reg;
+ region = set->rs[i].region;
+ r = search_in_range(reg, str, end, start, ep, orig_range, region, option, mps[i]);
+ if (r > 0) {
+ if (str + r < ep) {
+ match_index = i;
+ *rmatch_pos = r;
+ if (lead == ONIG_REGSET_PRIORITY_TO_REGEX_ORDER)
+ break;
+
+ ep = str + r;
+ }
+ }
+ else if (r == 0) {
+ match_index = i;
+ *rmatch_pos = r;
+ break;
+ }
+ }
+
+ return match_index;
+}
+
+extern int
+onig_regset_search_with_param(OnigRegSet* set,
+ const UChar* str, const UChar* end,
+ const UChar* start, const UChar* range,
+ OnigRegSetLead lead, OnigOptionType option, OnigMatchParam* mps[],
+ int* rmatch_pos)
+{
+ int r;
+ int i;
+ UChar *s, *prev;
+ regex_t* reg;
+ OnigEncoding enc;
+ OnigRegion* region;
+ MatchArg* msas;
+ const UChar *orig_start = start;
+ const UChar *orig_range = range;
+
+ if (set->n == 0)
+ return ONIG_MISMATCH;
+
+ if (IS_POSIX_REGION(option))
+ return ONIGERR_INVALID_ARGUMENT;
+
+ r = 0;
+ enc = set->enc;
+ msas = (MatchArg* )NULL;
+
+ for (i = 0; i < set->n; i++) {
+ reg = set->rs[i].reg;
+ region = set->rs[i].region;
+ ADJUST_MATCH_PARAM(reg, mps[i]);
+ if (IS_NOT_NULL(region)) {
+ r = onig_region_resize_clear(region, reg->num_mem + 1);
+ if (r != 0) goto finish_no_msa;
+ }
+ }
+
+ if (start > end || start < str) goto mismatch_no_msa;
+ if (str < end) {
+ /* forward search only */
+ if (range <= start)
+ return ONIGERR_INVALID_ARGUMENT;
+ }
+
+ if (ONIG_IS_OPTION_ON(option, ONIG_OPTION_CHECK_VALIDITY_OF_STRING)) {
+ if (! ONIGENC_IS_VALID_MBC_STRING(enc, str, end)) {
+ r = ONIGERR_INVALID_WIDE_CHAR_VALUE;
+ goto finish_no_msa;
+ }
+ }
+
+ if (set->anchor != OPTIMIZE_NONE && str < end) {
+ UChar *min_semi_end, *max_semi_end;
+
+ if ((set->anchor & ANCR_BEGIN_POSITION) != 0) {
+ /* search start-position only */
+ begin_position:
+ range = start + 1;
+ }
+ else if ((set->anchor & ANCR_BEGIN_BUF) != 0) {
+ /* search str-position only */
+ if (start != str) goto mismatch_no_msa;
+ range = str + 1;
+ }
+ else if ((set->anchor & ANCR_END_BUF) != 0) {
+ min_semi_end = max_semi_end = (UChar* )end;
+
+ end_buf:
+ if ((OnigLen )(max_semi_end - str) < set->anc_dmin)
+ goto mismatch_no_msa;
+
+ if ((OnigLen )(min_semi_end - start) > set->anc_dmax) {
+ start = min_semi_end - set->anc_dmax;
+ if (start < end)
+ start = onigenc_get_right_adjust_char_head(enc, str, start);
+ }
+ if ((OnigLen )(max_semi_end - (range - 1)) < set->anc_dmin) {
+ range = max_semi_end - set->anc_dmin + 1;
+ }
+ if (start > range) goto mismatch_no_msa;
+ }
+ else if ((set->anchor & ANCR_SEMI_END_BUF) != 0) {
+ UChar* pre_end = ONIGENC_STEP_BACK(enc, str, end, 1);
+
+ max_semi_end = (UChar* )end;
+ if (ONIGENC_IS_MBC_NEWLINE(enc, pre_end, end)) {
+ min_semi_end = pre_end;
+
+#ifdef USE_CRNL_AS_LINE_TERMINATOR
+ pre_end = ONIGENC_STEP_BACK(enc, str, pre_end, 1);
+ if (IS_NOT_NULL(pre_end) &&
+ ONIGENC_IS_MBC_CRNL(enc, pre_end, end)) {
+ min_semi_end = pre_end;
+ }
+#endif
+ if (min_semi_end > str && start <= min_semi_end) {
+ goto end_buf;
+ }
+ }
+ else {
+ min_semi_end = (UChar* )end;
+ goto end_buf;
+ }
+ }
+ else if ((set->anchor & ANCR_ANYCHAR_INF_ML) != 0) {
+ goto begin_position;
+ }
+ }
+ else if (str == end) { /* empty string */
+ start = end = str;
+ s = (UChar* )start;
+ prev = (UChar* )NULL;
+
+ msas = (MatchArg* )xmalloc(sizeof(*msas) * set->n);
+ CHECK_NULL_RETURN_MEMERR(msas);
+ for (i = 0; i < set->n; i++) {
+ reg = set->rs[i].reg;
+ MATCH_ARG_INIT(msas[i], reg, option, set->rs[i].region, start, mps[i]);
+ }
+ for (i = 0; i < set->n; i++) {
+ reg = set->rs[i].reg;
+ if (reg->threshold_len == 0) {
+ REGSET_MATCH_AND_RETURN_CHECK(end);
+ }
+ }
+
+ goto mismatch;
+ }
+
+ if (lead == ONIG_REGSET_POSITION_LEAD) {
+ msas = (MatchArg* )xmalloc(sizeof(*msas) * set->n);
+ CHECK_NULL_RETURN_MEMERR(msas);
+
+ for (i = 0; i < set->n; i++) {
+ MATCH_ARG_INIT(msas[i], set->rs[i].reg, option, set->rs[i].region,
+ orig_start, mps[i]);
+ }
+
+ r = regset_search_body_position_lead(set, str, end, start, range,
+ orig_range, option, msas, rmatch_pos);
+ }
+ else {
+ r = regset_search_body_regex_lead(set, str, end, start, orig_range,
+ lead, option, mps, rmatch_pos);
+ }
+ if (r < 0) goto finish;
+ else goto match2;
+
+ mismatch:
+ r = ONIG_MISMATCH;
+ finish:
+ for (i = 0; i < set->n; i++) {
+ if (IS_NOT_NULL(msas))
+ MATCH_ARG_FREE(msas[i]);
+ if (IS_FIND_NOT_EMPTY(set->rs[i].reg->options) &&
+ IS_NOT_NULL(set->rs[i].region)) {
+ onig_region_clear(set->rs[i].region);
+ }
+ }
+ if (IS_NOT_NULL(msas)) xfree(msas);
+ return r;
+
+ mismatch_no_msa:
+ r = ONIG_MISMATCH;
+ finish_no_msa:
+ return r;
+
+ match:
+ *rmatch_pos = (int )(s - str);
+ match2:
+ for (i = 0; i < set->n; i++) {
+ if (IS_NOT_NULL(msas))
+ MATCH_ARG_FREE(msas[i]);
+ if (IS_FIND_NOT_EMPTY(set->rs[i].reg->options) &&
+ IS_NOT_NULL(set->rs[i].region)) {
+ onig_region_clear(set->rs[i].region);
+ }
+ }
+ if (IS_NOT_NULL(msas)) xfree(msas);
+ return r; /* regex index */
+}
+
+extern int
+onig_regset_search(OnigRegSet* set, const UChar* str, const UChar* end,
+ const UChar* start, const UChar* range,
+ OnigRegSetLead lead, OnigOptionType option, int* rmatch_pos)
{
- UChar *s, *end;
+ int r;
+ int i;
+ OnigMatchParam* mp;
+ OnigMatchParam** mps;
+
+ mps = (OnigMatchParam** )xmalloc((sizeof(OnigMatchParam*) + sizeof(OnigMatchParam)) * set->n);
+ CHECK_NULL_RETURN_MEMERR(mps);
+
+ mp = (OnigMatchParam* )(mps + set->n);
+
+ for (i = 0; i < set->n; i++) {
+ onig_initialize_match_param(mp + i);
+ mps[i] = mp + i;
+ }
+
+ r = onig_regset_search_with_param(set, str, end, start, range, lead, option, mps,
+ rmatch_pos);
+ for (i = 0; i < set->n; i++)
+ onig_free_match_param_content(mp + i);
+
+ xfree(mps);
+
+ return r;
+}
+
+static UChar*
+slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
+ const UChar* text, const UChar* text_end, UChar* text_range)
+{
+ UChar *t, *p, *s, *end;
end = (UChar* )text_end;
end -= target_end - target - 1;
s = (UChar* )text;
while (s < end) {
+ if (*s == *target) {
+ p = s + 1;
+ t = target + 1;
+ while (t < target_end) {
+ if (*t != *p++)
+ break;
+ t++;
+ }
+ if (t == target_end)
+ return s;
+ }
+ s += enclen(enc, s);
+ }
+
+ return (UChar* )NULL;
+}
+
+static int
+str_lower_case_match(OnigEncoding enc, int case_fold_flag,
+ const UChar* t, const UChar* tend,
+ const UChar* p, const UChar* end)
+{
+ int lowlen;
+ UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
+
+ while (t < tend) {
+ if (p >= end) return 0;
+ lowlen = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, lowbuf);
+ q = lowbuf;
+ while (lowlen > 0) {
+ if (t >= tend) return 0;
+ if (*t++ != *q++) return 0;
+ lowlen--;
+ }
+ }
+
+ return 1;
+}
+
+static UChar*
+slow_search_ic(OnigEncoding enc, int case_fold_flag,
+ UChar* target, UChar* target_end,
+ const UChar* text, const UChar* text_end, UChar* text_range)
+{
+ UChar *s;
+
+ s = (UChar* )text;
+
+ while (s < text_range) {
if (str_lower_case_match(enc, case_fold_flag, target, target_end,
s, text_end))
return s;
return (UChar* )NULL;
}
-static UChar*
-sunday_quick_search_case_fold(regex_t* reg,
- const UChar* target, const UChar* target_end,
- const UChar* text, const UChar* text_end,
- const UChar* text_range)
-{
- const UChar *s, *se, *end;
- const UChar *tail;
- int skip, tlen1;
- int map_offset;
- int case_fold_flag;
- OnigEncoding enc;
-
-#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr,
- "sunday_quick_search_case_fold: text: %p, text_end: %p, text_range: %p\n", text, text_end, text_range);
-#endif
-
- enc = reg->enc;
- case_fold_flag = reg->case_fold_flag;
-
- tail = target_end - 1;
- tlen1 = (int )(tail - target);
- end = text_range;
- if (end + tlen1 > text_end)
- end = text_end - tlen1;
-
- map_offset = reg->map_offset;
- s = text;
-
- while (s < end) {
- if (str_lower_case_match(enc, case_fold_flag, target, target_end,
- s, text_end))
- return (UChar* )s;
-
- se = s + tlen1;
- if (se + map_offset >= text_end) break;
- skip = reg->map[*(se + map_offset)];
-#if 0
- p = s;
- do {
- s += enclen(enc, s);
- } while ((s - p) < skip && s < end);
-#else
- /* This is faster than prev code for long text. ex: /(?i)Twain/ */
- s += skip;
- if (s < end)
- s = onigenc_get_right_adjust_char_head(enc, text, s);
-#endif
- }
-
- return (UChar* )NULL;
-}
-
static UChar*
map_search(OnigEncoding enc, UChar map[],
const UChar* text, const UChar* text_range)
}
static int
-forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
- UChar* range, UChar** low, UChar** high, UChar** low_prev)
+forward_search(regex_t* reg, const UChar* str, const UChar* end, UChar* start,
+ UChar* range, UChar** low, UChar** high, UChar** low_prev)
{
UChar *p, *pprev = (UChar* )NULL;
#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr, "forward_search_range: str: %p, end: %p, s: %p, range: %p\n",
- str, end, s, range);
+ fprintf(stderr, "forward_search: str: %p, end: %p, start: %p, range: %p\n",
+ str, end, start, range);
#endif
- p = s;
- if (reg->dmin > 0) {
+ p = start;
+ if (reg->dist_min != 0) {
+ if (end - p <= reg->dist_min)
+ return 0; /* fail */
+
if (ONIGENC_IS_SINGLEBYTE(reg->enc)) {
- p += reg->dmin;
+ p += reg->dist_min;
}
else {
- UChar *q = p + reg->dmin;
-
- if (q >= end) return 0; /* fail */
+ UChar *q = p + reg->dist_min;
while (p < q) p += enclen(reg->enc, p);
}
}
reg->exact, reg->exact_end, p, end, range);
break;
- case OPTIMIZE_STR_CASE_FOLD_FAST:
- p = sunday_quick_search_case_fold(reg, reg->exact, reg->exact_end, p, end,
- range);
- break;
-
case OPTIMIZE_STR_FAST:
p = sunday_quick_search(reg, reg->exact, reg->exact_end, p, end, range);
break;
}
if (p && p < range) {
- if (p - reg->dmin < s) {
+ if (p - start < reg->dist_min) {
retry_gate:
pprev = p;
p += enclen(reg->enc, p);
switch (reg->sub_anchor) {
case ANCR_BEGIN_LINE:
if (!ON_STR_BEGIN(p)) {
- prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : str), p);
+ prev = onigenc_get_prev_char_head(reg->enc, (pprev ? pprev : str), p);
if (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end))
goto retry_gate;
}
#endif
)
goto retry_gate;
+
break;
}
}
- if (reg->dmax == 0) {
+ if (reg->dist_max == 0) {
*low = p;
if (low_prev) {
- if (*low > s)
- *low_prev = onigenc_get_prev_char_head(reg->enc, s, p);
+ if (*low > start)
+ *low_prev = onigenc_get_prev_char_head(reg->enc, start, p);
else
*low_prev = onigenc_get_prev_char_head(reg->enc,
(pprev ? pprev : str), p);
}
+ *high = p;
}
else {
- if (reg->dmax != INFINITE_LEN) {
- if (p - str < reg->dmax) {
+ if (reg->dist_max != INFINITE_LEN) {
+ if (p - str < reg->dist_max) {
*low = (UChar* )str;
if (low_prev)
*low_prev = onigenc_get_prev_char_head(reg->enc, str, *low);
}
else {
- *low = p - reg->dmax;
- if (*low > s) {
- *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s,
+ *low = p - reg->dist_max;
+ if (*low > start) {
+ *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, start,
*low, (const UChar** )low_prev);
- if (low_prev && IS_NULL(*low_prev))
- *low_prev = onigenc_get_prev_char_head(reg->enc,
- (pprev ? pprev : s), *low);
}
else {
if (low_prev)
}
}
}
+ /* no needs to adjust *high, *high is used as range check only */
+ if (p - str < reg->dist_min)
+ *high = (UChar* )str;
+ else
+ *high = p - reg->dist_min;
}
- /* no needs to adjust *high, *high is used as range check only */
- *high = p - reg->dmin;
#ifdef ONIG_DEBUG_SEARCH
fprintf(stderr,
- "forward_search_range success: low: %d, high: %d, dmin: %d, dmax: %d\n",
- (int )(*low - str), (int )(*high - str), reg->dmin, reg->dmax);
+ "forward_search success: low: %d, high: %d, dmin: %u, dmax: %u\n",
+ (int )(*low - str), (int )(*high - str),
+ reg->dist_min, reg->dist_max);
#endif
return 1; /* success */
}
static int
-backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
- UChar* s, const UChar* range, UChar* adjrange,
- UChar** low, UChar** high)
+backward_search(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
+ const UChar* range, UChar* adjrange, UChar** low, UChar** high)
{
UChar *p;
- if (range == 0) goto fail;
-
- range += reg->dmin;
p = s;
retry:
break;
case OPTIMIZE_STR_CASE_FOLD:
- case OPTIMIZE_STR_CASE_FOLD_FAST:
p = slow_search_backward_ic(reg->enc, reg->case_fold_flag,
reg->exact, reg->exact_end,
range, adjrange, end, p);
}
}
- /* no needs to adjust *high, *high is used as range check only */
- if (reg->dmax != INFINITE_LEN) {
- *low = p - reg->dmax;
- *high = p - reg->dmin;
+ if (reg->dist_max != INFINITE_LEN) {
+ if (p - str < reg->dist_max)
+ *low = (UChar* )str;
+ else
+ *low = p - reg->dist_max;
+
+ if (reg->dist_min != 0) {
+ if (p - str < reg->dist_min)
+ *high = (UChar* )str;
+ else
+ *high = p - reg->dist_min;
+ }
+ else {
+ *high = p;
+ }
+
*high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high);
}
#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr, "backward_search_range: low: %d, high: %d\n",
+ fprintf(stderr, "backward_search: low: %d, high: %d\n",
(int )(*low - str), (int )(*high - str));
#endif
return 1; /* success */
fail:
#ifdef ONIG_DEBUG_SEARCH
- fprintf(stderr, "backward_search_range: fail.\n");
+ fprintf(stderr, "backward_search: fail.\n");
#endif
return 0; /* fail */
}
{
int r;
OnigMatchParam mp;
+ const UChar* data_range;
onig_initialize_match_param(&mp);
- r = onig_search_with_param(reg, str, end, start, range, region, option, &mp);
+
+ /* The following is an expanded code of onig_search_with_param() */
+ if (range > start)
+ data_range = range;
+ else
+ data_range = end;
+
+ r = search_in_range(reg, str, end, start, range, data_range, region,
+ option, &mp);
+
onig_free_match_param_content(&mp);
return r;
}
-extern int
-onig_search_with_param(regex_t* reg, const UChar* str, const UChar* end,
- const UChar* start, const UChar* range, OnigRegion* region,
- OnigOptionType option, OnigMatchParam* mp)
+static int
+search_in_range(regex_t* reg, const UChar* str, const UChar* end,
+ const UChar* start, const UChar* range, /* match start range */
+ const UChar* data_range, /* subject string range */
+ OnigRegion* region,
+ OnigOptionType option, OnigMatchParam* mp)
{
int r;
UChar *s, *prev;
MatchArg msa;
const UChar *orig_start = start;
- const UChar *orig_range = range;
#ifdef ONIG_DEBUG_SEARCH
fprintf(stderr,
min_semi_end = max_semi_end = (UChar* )end;
end_buf:
- if ((OnigLen )(max_semi_end - str) < reg->anchor_dmin)
+ if ((OnigLen )(max_semi_end - str) < reg->anc_dist_min)
goto mismatch_no_msa;
if (range > start) {
- if ((OnigLen )(min_semi_end - start) > reg->anchor_dmax) {
- start = min_semi_end - reg->anchor_dmax;
+ if (reg->anc_dist_max != INFINITE_LEN &&
+ min_semi_end - start > reg->anc_dist_max) {
+ start = min_semi_end - reg->anc_dist_max;
if (start < end)
start = onigenc_get_right_adjust_char_head(reg->enc, str, start);
}
- if ((OnigLen )(max_semi_end - (range - 1)) < reg->anchor_dmin) {
- range = max_semi_end - reg->anchor_dmin + 1;
+ if (max_semi_end - (range - 1) < reg->anc_dist_min) {
+ if (max_semi_end - str + 1 < reg->anc_dist_min)
+ goto mismatch_no_msa;
+ else
+ range = max_semi_end - reg->anc_dist_min + 1;
}
if (start > range) goto mismatch_no_msa;
Backward search is used. */
}
else {
- if ((OnigLen )(min_semi_end - range) > reg->anchor_dmax) {
- range = min_semi_end - reg->anchor_dmax;
+ if (reg->anc_dist_max != INFINITE_LEN &&
+ min_semi_end - range > reg->anc_dist_max) {
+ range = min_semi_end - reg->anc_dist_max;
}
- if ((OnigLen )(max_semi_end - start) < reg->anchor_dmin) {
- start = max_semi_end - reg->anchor_dmin;
- start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start);
+ if (max_semi_end - start < reg->anc_dist_min) {
+ if (max_semi_end - str < reg->anc_dist_min)
+ goto mismatch_no_msa;
+ else {
+ start = max_semi_end - reg->anc_dist_min;
+ start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start);
+ }
}
if (range > start) goto mismatch_no_msa;
}
if (reg->optimize != OPTIMIZE_NONE) {
UChar *sch_range, *low, *high, *low_prev;
- sch_range = (UChar* )range;
- if (reg->dmax != 0) {
- if (reg->dmax == INFINITE_LEN)
+ if (reg->dist_max != 0) {
+ if (reg->dist_max == INFINITE_LEN)
sch_range = (UChar* )end;
else {
- sch_range += reg->dmax;
- if (sch_range > end) sch_range = (UChar* )end;
+ if ((end - range) < reg->dist_max)
+ sch_range = (UChar* )end;
+ else {
+ sch_range = (UChar* )range + reg->dist_max;
+ }
}
}
+ else
+ sch_range = (UChar* )range;
if ((end - start) < reg->threshold_len)
goto mismatch;
- if (reg->dmax != INFINITE_LEN) {
+ if (reg->dist_max != INFINITE_LEN) {
do {
- if (! forward_search_range(reg, str, end, s, sch_range,
- &low, &high, &low_prev)) goto mismatch;
+ if (! forward_search(reg, str, end, s, sch_range, &low, &high,
+ &low_prev)) goto mismatch;
if (s < low) {
s = low;
prev = low_prev;
}
while (s <= high) {
- MATCH_AND_RETURN_CHECK(orig_range);
+ MATCH_AND_RETURN_CHECK(data_range);
prev = s;
s += enclen(reg->enc, s);
}
goto mismatch;
}
else { /* check only. */
- if (! forward_search_range(reg, str, end, s, sch_range,
- &low, &high, (UChar** )NULL)) goto mismatch;
+ if (! forward_search(reg, str, end, s, sch_range, &low, &high,
+ (UChar** )NULL)) goto mismatch;
if ((reg->anchor & ANCR_ANYCHAR_INF) != 0) {
do {
- MATCH_AND_RETURN_CHECK(orig_range);
+ MATCH_AND_RETURN_CHECK(data_range);
prev = s;
s += enclen(reg->enc, s);
}
do {
- MATCH_AND_RETURN_CHECK(orig_range);
+ MATCH_AND_RETURN_CHECK(data_range);
prev = s;
s += enclen(reg->enc, s);
} while (s < range);
if (s == range) { /* because empty match with /$/. */
- MATCH_AND_RETURN_CHECK(orig_range);
+ MATCH_AND_RETURN_CHECK(data_range);
}
}
else { /* backward search */
if (reg->optimize != OPTIMIZE_NONE) {
UChar *low, *high, *adjrange, *sch_start;
+ const UChar *min_range;
+
+ if ((end - range) < reg->threshold_len) goto mismatch;
if (range < end)
adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range);
else
adjrange = (UChar* )end;
- if (reg->dmax != INFINITE_LEN &&
- (end - range) >= reg->threshold_len) {
+ if (end - range > reg->dist_min)
+ min_range = range + reg->dist_min;
+ else
+ min_range = end;
+
+ if (reg->dist_max != INFINITE_LEN) {
do {
- sch_start = s + reg->dmax;
- if (sch_start > end) sch_start = (UChar* )end;
- if (backward_search_range(reg, str, end, sch_start, range, adjrange,
- &low, &high) <= 0)
+ if (end - s > reg->dist_max)
+ sch_start = s + reg->dist_max;
+ else {
+ sch_start = onigenc_get_prev_char_head(reg->enc, str, end);
+ }
+
+ if (backward_search(reg, str, end, sch_start, min_range, adjrange,
+ &low, &high) <= 0)
goto mismatch;
if (s > high)
goto mismatch;
}
else { /* check only. */
- if ((end - range) < reg->threshold_len) goto mismatch;
+ sch_start = onigenc_get_prev_char_head(reg->enc, str, end);
- sch_start = s;
- if (reg->dmax != 0) {
- if (reg->dmax == INFINITE_LEN)
- sch_start = (UChar* )end;
- else {
- sch_start += reg->dmax;
- if (sch_start > end) sch_start = (UChar* )end;
- else
- sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc,
- start, sch_start);
- }
- }
- if (backward_search_range(reg, str, end, sch_start, range, adjrange,
- &low, &high) <= 0) goto mismatch;
+ if (backward_search(reg, str, end, sch_start, min_range, adjrange,
+ &low, &high) <= 0) goto mismatch;
}
}
return (int )(s - str);
}
+extern int
+onig_search_with_param(regex_t* reg, const UChar* str, const UChar* end,
+ const UChar* start, const UChar* range, OnigRegion* region,
+ OnigOptionType option, OnigMatchParam* mp)
+{
+ const UChar* data_range;
+
+ if (range > start)
+ data_range = range;
+ else
+ data_range = end;
+
+ return search_in_range(reg, str, end, start, range, data_range, region,
+ option, mp);
+}
+
extern int
onig_scan(regex_t* reg, const UChar* str, const UChar* end,
OnigRegion* region, OnigOptionType option,
*to = *from;
}
+extern int
+onig_regset_new(OnigRegSet** rset, int n, regex_t* regs[])
+{
+#define REGSET_INITIAL_ALLOC_SIZE 10
+
+ int i;
+ int r;
+ int alloc;
+ OnigRegSet* set;
+ RR* rs;
+
+ *rset = 0;
+
+ set = (OnigRegSet* )xmalloc(sizeof(*set));
+ CHECK_NULL_RETURN_MEMERR(set);
+
+ alloc = n > REGSET_INITIAL_ALLOC_SIZE ? n : REGSET_INITIAL_ALLOC_SIZE;
+ rs = (RR* )xmalloc(sizeof(set->rs[0]) * alloc);
+ if (IS_NULL(rs)) {
+ xfree(set);
+ return ONIGERR_MEMORY;
+ }
+
+ set->rs = rs;
+ set->n = 0;
+ set->alloc = alloc;
+
+ for (i = 0; i < n; i++) {
+ regex_t* reg = regs[i];
+
+ r = onig_regset_add(set, reg);
+ if (r != 0) {
+ for (i = 0; i < set->n; i++) {
+ OnigRegion* region = set->rs[i].region;
+ if (IS_NOT_NULL(region))
+ onig_region_free(region, 1);
+ }
+ xfree(set->rs);
+ xfree(set);
+ return r;
+ }
+ }
+
+ *rset = set;
+ return 0;
+}
+
+static void
+update_regset_by_reg(OnigRegSet* set, regex_t* reg)
+{
+ if (set->n == 1) {
+ set->enc = reg->enc;
+ set->anchor = reg->anchor;
+ set->anc_dmin = reg->anc_dist_min;
+ set->anc_dmax = reg->anc_dist_max;
+ set->all_low_high =
+ (reg->optimize == OPTIMIZE_NONE || reg->dist_max == INFINITE_LEN) ? 0 : 1;
+ set->anychar_inf = (reg->anchor & ANCR_ANYCHAR_INF) != 0 ? 1 : 0;
+ }
+ else {
+ int anchor;
+
+ anchor = set->anchor & reg->anchor;
+ if (anchor != 0) {
+ OnigLen anc_dmin;
+ OnigLen anc_dmax;
+
+ anc_dmin = set->anc_dmin;
+ anc_dmax = set->anc_dmax;
+ if (anc_dmin > reg->anc_dist_min) anc_dmin = reg->anc_dist_min;
+ if (anc_dmax < reg->anc_dist_max) anc_dmax = reg->anc_dist_max;
+ set->anc_dmin = anc_dmin;
+ set->anc_dmax = anc_dmax;
+ }
+
+ set->anchor = anchor;
+
+ if (reg->optimize == OPTIMIZE_NONE || reg->dist_max == INFINITE_LEN)
+ set->all_low_high = 0;
+
+ if ((reg->anchor & ANCR_ANYCHAR_INF) != 0)
+ set->anychar_inf = 1;
+ }
+}
+
+extern int
+onig_regset_add(OnigRegSet* set, regex_t* reg)
+{
+ OnigRegion* region;
+
+ if (IS_FIND_LONGEST(reg->options))
+ return ONIGERR_INVALID_ARGUMENT;
+
+ if (set->n != 0 && reg->enc != set->enc)
+ return ONIGERR_INVALID_ARGUMENT;
+
+ if (set->n >= set->alloc) {
+ RR* nrs;
+ int new_alloc;
+
+ new_alloc = set->alloc * 2;
+ nrs = (RR* )xrealloc(set->rs, sizeof(set->rs[0]) * new_alloc);
+ CHECK_NULL_RETURN_MEMERR(nrs);
+
+ set->rs = nrs;
+ set->alloc = new_alloc;
+ }
+
+ region = onig_region_new();
+ CHECK_NULL_RETURN_MEMERR(region);
+
+ set->rs[set->n].reg = reg;
+ set->rs[set->n].region = region;
+ set->n++;
+
+ update_regset_by_reg(set, reg);
+ return 0;
+}
+
+extern int
+onig_regset_replace(OnigRegSet* set, int at, regex_t* reg)
+{
+ int i;
+
+ if (at < 0 || at >= set->n)
+ return ONIGERR_INVALID_ARGUMENT;
+
+ if (IS_NULL(reg)) {
+ onig_region_free(set->rs[at].region, 1);
+ for (i = at; i < set->n - 1; i++) {
+ set->rs[i].reg = set->rs[i+1].reg;
+ set->rs[i].region = set->rs[i+1].region;
+ }
+ set->n--;
+ }
+ else {
+ if (IS_FIND_LONGEST(reg->options))
+ return ONIGERR_INVALID_ARGUMENT;
+
+ if (set->n > 1 && reg->enc != set->enc)
+ return ONIGERR_INVALID_ARGUMENT;
+
+ set->rs[at].reg = reg;
+ }
+
+ for (i = 0; i < set->n; i++)
+ update_regset_by_reg(set, set->rs[i].reg);
+
+ return 0;
+}
+
+extern void
+onig_regset_free(OnigRegSet* set)
+{
+ int i;
+
+ for (i = 0; i < set->n; i++) {
+ regex_t* reg;
+ OnigRegion* region;
+
+ reg = set->rs[i].reg;
+ region = set->rs[i].region;
+ onig_free(reg);
+ if (IS_NOT_NULL(region))
+ onig_region_free(region, 1);
+ }
+
+ xfree(set->rs);
+ xfree(set);
+}
+
+extern int
+onig_regset_number_of_regex(OnigRegSet* set)
+{
+ return set->n;
+}
+
+extern regex_t*
+onig_regset_get_regex(OnigRegSet* set, int at)
+{
+ if (at < 0 || at >= set->n)
+ return (regex_t* )0;
+
+ return set->rs[at].reg;
+}
+
+extern OnigRegion*
+onig_regset_get_region(OnigRegSet* set, int at)
+{
+ if (at < 0 || at >= set->n)
+ return (OnigRegion* )0;
+
+ return set->rs[at].region;
+}
+
+
#ifdef USE_DIRECT_THREADED_CODE
extern int
onig_init_for_match_at(regex_t* reg)
const UChar* str;
StackType* stk_base;
int i;
+ StackIndex* mem_start_stk;
+ StackIndex* mem_end_stk;
i = mem_num;
reg = a->regex;
str = a->string;
stk_base = a->stk_base;
+ mem_start_stk = a->mem_start_stk;
+ mem_end_stk = a->mem_end_stk;
if (i > 0) {
if (a->mem_end_stk[i] != INVALID_STACK_INDEX) {
- if (MEM_STATUS_AT(reg->bt_mem_start, i))
- *begin = (int )(STACK_AT(a->mem_start_stk[i])->u.mem.pstr - str);
- else
- *begin = (int )((UChar* )((void* )a->mem_start_stk[i]) - str);
-
- *end = (int )((MEM_STATUS_AT(reg->bt_mem_end, i)
- ? STACK_AT(a->mem_end_stk[i])->u.mem.pstr
- : (UChar* )((void* )a->mem_end_stk[i])) - str);
+ *begin = (int )(STACK_MEM_START(reg, i) - str);
+ *end = (int )(STACK_MEM_END(reg, i) - str);
}
else {
*begin = *end = ONIG_REGION_NOTPOS;
}
}
- else if (i == 0) {
-#if 0
- *begin = a->start - str;
- *end = a->current - str;
-#else
- return ONIGERR_INVALID_ARGUMENT;
-#endif
- }
else
return ONIGERR_INVALID_ARGUMENT;
return ONIG_MISMATCH;
}
-#if 0
-extern int
-onig_builtin_success(OnigCalloutArgs* args ARG_UNUSED, void* user_data ARG_UNUSED)
-{
- return ONIG_CALLOUT_SUCCESS;
-}
-#endif
-
extern int
onig_builtin_error(OnigCalloutArgs* args, void* user_data ARG_UNUSED)
{
regext.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
}
}
-#if 0
static void
conv_ext0le32(const UChar* s, const UChar* end, UChar* conv)
{
s += 2;
}
}
-#endif
static int
conv_encoding(OnigEncoding from, OnigEncoding to, const UChar* s, const UChar* end,
reggnu.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regint.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#endif
#endif
-#if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
- (defined(__ppc__) && defined(__APPLE__)) || \
- defined(__x86_64) || defined(__x86_64__) || \
- defined(__mc68020__)
-#define PLATFORM_UNALIGNED_WORD_ACCESS
-#endif
-
+#ifndef ONIG_DISABLE_DIRECT_THREADING
#ifdef __GNUC__
#define USE_GOTO_LABELS_AS_VALUES
#endif
+#endif
/* config */
/* spec. config */
#define USE_VARIABLE_META_CHARS
#define USE_POSIX_API_REGION_OPTION
#define USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
+/* #define USE_REPEAT_AND_EMPTY_CHECK_LOCAL_VAR */
+
#include "regenc.h"
#define CHAR_MAP_SIZE 256
#define INFINITE_LEN ONIG_INFINITE_DISTANCE
-#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
-
-#define PLATFORM_GET_INC(val,p,type) do{\
- val = *(type* )p;\
- (p) += sizeof(type);\
-} while(0)
-
-#else
-
-#define PLATFORM_GET_INC(val,p,type) do{\
- xmemcpy(&val, (p), sizeof(type));\
- (p) += sizeof(type);\
-} while(0)
-
-/* sizeof(OnigCodePoint) */
-#ifdef SIZEOF_SIZE_T
-# define WORD_ALIGNMENT_SIZE SIZEOF_SIZE_T
-#else
-# define WORD_ALIGNMENT_SIZE SIZEOF_LONG
-#endif
-
-#define GET_ALIGNMENT_PAD_SIZE(addr,pad_size) do {\
- (pad_size) = WORD_ALIGNMENT_SIZE - ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
- if ((pad_size) == WORD_ALIGNMENT_SIZE) (pad_size) = 0;\
-} while (0)
-
-#define ALIGNMENT_RIGHT(addr) do {\
- (addr) += (WORD_ALIGNMENT_SIZE - 1);\
- (addr) -= ((uintptr_t )(addr) % WORD_ALIGNMENT_SIZE);\
-} while (0)
-
-#endif /* PLATFORM_UNALIGNED_WORD_ACCESS */
-
#ifdef USE_CALLOUT
typedef struct {
- int flag;
- OnigCalloutOf of;
- int in;
- int name_id;
- const UChar* tag_start;
- const UChar* tag_end;
+ int flag;
+ OnigCalloutOf of;
+ int in;
+ int name_id;
+ const UChar* tag_start;
+ const UChar* tag_end;
OnigCalloutType type;
OnigCalloutFunc start_func;
OnigCalloutFunc end_func;
OPTIMIZE_STR, /* Slow Search */
OPTIMIZE_STR_FAST, /* Sunday quick search / BMH */
OPTIMIZE_STR_FAST_STEP_FORWARD, /* Sunday quick search / BMH */
- OPTIMIZE_STR_CASE_FOLD_FAST, /* Sunday quick search / BMH (ignore case) */
OPTIMIZE_STR_CASE_FOLD, /* Slow Search (ignore case) */
OPTIMIZE_MAP /* char map */
};
#define MEM_STATUS_AT0(stats,n) \
((n) > 0 && (n) < (int )MEM_STATUS_BITS_NUM ? ((stats) & ((MemStatusType )1 << n)) : ((stats) & 1))
+#define MEM_STATUS_IS_ALL_ON(stats) (((stats) & 1) != 0)
+
#define MEM_STATUS_ON(stats,n) do {\
if ((n) < (int )MEM_STATUS_BITS_NUM) {\
if ((n) != 0)\
(stats) |= ((MemStatusType )1 << (n));\
} while (0)
+#define MEM_STATUS_LIMIT_AT(stats,n) \
+ ((n) < (int )MEM_STATUS_BITS_NUM ? ((stats) & ((MemStatusType )1 << n)) : 0)
+#define MEM_STATUS_LIMIT_ON(stats,n) do {\
+ if ((n) < (int )MEM_STATUS_BITS_NUM && (n) != 0) {\
+ (stats) |= ((MemStatusType )1 << (n));\
+ }\
+} while (0)
-#define INT_MAX_LIMIT ((1UL << (SIZEOF_INT * 8 - 1)) - 1)
#define IS_CODE_WORD_ASCII(enc,code) \
(ONIGENC_IS_CODE_ASCII(code) && ONIGENC_IS_CODE_WORD(enc,code))
/* bitset */
#define BITS_PER_BYTE 8
#define SINGLE_BYTE_SIZE (1 << BITS_PER_BYTE)
-#define BITS_IN_ROOM (sizeof(Bits) * BITS_PER_BYTE)
+#define BITS_IN_ROOM 32 /* 4 * BITS_PER_BYTE */
#define BITSET_SIZE (SINGLE_BYTE_SIZE / BITS_IN_ROOM)
-#ifdef PLATFORM_UNALIGNED_WORD_ACCESS
-typedef unsigned int Bits;
-#else
-typedef unsigned char Bits;
-#endif
-typedef Bits BitSet[BITSET_SIZE];
-typedef Bits* BitSetRef;
+typedef uint32_t Bits;
+typedef Bits BitSet[BITSET_SIZE];
+typedef Bits* BitSetRef;
#define SIZE_BITSET sizeof(BitSet)
for (i = 0; i < (int )BITSET_SIZE; i++) { (bs)[i] = 0; } \
} while (0)
-#define BS_ROOM(bs,pos) (bs)[pos / BITS_IN_ROOM]
-#define BS_BIT(pos) (1 << (pos % BITS_IN_ROOM))
+#define BS_ROOM(bs,pos) (bs)[(unsigned int )(pos) >> 5]
+#define BS_BIT(pos) (1u << ((unsigned int )(pos) & 0x1f))
#define BITSET_AT(bs, pos) (BS_ROOM(bs,pos) & BS_BIT(pos))
#define BITSET_SET_BIT(bs, pos) BS_ROOM(bs,pos) |= BS_BIT(pos)
#define BB_INIT(buf,size) bbuf_init((BBuf* )(buf), (size))
+/*
#define BB_SIZE_INC(buf,inc) do{\
(buf)->alloc += (inc);\
(buf)->p = (UChar* )xrealloc((buf)->p, (buf)->alloc);\
if (IS_NULL((buf)->p)) return(ONIGERR_MEMORY);\
} while (0)
+*/
#define BB_EXPAND(buf,low) do{\
do { (buf)->alloc *= 2; } while ((buf)->alloc < (unsigned int )low);\
/* operation code */
enum OpCode {
- OP_FINISH = 0, /* matching process terminator (no more alternative) */
- OP_END = 1, /* pattern code terminator (success end) */
-
- OP_EXACT1 = 2, /* single byte, N = 1 */
- OP_EXACT2, /* single byte, N = 2 */
- OP_EXACT3, /* single byte, N = 3 */
- OP_EXACT4, /* single byte, N = 4 */
- OP_EXACT5, /* single byte, N = 5 */
- OP_EXACTN, /* single byte */
- OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
- OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
- OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
- OP_EXACTMB2N, /* mb-length = 2 */
- OP_EXACTMB3N, /* mb-length = 3 */
- OP_EXACTMBN, /* other length */
-
- OP_EXACT1_IC, /* single byte, N = 1, ignore case */
- OP_EXACTN_IC, /* single byte, ignore case */
-
+ OP_FINISH = 0, /* matching process terminator (no more alternative) */
+ OP_END = 1, /* pattern code terminator (success end) */
+ OP_STR_1 = 2, /* single byte, N = 1 */
+ OP_STR_2, /* single byte, N = 2 */
+ OP_STR_3, /* single byte, N = 3 */
+ OP_STR_4, /* single byte, N = 4 */
+ OP_STR_5, /* single byte, N = 5 */
+ OP_STR_N, /* single byte */
+ OP_STR_MB2N1, /* mb-length = 2 N = 1 */
+ OP_STR_MB2N2, /* mb-length = 2 N = 2 */
+ OP_STR_MB2N3, /* mb-length = 2 N = 3 */
+ OP_STR_MB2N, /* mb-length = 2 */
+ OP_STR_MB3N, /* mb-length = 3 */
+ OP_STR_MBN, /* other length */
+ OP_STR_1_IC, /* single byte, N = 1, ignore case */
+ OP_STR_N_IC, /* single byte, ignore case */
OP_CCLASS,
OP_CCLASS_MB,
OP_CCLASS_MIX,
OP_CCLASS_NOT,
OP_CCLASS_MB_NOT,
OP_CCLASS_MIX_NOT,
-
OP_ANYCHAR, /* "." */
OP_ANYCHAR_ML, /* "." multi-line */
OP_ANYCHAR_STAR, /* ".*" */
OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
OP_ANYCHAR_STAR_PEEK_NEXT,
OP_ANYCHAR_ML_STAR_PEEK_NEXT,
-
OP_WORD,
OP_WORD_ASCII,
OP_NO_WORD,
OP_NO_WORD_BOUNDARY,
OP_WORD_BEGIN,
OP_WORD_END,
-
OP_TEXT_SEGMENT_BOUNDARY,
-
OP_BEGIN_BUF,
OP_END_BUF,
OP_BEGIN_LINE,
OP_END_LINE,
OP_SEMI_END_BUF,
OP_BEGIN_POSITION,
-
OP_BACKREF1,
OP_BACKREF2,
OP_BACKREF_N,
OP_BACKREF_WITH_LEVEL_IC, /* \k<xxx+n>, \k<xxx-n> */
OP_BACKREF_CHECK, /* (?(n)), (?('name')) */
OP_BACKREF_CHECK_WITH_LEVEL, /* (?(n-level)), (?('name-level')) */
-
- OP_MEMORY_START,
- OP_MEMORY_START_PUSH, /* push back-tracker to stack */
- OP_MEMORY_END_PUSH, /* push back-tracker to stack */
- OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
- OP_MEMORY_END,
- OP_MEMORY_END_REC, /* push marker to stack */
-
+ OP_MEM_START,
+ OP_MEM_START_PUSH, /* push back-tracker to stack */
+ OP_MEM_END_PUSH, /* push back-tracker to stack */
+#ifdef USE_CALL
+ OP_MEM_END_PUSH_REC, /* push back-tracker to stack */
+#endif
+ OP_MEM_END,
+#ifdef USE_CALL
+ OP_MEM_END_REC, /* push marker to stack */
+#endif
OP_FAIL, /* pop stack and move */
OP_JUMP,
OP_PUSH,
OP_PUSH_SUPER,
OP_POP_OUT,
#ifdef USE_OP_PUSH_OR_JUMP_EXACT
- OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
+ OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
#endif
- OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
- OP_REPEAT, /* {n,m} */
- OP_REPEAT_NG, /* {n,m}? (non greedy) */
+ OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
+ OP_REPEAT, /* {n,m} */
+ OP_REPEAT_NG, /* {n,m}? (non greedy) */
OP_REPEAT_INC,
- OP_REPEAT_INC_NG, /* non greedy */
- OP_REPEAT_INC_SG, /* search and get in stack */
- OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
+ OP_REPEAT_INC_NG, /* non greedy */
OP_EMPTY_CHECK_START, /* null loop checker start */
OP_EMPTY_CHECK_END, /* null loop checker end */
OP_EMPTY_CHECK_END_MEMST, /* null loop checker end (with capture status) */
+#ifdef USE_CALL
OP_EMPTY_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
-
+#endif
OP_PREC_READ_START, /* (?=...) start */
OP_PREC_READ_END, /* (?=...) end */
OP_PREC_READ_NOT_START, /* (?!...) start */
OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
OP_LOOK_BEHIND_NOT_START, /* (?<!...) start */
OP_LOOK_BEHIND_NOT_END, /* (?<!...) end */
-
- OP_CALL, /* \g<name> */
- OP_RETURN,
OP_PUSH_SAVE_VAL,
OP_UPDATE_VAR,
+#ifdef USE_CALL
+ OP_CALL, /* \g<name> */
+ OP_RETURN,
+#endif
#ifdef USE_CALLOUT
OP_CALLOUT_CONTENTS, /* (?{...}) (?{{...}}) */
OP_CALLOUT_NAME, /* (*name) (*name[tag](args...)) */
};
enum SaveType {
- SAVE_KEEP = 0, /* SAVE S */
- SAVE_S = 1,
+ SAVE_KEEP = 0, /* SAVE S */
+ SAVE_S = 1,
SAVE_RIGHT_RANGE = 2,
};
#define SIZE_UPDATE_VAR_TYPE sizeof(UpdateVarType)
#define SIZE_MODE sizeof(ModeType)
-#define GET_RELADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, RelAddrType)
-#define GET_ABSADDR_INC(addr,p) PLATFORM_GET_INC(addr, p, AbsAddrType)
-#define GET_LENGTH_INC(len,p) PLATFORM_GET_INC(len, p, LengthType)
-#define GET_MEMNUM_INC(num,p) PLATFORM_GET_INC(num, p, MemNumType)
-#define GET_REPEATNUM_INC(num,p) PLATFORM_GET_INC(num, p, RepeatNumType)
-#define GET_OPTION_INC(option,p) PLATFORM_GET_INC(option, p, OnigOptionType)
-#define GET_POINTER_INC(ptr,p) PLATFORM_GET_INC(ptr, p, PointerType)
-#define GET_SAVE_TYPE_INC(type,p) PLATFORM_GET_INC(type, p, SaveType)
-#define GET_UPDATE_VAR_TYPE_INC(type,p) PLATFORM_GET_INC(type, p, UpdateVarType)
-#define GET_MODE_INC(mode,p) PLATFORM_GET_INC(mode, p, ModeType)
-
/* code point's address must be aligned address. */
#define GET_CODE_POINT(code,p) code = *((OnigCodePoint* )(p))
-#define GET_BYTE_INC(byte,p) do{\
- byte = *(p);\
- (p)++;\
-} while(0)
/* op-code + arg size */
-#if 0
-#define SIZE_OP_ANYCHAR_STAR SIZE_OPCODE
-#define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT (SIZE_OPCODE + 1)
-#define SIZE_OP_JUMP (SIZE_OPCODE + SIZE_RELADDR)
-#define SIZE_OP_PUSH (SIZE_OPCODE + SIZE_RELADDR)
-#define SIZE_OP_PUSH_SUPER (SIZE_OPCODE + SIZE_RELADDR)
-#define SIZE_OP_POP_OUT SIZE_OPCODE
-#ifdef USE_OP_PUSH_OR_JUMP_EXACT
-#define SIZE_OP_PUSH_OR_JUMP_EXACT1 (SIZE_OPCODE + SIZE_RELADDR + 1)
-#endif
-#define SIZE_OP_PUSH_IF_PEEK_NEXT (SIZE_OPCODE + SIZE_RELADDR + 1)
-#define SIZE_OP_REPEAT_INC (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_REPEAT_INC_NG (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_WORD_BOUNDARY (SIZE_OPCODE + SIZE_MODE)
-#define SIZE_OP_PREC_READ_START SIZE_OPCODE
-#define SIZE_OP_PREC_READ_NOT_START (SIZE_OPCODE + SIZE_RELADDR)
-#define SIZE_OP_PREC_READ_END SIZE_OPCODE
-#define SIZE_OP_PREC_READ_NOT_END SIZE_OPCODE
-#define SIZE_OP_FAIL SIZE_OPCODE
-#define SIZE_OP_MEMORY_START (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_MEMORY_START_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_MEMORY_END_PUSH (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_MEMORY_END_PUSH_REC (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_MEMORY_END (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_MEMORY_END_REC (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_ATOMIC_START SIZE_OPCODE
-#define SIZE_OP_ATOMIC_END SIZE_OPCODE
-#define SIZE_OP_EMPTY_CHECK_START (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_EMPTY_CHECK_END (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_LOOK_BEHIND (SIZE_OPCODE + SIZE_LENGTH)
-#define SIZE_OP_LOOK_BEHIND_NOT_START (SIZE_OPCODE + SIZE_RELADDR + SIZE_LENGTH)
-#define SIZE_OP_LOOK_BEHIND_NOT_END SIZE_OPCODE
-#define SIZE_OP_CALL (SIZE_OPCODE + SIZE_ABSADDR)
-#define SIZE_OP_RETURN SIZE_OPCODE
-#define SIZE_OP_PUSH_SAVE_VAL (SIZE_OPCODE + SIZE_SAVE_TYPE + SIZE_MEMNUM)
-#define SIZE_OP_UPDATE_VAR (SIZE_OPCODE + SIZE_UPDATE_VAR_TYPE + SIZE_MEMNUM)
-
-#ifdef USE_CALLOUT
-#define SIZE_OP_CALLOUT_CONTENTS (SIZE_OPCODE + SIZE_MEMNUM)
-#define SIZE_OP_CALLOUT_NAME (SIZE_OPCODE + SIZE_MEMNUM + SIZE_MEMNUM)
-#endif
-
-#else /* if 0 */
/* for relative address increment to go next op. */
-#define SIZE_INC_OP 1
-
-#define SIZE_OP_ANYCHAR_STAR 1
-#define SIZE_OP_ANYCHAR_STAR_PEEK_NEXT 1
-#define SIZE_OP_JUMP 1
-#define SIZE_OP_PUSH 1
-#define SIZE_OP_PUSH_SUPER 1
-#define SIZE_OP_POP_OUT 1
+#define SIZE_INC 1
+
+#define OPSIZE_ANYCHAR_STAR 1
+#define OPSIZE_ANYCHAR_STAR_PEEK_NEXT 1
+#define OPSIZE_JUMP 1
+#define OPSIZE_PUSH 1
+#define OPSIZE_PUSH_SUPER 1
+#define OPSIZE_POP_OUT 1
#ifdef USE_OP_PUSH_OR_JUMP_EXACT
-#define SIZE_OP_PUSH_OR_JUMP_EXACT1 1
-#endif
-#define SIZE_OP_PUSH_IF_PEEK_NEXT 1
-#define SIZE_OP_REPEAT 1
-#define SIZE_OP_REPEAT_INC 1
-#define SIZE_OP_REPEAT_INC_NG 1
-#define SIZE_OP_WORD_BOUNDARY 1
-#define SIZE_OP_PREC_READ_START 1
-#define SIZE_OP_PREC_READ_NOT_START 1
-#define SIZE_OP_PREC_READ_END 1
-#define SIZE_OP_PREC_READ_NOT_END 1
-#define SIZE_OP_BACKREF 1
-#define SIZE_OP_FAIL 1
-#define SIZE_OP_MEMORY_START 1
-#define SIZE_OP_MEMORY_START_PUSH 1
-#define SIZE_OP_MEMORY_END_PUSH 1
-#define SIZE_OP_MEMORY_END_PUSH_REC 1
-#define SIZE_OP_MEMORY_END 1
-#define SIZE_OP_MEMORY_END_REC 1
-#define SIZE_OP_ATOMIC_START 1
-#define SIZE_OP_ATOMIC_END 1
-#define SIZE_OP_EMPTY_CHECK_START 1
-#define SIZE_OP_EMPTY_CHECK_END 1
-#define SIZE_OP_LOOK_BEHIND 1
-#define SIZE_OP_LOOK_BEHIND_NOT_START 1
-#define SIZE_OP_LOOK_BEHIND_NOT_END 1
-#define SIZE_OP_CALL 1
-#define SIZE_OP_RETURN 1
-#define SIZE_OP_PUSH_SAVE_VAL 1
-#define SIZE_OP_UPDATE_VAR 1
+#define OPSIZE_PUSH_OR_JUMP_EXACT1 1
+#endif
+#define OPSIZE_PUSH_IF_PEEK_NEXT 1
+#define OPSIZE_REPEAT 1
+#define OPSIZE_REPEAT_INC 1
+#define OPSIZE_REPEAT_INC_NG 1
+#define OPSIZE_WORD_BOUNDARY 1
+#define OPSIZE_PREC_READ_START 1
+#define OPSIZE_PREC_READ_NOT_START 1
+#define OPSIZE_PREC_READ_END 1
+#define OPSIZE_PREC_READ_NOT_END 1
+#define OPSIZE_BACKREF 1
+#define OPSIZE_FAIL 1
+#define OPSIZE_MEM_START 1
+#define OPSIZE_MEM_START_PUSH 1
+#define OPSIZE_MEM_END_PUSH 1
+#define OPSIZE_MEM_END_PUSH_REC 1
+#define OPSIZE_MEM_END 1
+#define OPSIZE_MEM_END_REC 1
+#define OPSIZE_ATOMIC_START 1
+#define OPSIZE_ATOMIC_END 1
+#define OPSIZE_EMPTY_CHECK_START 1
+#define OPSIZE_EMPTY_CHECK_END 1
+#define OPSIZE_LOOK_BEHIND 1
+#define OPSIZE_LOOK_BEHIND_NOT_START 1
+#define OPSIZE_LOOK_BEHIND_NOT_END 1
+#define OPSIZE_CALL 1
+#define OPSIZE_RETURN 1
+#define OPSIZE_PUSH_SAVE_VAL 1
+#define OPSIZE_UPDATE_VAR 1
#ifdef USE_CALLOUT
-#define SIZE_OP_CALLOUT_CONTENTS 1
-#define SIZE_OP_CALLOUT_NAME 1
+#define OPSIZE_CALLOUT_CONTENTS 1
+#define OPSIZE_CALLOUT_NAME 1
#endif
-#endif /* if 0 */
#define MC_ESC(syn) (syn)->meta_char_table.esc
} repeat; /* REPEAT, REPEAT_NG */
struct {
MemNumType id;
- } repeat_inc; /* REPEAT_INC, REPEAT_INC_SG, REPEAT_INC_NG, REPEAT_INC_NG_SG */
+ } repeat_inc; /* REPEAT_INC, REPEAT_INC_NG */
struct {
MemNumType mem;
} empty_check_start;
#endif
} RegexExt;
+typedef struct {
+ int lower;
+ int upper;
+ union {
+ Operation* pcode; /* address of repeated body */
+ int offset;
+ } u;
+} RepeatRange;
+
struct re_pattern_buffer {
/* common members of BBuf(bytes-buffer) */
Operation* ops;
#ifdef USE_DIRECT_THREADED_CODE
enum OpCode* ocs;
#endif
- Operation* ops_curr;
- unsigned int ops_used; /* used space for ops */
- unsigned int ops_alloc; /* allocated space for ops */
+ Operation* ops_curr;
+ unsigned int ops_used; /* used space for ops */
+ unsigned int ops_alloc; /* allocated space for ops */
unsigned char* string_pool;
unsigned char* string_pool_end;
- int num_mem; /* used memory(...) num counted from 1 */
- int num_repeat; /* OP_REPEAT/OP_REPEAT_NG id-counter */
- int num_null_check; /* OP_EMPTY_CHECK_START/END id counter */
- int num_call; /* number of subexp call */
- unsigned int capture_history; /* (?@...) flag (1-31) */
- unsigned int bt_mem_start; /* need backtrack flag */
- unsigned int bt_mem_end; /* need backtrack flag */
- int stack_pop_level;
- int repeat_range_alloc;
- OnigRepeatRange* repeat_range;
-
- OnigEncoding enc;
- OnigOptionType options;
- OnigSyntaxType* syntax;
- OnigCaseFoldType case_fold_flag;
- void* name_table;
+ int num_mem; /* used memory(...) num counted from 1 */
+ int num_repeat; /* OP_REPEAT/OP_REPEAT_NG id-counter */
+ int num_empty_check; /* OP_EMPTY_CHECK_START/END id counter */
+ int num_call; /* number of subexp call */
+ MemStatusType capture_history; /* (?@...) flag (1-31) */
+ MemStatusType push_mem_start; /* need backtrack flag */
+ MemStatusType push_mem_end; /* need backtrack flag */
+ MemStatusType empty_status_mem;
+ int stack_pop_level;
+ int repeat_range_alloc;
+ RepeatRange* repeat_range;
+
+ OnigEncoding enc;
+ OnigOptionType options;
+ OnigSyntaxType* syntax;
+ OnigCaseFoldType case_fold_flag;
+ void* name_table;
/* optimization info (string search, char-map and anchors) */
int optimize; /* optimize flag */
int threshold_len; /* search str-length for apply optimize */
int anchor; /* BEGIN_BUF, BEGIN_POS, (SEMI_)END_BUF */
- OnigLen anchor_dmin; /* (SEMI_)END_BUF anchor distance */
- OnigLen anchor_dmax; /* (SEMI_)END_BUF anchor distance */
+ OnigLen anc_dist_min; /* (SEMI_)END_BUF anchor distance */
+ OnigLen anc_dist_max; /* (SEMI_)END_BUF anchor distance */
int sub_anchor; /* start-anchor for exact or map */
unsigned char *exact;
unsigned char *exact_end;
unsigned char map[CHAR_MAP_SIZE]; /* used as BMH skip or char-map */
int map_offset;
- OnigLen dmin; /* min-distance of exact or map */
- OnigLen dmax; /* max-distance of exact or map */
+ OnigLen dist_min; /* min-distance of exact or map */
+ OnigLen dist_max; /* max-distance of exact or map */
RegexExt* extp;
};
regparse.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return 0;
}
+#ifdef ONIG_DEBUG_PARSE
+#define INC_PARSE_DEPTH(d) do {\
+ (d)++;\
+ if (env->max_parse_depth < (d)) env->max_parse_depth = d;\
+ if ((d) > ParseDepthLimit) \
+ return ONIGERR_PARSE_DEPTH_LIMIT_OVER;\
+} while (0)
+#else
+#define INC_PARSE_DEPTH(d) do {\
+ (d)++;\
+ if ((d) > ParseDepthLimit) \
+ return ONIGERR_PARSE_DEPTH_LIMIT_OVER;\
+} while (0)
+#endif
+
+#define DEC_PARSE_DEPTH(d) (d)--
+
+
static int
bbuf_init(BBuf* buf, int size)
{
return 0;
}
-static int backref_rel_to_abs(int rel_no, ScanEnv* env)
+static int
+backref_rel_to_abs(int rel_no, ScanEnv* env)
{
if (rel_no > 0) {
return env->num_mem + rel_no;
}
}
-#if 0
-static void
-bitset_set_all(BitSetRef bs)
-{
- int i;
- for (i = 0; i < BITSET_SIZE; i++) { bs[i] = ~((Bits )0); }
-}
-#endif
-
static void
bitset_invert(BitSetRef bs)
{
{
int nid = env->save_num;
-#if 0
- if (IS_NULL(env->saves)) {
- int n = 10;
- env->saves = (SaveItem* )xmalloc(sizeof(SaveItem) * n);
- CHECK_NULL_RETURN_MEMERR(env->saves);
- env->save_alloc_num = n;
- }
- else if (env->save_alloc_num <= nid) {
- int n = env->save_alloc_num * 2;
- SaveItem* p = (SaveItem* )xrealloc(env->saves, sizeof(SaveItem) * n);
- CHECK_NULL_RETURN_MEMERR(p);
- env->saves = p;
- env->save_alloc_num = n;
- }
-
- env->saves[nid].type = type;
-#endif
-
env->save_num++;
*id = nid;
return 0;
c = ONIGENC_MBC_TO_CODE(enc, p, end); \
pfetch_prev = p; \
p += ONIGENC_MBC_ENC_LEN(enc, p); \
- if(UNEXPECTED(p > end)) p = end; \
} while (0)
#define PINC_S do { \
p += ONIGENC_MBC_ENC_LEN(enc, p); \
- if(UNEXPECTED(p > end)) p = end; \
} while (0)
#define PFETCH_S(c) do { \
c = ONIGENC_MBC_TO_CODE(enc, p, end); \
p += ONIGENC_MBC_ENC_LEN(enc, p); \
- if(UNEXPECTED(p > end)) p = end; \
} while (0)
#define PPEEK (p < end ? ONIGENC_MBC_TO_CODE(enc, p, end) : PEND_VALUE)
str_end_hash(st_str_end_key* x)
{
UChar *p;
- int val = 0;
+ unsigned val = 0;
p = x->s;
while (p < x->end) {
- val = val * 997 + (int )*p++;
+ val = val * 997 + (unsigned )*p++;
}
- return val + (val >> 5);
+ return (int) (val + (val >> 5));
}
extern hash_table_type*
callout_name_table_hash(st_callout_name_key* x)
{
UChar *p;
- int val = 0;
+ unsigned int val = 0;
p = x->s;
while (p < x->end) {
- val = val * 997 + (int )*p++;
+ val = val * 997 + (unsigned int )*p++;
}
/* use intptr_t for escape warning in Windows */
- return val + (val >> 5) + ((intptr_t )x->enc & 0xffff) + x->type;
+ return (int )(val + (val >> 5) + ((intptr_t )x->enc & 0xffff) + x->type);
}
extern hash_table_type*
static void
scan_env_clear(ScanEnv* env)
{
- MEM_STATUS_CLEAR(env->capture_history);
- MEM_STATUS_CLEAR(env->bt_mem_start);
- MEM_STATUS_CLEAR(env->bt_mem_end);
+ MEM_STATUS_CLEAR(env->cap_history);
+ MEM_STATUS_CLEAR(env->backtrack_mem);
MEM_STATUS_CLEAR(env->backrefed_mem);
env->error = (UChar* )NULL;
env->error_end = (UChar* )NULL;
xmemset(env->mem_env_static, 0, sizeof(env->mem_env_static));
env->parse_depth = 0;
+#ifdef ONIG_DEBUG_PARSE
+ env->max_parse_depth = 0;
+#endif
+ env->backref_num = 0;
env->keep_num = 0;
env->save_num = 0;
env->save_alloc_num = 0;
}
for (i = env->num_mem + 1; i < alloc; i++) {
- p[i].node = NULL_NODE;
-#if 0
- p[i].in = 0;
- p[i].recursion = 0;
-#endif
+ p[i].mem_node = NULL_NODE;
+ p[i].empty_repeat_node = NULL_NODE;
}
env->mem_env_dynamic = p;
scan_env_set_mem_node(ScanEnv* env, int num, Node* node)
{
if (env->num_mem >= num)
- SCANENV_MEMENV(env)[num].node = node;
+ SCANENV_MEMENV(env)[num].mem_node = node;
else
return ONIGERR_PARSER_BUG;
return 0;
static Node*
node_new_anychar(void)
{
- Node* node = node_new_ctype(CTYPE_ANYCHAR, 0, ONIG_OPTION_NONE);
+ Node* node = node_new_ctype(CTYPE_ANYCHAR, FALSE, ONIG_OPTION_NONE);
return node;
}
return node_new_list(left, right);
}
-extern Node*
-onig_node_list_add(Node* list, Node* x)
-{
- Node *n;
-
- n = onig_node_new_list(x, NULL);
- if (IS_NULL(n)) return NULL_NODE;
-
- if (IS_NOT_NULL(list)) {
- while (IS_NOT_NULL(NODE_CDR(list)))
- list = NODE_CDR(list);
-
- NODE_CDR(list) = n;
- }
-
- return n;
-}
-
extern Node*
onig_node_new_alt(Node* left, Node* right)
{
for (i = 0; i < back_num; i++) {
if (backrefs[i] <= env->num_mem &&
- IS_NULL(SCANENV_MEMENV(env)[backrefs[i]].node)) {
+ IS_NULL(SCANENV_MEMENV(env)[backrefs[i]].mem_node)) {
NODE_STATUS_ADD(node, RECURSION); /* /...(\1).../ */
break;
}
for (i = 0; i < back_num; i++)
p[i] = backrefs[i];
}
+
+ env->backref_num++;
return node;
}
CHECK_NULL_RETURN(node);
NODE_SET_TYPE(node, NODE_QUANT);
- QUANT_(node)->lower = lower;
- QUANT_(node)->upper = upper;
- QUANT_(node)->greedy = 1;
- QUANT_(node)->emptiness = BODY_IS_NOT_EMPTY;
- QUANT_(node)->head_exact = NULL_NODE;
- QUANT_(node)->next_head_exact = NULL_NODE;
- QUANT_(node)->is_refered = 0;
+ QUANT_(node)->lower = lower;
+ QUANT_(node)->upper = upper;
+ QUANT_(node)->greedy = 1;
+ QUANT_(node)->emptiness = BODY_IS_NOT_EMPTY;
+ QUANT_(node)->head_exact = NULL_NODE;
+ QUANT_(node)->next_head_exact = NULL_NODE;
+ QUANT_(node)->include_referred = 0;
if (by_number != 0)
NODE_STATUS_ADD(node, BY_NUMBER);
ns[1] = NULL_NODE;
r = ONIGERR_MEMORY;
- ns[0] = onig_node_new_anchor(ANCR_NO_TEXT_SEGMENT_BOUNDARY, 0);
+ ns[0] = onig_node_new_anchor(ANCR_NO_TEXT_SEGMENT_BOUNDARY, FALSE);
if (IS_NULL(ns[0])) goto err;
r = node_new_true_anychar(&ns[1], env);
ns[0] = x;
ns[1] = NULL_NODE;
- x = node_new_quantifier(0, INFINITE_REPEAT, 1);
+ x = node_new_quantifier(0, INFINITE_REPEAT, TRUE);
if (IS_NULL(x)) goto err;
NODE_BODY(x) = ns[0];
ns[0] = x;
- x = node_new_quantifier(lower, upper, 0);
+ x = node_new_quantifier(lower, upper, FALSE);
if (IS_NULL(x)) goto err0;
NODE_BODY(x) = ns[0];
x = make_alt(2, ns);
if (IS_NULL(x)) goto err0;
- if (is_range_cutter != 0)
+ if (is_range_cutter != FALSE)
NODE_STATUS_ADD(x, SUPER);
*node = x;
ns[0] = NULL_NODE; ns[1] = x;
- r = node_new_update_var_gimmick(&ns[0], UPDATE_VAR_RIGHT_RANGE_INIT, 0, env);
+#define ID_NOT_USED_DONT_CARE_ME 0
+
+ r = node_new_update_var_gimmick(&ns[0], UPDATE_VAR_RIGHT_RANGE_INIT,
+ ID_NOT_USED_DONT_CARE_ME, env);
if (r != 0) goto err;
x = make_alt(2, ns);
id1 = GIMMICK_(ns[0])->id;
r = make_absent_engine(&ns[1], id1, absent, body, lower, upper, possessive,
- 0, env);
+ FALSE, env);
if (r != 0) goto err;
ns[2] = ns[3] = NULL_NODE;
if (expr == NULL_NODE) {
/* default expr \O* */
- quant = node_new_quantifier(0, INFINITE_REPEAT, 0);
+ quant = node_new_quantifier(0, INFINITE_REPEAT, FALSE);
if (IS_NULL(quant)) goto err0;
r = node_new_true_anychar(&body, env);
return onig_node_str_cat(node, s, s + 1);
}
-extern void
-onig_node_conv_to_str_node(Node* node, int flag)
-{
- NODE_SET_TYPE(node, NODE_STRING);
- STR_(node)->flag = flag;
- STR_(node)->capacity = 0;
- STR_(node)->s = STR_(node)->buf;
- STR_(node)->end = STR_(node)->buf;
-}
-
extern void
onig_node_str_clear(Node* node)
{
xfree(STR_(node)->s);
}
- STR_(node)->capacity = 0;
STR_(node)->flag = 0;
STR_(node)->s = STR_(node)->buf;
STR_(node)->end = STR_(node)->buf;
+ STR_(node)->capacity = 0;
+ STR_(node)->case_min_len = 0;
}
static Node*
CHECK_NULL_RETURN(node);
NODE_SET_TYPE(node, NODE_STRING);
- STR_(node)->capacity = 0;
STR_(node)->flag = 0;
STR_(node)->s = STR_(node)->buf;
STR_(node)->end = STR_(node)->buf;
+ STR_(node)->capacity = 0;
+ STR_(node)->case_min_len = 0;
+
if (onig_node_str_cat(node, s, end)) {
onig_node_free(node);
return NULL;
}
static Node*
-node_new_str_raw(UChar* s, UChar* end)
+node_new_str_crude(UChar* s, UChar* end)
{
Node* node = node_new_str(s, end);
CHECK_NULL_RETURN(node);
- NODE_STRING_SET_RAW(node);
+ NODE_STRING_SET_CRUDE(node);
return node;
}
}
static Node*
-node_new_str_raw_char(UChar c)
+node_new_str_crude_char(UChar c)
{
int i;
UChar p[1];
Node* node;
p[0] = c;
- node = node_new_str_raw(p, p + 1);
+ node = node_new_str_crude(p, p + 1);
/* clear buf tail */
for (i = 1; i < NODE_STRING_BUF_SIZE; i++)
if (p && p > sn->s) { /* can be split. */
rn = node_new_str(p, sn->end);
CHECK_NULL_RETURN(rn);
- if (NODE_STRING_IS_RAW(node))
- NODE_STRING_SET_RAW(rn);
+ if (NODE_STRING_IS_CRUDE(node))
+ NODE_STRING_SET_CRUDE(rn);
sn->end = (UChar* )p;
}
return 0;
}
-extern int
-onig_scan_unsigned_number(UChar** src, const UChar* end, OnigEncoding enc)
+static int
+scan_number(UChar** src, const UChar* end, OnigEncoding enc)
{
- unsigned int num, val;
+ int num, val;
OnigCodePoint c;
UChar* p = *src;
PFETCH_READY;
while (! PEND) {
PFETCH(c);
if (IS_CODE_DIGIT_ASCII(enc, c)) {
- val = (unsigned int )DIGITVAL(c);
- if ((INT_MAX_LIMIT - val) / 10UL < num)
+ val = (int )DIGITVAL(c);
+ if ((INT_MAX - val) / 10 < num)
return -1; /* overflow */
num = num * 10 + val;
}
static int
-scan_unsigned_hexadecimal_number(UChar** src, UChar* end, int minlen,
- int maxlen, OnigEncoding enc)
+scan_hexadecimal_number(UChar** src, UChar* end, int minlen, int maxlen,
+ OnigEncoding enc, OnigCodePoint* rcode)
{
+ OnigCodePoint code;
OnigCodePoint c;
- unsigned int num, val;
+ unsigned int val;
int n;
UChar* p = *src;
PFETCH_READY;
- num = 0;
+ code = 0;
n = 0;
while (! PEND && n < maxlen) {
PFETCH(c);
if (IS_CODE_XDIGIT_ASCII(enc, c)) {
n++;
- val = (unsigned int )XDIGITVAL(enc,c);
- if ((INT_MAX_LIMIT - val) / 16UL < num)
+ val = (unsigned int )XDIGITVAL(enc, c);
+ if ((UINT_MAX - val) / 16UL < code)
return ONIGERR_TOO_BIG_NUMBER; /* overflow */
- num = (num << 4) + XDIGITVAL(enc,c);
+ code = (code << 4) + val;
}
else {
PUNFETCH;
if (n < minlen)
return ONIGERR_INVALID_CODE_POINT_VALUE;
+ *rcode = code;
*src = p;
- return num;
+ return ONIG_NORMAL;
}
static int
-scan_unsigned_octal_number(UChar** src, UChar* end, int maxlen,
- OnigEncoding enc)
+scan_octal_number(UChar** src, UChar* end, int minlen, int maxlen,
+ OnigEncoding enc, OnigCodePoint* rcode)
{
+ OnigCodePoint code;
OnigCodePoint c;
- unsigned int num, val;
+ unsigned int val;
+ int n;
UChar* p = *src;
PFETCH_READY;
- num = 0;
- while (! PEND && maxlen-- != 0) {
+ code = 0;
+ n = 0;
+ while (! PEND && n < maxlen) {
PFETCH(c);
if (IS_CODE_DIGIT_ASCII(enc, c) && c < '8') {
- val = ODIGITVAL(c);
- if ((INT_MAX_LIMIT - val) / 8UL < num)
- return -1; /* overflow */
+ n++;
+ val = (unsigned int )ODIGITVAL(c);
+ if ((UINT_MAX - val) / 8UL < code)
+ return ONIGERR_TOO_BIG_NUMBER; /* overflow */
- num = (num << 3) + val;
+ code = (code << 3) + val;
}
else {
PUNFETCH;
break;
}
}
+
+ if (n < minlen)
+ return ONIGERR_INVALID_CODE_POINT_VALUE;
+
+ *rcode = code;
*src = p;
- return num;
+ return ONIG_NORMAL;
}
{RQ_ASIS, RQ_PQ_Q, RQ_DEL, RQ_AQ, RQ_AQ, RQ_DEL} /* '+?' */
};
-extern void
-onig_reduce_nested_quantifier(Node* pnode, Node* cnode)
+extern int
+onig_reduce_nested_quantifier(Node* pnode)
{
int pnum, cnum;
QuantNode *p, *c;
+ Node* cnode;
+
+ cnode = NODE_BODY(pnode);
p = QUANT_(pnode);
c = QUANT_(cnode);
pnum = quantifier_type_num(p);
cnum = quantifier_type_num(c);
if (pnum < 0 || cnum < 0) {
- if ((p->lower == p->upper) && ! IS_INFINITE_REPEAT(p->upper)) {
- if ((c->lower == c->upper) && ! IS_INFINITE_REPEAT(c->upper)) {
- int n = onig_positive_int_multiply(p->lower, c->lower);
- if (n >= 0) {
- p->lower = p->upper = n;
- NODE_BODY(pnode) = NODE_BODY(cnode);
- goto remove_cnode;
- }
- }
+ if (p->lower == p->upper && c->lower == c->upper) {
+ int n = onig_positive_int_multiply(p->lower, c->lower);
+ if (n < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
+
+ p->lower = p->upper = n;
+ NODE_BODY(pnode) = NODE_BODY(cnode);
+ goto remove_cnode;
}
- return ;
+ return 0;
}
switch(ReduceTypeTable[cnum][pnum]) {
case RQ_DEL:
*pnode = *cnode;
+ goto remove_cnode;
break;
case RQ_A:
NODE_BODY(pnode) = NODE_BODY(cnode);
p->lower = 0; p->upper = INFINITE_REPEAT; p->greedy = 1;
+ goto remove_cnode;
break;
case RQ_AQ:
NODE_BODY(pnode) = NODE_BODY(cnode);
p->lower = 0; p->upper = INFINITE_REPEAT; p->greedy = 0;
+ goto remove_cnode;
break;
case RQ_QQ:
NODE_BODY(pnode) = NODE_BODY(cnode);
p->lower = 0; p->upper = 1; p->greedy = 0;
+ goto remove_cnode;
break;
case RQ_P_QQ:
- NODE_BODY(pnode) = cnode;
p->lower = 0; p->upper = 1; p->greedy = 0;
c->lower = 1; c->upper = INFINITE_REPEAT; c->greedy = 1;
- return ;
break;
case RQ_PQ_Q:
- NODE_BODY(pnode) = cnode;
p->lower = 0; p->upper = 1; p->greedy = 1;
c->lower = 1; c->upper = INFINITE_REPEAT; c->greedy = 0;
- return ;
break;
case RQ_ASIS:
- NODE_BODY(pnode) = cnode;
- return ;
break;
}
+ return 0;
+
remove_cnode:
NODE_BODY(cnode) = NULL_NODE;
onig_node_free(cnode);
+ return 0;
}
static int
alen = ONIGENC_CODE_TO_MBC(env->enc, 0x0a, buf + dlen);
if (alen < 0) return alen;
- crnl = node_new_str_raw(buf, buf + dlen + alen);
+ crnl = node_new_str_crude(buf, buf + dlen + alen);
CHECK_NULL_RETURN_MEMERR(crnl);
ncc = node_new_cclass();
if (r != 0) goto err1;
}
- x = node_new_bag_if_else(crnl, 0, ncc);
+ x = node_new_bag_if_else(crnl, NULL_NODE, ncc);
if (IS_NULL(x)) goto err1;
*node = x;
enum TokenSyms {
TK_EOT = 0, /* end of token */
- TK_RAW_BYTE = 1,
+ TK_CRUDE_BYTE = 1,
TK_CHAR,
TK_STRING,
TK_CODE_POINT,
TK_ALT,
TK_SUBEXP_OPEN,
TK_SUBEXP_CLOSE,
- TK_CC_OPEN,
+ TK_OPEN_CC,
TK_QUOTE_OPEN,
TK_CHAR_PROPERTY, /* \p{...}, \P{...} */
TK_KEEP, /* \K */
/* in cc */
TK_CC_CLOSE,
TK_CC_RANGE,
- TK_POSIX_BRACKET_OPEN,
- TK_CC_AND, /* && */
- TK_CC_CC_OPEN /* [ */
+ TK_CC_POSIX_BRACKET_OPEN,
+ TK_CC_AND, /* && */
+ TK_CC_OPEN_CC /* [ */
};
typedef struct {
UChar* backp;
union {
UChar* s;
- int c;
+ UChar byte;
OnigCodePoint code;
int anchor;
int subtype;
static int
-fetch_interval_quantifier(UChar** src, UChar* end, PToken* tok, ScanEnv* env)
+fetch_interval(UChar** src, UChar* end, PToken* tok, ScanEnv* env)
{
int low, up, syn_allow, non_low = 0;
int r = 0;
}
}
- low = onig_scan_unsigned_number(&p, end, env->enc);
+ low = scan_number(&p, end, env->enc);
if (low < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
if (low > ONIG_MAX_REPEAT_NUM)
return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
PFETCH(c);
if (c == ',') {
UChar* prev = p;
- up = onig_scan_unsigned_number(&p, end, env->enc);
+ up = scan_number(&p, end, env->enc);
if (up < 0) return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
if (up > ONIG_MAX_REPEAT_NUM)
return ONIGERR_TOO_BIG_NUMBER_FOR_REPEAT_RANGE;
if (PEND) goto invalid;
PFETCH(c);
if (IS_SYNTAX_OP(env->syntax, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) {
- if (c != MC_ESC(env->syntax)) goto invalid;
+ if (c != MC_ESC(env->syntax) || PEND) goto invalid;
PFETCH(c);
}
if (c != '}') goto invalid;
PFETCH(c);
if (! IS_CODE_DIGIT_ASCII(enc, c)) goto err;
PUNFETCH;
- level = onig_scan_unsigned_number(&p, end, enc);
+ level = scan_number(&p, end, enc);
if (level < 0) return ONIGERR_TOO_BIG_NUMBER;
*rlevel = (level * flag);
exist_level = 1;
end:
if (r == 0) {
if (*num_type != IS_NOT_NUM) {
- *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
+ *rback_num = scan_number(&pnum_head, name_end, enc);
if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
else if (*rback_num == 0) {
if (*num_type == IS_REL_NUM)
static int
fetch_name(OnigCodePoint start_code, UChar** src, UChar* end,
UChar** rname_end, ScanEnv* env, int* rback_num,
- enum REF_NUM* num_type, int ref)
+ enum REF_NUM* num_type, int is_ref)
{
int r, sign;
int digit_count;
return ONIGERR_EMPTY_GROUP_NAME;
if (IS_CODE_DIGIT_ASCII(enc, c)) {
- if (ref == 1)
+ if (is_ref == TRUE)
*num_type = IS_ABS_NUM;
else {
r = ONIGERR_INVALID_GROUP_NAME;
digit_count++;
}
else if (c == '-') {
- if (ref == 1) {
+ if (is_ref == TRUE) {
*num_type = IS_REL_NUM;
sign = -1;
pnum_head = p;
}
}
else if (c == '+') {
- if (ref == 1) {
+ if (is_ref == TRUE) {
*num_type = IS_REL_NUM;
sign = 1;
pnum_head = p;
}
if (*num_type != IS_NOT_NUM) {
- *rback_num = onig_scan_unsigned_number(&pnum_head, name_end, enc);
+ *rback_num = scan_number(&pnum_head, name_end, enc);
if (*rback_num < 0) return ONIGERR_TOO_BIG_NUMBER;
else if (*rback_num == 0) {
if (*num_type == IS_REL_NUM) {
static int
fetch_token_in_cc(PToken* tok, UChar** src, UChar* end, ScanEnv* env)
{
- int num;
+ int r;
+ OnigCodePoint code;
OnigCodePoint c, c2;
OnigSyntaxType* syn = env->syntax;
OnigEncoding enc = env->enc;
PFETCH(c);
tok->type = TK_CHAR;
tok->base = 0;
- tok->u.c = c;
+ tok->u.code = c;
tok->escaped = 0;
if (c == ']') {
PFETCH(c);
tok->escaped = 1;
- tok->u.c = c;
+ tok->u.code = c;
switch (c) {
case 'w':
tok->type = TK_CHAR_TYPE;
prev = p;
if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
PINC;
- num = scan_unsigned_octal_number(&p, end, 11, enc);
- if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
+ r = scan_octal_number(&p, end, 0, 11, enc, &code);
+ if (r < 0) return r;
if (!PEND) {
c2 = PPEEK;
if (IS_CODE_DIGIT_ASCII(enc, c2))
PINC;
tok->type = TK_CODE_POINT;
tok->base = 8;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
else {
/* can't read nothing or invalid format */
prev = p;
if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
PINC;
- num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
- if (num < 0) {
- if (num == ONIGERR_TOO_BIG_NUMBER)
- return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
- else
- return num;
- }
+ r = scan_hexadecimal_number(&p, end, 0, 8, enc, &code);
+ if (r < 0) return r;
if (!PEND) {
c2 = PPEEK;
if (IS_CODE_XDIGIT_ASCII(enc, c2))
PINC;
tok->type = TK_CODE_POINT;
tok->base = 16;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
else {
/* can't read nothing or invalid format */
}
}
else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
- num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
- if (num < 0) return num;
+ r = scan_hexadecimal_number(&p, end, 0, 2, enc, &code);
+ if (r < 0) return r;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
- tok->type = TK_RAW_BYTE;
+ tok->type = TK_CRUDE_BYTE;
tok->base = 16;
- tok->u.c = num;
+ tok->u.byte = (UChar )code;
}
break;
prev = p;
if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
- num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
- if (num < 0) return num;
+ r = scan_hexadecimal_number(&p, end, 4, 4, enc, &code);
+ if (r < 0) return r;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
tok->type = TK_CODE_POINT;
tok->base = 16;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
break;
if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
PUNFETCH;
prev = p;
- num = scan_unsigned_octal_number(&p, end, 3, enc);
- if (num < 0 || num >= 256) return ONIGERR_TOO_BIG_NUMBER;
+ r = scan_octal_number(&p, end, 0, 3, enc, &code);
+ if (r < 0) return r;
+ if (code >= 256) return ONIGERR_TOO_BIG_NUMBER;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
- tok->type = TK_RAW_BYTE;
+ tok->type = TK_CRUDE_BYTE;
tok->base = 8;
- tok->u.c = num;
+ tok->u.byte = (UChar )code;
}
break;
default:
PUNFETCH;
- num = fetch_escaped_value(&p, end, env, &c2);
- if (num < 0) return num;
- if (tok->u.c != c2) {
+ r = fetch_escaped_value(&p, end, env, &c2);
+ if (r < 0) return r;
+ if (tok->u.code != c2) {
tok->u.code = c2;
tok->type = TK_CODE_POINT;
}
PINC;
if (str_exist_check_with_esc(send, 2, p, end,
(OnigCodePoint )']', enc, syn)) {
- tok->type = TK_POSIX_BRACKET_OPEN;
+ tok->type = TK_CC_POSIX_BRACKET_OPEN;
}
else {
PUNFETCH;
else {
cc_in_cc:
if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_CCLASS_SET_OP)) {
- tok->type = TK_CC_CC_OPEN;
+ tok->type = TK_CC_OPEN_CC;
}
else {
CC_ESC_WARN(env, (UChar* )"[");
static int
fetch_token(PToken* tok, UChar** src, UChar* end, ScanEnv* env)
{
- int r, num;
+ int r;
+ OnigCodePoint code;
OnigCodePoint c;
OnigEncoding enc = env->enc;
OnigSyntaxType* syn = env->syntax;
tok->backp = p;
PFETCH(c);
- tok->u.c = c;
+ tok->u.code = c;
tok->escaped = 1;
switch (c) {
case '*':
case '{':
if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_BRACE_INTERVAL)) break;
- r = fetch_interval_quantifier(&p, end, tok, env);
+ r = fetch_interval(&p, end, tok, env);
if (r < 0) return r; /* error */
if (r == 0) goto greedy_check2;
else if (r == 2) { /* {n} */
prev = p;
if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_O_BRACE_OCTAL)) {
PINC;
- num = scan_unsigned_octal_number(&p, end, 11, enc);
- if (num < 0) return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
+ r = scan_octal_number(&p, end, 0, 11, enc, &code);
+ if (r < 0) return r;
if (!PEND) {
if (IS_CODE_DIGIT_ASCII(enc, PPEEK))
return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
if ((p > prev + enclen(enc, prev)) && !PEND && PPEEK_IS('}')) {
PINC;
tok->type = TK_CODE_POINT;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
else {
/* can't read nothing or invalid format */
prev = p;
if (PPEEK_IS('{') && IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_BRACE_HEX8)) {
PINC;
- num = scan_unsigned_hexadecimal_number(&p, end, 0, 8, enc);
- if (num < 0) {
- if (num == ONIGERR_TOO_BIG_NUMBER)
- return ONIGERR_TOO_BIG_WIDE_CHAR_VALUE;
- else
- return num;
- }
+ r = scan_hexadecimal_number(&p, end, 0, 8, enc, &code);
+ if (r < 0) return r;
if (!PEND) {
if (IS_CODE_XDIGIT_ASCII(enc, PPEEK))
return ONIGERR_TOO_LONG_WIDE_CHAR_VALUE;
if ((p > prev + enclen(enc, prev)) && !PEND && PPEEK_IS('}')) {
PINC;
tok->type = TK_CODE_POINT;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
else {
/* can't read nothing or invalid format */
}
}
else if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_X_HEX2)) {
- num = scan_unsigned_hexadecimal_number(&p, end, 0, 2, enc);
- if (num < 0) return num;
+ r = scan_hexadecimal_number(&p, end, 0, 2, enc, &code);
+ if (r < 0) return r;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
- tok->type = TK_RAW_BYTE;
+ tok->type = TK_CRUDE_BYTE;
tok->base = 16;
- tok->u.c = num;
+ tok->u.byte = (UChar )code;
}
break;
prev = p;
if (IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_U_HEX4)) {
- num = scan_unsigned_hexadecimal_number(&p, end, 4, 4, enc);
- if (num < 0) return num;
+ r = scan_hexadecimal_number(&p, end, 4, 4, enc, &code);
+ if (r < 0) return r;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
tok->type = TK_CODE_POINT;
tok->base = 16;
- tok->u.code = (OnigCodePoint )num;
+ tok->u.code = code;
}
break;
case '5': case '6': case '7': case '8': case '9':
PUNFETCH;
prev = p;
- num = onig_scan_unsigned_number(&p, end, enc);
- if (num < 0 || num > ONIG_MAX_BACKREF_NUM) {
+ r = scan_number(&p, end, enc);
+ if (r < 0 || r > ONIG_MAX_BACKREF_NUM) {
goto skip_backref;
}
if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_DECIMAL_BACKREF) &&
- (num <= env->num_mem || num <= 9)) { /* This spec. from GNU regex */
+ (r <= env->num_mem || r <= 9)) { /* This spec. from GNU regex */
if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
- if (num > env->num_mem || IS_NULL(SCANENV_MEMENV(env)[num].node))
+ if (r > env->num_mem || IS_NULL(SCANENV_MEMENV(env)[r].mem_node))
return ONIGERR_INVALID_BACKREF;
}
tok->type = TK_BACKREF;
tok->u.backref.num = 1;
- tok->u.backref.ref1 = num;
+ tok->u.backref.ref1 = r;
tok->u.backref.by_name = 0;
#ifdef USE_BACKREF_WITH_LEVEL
tok->u.backref.exist_level = 0;
case '0':
if (IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_OCTAL3)) {
prev = p;
- num = scan_unsigned_octal_number(&p, end, (c == '0' ? 2:3), enc);
- if (num < 0 || num >= 256) return ONIGERR_TOO_BIG_NUMBER;
+ r = scan_octal_number(&p, end, 0, (c == '0' ? 2:3), enc, &code);
+ if (r < 0 || r >= 256) return ONIGERR_TOO_BIG_NUMBER;
if (p == prev) { /* can't read nothing. */
- num = 0; /* but, it's not error */
+ code = 0; /* but, it's not error */
}
- tok->type = TK_RAW_BYTE;
+ tok->type = TK_CRUDE_BYTE;
tok->base = 8;
- tok->u.c = num;
+ tok->u.byte = (UChar )code;
}
else if (c != '0') {
PINC;
if (r == 1) tok->u.backref.exist_level = 1;
else tok->u.backref.exist_level = 0;
#else
- r = fetch_name(c, &p, end, &name_end, env, &back_num, &num_type, 1);
+ r = fetch_name(c, &p, end, &name_end, env, &back_num, &num_type, TRUE);
#endif
if (r < 0) return r;
if (IS_SYNTAX_BV(syn, ONIG_SYN_STRICT_CHECK_BACKREF)) {
if (back_num > env->num_mem ||
- IS_NULL(SCANENV_MEMENV(env)[back_num].node))
+ IS_NULL(SCANENV_MEMENV(env)[back_num].mem_node))
return ONIGERR_INVALID_BACKREF;
}
tok->type = TK_BACKREF;
tok->u.backref.ref1 = back_num;
}
else {
- num = name_to_group_numbers(env, prev, name_end, &backs);
+ int num = name_to_group_numbers(env, prev, name_end, &backs);
if (num <= 0) {
return ONIGERR_UNDEFINED_NAME_REFERENCE;
}
int i;
for (i = 0; i < num; i++) {
if (backs[i] > env->num_mem ||
- IS_NULL(SCANENV_MEMENV(env)[backs[i]].node))
+ IS_NULL(SCANENV_MEMENV(env)[backs[i]].mem_node))
return ONIGERR_INVALID_BACKREF;
}
}
prev = p;
r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env,
- &gnum, &num_type, 1);
+ &gnum, &num_type, TRUE);
if (r < 0) return r;
if (num_type != IS_NOT_NUM) {
OnigCodePoint c2;
PUNFETCH;
- num = fetch_escaped_value(&p, end, env, &c2);
- if (num < 0) return num;
- /* set_raw: */
- if (tok->u.c != c2) {
+ r = fetch_escaped_value(&p, end, env, &c2);
+ if (r < 0) return r;
+ if (tok->u.code != c2) {
tok->type = TK_CODE_POINT;
tok->u.code = c2;
}
else { /* string */
- int len;
- SAFE_ENC_LEN(enc, tok->backp, end, len);
- p = tok->backp + len;
+ p = tok->backp + enclen(enc, tok->backp);
}
}
break;
}
}
else {
- tok->u.c = c;
+ tok->u.code = c;
tok->escaped = 0;
#ifdef USE_VARIABLE_META_CHARS
case '{':
if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACE_INTERVAL)) break;
- r = fetch_interval_quantifier(&p, end, tok, env);
+ r = fetch_interval(&p, end, tok, env);
if (r < 0) return r; /* error */
if (r == 0) goto greedy_check2;
else if (r == 2) { /* {n} */
{
PINC;
name = p;
- r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env, &gnum,
- &num_type, 0);
+ r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env,
+ &gnum, &num_type, FALSE);
if (r < 0) return r;
tok->type = TK_CALL;
{
name = p;
r = fetch_name((OnigCodePoint )'(', &p, end, &name_end, env,
- &gnum, &num_type, 1);
+ &gnum, &num_type, TRUE);
if (r < 0) return r;
if (num_type == IS_NOT_NUM) {
case '[':
if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_BRACKET_CC)) break;
- tok->type = TK_CC_OPEN;
+ tok->type = TK_OPEN_CC;
break;
case ']':
int c, r;
int ascii_mode;
+ int is_single;
const OnigCodePoint *ranges;
OnigCodePoint limit;
OnigCodePoint sb_out;
}
r = 0;
+ is_single = ONIGENC_IS_SINGLEBYTE(enc);
limit = ascii_mode ? ASCII_LIMIT : SINGLE_BYTE_SIZE;
switch (ctype) {
case ONIGENC_CTYPE_ALNUM:
if (not != 0) {
for (c = 0; c < (int )limit; c++) {
- if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
- BITSET_SET_BIT(cc->bs, c);
+ if (is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1) {
+ if (! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
+ BITSET_SET_BIT(cc->bs, c);
+ }
}
for (c = limit; c < SINGLE_BYTE_SIZE; c++) {
- BITSET_SET_BIT(cc->bs, c);
+ if (is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1)
+ BITSET_SET_BIT(cc->bs, c);
}
- ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
+ if (is_single == 0)
+ ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
}
else {
for (c = 0; c < (int )limit; c++) {
- if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
- BITSET_SET_BIT(cc->bs, c);
+ if (is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1) {
+ if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
+ BITSET_SET_BIT(cc->bs, c);
+ }
}
}
break;
case ONIGENC_CTYPE_WORD:
if (not != 0) {
for (c = 0; c < (int )limit; c++) {
- if (ONIGENC_CODE_TO_MBCLEN(enc, c) > 0 /* check invalid code point */
+ /* check invalid code point */
+ if ((is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1)
&& ! ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
BITSET_SET_BIT(cc->bs, c);
}
for (c = limit; c < SINGLE_BYTE_SIZE; c++) {
- if (ONIGENC_CODE_TO_MBCLEN(enc, c) > 0)
+ if (is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1)
BITSET_SET_BIT(cc->bs, c);
}
+ if (ascii_mode != 0 && is_single == 0)
+ ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
}
else {
for (c = 0; c < (int )limit; c++) {
- if (ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
+ if ((is_single != 0 || ONIGENC_CODE_TO_MBCLEN(enc, c) == 1)
+ && ONIGENC_IS_CODE_CTYPE(enc, (OnigCodePoint )c, ctype))
BITSET_SET_BIT(cc->bs, c);
}
- if (ascii_mode == 0)
+ if (ascii_mode == 0 && is_single == 0)
ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
}
break;
{
int r;
OnigCodePoint c;
- OnigEncoding enc = env->enc;
- UChar *prev, *start, *p = *src;
+ OnigEncoding enc;
+ UChar *prev, *start, *p;
- r = 0;
+ p = *src;
+ enc = env->enc;
+ r = ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
start = prev = p;
while (!PEND) {
PFETCH_S(c);
if (c == '}') {
r = ONIGENC_PROPERTY_NAME_TO_CTYPE(enc, start, prev);
- if (r < 0) break;
+ if (r >= 0) {
+ *src = p;
+ }
+ else {
+ onig_scan_env_set_error_string(env, r, *src, prev);
+ }
- *src = p;
return r;
}
else if (c == '(' || c == ')' || c == '{' || c == '|') {
- r = ONIGERR_INVALID_CHAR_PROPERTY_NAME;
break;
}
}
- onig_scan_env_set_error_string(env, r, *src, prev);
return r;
}
*np = node_new_cclass();
CHECK_NULL_RETURN_MEMERR(*np);
cc = CCLASS_(*np);
- r = add_ctype_to_cc(cc, ctype, 0, env);
+ r = add_ctype_to_cc(cc, ctype, FALSE, env);
if (r != 0) return r;
if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
}
-enum CCSTATE {
- CCS_VALUE,
- CCS_RANGE,
- CCS_COMPLETE,
- CCS_START
-};
+typedef enum {
+ CS_VALUE,
+ CS_RANGE,
+ CS_COMPLETE,
+ CS_START
+} CSTATE;
-enum CCVALTYPE {
- CCV_SB,
- CCV_CODE_POINT,
- CCV_CLASS
-};
+typedef enum {
+ CV_UNDEF,
+ CV_SB,
+ CV_MB,
+ CV_CPROP
+} CVAL;
static int
-next_state_class(CClassNode* cc, OnigCodePoint* vs, enum CCVALTYPE* type,
- enum CCSTATE* state, ScanEnv* env)
+cc_cprop_next(CClassNode* cc, OnigCodePoint* pcode, CVAL* val, CSTATE* state,
+ ScanEnv* env)
{
int r;
- if (*state == CCS_RANGE)
+ if (*state == CS_RANGE)
return ONIGERR_CHAR_CLASS_VALUE_AT_END_OF_RANGE;
- if (*state == CCS_VALUE && *type != CCV_CLASS) {
- if (*type == CCV_SB)
- BITSET_SET_BIT(cc->bs, (int )(*vs));
- else if (*type == CCV_CODE_POINT) {
- r = add_code_range(&(cc->mbuf), env, *vs, *vs);
+ if (*state == CS_VALUE) {
+ if (*val == CV_SB)
+ BITSET_SET_BIT(cc->bs, (int )(*pcode));
+ else if (*val == CV_MB) {
+ r = add_code_range(&(cc->mbuf), env, *pcode, *pcode);
if (r < 0) return r;
}
}
- *state = CCS_VALUE;
- *type = CCV_CLASS;
+ *state = CS_VALUE;
+ *val = CV_CPROP;
return 0;
}
static int
-next_state_val(CClassNode* cc, OnigCodePoint *from, OnigCodePoint to,
- int* from_israw, int to_israw,
- enum CCVALTYPE intype, enum CCVALTYPE* type,
- enum CCSTATE* state, ScanEnv* env)
+cc_char_next(CClassNode* cc, OnigCodePoint *from, OnigCodePoint to,
+ int* from_raw, int to_raw, CVAL intype, CVAL* type,
+ CSTATE* state, ScanEnv* env)
{
int r;
switch (*state) {
- case CCS_VALUE:
- if (*type == CCV_SB) {
+ case CS_VALUE:
+ if (*type == CV_SB) {
if (*from > 0xff)
return ONIGERR_INVALID_CODE_POINT_VALUE;
BITSET_SET_BIT(cc->bs, (int )(*from));
}
- else if (*type == CCV_CODE_POINT) {
+ else if (*type == CV_MB) {
r = add_code_range(&(cc->mbuf), env, *from, *from);
if (r < 0) return r;
}
break;
- case CCS_RANGE:
+ case CS_RANGE:
if (intype == *type) {
- if (intype == CCV_SB) {
+ if (intype == CV_SB) {
if (*from > 0xff || to > 0xff)
return ONIGERR_INVALID_CODE_POINT_VALUE;
if (r < 0) return r;
}
ccs_range_end:
- *state = CCS_COMPLETE;
+ *state = CS_COMPLETE;
break;
- case CCS_COMPLETE:
- case CCS_START:
- *state = CCS_VALUE;
+ case CS_COMPLETE:
+ case CS_START:
+ *state = CS_VALUE;
break;
default:
break;
}
- *from_israw = to_israw;
- *from = to;
- *type = intype;
+ *from_raw = to_raw;
+ *from = to;
+ *type = intype;
return 0;
}
}
static int
-parse_char_class(Node** np, PToken* tok, UChar** src, UChar* end, ScanEnv* env)
+parse_cc(Node** np, PToken* tok, UChar** src, UChar* end, ScanEnv* env)
{
int r, neg, len, fetched, and_start;
- OnigCodePoint v, vs;
+ OnigCodePoint in_code, curr_code;
UChar *p;
Node* node;
CClassNode *cc, *prev_cc;
CClassNode work_cc;
-
- enum CCSTATE state;
- enum CCVALTYPE val_type, in_type;
- int val_israw, in_israw;
+ int curr_raw, in_raw;
+ CSTATE state;
+ CVAL in_type;
+ CVAL curr_type;
*np = NULL_NODE;
- env->parse_depth++;
- if (env->parse_depth > ParseDepthLimit)
- return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
+ INC_PARSE_DEPTH(env->parse_depth);
prev_cc = (CClassNode* )NULL;
r = fetch_token_in_cc(tok, src, end, env);
- if (r == TK_CHAR && tok->u.c == '^' && tok->escaped == 0) {
+ if (r == TK_CHAR && tok->u.code == (OnigCodePoint )'^' && tok->escaped == 0) {
neg = 1;
r = fetch_token_in_cc(tok, src, end, env);
}
cc = CCLASS_(node);
and_start = 0;
- state = CCS_START;
+ state = CS_START;
+ curr_type = CV_UNDEF;
+
p = *src;
while (r != TK_CC_CLOSE) {
fetched = 0;
switch (r) {
case TK_CHAR:
any_char_in:
- len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.c);
- if (len > 1) {
- in_type = CCV_CODE_POINT;
- }
- else if (len < 0) {
+ len = ONIGENC_CODE_TO_MBCLEN(env->enc, tok->u.code);
+ if (len < 0) {
r = len;
goto err;
}
- else {
- /* sb_char: */
- in_type = CCV_SB;
- }
- v = (OnigCodePoint )tok->u.c;
- in_israw = 0;
+ in_type = (len == 1) ? CV_SB : CV_MB;
+ in_code = tok->u.code;
+ in_raw = 0;
goto val_entry2;
break;
- case TK_RAW_BYTE:
+ case TK_CRUDE_BYTE:
/* tok->base != 0 : octal or hexadec. */
if (! ONIGENC_IS_SINGLEBYTE(env->enc) && tok->base != 0) {
int i, j;
UChar* psave = p;
int base = tok->base;
- buf[0] = tok->u.c;
+ buf[0] = tok->u.byte;
for (i = 1; i < ONIGENC_MBC_MAXLEN(env->enc); i++) {
r = fetch_token_in_cc(tok, &p, end, env);
if (r < 0) goto err;
- if (r != TK_RAW_BYTE || tok->base != base) {
+ if (r != TK_CRUDE_BYTE || tok->base != base) {
fetched = 1;
break;
}
- buf[i] = tok->u.c;
+ buf[i] = tok->u.byte;
}
if (i < ONIGENC_MBC_MINLEN(env->enc)) {
}
if (i == 1) {
- v = (OnigCodePoint )buf[0];
- goto raw_single;
+ in_code = (OnigCodePoint )buf[0];
+ goto crude_single;
}
else {
- v = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);
- in_type = CCV_CODE_POINT;
+ in_code = ONIGENC_MBC_TO_CODE(env->enc, buf, bufe);
+ in_type = CV_MB;
}
}
else {
- v = (OnigCodePoint )tok->u.c;
- raw_single:
- in_type = CCV_SB;
+ in_code = (OnigCodePoint )tok->u.byte;
+ crude_single:
+ in_type = CV_SB;
}
- in_israw = 1;
+ in_raw = 1;
goto val_entry2;
break;
case TK_CODE_POINT:
- v = tok->u.code;
- in_israw = 1;
+ in_code = tok->u.code;
+ in_raw = 1;
val_entry:
- len = ONIGENC_CODE_TO_MBCLEN(env->enc, v);
+ len = ONIGENC_CODE_TO_MBCLEN(env->enc, in_code);
if (len < 0) {
- if (state != CCS_RANGE ||
+ if (state != CS_RANGE ||
! IS_SYNTAX_BV(env->syntax,
ONIG_SYN_ALLOW_INVALID_CODE_END_OF_RANGE_IN_CC) ||
- v < 0x100 || ONIGENC_MBC_MAXLEN(env->enc) == 1) {
+ in_code < 0x100 || ONIGENC_MBC_MAXLEN(env->enc) == 1) {
r = len;
goto err;
}
}
- in_type = (len == 1 ? CCV_SB : CCV_CODE_POINT);
+ in_type = (len == 1 ? CV_SB : CV_MB);
val_entry2:
- r = next_state_val(cc, &vs, v, &val_israw, in_israw, in_type, &val_type,
- &state, env);
+ r = cc_char_next(cc, &curr_code, in_code, &curr_raw, in_raw, in_type,
+ &curr_type, &state, env);
if (r != 0) goto err;
break;
- case TK_POSIX_BRACKET_OPEN:
+ case TK_CC_POSIX_BRACKET_OPEN:
r = parse_posix_bracket(cc, &p, end, env);
if (r < 0) goto err;
if (r == 1) { /* is not POSIX bracket */
CC_ESC_WARN(env, (UChar* )"[");
p = tok->backp;
- v = (OnigCodePoint )tok->u.c;
- in_israw = 0;
+ in_code = tok->u.code;
+ in_raw = 0;
goto val_entry;
}
- goto next_class;
+ goto next_cprop;
break;
case TK_CHAR_TYPE:
r = add_ctype_to_cc(cc, tok->u.prop.ctype, tok->u.prop.not, env);
if (r != 0) goto err;
- next_class:
- r = next_state_class(cc, &vs, &val_type, &state, env);
+ next_cprop:
+ r = cc_cprop_next(cc, &curr_code, &curr_type, &state, env);
if (r != 0) goto err;
break;
}
r = add_ctype_to_cc(cc, ctype, tok->u.prop.not, env);
if (r != 0) goto err;
- goto next_class;
+ goto next_cprop;
}
break;
case TK_CC_RANGE:
- if (state == CCS_VALUE) {
+ if (state == CS_VALUE) {
r = fetch_token_in_cc(tok, &p, end, env);
if (r < 0) goto err;
+
fetched = 1;
if (r == TK_CC_CLOSE) { /* allow [x-] */
range_end_val:
- v = (OnigCodePoint )'-';
- in_israw = 0;
+ in_code = (OnigCodePoint )'-';
+ in_raw = 0;
goto val_entry;
}
else if (r == TK_CC_AND) {
goto range_end_val;
}
- if (val_type == CCV_CLASS) {
+ if (curr_type == CV_CPROP) {
r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
goto err;
}
- state = CCS_RANGE;
+ state = CS_RANGE;
}
- else if (state == CCS_START) {
+ else if (state == CS_START) {
/* [-xa] is allowed */
- v = (OnigCodePoint )tok->u.c;
- in_israw = 0;
+ in_code = tok->u.code;
+ in_raw = 0;
r = fetch_token_in_cc(tok, &p, end, env);
if (r < 0) goto err;
+
fetched = 1;
/* [--x] or [a&&-x] is warned. */
if (r == TK_CC_RANGE || and_start != 0)
goto val_entry;
}
- else if (state == CCS_RANGE) {
+ else if (state == CS_RANGE) {
CC_ESC_WARN(env, (UChar* )"-");
- goto any_char_in; /* [!--x] is allowed */
+ goto any_char_in; /* [!--] is allowed */
}
- else { /* CCS_COMPLETE */
+ else { /* CS_COMPLETE */
r = fetch_token_in_cc(tok, &p, end, env);
if (r < 0) goto err;
+
fetched = 1;
- if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */
+ if (r == TK_CC_CLOSE)
+ goto range_end_val; /* allow [a-b-] */
else if (r == TK_CC_AND) {
CC_ESC_WARN(env, (UChar* )"-");
goto range_end_val;
}
break;
- case TK_CC_CC_OPEN: /* [ */
+ case TK_CC_OPEN_CC: /* [ */
{
Node *anode;
CClassNode* acc;
- r = parse_char_class(&anode, tok, &p, end, env);
+ if (state == CS_VALUE) {
+ r = cc_char_next(cc, &curr_code, 0, &curr_raw, 0, curr_type, &curr_type,
+ &state, env);
+ if (r != 0) goto err;
+ }
+ state = CS_COMPLETE;
+
+ r = parse_cc(&anode, tok, &p, end, env);
if (r != 0) {
onig_node_free(anode);
goto cc_open_err;
case TK_CC_AND: /* && */
{
- if (state == CCS_VALUE) {
- r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,
- &val_type, &state, env);
+ if (state == CS_VALUE) {
+ r = cc_char_next(cc, &curr_code, 0, &curr_raw, 0, curr_type, &curr_type,
+ &state, env);
if (r != 0) goto err;
}
/* initialize local variables */
and_start = 1;
- state = CCS_START;
+ state = CS_START;
if (IS_NOT_NULL(prev_cc)) {
r = and_cclass(prev_cc, cc, env->enc);
}
}
- if (state == CCS_VALUE) {
- r = next_state_val(cc, &vs, 0, &val_israw, 0, val_type,
- &val_type, &state, env);
+ if (state == CS_VALUE) {
+ r = cc_char_next(cc, &curr_code, 0, &curr_raw, 0, curr_type, &curr_type,
+ &state, env);
if (r != 0) goto err;
}
}
}
*src = p;
- env->parse_depth--;
+ DEC_PARSE_DEPTH(env->parse_depth);
return 0;
err:
return r;
}
-static int parse_subexp(Node** top, PToken* tok, int term,
- UChar** src, UChar* end, ScanEnv* env, int group_head);
+static int parse_alts(Node** top, PToken* tok, int term,
+ UChar** src, UChar* end, ScanEnv* env, int group_head);
#ifdef USE_CALLOUT
static int
parse_callout_args(int skip_mode, int cterm, UChar** src, UChar* end,
- unsigned int types[], OnigValue vals[], ScanEnv* env)
+ int max_arg_num, unsigned int types[], OnigValue vals[],
+ ScanEnv* env)
{
#define MAX_CALLOUT_ARG_BYTE_LENGTH 128
if (PEND) return ONIGERR_INVALID_CALLOUT_PATTERN;
+ c = 0;
n = 0;
while (n < ONIG_CALLOUT_MAX_ARGS_NUM) {
- c = 0;
cn = 0;
esc = 0;
eesc = 0;
size_t clen;
add_char:
- if (skip_mode == 0) {
+ if (skip_mode == FALSE) {
clen = p - e;
if (bufend + clen > buf + MAX_CALLOUT_ARG_BYTE_LENGTH)
return ONIGERR_INVALID_CALLOUT_ARG; /* too long argument */
}
if (cn != 0) {
- if (skip_mode == 0) {
+ if (max_arg_num >= 0 && n >= max_arg_num)
+ return ONIGERR_INVALID_CALLOUT_ARG;
+
+ if (skip_mode == FALSE) {
if ((types[n] & ONIG_TYPE_LONG) != 0) {
int fixed = 0;
if (cn > 0) {
/* read for single check only */
save = p;
- arg_num = parse_callout_args(1, '}', &p, end, 0, 0, env);
+ arg_num = parse_callout_args(TRUE, '}', &p, end, -1, NULL, NULL, env);
if (arg_num < 0) return arg_num;
is_not_single = PPEEK_IS(cterm) ? 0 : 1;
types[i] = get_callout_arg_type_by_name_id(name_id, i);
}
- arg_num = parse_callout_args(0, '}', &p, end, types, vals, env);
+ arg_num = parse_callout_args(FALSE, '}', &p, end, max_arg_num, types, vals, env);
if (arg_num < 0) return arg_num;
if (PEND) return ONIGERR_END_PATTERN_IN_GROUP;
group:
r = fetch_token(tok, &p, end, env);
if (r < 0) return r;
- r = parse_subexp(np, tok, term, &p, end, env, 0);
+ r = parse_alts(np, tok, term, &p, end, env, FALSE);
if (r < 0) return r;
*src = p;
return 1; /* group */
break;
case '=':
- *np = onig_node_new_anchor(ANCR_PREC_READ, 0);
+ *np = onig_node_new_anchor(ANCR_PREC_READ, FALSE);
break;
case '!': /* preceding read */
- *np = onig_node_new_anchor(ANCR_PREC_READ_NOT, 0);
+ *np = onig_node_new_anchor(ANCR_PREC_READ_NOT, FALSE);
break;
case '>': /* (?>...) stop backtrack */
*np = node_new_bag(BAG_STOP_BACKTRACK);
if (PEND) return ONIGERR_END_PATTERN_WITH_UNMATCHED_PARENTHESIS;
PFETCH(c);
if (c == '=')
- *np = onig_node_new_anchor(ANCR_LOOK_BEHIND, 0);
+ *np = onig_node_new_anchor(ANCR_LOOK_BEHIND, FALSE);
else if (c == '!')
- *np = onig_node_new_anchor(ANCR_LOOK_BEHIND_NOT, 0);
+ *np = onig_node_new_anchor(ANCR_LOOK_BEHIND_NOT, FALSE);
else {
if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_QMARK_LT_NAMED_GROUP)) {
UChar *name;
named_group2:
name = p;
r = fetch_name((OnigCodePoint )c, &p, end, &name_end, env, &num,
- &num_type, 0);
+ &num_type, FALSE);
if (r < 0) return r;
num = scan_env_add_mem_entry(env);
CHECK_NULL_RETURN_MEMERR(*np);
BAG_(*np)->m.regnum = num;
if (list_capture != 0)
- MEM_STATUS_ON_SIMPLE(env->capture_history, num);
+ MEM_STATUS_ON_SIMPLE(env->cap_history, num);
env->num_named++;
}
else {
r = fetch_token(tok, &p, end, env);
if (r < 0) return r;
- r = parse_subexp(&absent, tok, term, &p, end, env, 1);
+ r = parse_alts(&absent, tok, term, &p, end, env, TRUE);
if (r < 0) {
onig_node_free(absent);
return r;
if (r == 1) exist_level = 1;
#else
r = fetch_name((OnigCodePoint )(is_enclosed != 0 ? c : '('),
- &p, end, &name_end, env, &back_num, &num_type, 1);
+ &p, end, &name_end, env, &back_num, &num_type, TRUE);
#endif
if (r < 0) {
if (is_enclosed == 0) {
if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_STRICT_CHECK_BACKREF)) {
if (back_num > env->num_mem ||
- IS_NULL(SCANENV_MEMENV(env)[back_num].node))
+ IS_NULL(SCANENV_MEMENV(env)[back_num].mem_node))
return ONIGERR_INVALID_BACKREF;
}
- condition = node_new_backref_checker(1, &back_num, 0,
+ condition = node_new_backref_checker(1, &back_num, FALSE,
#ifdef USE_BACKREF_WITH_LEVEL
exist_level, level,
#endif
int i;
for (i = 0; i < num; i++) {
if (backs[i] > env->num_mem ||
- IS_NULL(SCANENV_MEMENV(env)[backs[i]].node))
+ IS_NULL(SCANENV_MEMENV(env)[backs[i]].mem_node))
return ONIGERR_INVALID_BACKREF;
}
}
- condition = node_new_backref_checker(num, backs, 1,
+ condition = node_new_backref_checker(num, backs, TRUE,
#ifdef USE_BACKREF_WITH_LEVEL
exist_level, level,
#endif
condition_is_checker = 0;
r = fetch_token(tok, &p, end, env);
if (r < 0) return r;
- r = parse_subexp(&condition, tok, term, &p, end, env, 0);
+ r = parse_alts(&condition, tok, term, &p, end, env, FALSE);
if (r < 0) {
onig_node_free(condition);
return r;
onig_node_free(condition);
return r;
}
- r = parse_subexp(&target, tok, term, &p, end, env, 1);
+ r = parse_alts(&target, tok, term, &p, end, env, TRUE);
if (r < 0) {
onig_node_free(condition);
onig_node_free(target);
return ONIGERR_GROUP_NUMBER_OVER_FOR_CAPTURE_HISTORY;
}
BAG_(*np)->m.regnum = num;
- MEM_STATUS_ON_SIMPLE(env->capture_history, num);
+ MEM_STATUS_ON_SIMPLE(env->cap_history, num);
}
else {
return ONIGERR_UNDEFINED_GROUP_OPTION;
case 'm':
if (IS_SYNTAX_OP2(env->syntax, ONIG_SYN_OP2_OPTION_PERL)) {
- OPTION_NEGATE(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? 1 : 0));
+ OPTION_NEGATE(option, ONIG_OPTION_SINGLELINE, (neg == 0 ? TRUE : FALSE));
}
else if (IS_SYNTAX_OP2(env->syntax,
ONIG_SYN_OP2_OPTION_ONIGURUMA|ONIG_SYN_OP2_OPTION_RUBY)) {
if (! ONIGENC_IS_UNICODE_ENCODING(enc))
return ONIGERR_UNDEFINED_GROUP_OPTION;
- OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER, 0);
- OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_WORD, 1);
+ OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER, FALSE);
+ OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_WORD, TRUE);
break;
#ifdef USE_UNICODE_WORD_BREAK
case 'w':
if (! ONIGENC_IS_UNICODE_ENCODING(enc))
return ONIGERR_UNDEFINED_GROUP_OPTION;
- OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_WORD, 0);
- OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER, 1);
+ OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_WORD, FALSE);
+ OPTION_NEGATE(option, ONIG_OPTION_TEXT_SEGMENT_EXTENDED_GRAPHEME_CLUSTER, TRUE);
break;
#endif
default:
env->options = option;
r = fetch_token(tok, &p, end, env);
if (r < 0) return r;
- r = parse_subexp(&target, tok, term, &p, end, env, 0);
+ r = parse_alts(&target, tok, term, &p, end, env, FALSE);
env->options = prev;
if (r < 0) {
onig_node_free(target);
CHECK_NULL_RETURN_MEMERR(*np);
r = fetch_token(tok, &p, end, env);
if (r < 0) return r;
- r = parse_subexp(&target, tok, term, &p, end, env, 0);
+ r = parse_alts(&target, tok, term, &p, end, env, FALSE);
if (r < 0) {
onig_node_free(target);
return r;
if (NODE_TYPE(*np) == NODE_BAG) {
if (BAG_(*np)->type == BAG_MEMORY) {
- /* Don't move this to previous of parse_subexp() */
+ /* Don't move this to previous of parse_alts() */
r = scan_env_set_mem_node(env, BAG_(*np)->m.regnum, *np);
if (r != 0) return r;
}
};
static int
-set_quantifier(Node* qnode, Node* target, int group, ScanEnv* env)
+assign_quantifier_body(Node* qnode, Node* target, int group, ScanEnv* env)
{
QuantNode* qn;
}
}
else {
+ int r;
+
NODE_BODY(qnode) = target;
- onig_reduce_nested_quantifier(qnode, target);
- goto q_exit;
+ r = onig_reduce_nested_quantifier(qnode);
+ return r;
}
}
break;
}
NODE_BODY(qnode) = target;
- q_exit:
return 0;
}
}
#endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
+#define ADD_CODE_INTO_CC(cc, code, enc) do {\
+ if (ONIGENC_MBC_MINLEN(enc) > 1 || ONIGENC_CODE_TO_MBCLEN(enc, code) != 1) {\
+ add_code_range_to_buf(&((cc)->mbuf), code, code);\
+ }\
+ else {\
+ BITSET_SET_BIT((cc)->bs, code);\
+ }\
+} while (0)
+
+extern int
+onig_new_cclass_with_code_list(Node** rnode, OnigEncoding enc,
+ int n, OnigCodePoint codes[])
+{
+ int i;
+ Node* node;
+ CClassNode* cc;
+
+ *rnode = NULL_NODE;
+
+ node = node_new_cclass();
+ CHECK_NULL_RETURN_MEMERR(node);
+
+ cc = CCLASS_(node);
+
+ for (i = 0; i < n; i++) {
+ ADD_CODE_INTO_CC(cc, codes[i], enc);
+ }
+
+ *rnode = node;
+ return 0;
+}
+
typedef struct {
ScanEnv* env;
CClassNode* cc;
IApplyCaseFoldArg* iarg;
ScanEnv* env;
CClassNode* cc;
- BitSetRef bs;
iarg = (IApplyCaseFoldArg* )arg;
env = iarg->env;
cc = iarg->cc;
- bs = cc->bs;
if (to_len == 1) {
int is_in = onig_is_code_in_cc(env->enc, from, cc);
#ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
if ((is_in != 0 && !IS_NCCLASS_NOT(cc)) ||
(is_in == 0 && IS_NCCLASS_NOT(cc))) {
- if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
- add_code_range(&(cc->mbuf), env, *to, *to);
- }
- else {
- BITSET_SET_BIT(bs, *to);
- }
+ ADD_CODE_INTO_CC(cc, *to, env->enc);
}
#else
if (is_in != 0) {
- if (ONIGENC_MBC_MINLEN(env->enc) > 1 || *to >= SINGLE_BYTE_SIZE) {
+ if (ONIGENC_MBC_MINLEN(env->enc) > 1 ||
+ ONIGENC_CODE_TO_MBCLEN(env->enc, *to) != 1) {
if (IS_NCCLASS_NOT(cc)) clear_not_flag_cclass(cc, env->enc);
add_code_range(&(cc->mbuf), env, *to, *to);
}
else {
if (IS_NCCLASS_NOT(cc)) {
- BITSET_CLEAR_BIT(bs, *to);
+ BITSET_CLEAR_BIT(cc->bs, *to);
}
else
- BITSET_SET_BIT(bs, *to);
+ BITSET_SET_BIT(cc->bs, *to);
}
}
#endif /* CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS */
else {
int r, i, len;
UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
- Node *snode = NULL_NODE;
if (onig_is_code_in_cc(env->enc, from, cc)
#ifdef CASE_FOLD_IS_APPLIED_INSIDE_NEGATIVE_CCLASS
&& !IS_NCCLASS_NOT(cc)
#endif
) {
+ int n, j, m, index;
+ Node* list_node;
+ Node* ns[3];
+
+ n = 0;
for (i = 0; i < to_len; i++) {
- len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);
- if (i == 0) {
- snode = onig_node_new_str(buf, buf + len);
- CHECK_NULL_RETURN_MEMERR(snode);
-
- /* char-class expanded multi-char only
- compare with string folded at match time. */
- NODE_STRING_SET_AMBIG(snode);
+ OnigCodePoint code;
+ Node* csnode;
+ CClassNode* cs_cc;
+
+ index = onigenc_unicode_fold1_key(&to[i]);
+ if (index >= 0) {
+ csnode = node_new_cclass();
+ cs_cc = CCLASS_(csnode);
+ if (IS_NULL(csnode)) {
+ err_free_ns:
+ for (j = 0; j < n; j++) onig_node_free(ns[j]);
+ return ONIGERR_MEMORY;
+ }
+ m = FOLDS1_UNFOLDS_NUM(index);
+ for (j = 0; j < m; j++) {
+ code = FOLDS1_UNFOLDS(index)[j];
+ ADD_CODE_INTO_CC(cs_cc, code, env->enc);
+ }
+ ADD_CODE_INTO_CC(cs_cc, to[i], env->enc);
+ ns[n++] = csnode;
}
else {
- r = onig_node_str_cat(snode, buf, buf + len);
- if (r < 0) {
- onig_node_free(snode);
- return r;
+ len = ONIGENC_CODE_TO_MBC(env->enc, to[i], buf);
+ if (n == 0 || NODE_TYPE(ns[n-1]) != NODE_STRING) {
+ csnode = onig_node_new_str(buf, buf + len);
+ if (IS_NULL(csnode)) goto err_free_ns;
+
+ NODE_STRING_SET_CASE_EXPANDED(csnode);
+ ns[n++] = csnode;
+ }
+ else {
+ r = onig_node_str_cat(ns[n-1], buf, buf + len);
+ if (r < 0) goto err_free_ns;
}
}
}
- *(iarg->ptail) = onig_node_new_alt(snode, NULL_NODE);
- CHECK_NULL_RETURN_MEMERR(*(iarg->ptail));
+ if (n == 1)
+ list_node = ns[0];
+ else
+ list_node = make_list(n, ns);
+
+ *(iarg->ptail) = onig_node_new_alt(list_node, NULL_NODE);
+ if (IS_NULL(*(iarg->ptail))) {
+ onig_node_free(list_node);
+ return ONIGERR_MEMORY;
+ }
iarg->ptail = &(NODE_CDR((*(iarg->ptail))));
}
}
env->options = BAG_(*np)->o.options;
r = fetch_token(tok, src, end, env);
if (r < 0) return r;
- r = parse_subexp(&target, tok, term, src, end, env, 0);
+ r = parse_alts(&target, tok, term, src, end, env, FALSE);
env->options = prev;
if (r < 0) {
onig_node_free(target);
if (! IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_UNMATCHED_CLOSE_SUBEXP))
return ONIGERR_UNMATCHED_CLOSE_PARENTHESIS;
- if (tok->escaped) goto tk_raw_byte;
+ if (tok->escaped) goto tk_crude_byte;
else goto tk_byte;
break;
}
break;
- case TK_RAW_BYTE:
- tk_raw_byte:
+ case TK_CRUDE_BYTE:
+ tk_crude_byte:
{
- *np = node_new_str_raw_char((UChar )tok->u.c);
+ *np = node_new_str_crude_char(tok->u.byte);
CHECK_NULL_RETURN_MEMERR(*np);
len = 1;
while (1) {
if (len >= ONIGENC_MBC_MINLEN(env->enc)) {
if (len == enclen(env->enc, STR_(*np)->s)) {
r = fetch_token(tok, src, end, env);
- goto tk_raw_byte_end;
+ goto tk_crude_byte_end;
}
}
r = fetch_token(tok, src, end, env);
if (r < 0) return r;
- if (r != TK_RAW_BYTE)
+ if (r != TK_CRUDE_BYTE)
return ONIGERR_TOO_SHORT_MULTI_BYTE_STRING;
- r = node_str_cat_char(*np, (UChar )tok->u.c);
+ r = node_str_cat_char(*np, tok->u.byte);
if (r < 0) return r;
len++;
}
- tk_raw_byte_end:
+ tk_crude_byte_end:
if (! ONIGENC_IS_VALID_MBC_STRING(env->enc, STR_(*np)->s, STR_(*np)->end))
return ONIGERR_INVALID_WIDE_CHAR_VALUE;
- NODE_STRING_CLEAR_RAW(*np);
+ NODE_STRING_CLEAR_CRUDE(*np);
goto string_end;
}
break;
len = ONIGENC_CODE_TO_MBC(env->enc, tok->u.code, buf);
if (len < 0) return len;
#ifdef NUMBERED_CHAR_IS_NOT_CASE_AMBIG
- *np = node_new_str_raw(buf, buf + len);
+ *np = node_new_str_crude(buf, buf + len);
#else
*np = node_new_str(buf, buf + len);
#endif
*np = node_new_cclass();
CHECK_NULL_RETURN_MEMERR(*np);
cc = CCLASS_(*np);
- add_ctype_to_cc(cc, tok->u.prop.ctype, 0, env);
+ add_ctype_to_cc(cc, tok->u.prop.ctype, FALSE, env);
if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
}
break;
if (r != 0) return r;
break;
- case TK_CC_OPEN:
+ case TK_OPEN_CC:
{
CClassNode* cc;
- r = parse_char_class(np, tok, src, end, env);
+ r = parse_cc(np, tok, src, end, env);
if (r != 0) return r;
cc = CCLASS_(*np);
case TK_ANYCHAR_ANYTIME:
*np = node_new_anychar();
CHECK_NULL_RETURN_MEMERR(*np);
- qn = node_new_quantifier(0, INFINITE_REPEAT, 0);
+ qn = node_new_quantifier(0, INFINITE_REPEAT, FALSE);
CHECK_NULL_RETURN_MEMERR(qn);
NODE_BODY(qn) = *np;
*np = qn;
if (is_invalid_quantifier_target(*tp))
return ONIGERR_TARGET_OF_REPEAT_OPERATOR_INVALID;
- parse_depth++;
- if (parse_depth > ParseDepthLimit)
- return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
+ INC_PARSE_DEPTH(parse_depth);
qn = node_new_quantifier(tok->u.repeat.lower, tok->u.repeat.upper,
r == TK_INTERVAL);
else {
target = *tp;
}
- r = set_quantifier(qn, target, group, env);
+ r = assign_quantifier_body(qn, target, group, env);
if (r < 0) {
onig_node_free(qn);
+ *tp = NULL_NODE;
return r;
}
Node *node, **headp;
*top = NULL;
+ INC_PARSE_DEPTH(env->parse_depth);
+
r = parse_exp(&node, tok, term, src, end, env, group_head);
if (r < 0) {
onig_node_free(node);
*top = node;
}
else {
- *top = node_new_list(node, NULL);
+ *top = node_new_list(node, NULL);
if (IS_NULL(*top)) {
onig_node_free(node);
return ONIGERR_MEMORY;
headp = &(NODE_CDR(*top));
while (r != TK_EOT && r != term && r != TK_ALT) {
- r = parse_exp(&node, tok, term, src, end, env, 0);
+ r = parse_exp(&node, tok, term, src, end, env, FALSE);
if (r < 0) {
onig_node_free(node);
return r;
}
}
+ DEC_PARSE_DEPTH(env->parse_depth);
return r;
}
/* term_tok: TK_EOT or TK_SUBEXP_CLOSE */
static int
-parse_subexp(Node** top, PToken* tok, int term, UChar** src, UChar* end,
- ScanEnv* env, int group_head)
+parse_alts(Node** top, PToken* tok, int term, UChar** src, UChar* end,
+ ScanEnv* env, int group_head)
{
int r;
Node *node, **headp;
*top = NULL;
- env->parse_depth++;
- if (env->parse_depth > ParseDepthLimit)
- return ONIGERR_PARSE_DEPTH_LIMIT_OVER;
+ INC_PARSE_DEPTH(env->parse_depth);
r = parse_branch(&node, tok, term, src, end, env, group_head);
if (r < 0) {
while (r == TK_ALT) {
r = fetch_token(tok, src, end, env);
if (r < 0) return r;
- r = parse_branch(&node, tok, term, src, end, env, 0);
+ r = parse_branch(&node, tok, term, src, end, env, FALSE);
if (r < 0) {
onig_node_free(node);
return r;
return ONIGERR_PARSER_BUG;
}
- env->parse_depth--;
+ DEC_PARSE_DEPTH(env->parse_depth);
return r;
}
r = fetch_token(&tok, src, end, env);
if (r < 0) return r;
- r = parse_subexp(top, &tok, TK_EOT, src, end, env, 0);
+ r = parse_alts(top, &tok, TK_EOT, src, end, env, FALSE);
if (r < 0) return r;
return 0;
regparse.h - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
#include "regint.h"
#define NODE_STRING_MARGIN 16
-#define NODE_STRING_BUF_SIZE 24 /* sizeof(CClassNode) - sizeof(int)*4 */
+#define NODE_STRING_BUF_SIZE 20 /* sizeof(CClassNode) - sizeof(int)*4 */
#define NODE_BACKREFS_SIZE 6
/* node type */
BODY_IS_EMPTY_POSSIBILITY_REC = 3
};
+struct _Node;
+
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
UChar* s;
UChar* end;
unsigned int flag;
- int capacity; /* (allocated size - 1) or 0: use buf[] */
UChar buf[NODE_STRING_BUF_SIZE];
+ int capacity; /* (allocated size - 1) or 0: use buf[] */
+ int case_min_len;
} StrNode;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
unsigned int flags;
BitSet bs;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* body;
int lower;
enum BodyEmptyType emptiness;
struct _Node* head_exact;
struct _Node* next_head_exact;
- int is_refered; /* include called node. don't eliminate even if {0} */
+ int include_referred; /* include called node. don't eliminate even if {0} */
} QuantNode;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* body;
enum BagType type;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* body; /* to BagNode : BAG_MEMORY */
int by_number;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
int back_num;
int back_static[NODE_BACKREFS_SIZE];
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* body;
int type;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* car;
struct _Node* cdr;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
int ctype;
int not;
typedef struct {
NodeType node_type;
int status;
+ struct _Node* parent;
enum GimmickType type;
int detail_type;
struct {
NodeType node_type;
int status;
+ struct _Node* parent;
struct _Node* body;
} base;
#define ANCR_ANYCHAR_INF_MASK (ANCR_ANYCHAR_INF | ANCR_ANYCHAR_INF_ML)
#define ANCR_END_BUF_MASK (ANCR_END_BUF | ANCR_SEMI_END_BUF)
-#define NODE_STRING_RAW (1<<0) /* by backslashed number */
-#define NODE_STRING_AMBIG (1<<1)
-#define NODE_STRING_GOOD_AMBIG (1<<2)
-#define NODE_STRING_DONT_GET_OPT_INFO (1<<3)
+#define NODE_STRING_CRUDE (1<<0)
+#define NODE_STRING_CASE_EXPANDED (1<<1)
+#define NODE_STRING_CASE_FOLD_MATCH (1<<2)
#define NODE_STRING_LEN(node) (int )((node)->u.str.end - (node)->u.str.s)
-#define NODE_STRING_SET_RAW(node) (node)->u.str.flag |= NODE_STRING_RAW
-#define NODE_STRING_CLEAR_RAW(node) (node)->u.str.flag &= ~NODE_STRING_RAW
-#define NODE_STRING_SET_AMBIG(node) (node)->u.str.flag |= NODE_STRING_AMBIG
-#define NODE_STRING_SET_GOOD_AMBIG(node) (node)->u.str.flag |= NODE_STRING_GOOD_AMBIG
-#define NODE_STRING_SET_DONT_GET_OPT_INFO(node) \
- (node)->u.str.flag |= NODE_STRING_DONT_GET_OPT_INFO
-#define NODE_STRING_IS_RAW(node) \
- (((node)->u.str.flag & NODE_STRING_RAW) != 0)
-#define NODE_STRING_IS_AMBIG(node) \
- (((node)->u.str.flag & NODE_STRING_AMBIG) != 0)
-#define NODE_STRING_IS_GOOD_AMBIG(node) \
- (((node)->u.str.flag & NODE_STRING_GOOD_AMBIG) != 0)
-#define NODE_STRING_IS_DONT_GET_OPT_INFO(node) \
- (((node)->u.str.flag & NODE_STRING_DONT_GET_OPT_INFO) != 0)
+#define NODE_STRING_SET_CRUDE(node) (node)->u.str.flag |= NODE_STRING_CRUDE
+#define NODE_STRING_CLEAR_CRUDE(node) (node)->u.str.flag &= ~NODE_STRING_CRUDE
+#define NODE_STRING_SET_CASE_EXPANDED(node) (node)->u.str.flag |= NODE_STRING_CASE_EXPANDED
+#define NODE_STRING_SET_CASE_FOLD_MATCH(node) (node)->u.str.flag |= NODE_STRING_CASE_FOLD_MATCH
+#define NODE_STRING_IS_CRUDE(node) \
+ (((node)->u.str.flag & NODE_STRING_CRUDE) != 0)
+#define NODE_STRING_IS_CASE_EXPANDED(node) \
+ (((node)->u.str.flag & NODE_STRING_CASE_EXPANDED) != 0)
+#define NODE_STRING_IS_CASE_FOLD_MATCH(node) \
+ (((node)->u.str.flag & NODE_STRING_CASE_FOLD_MATCH) != 0)
#define BACKREFS_P(br) \
(IS_NOT_NULL((br)->back_dynamic) ? (br)->back_dynamic : (br)->back_static)
#define NODE_ST_FIXED_OPTION (1<<18)
#define NODE_ST_PROHIBIT_RECURSION (1<<19)
#define NODE_ST_SUPER (1<<20)
+#define NODE_ST_EMPTY_STATUS_CHECK (1<<21)
#define NODE_STATUS(node) (((Node* )node)->u.base.status)
((NODE_STATUS(node) & NODE_ST_PROHIBIT_RECURSION) != 0)
#define NODE_IS_STRICT_REAL_REPEAT(node) \
((NODE_STATUS(node) & NODE_ST_STRICT_REAL_REPEAT) != 0)
+#define NODE_IS_EMPTY_STATUS_CHECK(node) \
+ ((NODE_STATUS(node) & NODE_ST_EMPTY_STATUS_CHECK) != 0)
+#define NODE_PARENT(node) ((node)->u.base.parent)
#define NODE_BODY(node) ((node)->u.base.body)
#define NODE_QUANT_BODY(node) ((node)->body)
#define NODE_BAG_BODY(node) ((node)->body)
(senv)->mem_env_dynamic : (senv)->mem_env_static)
typedef struct {
- Node* node;
-#if 0
- int in;
- int recursion;
-#endif
+ Node* mem_node;
+ Node* empty_repeat_node;
} MemEnv;
typedef struct {
OnigCaseFoldType case_fold_flag;
OnigEncoding enc;
OnigSyntaxType* syntax;
- MemStatusType capture_history;
- MemStatusType bt_mem_start;
- MemStatusType bt_mem_end;
+ MemStatusType cap_history;
+ MemStatusType backtrack_mem; /* backtrack/recursion */
MemStatusType backrefed_mem;
UChar* pattern;
UChar* pattern_end;
MemEnv mem_env_static[SCANENV_MEMENV_SIZE];
MemEnv* mem_env_dynamic;
unsigned int parse_depth;
-
+#ifdef ONIG_DEBUG_PARSE
+ unsigned int max_parse_depth;
+#endif
+ int backref_num;
int keep_num;
int save_num;
int save_alloc_num;
extern int onig_strncmp P_((const UChar* s1, const UChar* s2, int n));
extern void onig_strcpy P_((UChar* dest, const UChar* src, const UChar* end));
extern void onig_scan_env_set_error_string P_((ScanEnv* env, int ecode, UChar* arg, UChar* arg_end));
-extern int onig_scan_unsigned_number P_((UChar** src, const UChar* end, OnigEncoding enc));
-extern void onig_reduce_nested_quantifier P_((Node* pnode, Node* cnode));
-extern void onig_node_conv_to_str_node P_((Node* node, int raw));
+extern int onig_reduce_nested_quantifier P_((Node* pnode));
extern int onig_node_str_cat P_((Node* node, const UChar* s, const UChar* end));
extern int onig_node_str_set P_((Node* node, const UChar* s, const UChar* end));
extern void onig_node_free P_((Node* node));
extern Node* onig_node_new_anchor P_((int type, int ascii_mode));
extern Node* onig_node_new_str P_((const UChar* s, const UChar* end));
extern Node* onig_node_new_list P_((Node* left, Node* right));
-extern Node* onig_node_list_add P_((Node* list, Node* x));
extern Node* onig_node_new_alt P_((Node* left, Node* right));
extern void onig_node_str_clear P_((Node* node));
extern int onig_names_free P_((regex_t* reg));
extern int onig_parse_tree P_((Node** root, const UChar* pattern, const UChar* end, regex_t* reg, ScanEnv* env));
extern int onig_free_shared_cclass_table P_((void));
extern int onig_is_code_in_cc P_((OnigEncoding enc, OnigCodePoint code, CClassNode* cc));
+extern int onig_new_cclass_with_code_list(Node** rnode, OnigEncoding enc, int n, OnigCodePoint codes[]);
extern OnigLen onig_get_tiny_min_len(Node* node, unsigned int inhibit_node_types, int* invalid_node);
#ifdef USE_CALLOUT
extern int onig_print_names(FILE*, regex_t*);
#endif
-#if (defined (__GNUC__) && __GNUC__ > 2 ) && !defined(DARWIN) && !defined(__hpux) && !defined(_AIX)
-# define UNEXPECTED(condition) __builtin_expect(condition, 0)
-#else
-# define UNEXPECTED(condition) (condition)
-#endif
-
-#define SAFE_ENC_LEN(enc, p, end, res) do { \
- int __res = enclen(enc, p); \
- if (UNEXPECTED(p + __res > end)) __res = end - p; \
- res = __res; \
-} while(0);
-
#endif /* REGPARSE_H */
regposerr.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regposix.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regsyntax.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regtrav.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2004 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
regversion.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
sjis.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
if ((code & 0xff00) != 0) *p++ = (UChar )(((code >> 8) & 0xff));
*p++ = (UChar )(code & 0xff);
-#if 0
- if (enclen(ONIG_ENCODING_SJIS, buf) != (p - buf))
- return REGERR_INVALID_CODE_POINT_VALUE;
-#endif
return (int )(p - buf);
}
}
}
-#if 0
-static int
-is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- return onigenc_mbn_is_mbc_ambiguous(ONIG_ENCODING_SJIS, flag, pp, end);
-
-}
-#endif
-
-#if 0
-static int
-is_code_ctype(OnigCodePoint code, unsigned int ctype)
-{
- if (code < 128)
- return ONIGENC_IS_ASCII_CODE_CTYPE(code, ctype);
- else {
- if (CTYPE_IS_WORD_GRAPH_PRINT(ctype)) {
- return (code_to_mbclen(code) > 1 ? TRUE : FALSE);
- }
- }
-
- return FALSE;
-}
-#endif
-
static UChar*
left_adjust_char_head(const UChar* start, const UChar* s)
{
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -pt -T -L ANSI-C -N onigenc_sjis_lookup_property_name --output-file gperf2.tmp sjis_prop.gperf */
+/* Command-line: gperf -pt -T -L ANSI-C -N onigenc_sjis_lookup_property_name --output-file gperf2.tmp sjis_prop.gperf */
/* Computed positions: -k'1,3' */
#if !((' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
unicode.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
for (fn = 0; fn < 2; fn++) {
int index;
cs[fn][0] = FOLDS2_FOLD(buk->index)[fn];
+ ncs[fn] = 1;
index = onigenc_unicode_fold1_key(&cs[fn][0]);
if (index >= 0) {
int m = FOLDS1_UNFOLDS_NUM(index);
for (i = 0; i < m; i++) {
cs[fn][i+1] = FOLDS1_UNFOLDS(index)[i];
}
- ncs[fn] = m + 1;
+ ncs[fn] += m;
}
- else
- ncs[fn] = 1;
}
for (i = 0; i < ncs[0]; i++) {
for (fn = 0; fn < 3; fn++) {
int index;
cs[fn][0] = FOLDS3_FOLD(buk->index)[fn];
+ ncs[fn] = 1;
index = onigenc_unicode_fold1_key(&cs[fn][0]);
if (index >= 0) {
int m = FOLDS1_UNFOLDS_NUM(index);
for (i = 0; i < m; i++) {
cs[fn][i+1] = FOLDS1_UNFOLDS(index)[i];
}
- ncs[fn] = m + 1;
+ ncs[fn] += m;
}
- else
- ncs[fn] = 1;
}
for (i = 0; i < ncs[0]; i++) {
/* unicode_egcb_data.c: Generated by make_unicode_egcb_data.py. */
/*-
- * Copyright (c) 2017-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2017-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* SUCH DAMAGE.
*/
-#define GRAPHEME_BREAK_PROPERTY_VERSION 12_1_0
+#define GRAPHEME_BREAK_PROPERTY_VERSION 120100
/*
CR
/* This file was converted by gperf_fold_key_conv.py
from gperf output file. */
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold1_key unicode_fold1_key.gperf */
+/* Command-line: gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold1_key unicode_fold1_key.gperf */
/* Computed positions: -k'1-3' */
/* This gperf source file was generated by make_unicode_fold_data.py */
/*-
- * Copyright (c) 2017-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2017-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
4026
};
- if (0 == 0)
+
{
int key = hash(codes);
/* This file was converted by gperf_fold_key_conv.py
from gperf output file. */
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold2_key unicode_fold2_key.gperf */
+/* Command-line: gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold2_key unicode_fold2_key.gperf */
/* Computed positions: -k'3,6' */
/* This gperf source file was generated by make_unicode_fold_data.py */
/*-
- * Copyright (c) 2017-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2017-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
129
};
- if (0 == 0)
+
{
int key = hash(codes);
/* This file was converted by gperf_fold_key_conv.py
from gperf output file. */
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold3_key unicode_fold3_key.gperf */
+/* Command-line: gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1 -N onigenc_unicode_fold3_key unicode_fold3_key.gperf */
/* Computed positions: -k'3,6,9' */
/* This gperf source file was generated by make_unicode_fold_data.py */
/*-
- * Copyright (c) 2017-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2017-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
0
};
- if (0 == 0)
+
{
int key = hash(codes);
/* This file was generated by make_unicode_fold_data.py. */
#include "regenc.h"
-#define UNICODE_CASEFOLD_VERSION 12_1_0
+#define UNICODE_CASEFOLD_VERSION 120100
OnigCodePoint OnigUnicodeFolds1[] = {
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -T -C -c -t -j1 -L ANSI-C --ignore-case --pic -Q unicode_prop_name_pool -N unicode_lookup_property_name --output-file gperf1.tmp unicode_property_data.gperf */
+/* Command-line: gperf -T -C -c -t -j1 -L ANSI-C --ignore-case --pic -Q unicode_prop_name_pool -N unicode_lookup_property_name --output-file gperf1.tmp unicode_property_data.gperf */
/* Computed positions: -k'1-3,5-6,12,16,$' */
#if !((' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
-#define UNICODE_PROPERTY_VERSION 12_1_0
+#define UNICODE_PROPERTY_VERSION 120100
+#define UNICODE_EMOJI_VERSION 1201
#define PROPERTY_NAME_MAX_SIZE 59
#define CODE_RANGES_NUM 568
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -T -C -c -t -j1 -L ANSI-C --ignore-case --pic -Q unicode_prop_name_pool -N unicode_lookup_property_name --output-file gperf2.tmp unicode_property_data_posix.gperf */
+/* Command-line: gperf -T -C -c -t -j1 -L ANSI-C --ignore-case --pic -Q unicode_prop_name_pool -N unicode_lookup_property_name --output-file gperf2.tmp unicode_property_data_posix.gperf */
/* Computed positions: -k'1,3' */
#if !((' ' == 32) && ('!' == 33) && ('"' == 34) && ('#' == 35) \
/* This file was converted by gperf_unfold_key_conv.py
from gperf output file. */
/* ANSI-C code produced by gperf version 3.1 */
-/* Command-line: /usr/local/bin/gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1,0 -N onigenc_unicode_unfold_key unicode_unfold_key.gperf */
+/* Command-line: gperf -n -C -T -c -t -j1 -L ANSI-C -F,-1,0 -N onigenc_unicode_unfold_key unicode_unfold_key.gperf */
/* Computed positions: -k'1-3' */
/* This gperf source file was generated by make_unicode_fold_data.py */
/*-
- * Copyright (c) 2017-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2017-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
{0x1e907, 4005, 1}
};
- if (0 == 0)
+
{
int key = hash(&code);
/* unicode_wb_data.c: Generated by make_unicode_wb_data.py. */
/*-
- * Copyright (c) 2019 K.Kosako <kkosako0 AT gmail DOT com>
+ * Copyright (c) 2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* SUCH DAMAGE.
*/
-#define WORD_BREAK_PROPERTY_VERSION 12_1_0
+#define WORD_BREAK_PROPERTY_VERSION 120100
/*
ALetter
utf16_be.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
}
static OnigCodePoint
-utf16be_mbc_to_code(const UChar* p, const UChar* end)
+utf16be_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
{
OnigCodePoint code;
if (UTF16_IS_SURROGATE_FIRST(*p)) {
- if (end - p < 4) return 0;
code = ((((p[0] - 0xd8) << 2) + ((p[1] & 0xc0) >> 6) + 1) << 16)
+ ((((p[1] & 0x3f) << 2) + (p[2] - 0xdc)) << 8)
+ p[3];
}
else {
- if (end - p < 2) return 0;
code = p[0] * 256 + p[1];
}
return code;
pp, end, fold);
}
-#if 0
-static int
-utf16be_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- const UChar* p = *pp;
-
- (*pp) += EncLen_UTF16[*p];
-
- if (*p == 0) {
- int c, v;
-
- p++;
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- return TRUE;
- }
-
- c = *p;
- v = ONIGENC_IS_UNICODE_ISO_8859_1_BIT_CTYPE(c, (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
-
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (c >= 0xaa && c <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
- return (v != 0 ? TRUE : FALSE);
- }
-
- return FALSE;
-}
-#endif
-
static UChar*
utf16be_left_adjust_char_head(const UChar* start, const UChar* s)
{
utf16_le.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
}
static OnigCodePoint
-utf16le_mbc_to_code(const UChar* p, const UChar* end)
+utf16le_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
{
OnigCodePoint code;
UChar c0 = *p;
UChar c1 = *(p+1);
if (UTF16_IS_SURROGATE_FIRST(c1)) {
- if (end - p < 4) return 0;
code = ((((c1 - 0xd8) << 2) + ((c0 & 0xc0) >> 6) + 1) << 16)
+ ((((c0 & 0x3f) << 2) + (p[3] - 0xdc)) << 8)
+ p[2];
fold);
}
-#if 0
-static int
-utf16le_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp,
- const UChar* end)
-{
- const UChar* p = *pp;
-
- (*pp) += EncLen_UTF16[*(p+1)];
-
- if (*(p+1) == 0) {
- int c, v;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- return TRUE;
- }
-
- c = *p;
- v = ONIGENC_IS_UNICODE_ISO_8859_1_BIT_CTYPE(c,
- (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (c >= 0xaa && c <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
- return (v != 0 ? TRUE : FALSE);
- }
-
- return FALSE;
-}
-#endif
-
static UChar*
utf16le_left_adjust_char_head(const UChar* start, const UChar* s)
{
utf32_be.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
static OnigCodePoint
utf32be_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
{
- if (end - p < 4) return 0;
return (OnigCodePoint )(((p[0] * 256 + p[1]) * 256 + p[2]) * 256 + p[3]);
}
fold);
}
-#if 0
-static int
-utf32be_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- const UChar* p = *pp;
-
- (*pp) += 4;
-
- if (*(p+2) == 0 && *(p+1) == 0 && *p == 0) {
- int c, v;
-
- p += 3;
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- return TRUE;
- }
-
- c = *p;
- v = ONIGENC_IS_UNICODE_ISO_8859_1_BIT_CTYPE(c,
- (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (c >= 0xaa && c <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
- return (v != 0 ? TRUE : FALSE);
- }
-
- return FALSE;
-}
-#endif
-
static UChar*
utf32be_left_adjust_char_head(const UChar* start, const UChar* s)
{
utf32_le.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2018 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
static OnigCodePoint
utf32le_mbc_to_code(const UChar* p, const UChar* end ARG_UNUSED)
{
- if (end - p < 4) return 0;
return (OnigCodePoint )(((p[3] * 256 + p[2]) * 256 + p[1]) * 256 + p[0]);
}
fold);
}
-#if 0
-static int
-utf32le_is_mbc_ambiguous(OnigCaseFoldType flag, const UChar** pp, const UChar* end)
-{
- const UChar* p = *pp;
-
- (*pp) += 4;
-
- if (*(p+1) == 0 && *(p+2) == 0 && *(p+3) == 0) {
- int c, v;
-
- if (*p == 0xdf && (flag & INTERNAL_ONIGENC_CASE_FOLD_MULTI_CHAR) != 0) {
- return TRUE;
- }
-
- c = *p;
- v = ONIGENC_IS_UNICODE_ISO_8859_1_BIT_CTYPE(c,
- (BIT_CTYPE_UPPER | BIT_CTYPE_LOWER));
- if ((v | BIT_CTYPE_LOWER) != 0) {
- /* 0xaa, 0xb5, 0xba are lower case letter, but can't convert. */
- if (c >= 0xaa && c <= 0xba)
- return FALSE;
- else
- return TRUE;
- }
- return (v != 0 ? TRUE : FALSE);
- }
-
- return FALSE;
-}
-#endif
-
static UChar*
utf32le_left_adjust_char_head(const UChar* start, const UChar* s)
{
utf8.c - Oniguruma (regular expression library)
**********************************************************************/
/*-
- * Copyright (c) 2002-2019 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
+ * Copyright (c) 2002-2019 K.Kosako
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
return TRUE;
}
-#if 0
-static int
-is_mbc_newline(const UChar* p, const UChar* end)
-{
- if (p < end) {
- if (*p == 0x0a) return 1;
-
-#ifdef USE_UNICODE_ALL_LINE_TERMINATORS
-#ifndef USE_CRNL_AS_LINE_TERMINATOR
- if (*p == 0x0d) return 1;
-#endif
- if (p + 1 < end) {
- if (*(p+1) == 0x85 && *p == 0xc2) /* U+0085 */
- return 1;
- if (p + 2 < end) {
- if ((*(p+2) == 0xa8 || *(p+2) == 0xa9)
- && *(p+1) == 0x80 && *p == 0xe2) /* U+2028, U+2029 */
- return 1;
- }
- }
-#endif
- }
-
- return 0;
-}
-#endif
-
static OnigCodePoint
mbc_to_code(const UChar* p, const UChar* end)
{
string(6) "中国"
[3]=>
string(3) "？"
- ["punct"]=>
- string(3) "？"
["wsp"]=>
string(2) " "
["word"]=>
string(6) "中国"
+ ["punct"]=>
+ string(3) "？"
}
projects / php / commitdiff