From: Seiji Masugata Date: Thu, 21 Dec 2006 17:37:53 +0000 (+0000) Subject: Synced PHP_5_2 Branch. X-Git-Tag: RELEASE_1_0_0RC1~573 X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=0d9e23c38bc9fca9eaaa5671ec17b25055ed0bc1;p=php Synced PHP_5_2 Branch. --- diff --git a/ext/mbstring/config.m4 b/ext/mbstring/config.m4 index c57f13cf06..e0698a2d62 100644 --- a/ext/mbstring/config.m4 +++ b/ext/mbstring/config.m4 @@ -171,6 +171,7 @@ AC_DEFUN([PHP_MBSTRING_SETUP_LIBMBFL], [ libmbfl/filters/mbfilter_euc_cn.c libmbfl/filters/mbfilter_euc_jp.c libmbfl/filters/mbfilter_euc_jp_win.c + libmbfl/filters/mbfilter_cp51932.c libmbfl/filters/mbfilter_euc_kr.c libmbfl/filters/mbfilter_euc_tw.c libmbfl/filters/mbfilter_htmlent.c @@ -282,7 +283,7 @@ if test "$PHP_MBSTRING" != "no"; then dnl libmbfl is required PHP_MBSTRING_SETUP_LIBMBFL PHP_MBSTRING_EXTENSION - PHP_INSTALL_HEADERS([ext/mbstring], [mbstring.h php_mbregex.h libmbfl/ libmbfl/mbfl/ oniguruma/]) + PHP_INSTALL_HEADERS([ext/mbstring], [mbstring.h php_mbregex.h libmbfl/config.h libmbfl/mbfl/eaw_table.h libmbfl/mbfl/mbfilter.h libmbfl/mbfl/mbfilter_8bit.h libmbfl/mbfl/mbfilter_pass.h libmbfl/mbfl/mbfilter_wchar.h libmbfl/mbfl/mbfl_allocators.h libmbfl/mbfl/mbfl_consts.h libmbfl/mbfl/mbfl_convert.h libmbfl/mbfl/mbfl_defs.h libmbfl/mbfl/mbfl_encoding.h libmbfl/mbfl/mbfl_filter_output.h libmbfl/mbfl/mbfl_ident.h libmbfl/mbfl/mbfl_language.h libmbfl/mbfl/mbfl_memory_device.h libmbfl/mbfl/mbfl_string.h oniguruma/oniguruma.h oniguruma/php_onig_compat.h]) fi # vim600: sts=2 sw=2 et diff --git a/ext/mbstring/libmbfl/filters/Makefile.am b/ext/mbstring/libmbfl/filters/Makefile.am index 0c71c8604b..9b2fda4c39 100644 --- a/ext/mbstring/libmbfl/filters/Makefile.am +++ b/ext/mbstring/libmbfl/filters/Makefile.am @@ -2,7 +2,7 @@ EXTRA_DIST=Makefile.bcc32 mk_sb_tbl.awk noinst_LTLIBRARIES=libmbfl_filters.la INCLUDES=-I../mbfl libmbfl_filters_la_LDFLAGS=-version-info $(SHLIB_VERSION) -libmbfl_filters_la_SOURCES=mbfilter_cp936.c mbfilter_hz.c mbfilter_euc_tw.c mbfilter_big5.c mbfilter_euc_jp.c mbfilter_jis.c mbfilter_iso8859_1.c mbfilter_iso8859_2.c mbfilter_cp1252.c mbfilter_cp1251.c mbfilter_ascii.c mbfilter_iso8859_3.c mbfilter_iso8859_4.c mbfilter_iso8859_5.c mbfilter_iso8859_6.c mbfilter_iso8859_7.c mbfilter_iso8859_8.c mbfilter_iso8859_9.c mbfilter_iso8859_10.c mbfilter_iso8859_13.c mbfilter_iso8859_14.c mbfilter_iso8859_15.c mbfilter_iso8859_16.c mbfilter_htmlent.c mbfilter_byte2.c mbfilter_byte4.c mbfilter_uuencode.c mbfilter_base64.c mbfilter_sjis.c mbfilter_7bit.c mbfilter_qprint.c mbfilter_ucs4.c mbfilter_ucs2.c mbfilter_utf32.c mbfilter_utf16.c mbfilter_utf8.c mbfilter_utf7.c mbfilter_utf7imap.c mbfilter_euc_jp_win.c mbfilter_cp932.c mbfilter_euc_cn.c mbfilter_euc_kr.c mbfilter_uhc.c mbfilter_iso2022_kr.c mbfilter_cp866.c mbfilter_koi8r.c mbfilter_armscii8.c html_entities.c cp932_table.h html_entities.h mbfilter_7bit.h mbfilter_ascii.h mbfilter_base64.h mbfilter_big5.h mbfilter_byte2.h mbfilter_byte4.h mbfilter_cp1251.h mbfilter_cp1252.h mbfilter_cp866.h mbfilter_cp932.h mbfilter_cp936.h mbfilter_euc_cn.h mbfilter_euc_jp.h mbfilter_euc_jp_win.h mbfilter_euc_kr.h mbfilter_euc_tw.h mbfilter_htmlent.h mbfilter_hz.h mbfilter_iso2022_kr.h mbfilter_iso8859_1.h mbfilter_iso8859_10.h mbfilter_iso8859_13.h mbfilter_iso8859_14.h mbfilter_iso8859_15.h mbfilter_iso8859_16.h mbfilter_iso8859_2.h mbfilter_iso8859_3.h mbfilter_iso8859_4.h mbfilter_iso8859_5.h mbfilter_iso8859_6.h mbfilter_iso8859_7.h mbfilter_iso8859_8.h mbfilter_iso8859_9.h mbfilter_jis.h mbfilter_koi8r.h mbfilter_armscii8.h mbfilter_qprint.h mbfilter_sjis.h mbfilter_ucs2.h mbfilter_ucs4.h mbfilter_uhc.h mbfilter_utf16.h mbfilter_utf32.h mbfilter_utf7.h mbfilter_utf7imap.h mbfilter_utf8.h mbfilter_uuencode.h unicode_prop.h unicode_table_big5.h unicode_table_cns11643.h unicode_table_cp1251.h unicode_table_cp1252.h unicode_table_cp866.h unicode_table_cp932_ext.h unicode_table_cp936.h unicode_table_iso8859_10.h unicode_table_iso8859_13.h unicode_table_iso8859_14.h unicode_table_iso8859_15.h unicode_table_iso8859_16.h unicode_table_iso8859_2.h unicode_table_iso8859_3.h unicode_table_iso8859_4.h unicode_table_iso8859_5.h unicode_table_iso8859_6.h unicode_table_iso8859_7.h unicode_table_iso8859_8.h unicode_table_iso8859_9.h unicode_table_jis.h unicode_table_koi8r.h unicode_table_armscii8.h unicode_table_uhc.h +libmbfl_filters_la_SOURCES=mbfilter_cp936.c mbfilter_hz.c mbfilter_euc_tw.c mbfilter_big5.c mbfilter_euc_jp.c mbfilter_jis.c mbfilter_iso8859_1.c mbfilter_iso8859_2.c mbfilter_cp1252.c mbfilter_cp1251.c mbfilter_ascii.c mbfilter_iso8859_3.c mbfilter_iso8859_4.c mbfilter_iso8859_5.c mbfilter_iso8859_6.c mbfilter_iso8859_7.c mbfilter_iso8859_8.c mbfilter_iso8859_9.c mbfilter_iso8859_10.c mbfilter_iso8859_13.c mbfilter_iso8859_14.c mbfilter_iso8859_15.c mbfilter_iso8859_16.c mbfilter_htmlent.c mbfilter_byte2.c mbfilter_byte4.c mbfilter_uuencode.c mbfilter_base64.c mbfilter_sjis.c mbfilter_7bit.c mbfilter_qprint.c mbfilter_ucs4.c mbfilter_ucs2.c mbfilter_utf32.c mbfilter_utf16.c mbfilter_utf8.c mbfilter_utf7.c mbfilter_utf7imap.c mbfilter_euc_jp_win.c mbfilter_cp932.c mbfilter_cp51932.c mbfilter_euc_cn.c mbfilter_euc_kr.c mbfilter_uhc.c mbfilter_iso2022_kr.c mbfilter_cp866.c mbfilter_koi8r.c mbfilter_armscii8.c html_entities.c cp932_table.h html_entities.h mbfilter_7bit.h mbfilter_ascii.h mbfilter_base64.h mbfilter_big5.h mbfilter_byte2.h mbfilter_byte4.h mbfilter_cp1251.h mbfilter_cp1252.h mbfilter_cp866.h mbfilter_cp932.h mbfilter_cp51932.h mbfilter_cp936.h mbfilter_euc_cn.h mbfilter_euc_jp.h mbfilter_euc_jp_win.h mbfilter_euc_kr.h mbfilter_euc_tw.h mbfilter_htmlent.h mbfilter_hz.h mbfilter_iso2022_kr.h mbfilter_iso8859_1.h mbfilter_iso8859_10.h mbfilter_iso8859_13.h mbfilter_iso8859_14.h mbfilter_iso8859_15.h mbfilter_iso8859_16.h mbfilter_iso8859_2.h mbfilter_iso8859_3.h mbfilter_iso8859_4.h mbfilter_iso8859_5.h mbfilter_iso8859_6.h mbfilter_iso8859_7.h mbfilter_iso8859_8.h mbfilter_iso8859_9.h mbfilter_jis.h mbfilter_koi8r.h mbfilter_armscii8.h mbfilter_qprint.h mbfilter_sjis.h mbfilter_ucs2.h mbfilter_ucs4.h mbfilter_uhc.h mbfilter_utf16.h mbfilter_utf32.h mbfilter_utf7.h mbfilter_utf7imap.h mbfilter_utf8.h mbfilter_uuencode.h unicode_prop.h unicode_table_big5.h unicode_table_cns11643.h unicode_table_cp1251.h unicode_table_cp1252.h unicode_table_cp866.h unicode_table_cp932_ext.h unicode_table_cp936.h unicode_table_iso8859_10.h unicode_table_iso8859_13.h unicode_table_iso8859_14.h unicode_table_iso8859_15.h unicode_table_iso8859_16.h unicode_table_iso8859_2.h unicode_table_iso8859_3.h unicode_table_iso8859_4.h unicode_table_iso8859_5.h unicode_table_iso8859_6.h unicode_table_iso8859_7.h unicode_table_iso8859_8.h unicode_table_iso8859_9.h unicode_table_jis.h unicode_table_koi8r.h unicode_table_armscii8.h unicode_table_uhc.h mbfilter_iso8859_2.c: unicode_table_iso8859_2.h diff --git a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c index e57ae69356..bc07f13438 100644 --- a/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c +++ b/ext/mbstring/libmbfl/filters/mbfilter_euc_jp_win.c @@ -60,7 +60,8 @@ static const unsigned char mblen_table_eucjp[] = { /* 0xA1-0xFE */ }; -static const char *mbfl_encoding_eucjp_win_aliases[] = {"eucJP-open", NULL}; +static const char *mbfl_encoding_eucjp_win_aliases[] = {"eucJP-open", + "eucJP-ms", NULL}; const struct mbfl_identify_vtbl vtbl_identify_eucjpwin = { mbfl_no_encoding_eucjp_win, diff --git a/ext/mbstring/libmbfl/filters/mbfilter_jis.c b/ext/mbstring/libmbfl/filters/mbfilter_jis.c index 58336d4e28..3657658ba8 100644 --- a/ext/mbstring/libmbfl/filters/mbfilter_jis.c +++ b/ext/mbstring/libmbfl/filters/mbfilter_jis.c @@ -58,6 +58,15 @@ const mbfl_encoding mbfl_encoding_2022jp = { MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE }; +const mbfl_encoding mbfl_encoding_2022jpms = { + mbfl_no_encoding_2022jpms, + "ISO-2022-JP-MS", + "ISO-2022-JP-MS", + NULL, + NULL, + MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_SHFTCODE +}; + const struct mbfl_identify_vtbl vtbl_identify_jis = { mbfl_no_encoding_jis, mbfl_filt_ident_common_ctor, @@ -72,6 +81,13 @@ const struct mbfl_identify_vtbl vtbl_identify_2022jp = { mbfl_filt_ident_2022jp }; +const struct mbfl_identify_vtbl vtbl_identify_2022jpms = { + mbfl_no_encoding_2022jpms, + mbfl_filt_ident_common_ctor, + mbfl_filt_ident_common_dtor, + mbfl_filt_ident_2022jp +}; + const struct mbfl_convert_vtbl vtbl_jis_wchar = { mbfl_no_encoding_jis, mbfl_no_encoding_wchar, @@ -108,6 +124,24 @@ const struct mbfl_convert_vtbl vtbl_wchar_2022jp = { mbfl_filt_conv_any_jis_flush }; +const struct mbfl_convert_vtbl vtbl_2022jpms_wchar = { + mbfl_no_encoding_2022jpms, + mbfl_no_encoding_wchar, + mbfl_filt_conv_common_ctor, + mbfl_filt_conv_common_dtor, + mbfl_filt_conv_jis_wchar, + mbfl_filt_conv_common_flush +}; + +const struct mbfl_convert_vtbl vtbl_wchar_2022jpms = { + mbfl_no_encoding_wchar, + mbfl_no_encoding_2022jpms, + mbfl_filt_conv_common_ctor, + mbfl_filt_conv_common_dtor, + mbfl_filt_conv_wchar_jis, + mbfl_filt_conv_any_jis_flush +}; + #define CK(statement) do { if ((statement) < 0) return (-1); } while (0) /* @@ -130,12 +164,10 @@ retry: filter->status += 2; } else if (c == 0x0e) { /* "kana in" */ filter->status = 0x20; + CK((*filter->output_function)(c, filter->data)); } else if (c == 0x0f) { /* "kana out" */ filter->status = 0; - } else if (filter->status == 0x10 && c == 0x5c) { /* YEN SIGN */ - CK((*filter->output_function)(0xa5, filter->data)); - } else if (filter->status == 0x10 && c == 0x7e) { /* OVER LINE */ - CK((*filter->output_function)(0x203e, filter->data)); + CK((*filter->output_function)(c, filter->data)); } else if (filter->status == 0x20 && c > 0x20 && c < 0x60) { /* kana */ CK((*filter->output_function)(0xff40 + c, filter->data)); } else if ((filter->status == 0x80 || filter->status == 0x90) && c > 0x20 && c < 0x7f) { /* kanji first char */ @@ -161,9 +193,34 @@ retry: s = (c1 - 0x21)*94 + c - 0x21; if (filter->status == 0x80) { if (s >= 0 && s < jisx0208_ucs_table_size) { + if ((filter->from)->no_encoding != + mbfl_no_encoding_2022jpms) { + w = jisx0208_ucs_table[s]; + } + else { + if ((c1 - 0x21) == 12) { + w = cp932ext1_ucs_table[s-12*94]; + } + else { + if (c1 >= 0x79 && c1 <= 0x7c) { + w = cp932ext2_ucs_table[s-(0x79-0x21)*94]; + } + else { w = jisx0208_ucs_table[s]; + } + } + } } else { + if ((filter->from)->no_encoding != + mbfl_no_encoding_2022jpms) { + w = 0; + } else { + if (c1 >= 0x79 && c1 <= 0x7c) { + w = cp932ext2_ucs_table[s-(0x79-0x21)*94]; + } else { w = 0; + } + } } if (w <= 0) { w = (c1 << 8) | c; @@ -287,7 +344,7 @@ retry: int mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter) { - int c1, s; + int c1, c2, s; s = 0; if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) { @@ -299,6 +356,18 @@ mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter) } else if (c >= ucs_r_jis_table_min && c < ucs_r_jis_table_max) { s = ucs_r_jis_table[c - ucs_r_jis_table_min]; } + if (s > 0x8080 && s < 0x10000 && + ((filter->to)->no_encoding == mbfl_no_encoding_2022jpms)) { + c1 = 0; + c2 = cp932ext2_ucs_table_max - cp932ext2_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */ + if (c == cp932ext2_ucs_table[c1]) { + s = ((c1/94 + 0x79) << 8) +(c1%94 + 0x21); + break; + } + c1++; + } + } if (s <= 0) { c1 = c & ~MBFL_WCSPLANE_MASK; if (c1 == MBFL_WCSPLANE_JIS0208) { @@ -327,8 +396,30 @@ mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter) } if (c == 0) { s = 0; - } else if (s <= 0) { + } else if (s <= 0 && ((filter->to)->no_encoding == + mbfl_no_encoding_2022jpms)) { s = -1; + c1 = 0; + c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext1 (13ku) */ + if (c == cp932ext1_ucs_table[c1]) { + s = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21); + break; + } + c1++; + } + if (s < 0 && ((filter->to)->no_encoding == + mbfl_no_encoding_2022jpms)) { + c1 = 0; + c2 = cp932ext2_ucs_table_max - cp932ext2_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */ + if (c == cp932ext2_ucs_table[c1]) { + s = ((c1/94 + 0x79) << 8) +(c1%94 + 0x21); + break; + } + c1++; + } + } } } if (s >= 0) { @@ -392,7 +483,7 @@ mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter) int mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter) { - int s; + int c1, c2, s; s = 0; if (c >= ucs_a1_jis_table_min && c < ucs_a1_jis_table_max) { @@ -428,9 +519,47 @@ mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter) s = 0; } else if (s <= 0) { s = -1; + if ( (filter->to)->no_encoding == + mbfl_no_encoding_2022jpms) { + c1 = 0; + c2 = cp932ext1_ucs_table_max - cp932ext1_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext1 (13ku) */ + if (c == cp932ext1_ucs_table[c1]) { + s = ((c1/94 + 0x2d) << 8) + (c1%94 + 0x21); + break; + } + c1++; + } + if ((filter->to)->no_encoding == + mbfl_no_encoding_2022jpms) { + c1 = 0; + c2 = cp932ext2_ucs_table_max - cp932ext2_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */ + if (c == cp932ext2_ucs_table[c1]) { + s = ((c1/94 + 0x79) << 8) +(c1%94 + 0x21); + break; + } + c1++; + } + } + } } - } else if ((s >= 0x80 && s < 0x2121) || (s > 0x8080)) { + } else if (((s >= 0x80 && s < 0x2121) && + (filter->to)->no_encoding != mbfl_no_encoding_2022jpms) || + (s > 0x8080)) { s = -1; + if ((filter->to)->no_encoding == + mbfl_no_encoding_2022jpms) { + c1 = 0; + c2 = cp932ext2_ucs_table_max - cp932ext2_ucs_table_min; + while (c1 < c2) { /* CP932 vendor ext3 (115ku - 119ku) */ + if (c == cp932ext2_ucs_table[c1]) { + s = ((c1/94 + 0x79) << 8) +(c1%94 + 0x21); + break; + } + c1++; + } + } } if (s >= 0) { if (s < 0x80) { /* ASCII */ @@ -441,6 +570,15 @@ mbfl_filt_conv_wchar_2022jp(int c, mbfl_convert_filter *filter) } filter->status = 0; CK((*filter->output_function)(s, filter->data)); + } else if (s < 0x100 && ((filter->to)->no_encoding == + mbfl_no_encoding_2022jpms)) { /* kana */ + if ((filter->status & 0xff00) != 0x100) { + CK((*filter->output_function)(0x1b, filter->data)); /* ESC */ + CK((*filter->output_function)(0x28, filter->data)); /* '(' */ + CK((*filter->output_function)(0x49, filter->data)); /* 'I' */ + } + filter->status = 0x100; + CK((*filter->output_function)(s & 0x7f, filter->data)); } else if (s < 0x10000) { /* X 0208 */ if ((filter->status & 0xff00) != 0x200) { CK((*filter->output_function)(0x1b, filter->data)); /* ESC */ diff --git a/ext/mbstring/libmbfl/filters/mbfilter_jis.h b/ext/mbstring/libmbfl/filters/mbfilter_jis.h index 0375307263..1ba244a282 100644 --- a/ext/mbstring/libmbfl/filters/mbfilter_jis.h +++ b/ext/mbstring/libmbfl/filters/mbfilter_jis.h @@ -34,12 +34,16 @@ extern const mbfl_encoding mbfl_encoding_jis; extern const mbfl_encoding mbfl_encoding_2022jp; +extern const mbfl_encoding mbfl_encoding_2022jpms; extern const struct mbfl_identify_vtbl vtbl_identify_2022jp; +extern const struct mbfl_identify_vtbl vtbl_identify_2022jpms; extern const struct mbfl_identify_vtbl vtbl_identify_jis; extern const struct mbfl_convert_vtbl vtbl_jis_wchar; extern const struct mbfl_convert_vtbl vtbl_wchar_jis; extern const struct mbfl_convert_vtbl vtbl_2022jp_wchar; extern const struct mbfl_convert_vtbl vtbl_wchar_2022jp; +extern const struct mbfl_convert_vtbl vtbl_2022jpms_wchar; +extern const struct mbfl_convert_vtbl vtbl_wchar_2022jpms; int mbfl_filt_conv_jis_wchar(int c, mbfl_convert_filter *filter); int mbfl_filt_conv_wchar_jis(int c, mbfl_convert_filter *filter); diff --git a/ext/mbstring/libmbfl/filters/unicode_table_jis.h b/ext/mbstring/libmbfl/filters/unicode_table_jis.h index 5671c4e851..e87dad93c8 100644 --- a/ext/mbstring/libmbfl/filters/unicode_table_jis.h +++ b/ext/mbstring/libmbfl/filters/unicode_table_jis.h @@ -36,13 +36,13 @@ static const unsigned short jisx0208_ucs_table[] = { 0xFF1F,0xFF01,0x309B,0x309C,0x00B4,0xFF40,0x00A8,0xFF3E, 0xFFE3,0xFF3F,0x30FD,0x30FE,0x309D,0x309E,0x3003,0x4EDD, 0x3005,0x3006,0x3007,0x30FC,0x2015,0x2010,0xFF0F,0xFF3C, - 0x301C,0x2016,0xFF5C,0x2026,0x2025,0x2018,0x2019,0x201C, + 0xFF5E,0x2225,0xFF5C,0x2026,0x2025,0x2018,0x2019,0x201C, 0x201D,0xFF08,0xFF09,0x3014,0x3015,0xFF3B,0xFF3D,0xFF5B, 0xFF5D,0x3008,0x3009,0x300A,0x300B,0x300C,0x300D,0x300E, - 0x300F,0x3010,0x3011,0xFF0B,0x2212,0x00B1,0x00D7,0x00F7, + 0x300F,0x3010,0x3011,0xFF0B,0xFF0D,0x00B1,0x00D7,0x00F7, 0xFF1D,0x2260,0xFF1C,0xFF1E,0x2266,0x2267,0x221E,0x2234, 0x2642,0x2640,0x00B0,0x2032,0x2033,0x2103,0xFFE5,0xFF04, - 0x00A2,0x00A3,0xFF05,0xFF03,0xFF06,0xFF0A,0xFF20,0x00A7, + 0xFFE0,0xFFE1,0xFF05,0xFF03,0xFF06,0xFF0A,0xFF20,0x00A7, 0x2606,0x2605,0x25CB,0x25CF,0x25CE,0x25C7, /* ku 2 */ @@ -51,7 +51,7 @@ static const unsigned short jisx0208_ucs_table[] = { 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x2208,0x220B,0x2286,0x2287,0x2282,0x2283,0x222A, 0x2229,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, - 0x0000,0x2227,0x2228,0x00AC,0x21D2,0x21D4,0x2200,0x2203, + 0x0000,0x2227,0x2228,0xFFE2,0x21D2,0x21D4,0x2200,0x2203, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x2220,0x22A5,0x2312,0x2202,0x2207, 0x2261,0x2252,0x226A,0x226B,0x221A,0x223D,0x221D,0x2235, @@ -1217,9 +1217,9 @@ static const unsigned short jisx0212_ucs_table[] = { /* ku 2 */ 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x02D8,0x02C7, - 0x00B8,0x02D9,0x02DD,0x00AF,0x02DB,0x02DA,0x007E,0x0384, + 0x00B8,0x02D9,0x02DD,0x00AF,0x02DB,0x02DA,0xFF5E,0x0384, 0x0385,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, - 0x0000,0x00A1,0x00A6,0x00BF,0x0000,0x0000,0x0000,0x0000, + 0x0000,0x00A1,0xFFE4,0x00BF,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, @@ -2471,7 +2471,7 @@ static const unsigned short ucs_a2_jis_table[] = { /* 2100h */ 0x0000,0x0000,0x0000,0x216E,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, - 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0xA2F1,0x0000, + 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x2d62,0x0000, 0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0xA2EF,0x0000,0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x2272,0x0000,0x0000,0x0000,0x0000, diff --git a/ext/mbstring/libmbfl/mbfl/mbfilter.c b/ext/mbstring/libmbfl/mbfl/mbfilter.c index ed674b2b65..a8e3b68d33 100644 --- a/ext/mbstring/libmbfl/mbfl/mbfilter.c +++ b/ext/mbstring/libmbfl/mbfl/mbfilter.c @@ -470,7 +470,6 @@ enum mbfl_no_encoding mbfl_encoding_detector_judge(mbfl_encoding_detector *ident n--; } - /* fallback judge */ if (encoding == mbfl_no_encoding_invalid) { n = identd->filter_list_size - 1; while (n >= 0) { @@ -631,7 +630,6 @@ mbfl_identify_encoding(mbfl_string *string, enum mbfl_no_encoding *elist, int el } } - /* cleanup */ /* dtors should be called in reverse order */ i = num; while (--i >= 0) { diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c index 78a995cd25..bcae8c4340 100644 --- a/ext/mbstring/libmbfl/mbfl/mbfl_convert.c +++ b/ext/mbstring/libmbfl/mbfl/mbfl_convert.c @@ -51,6 +51,7 @@ #include "filters/mbfilter_euc_kr.h" #include "filters/mbfilter_iso2022_kr.h" #include "filters/mbfilter_sjis.h" +#include "filters/mbfilter_cp51932.h" #include "filters/mbfilter_jis.h" #include "filters/mbfilter_euc_jp.h" #include "filters/mbfilter_euc_jp_win.h" @@ -104,10 +105,14 @@ const struct mbfl_convert_vtbl *mbfl_convert_filter_list[] = { &vtbl_wchar_eucjp, &vtbl_sjis_wchar, &vtbl_wchar_sjis, + &vtbl_cp51932_wchar, + &vtbl_wchar_cp51932, &vtbl_jis_wchar, &vtbl_wchar_jis, &vtbl_2022jp_wchar, &vtbl_wchar_2022jp, + &vtbl_2022jpms_wchar, + &vtbl_wchar_2022jpms, &vtbl_eucjpwin_wchar, &vtbl_wchar_eucjpwin, &vtbl_sjiswin_wchar, diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_encoding.c b/ext/mbstring/libmbfl/mbfl/mbfl_encoding.c index 64d8cb4e47..3c18fca974 100644 --- a/ext/mbstring/libmbfl/mbfl/mbfl_encoding.c +++ b/ext/mbstring/libmbfl/mbfl/mbfl_encoding.c @@ -57,6 +57,7 @@ #include "filters/mbfilter_euc_kr.h" #include "filters/mbfilter_iso2022_kr.h" #include "filters/mbfilter_sjis.h" +#include "filters/mbfilter_cp51932.h" #include "filters/mbfilter_jis.h" #include "filters/mbfilter_euc_jp.h" #include "filters/mbfilter_euc_jp_win.h" @@ -149,8 +150,10 @@ static const mbfl_encoding *mbfl_encoding_ptr_list[] = { &mbfl_encoding_sjis, &mbfl_encoding_eucjp_win, &mbfl_encoding_sjis_win, + &mbfl_encoding_cp51932, &mbfl_encoding_jis, &mbfl_encoding_2022jp, + &mbfl_encoding_2022jpms, &mbfl_encoding_cp1252, &mbfl_encoding_8859_1, &mbfl_encoding_8859_2, diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_encoding.h b/ext/mbstring/libmbfl/mbfl/mbfl_encoding.h index d38b1d7d4a..af93b6ea50 100644 --- a/ext/mbstring/libmbfl/mbfl/mbfl_encoding.h +++ b/ext/mbstring/libmbfl/mbfl/mbfl_encoding.h @@ -70,8 +70,10 @@ enum mbfl_no_encoding { mbfl_no_encoding_eucjp_win, mbfl_no_encoding_sjis_win, mbfl_no_encoding_sjis_mac, + mbfl_no_encoding_cp51932, mbfl_no_encoding_jis, mbfl_no_encoding_2022jp, + mbfl_no_encoding_2022jpms, mbfl_no_encoding_cp1252, mbfl_no_encoding_8859_1, mbfl_no_encoding_8859_2, diff --git a/ext/mbstring/libmbfl/mbfl/mbfl_ident.c b/ext/mbstring/libmbfl/mbfl/mbfl_ident.c index 72ccef68d2..40c964c205 100644 --- a/ext/mbstring/libmbfl/mbfl/mbfl_ident.c +++ b/ext/mbstring/libmbfl/mbfl/mbfl_ident.c @@ -106,6 +106,7 @@ static const struct mbfl_identify_vtbl *mbfl_identify_filter_list[] = { &vtbl_identify_sjiswin, &vtbl_identify_jis, &vtbl_identify_2022jp, + &vtbl_identify_2022jpms, &vtbl_identify_euccn, &vtbl_identify_cp936, &vtbl_identify_hz, diff --git a/ext/mbstring/mb_gpc.c b/ext/mbstring/mb_gpc.c index d5adaddd0e..46b1f41283 100644 --- a/ext/mbstring/mb_gpc.c +++ b/ext/mbstring/mb_gpc.c @@ -153,6 +153,8 @@ MBSTRING_API SAPI_TREAT_DATA_FUNC(mbstr_treat_data) info.num_from_encodings = MBSTRG(http_input_list_size); info.from_language = MBSTRG(language); + MBSTRG(illegalchars) = 0; + detected = _php_mb_encoding_handler_ex(&info, array_ptr, res TSRMLS_CC); MBSTRG(http_input_identify) = detected; @@ -331,6 +333,7 @@ enum mbfl_no_encoding _php_mb_encoding_handler_ex(const php_mb_encoding_handler_ out: if (convd != NULL) { + MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd); mbfl_buffer_converter_delete(convd); } if (val_list != NULL) { diff --git a/ext/mbstring/mbstring.c b/ext/mbstring/mbstring.c index 41a30e9990..fd6143acc7 100644 --- a/ext/mbstring/mbstring.c +++ b/ext/mbstring/mbstring.c @@ -60,6 +60,7 @@ #include "ext/standard/url.h" #include "main/php_output.h" #include "ext/standard/info.h" + #include "libmbfl/mbfl/mbfl_allocators.h" #include "php_variables.h" @@ -936,7 +937,10 @@ PHP_RINIT_FUNCTION(mbstring) MBSTRG(current_http_output_encoding) = MBSTRG(http_output_encoding); MBSTRG(current_filter_illegal_mode) = MBSTRG(filter_illegal_mode); MBSTRG(current_filter_illegal_substchar) = MBSTRG(filter_illegal_substchar); - MBSTRG(illegalchars) = 0; + + if (!MBSTRG(encoding_translation)) { + MBSTRG(illegalchars) = 0; + } n = 0; if (MBSTRG(detect_order_list)) { @@ -967,14 +971,14 @@ PHP_RINIT_FUNCTION(mbstring) zend_hash_find(EG(function_table), p->ovld_func, strlen(p->ovld_func)+1 , (void **)&func); if (zend_hash_find(EG(function_table), p->orig_func, strlen(p->orig_func)+1, (void **)&orig) != SUCCESS) { - php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't find function %s", p->orig_func); + php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't find function %s.", p->orig_func); return FAILURE; } else { zend_hash_add(EG(function_table), p->save_func, strlen(p->save_func)+1, orig, sizeof(zend_function), NULL); if (zend_hash_update(EG(function_table), p->orig_func, strlen(p->orig_func)+1, func, sizeof(zend_function), NULL) == FAILURE) { - php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't replace function %s", p->orig_func); + php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't replace function %s.", p->orig_func); return FAILURE; } } @@ -1341,7 +1345,7 @@ PHP_FUNCTION(mb_substitute_character) MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR; MBSTRG(current_filter_illegal_substchar) = Z_LVAL_PP(arg1); } else { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character."); RETVAL_FALSE; } } @@ -1352,7 +1356,7 @@ PHP_FUNCTION(mb_substitute_character) MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR; MBSTRG(current_filter_illegal_substchar) = Z_LVAL_PP(arg1); } else { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character."); RETVAL_FALSE; } break; @@ -1599,11 +1603,11 @@ PHP_FUNCTION(mb_strpos) } if (offset < 0 || (unsigned long)offset > haystack.len) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset is out of range"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset not contained in string."); RETURN_FALSE; } if (needle.len == 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty needle"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty delimiter."); RETURN_FALSE; } @@ -1615,16 +1619,16 @@ PHP_FUNCTION(mb_strpos) case 1: break; case 2: - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Needle has non-positive length"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Needle has not positive length."); break; case 4: - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding or conversion error"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding or conversion error."); break; case 8: - php_error_docref(NULL TSRMLS_CC, E_NOTICE, "Argument is empty"); + php_error_docref(NULL TSRMLS_CC, E_NOTICE, "Argument is empty."); break; default: - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown error in mb_strpos"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown error in mb_strpos."); break; } RETVAL_FALSE; @@ -1637,10 +1641,13 @@ PHP_FUNCTION(mb_strpos) PHP_FUNCTION(mb_strrpos) { int n; - long offset = 0; mbfl_string haystack, needle; char *enc_name = NULL; int enc_name_len; + zval **zoffset; + long offset = 0, str_flg; + char *enc_name2 = NULL; + int enc_name_len2; mbfl_string_init(&haystack); mbfl_string_init(&needle); @@ -1649,10 +1656,51 @@ PHP_FUNCTION(mb_strrpos) needle.no_language = MBSTRG(current_language); needle.no_encoding = MBSTRG(current_internal_encoding); - if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|ls", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &offset, &enc_name, &enc_name_len) == FAILURE) { + if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|Zs", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &zoffset, &enc_name, &enc_name_len) == FAILURE) { RETURN_FALSE; } + if(ZEND_NUM_ARGS() >= 3) { + if (Z_TYPE_PP(zoffset) == IS_STRING) { + enc_name2 = Z_STRVAL_PP(zoffset); + enc_name_len2 = Z_STRLEN_PP(zoffset); + str_flg = 1; + + if (enc_name2 != NULL) { + switch (*enc_name2) { + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + case '8': + case '9': + case ' ': + case '-': + case '.': + break; + default : + str_flg = 0; + break; + } + } + + if(str_flg) { + convert_to_long_ex(zoffset); + offset = Z_LVAL_PP(zoffset); + } else { + enc_name = enc_name2; + enc_name_len = enc_name_len2; + } + } else { + convert_to_long_ex(zoffset); + offset = Z_LVAL_PP(zoffset); + } + } + if (enc_name != NULL) { haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name); if (haystack.no_encoding == mbfl_no_encoding_invalid) { @@ -1662,11 +1710,9 @@ PHP_FUNCTION(mb_strrpos) } if (haystack.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty haystack"); RETURN_FALSE; } if (needle.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty needle"); RETURN_FALSE; } n = mbfl_strpos(&haystack, &needle, offset, 1); @@ -1720,6 +1766,10 @@ PHP_FUNCTION(mb_strripos) RETURN_FALSE; } + if(offset > old_haystack_len){ + RETURN_FALSE; + } + n = php_mb_stripos(1, old_haystack, old_haystack_len, old_needle, old_needle_len, offset, from_encoding TSRMLS_CC); if (n >= 0) { @@ -1759,12 +1809,8 @@ PHP_FUNCTION(mb_strstr) } } - if (haystack.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty haystack"); - RETURN_FALSE; - } if (needle.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty needle"); + php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty delimiter."); RETURN_FALSE; } n = mbfl_strpos(&haystack, &needle, 0, 0); @@ -1821,8 +1867,10 @@ PHP_FUNCTION(mb_strrchr) } } + if (haystack.len <= 0) { + RETURN_FALSE; + } if (needle.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty needle"); RETURN_FALSE; } n = mbfl_strpos(&haystack, &needle, 0, 1); @@ -1870,6 +1918,11 @@ PHP_FUNCTION(mb_stristr) RETURN_FALSE; } + if(!needle.len){ + php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty delimiter."); + RETURN_FALSE; + } + haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding); if (haystack.no_encoding == mbfl_no_encoding_invalid) { php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", from_encoding); @@ -1983,7 +2036,7 @@ PHP_FUNCTION(mb_substr_count) } if (needle.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty needle"); + php_error_docref(NULL TSRMLS_CC, E_WARNING,"Empty substring."); RETURN_FALSE; } @@ -2166,7 +2219,7 @@ PHP_FUNCTION(mb_strcut) if (((unsigned) from + (unsigned) len) > Z_STRLEN_PP(arg1)) { len = Z_STRLEN_PP(arg1) - from; } - + ret = mbfl_strcut(&string, &result, from, len); if (ret != NULL) { RETVAL_STRINGL(ret->val, ret->len, 0); /* the string is already strdup()'ed */ @@ -2526,7 +2579,7 @@ PHP_FUNCTION(mb_detect_encoding) mbfl_string string; const char *ret; enum mbfl_no_encoding *elist; - int size, *list, strict; + int size, *list, strict = 0; if (ZEND_NUM_ARGS() == 1) { if (zend_get_parameters_ex(1, &arg_str) == FAILURE) { @@ -3904,8 +3957,8 @@ PHP_FUNCTION(mb_get_info) if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_body_encoding)) != NULL) { RETVAL_STRING(name, 1); } - } else if (!strcasecmp("illegal_chars", typ)) { - RETVAL_LONG(MBSTRG(illegalchars)); + } else if (!strcasecmp("illegal_chars", typ)) { + RETVAL_LONG(MBSTRG(illegalchars)); } else if (!strcasecmp("encoding_translation", typ)) { if (MBSTRG(encoding_translation)) { RETVAL_STRING("On", 1); @@ -4019,8 +4072,13 @@ PHP_FUNCTION(mb_check_encoding) if (ret != NULL) { MBSTRG(illegalchars) += illegalchars; - efree(ret->val); - RETURN_BOOL(illegalchars == 0); + if (illegalchars == 0 && strncmp(string.val, ret->val, string.len) == 0) { + efree(ret->val); + RETURN_TRUE; + } else { + efree(ret->val); + RETURN_FALSE; + } } else { RETURN_FALSE; } @@ -4297,7 +4355,6 @@ MBSTRING_API int php_mb_stripos(int mode, char *old_haystack, int old_haystack_l } if (haystack.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty haystack"); break; } @@ -4308,7 +4365,6 @@ MBSTRING_API int php_mb_stripos(int mode, char *old_haystack, int old_haystack_l } if (needle.len <= 0) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty needle"); break; } @@ -4319,7 +4375,7 @@ MBSTRING_API int php_mb_stripos(int mode, char *old_haystack, int old_haystack_l } if (offset < 0 || (unsigned long)offset > haystack.len) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset is out of range"); + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset not contained in string."); break; }