ICU-10128 encode new properties bpt & bpb in ubidi.icu format version 2.1

X-SVN-Rev: 33557

diff --git a/tools/unicode/c/genprops/bidipropsbuilder.cpp b/tools/unicode/c/genprops/bidipropsbuilder.cpp
index 30fbbce43819d02b7e88c4bfb6bcf73d9522e114..6f644a93eecfd74cd14e0d7fefabc7b540d670a4 100644 (file)
--- a/tools/unicode/c/genprops/bidipropsbuilder.cpp
+++ b/tools/unicode/c/genprops/bidipropsbuilder.cpp
@@ -1,7 +1,7 @@
 /*
 *******************************************************************************
 *
-*   Copyright (C) 2004-2012, International Business Machines
+*   Copyright (C) 2004-2013, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
@@ -43,7 +43,7 @@ the udata API for loading ICU data. Especially, a UDataInfo structure
 precedes the actual data. It contains platform properties values and the
 file format version.
 
-The following is a description of format version 2.0 .
+The following is a description of format version 2.1 .
 
 The file contains the following structures:
 
@@ -80,7 +80,7 @@ Bits
     12  is mirrored
     11  Bidi_Control
     10  Join_Control
- 9.. 8  reserved (set to 0)
+ 9.. 8  Bidi_Paired_Bracket_Type(bpt) -- new in format version 2.1
  7.. 5  Joining_Type
  4.. 0  BiDi category
 
@@ -120,6 +120,11 @@ containing the Joining_Group value.
 
 All code points outside of this range have No_Joining_Group (0).
 
+--- Changes in format version 2.1 ---
+
+Addition of Bidi_Paired_Bracket_Type(bpt) values.
+(Trie data bits 9..8 were reserved.)
+
 --- Changes in format version 2 ---
 
 Change from UTrie to UTrie2.
@@ -140,7 +145,7 @@ static UDataInfo dataInfo={
 
     /* dataFormat="BiDi" */
     { UBIDI_FMT_0, UBIDI_FMT_1, UBIDI_FMT_2, UBIDI_FMT_3 },
-    { 2, 0, 0, 0 },                             /* formatVersion */
+    { 2, 1, 0, 0 },                             /* formatVersion */
     { 6, 0, 0, 0 }                              /* dataVersion */
 };
 
@@ -252,6 +257,15 @@ BiDiPropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
     UChar32 start=props.start;
     UChar32 end=props.end;
 
+    // The runtime code relies on this invariant for returning both bmg and bpb
+    // from the same data.
+    int32_t bpt=props.getIntProp(UCHAR_BIDI_PAIRED_BRACKET_TYPE);
+    if(!(bpt==0 ? props.bpb==U_SENTINEL : props.bpb==props.bmg)) {
+        fprintf(stderr,
+                "genprops error: invariant not true: "
+                "if(bpt==None) then bpb=<none> else bpb=bmg\n");
+        return;
+    }
     int32_t delta=encodeBidiMirroringGlyph(start, end, props.bmg, errorCode);
     uint32_t value=(uint32_t)delta<<UBIDI_MIRROR_DELTA_SHIFT;
     if(props.binProps[UCHAR_BIDI_MIRRORED]) {
@@ -263,6 +277,7 @@ BiDiPropsBuilder::setProps(const UniProps &props, const UnicodeSet &newValues,
     if(props.binProps[UCHAR_JOIN_CONTROL]) {
         value|=U_MASK(UBIDI_JOIN_CONTROL_SHIFT);
     }
+    value|=(uint32_t)bpt<<UBIDI_BPT_SHIFT;
     value|=(uint32_t)props.getIntProp(UCHAR_JOINING_TYPE)<<UBIDI_JT_SHIFT;
     value|=(uint32_t)props.getIntProp(UCHAR_BIDI_CLASS);
     utrie2_setRange32(pTrie, start, end, value, TRUE, &errorCode);

diff --git a/tools/unicode/c/genprops/pnames_data.h b/tools/unicode/c/genprops/pnames_data.h
index 24ba09f60c2e049c550cffccac6398b17c15dbb6..712dbaa21270112865fe18f02e4bc705bb6f469c 100644 (file)
--- a/tools/unicode/c/genprops/pnames_data.h
+++ b/tools/unicode/c/genprops/pnames_data.h
@@ -1,18 +1,18 @@
 /**
- * Copyright (C) 2002-2012, International Business Machines Corporation and
+ * Copyright (C) 2002-2013, International Business Machines Corporation and
  * others. All Rights Reserved.
  *
  * machine-generated by: icu/tools/unicode/py/preparseucd.py
  */
 
-#define UNICODE_VERSION { 6, 2, 0, 0 }
+#define UNICODE_VERSION { 6, 3, 0, 0 }
 
 static const Value VALUES_binprop[2] = {
     Value(0, "N No F False"),
     Value(1, "Y Yes T True"),
 };
 
-static const Value VALUES_bc[19] = {
+static const Value VALUES_bc[23] = {
     Value(U_LEFT_TO_RIGHT, "L Left_To_Right"),
     Value(U_RIGHT_TO_LEFT, "R Right_To_Left"),
     Value(U_EUROPEAN_NUMBER, "EN European_Number"),
@@ -32,6 +32,10 @@ static const Value VALUES_bc[19] = {
     Value(U_POP_DIRECTIONAL_FORMAT, "PDF Pop_Directional_Format"),
     Value(U_DIR_NON_SPACING_MARK, "NSM Nonspacing_Mark"),
     Value(U_BOUNDARY_NEUTRAL, "BN Boundary_Neutral"),
+    Value(U_FIRST_STRONG_ISOLATE, "FSI First_Strong_Isolate"),
+    Value(U_LEFT_TO_RIGHT_ISOLATE, "LRI Left_To_Right_Isolate"),
+    Value(U_RIGHT_TO_LEFT_ISOLATE, "RLI Right_To_Left_Isolate"),
+    Value(U_POP_DIRECTIONAL_ISOLATE, "PDI Pop_Directional_Isolate"),
 };
 
 static const Value VALUES_blk[221] = {
@@ -848,7 +852,7 @@ static const Value VALUES_SB[15] = {
     Value(U_SB_SCONTINUE, "SC SContinue"),
 };
 
-static const Value VALUES_WB[14] = {
+static const Value VALUES_WB[17] = {
     Value(U_WB_OTHER, "XX Other"),
     Value(U_WB_ALETTER, "LE ALetter"),
     Value(U_WB_FORMAT, "FO Format"),
@@ -863,6 +867,15 @@ static const Value VALUES_WB[14] = {
     Value(U_WB_MIDNUMLET, "MB MidNumLet"),
     Value(U_WB_NEWLINE, "NL Newline"),
     Value(U_WB_REGIONAL_INDICATOR, "RI Regional_Indicator"),
+    Value(U_WB_HEBREW_LETTER, "HL Hebrew_Letter"),
+    Value(U_WB_SINGLE_QUOTE, "SQ Single_Quote"),
+    Value(U_WB_DOUBLE_QUOTE, "DQ Double_Quote"),
+};
+
+static const Value VALUES_bpt[3] = {
+    Value(U_BPT_NONE, "n None"),
+    Value(U_BPT_OPEN, "o Open"),
+    Value(U_BPT_CLOSE, "c Close"),
 };
 
 static const Value VALUES_gcm[38] = {
@@ -906,7 +919,7 @@ static const Value VALUES_gcm[38] = {
     Value((int32_t)U_GC_ZS_MASK, "Zs Space_Separator"),
 };
 
-static const Property PROPERTIES[94] = {
+static const Property PROPERTIES[96] = {
     Property(UCHAR_ALPHABETIC, "Alpha Alphabetic"),
     Property(UCHAR_ASCII_HEX_DIGIT, "AHex ASCII_Hex_Digit"),
     Property(UCHAR_BIDI_CONTROL, "Bidi_C Bidi_Control"),
@@ -964,7 +977,7 @@ static const Property PROPERTIES[94] = {
     Property(UCHAR_CHANGES_WHEN_CASEFOLDED, "CWCF Changes_When_Casefolded"),
     Property(UCHAR_CHANGES_WHEN_CASEMAPPED, "CWCM Changes_When_Casemapped"),
     Property(UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED, "CWKCF Changes_When_NFKC_Casefolded"),
-    Property(UCHAR_BIDI_CLASS, "bc Bidi_Class", VALUES_bc, 19),
+    Property(UCHAR_BIDI_CLASS, "bc Bidi_Class", VALUES_bc, 23),
     Property(UCHAR_BLOCK, "blk Block", VALUES_blk, 221),
     Property(UCHAR_CANONICAL_COMBINING_CLASS, "ccc Canonical_Combining_Class", VALUES_ccc, 57),
     Property(UCHAR_DECOMPOSITION_TYPE, "dt Decomposition_Type", VALUES_dt, 18),
@@ -984,7 +997,8 @@ static const Property PROPERTIES[94] = {
     Property(UCHAR_TRAIL_CANONICAL_COMBINING_CLASS, "tccc Trail_Canonical_Combining_Class", VALUES_tccc, 57),
     Property(UCHAR_GRAPHEME_CLUSTER_BREAK, "GCB Grapheme_Cluster_Break", VALUES_GCB, 13),
     Property(UCHAR_SENTENCE_BREAK, "SB Sentence_Break", VALUES_SB, 15),
-    Property(UCHAR_WORD_BREAK, "WB Word_Break", VALUES_WB, 14),
+    Property(UCHAR_WORD_BREAK, "WB Word_Break", VALUES_WB, 17),
+    Property(UCHAR_BIDI_PAIRED_BRACKET_TYPE, "bpt Bidi_Paired_Bracket_Type", VALUES_bpt, 3),
     Property(UCHAR_GENERAL_CATEGORY_MASK, "gcm General_Category_Mask", VALUES_gcm, 38),
     Property(UCHAR_NUMERIC_VALUE, "nv Numeric_Value"),
     Property(UCHAR_AGE, "age Age"),
@@ -1000,6 +1014,7 @@ static const Property PROPERTIES[94] = {
     Property(UCHAR_TITLECASE_MAPPING, "tc Titlecase_Mapping"),
     Property(UCHAR_UNICODE_1_NAME, "na1 Unicode_1_Name"),
     Property(UCHAR_UPPERCASE_MAPPING, "uc Uppercase_Mapping"),
+    Property(UCHAR_BIDI_PAIRED_BRACKET, "bpb Bidi_Paired_Bracket"),
     Property(UCHAR_SCRIPT_EXTENSIONS, "scx Script_Extensions"),
 };
 

diff --git a/tools/unicode/c/genuca/genuca.cpp b/tools/unicode/c/genuca/genuca.cpp
index b854159893a48f872242d8c4fb4f19f950608b3e..c1f210c08618040605390b8e4a492a9cdf5c9838 100644 (file)
--- a/tools/unicode/c/genuca/genuca.cpp
+++ b/tools/unicode/c/genuca/genuca.cpp
@@ -1,7 +1,7 @@
 /*
 *******************************************************************************
 *
-*   Copyright (C) 2000-2012, International Business Machines
+*   Copyright (C) 2000-2013, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *
 *******************************************************************************
@@ -705,7 +705,7 @@ UCAElements *readAnElement(FILE *data, tempUCATable *t, UCAConstants *consts, Le
                 u_versionToString(UCAVersion, uca);
                 u_versionToString(UCDVersion, ucd);
                 // Warning, not error, to permit bootstrapping during a version upgrade.
-                fprintf(stderr, "warning: UCA version %s != UCD version %s (temporarily change the FractionalUCA.txt UCA version during Unicode version upgrade)\n", uca, ucd);
+                fprintf(stderr, "warning: UCA version %s != UCD version %s\n", uca, ucd);
                 // *status = U_INVALID_FORMAT_ERROR;
                 // return NULL;
               }

diff --git a/tools/unicode/py/preparseucd.py b/tools/unicode/py/preparseucd.py
index 3030d712c1ff3ad91255463d64c815a6de8a4d8c..f64a8f705d79e4ab222a10692c04280a86d33998 100755 (executable)
--- a/tools/unicode/py/preparseucd.py
+++ b/tools/unicode/py/preparseucd.py
@@ -1781,7 +1781,7 @@ _ublock_re = re.compile(" *(UBLOCK_[0-9A-Z_]+) *= *[0-9]+,")
 # Sample line to match:
 #    U_EA_AMBIGUOUS,
 _prop_and_value_re = re.compile(
-    " *(U_(DT|EA|GCB|HST|LB|JG|JT|NT|SB|WB)_([0-9A-Z_]+))")
+    " *(U_(BPT|DT|EA|GCB|HST|LB|JG|JT|NT|SB|WB)_([0-9A-Z_]+))")
 
 # Sample line to match if it has matched _prop_and_value_re
 # (we want to exclude aliases):