ICU-8972 various small fixes & improvements

X-SVN-Rev: 31136

diff --git a/icu4c/source/common/charstr.cpp b/icu4c/source/common/charstr.cpp
index cdd58281b38a425c7d0a90c881cb92ac0395b7af..76723d9790c1f7bdbdaf3a6a8c4ec47411c8c7e0 100644 (file)
--- a/icu4c/source/common/charstr.cpp
+++ b/icu4c/source/common/charstr.cpp
@@ -1,6 +1,6 @@
 /*
 *******************************************************************************
-*   Copyright (C) 2010, International Business Machines
+*   Copyright (C) 2010-2011, International Business Machines
 *   Corporation and others.  All Rights Reserved.
 *******************************************************************************
 *   file name:  charstr.cpp
@@ -127,4 +127,19 @@ UBool CharString::ensureCapacity(int32_t capacity,
     return TRUE;
 }
 
+CharString &CharString::appendPathPart(const StringPiece &s, UErrorCode &errorCode) {
+    if(U_FAILURE(errorCode)) {
+        return *this;
+    }
+    if(s.length()==0) {
+        return *this;
+    }
+    char c;
+    if(len>0 && (c=buffer[len-1])!=U_FILE_SEP_CHAR && c!=U_FILE_ALT_SEP_CHAR) {
+        append(U_FILE_SEP_CHAR, errorCode);
+    }
+    append(s, errorCode);
+    return *this;
+}
+
 U_NAMESPACE_END

diff --git a/icu4c/source/common/charstr.h b/icu4c/source/common/charstr.h
index da391e245d0b7a8e666b12c24f108f4fc34e39cb..b655361527913cb7fdc05c30b8262f20d893d2af 100644 (file)
--- a/icu4c/source/common/charstr.h
+++ b/icu4c/source/common/charstr.h
@@ -107,6 +107,13 @@ public:
 
     CharString &appendInvariantChars(const UnicodeString &s, UErrorCode &errorCode);
 
+    /**
+     * Appends a filename/path part, e.g., a directory name.
+     * First appends a U_FILE_SEP_CHAR if necessary.
+     * Does nothing if s is empty.
+     */
+    CharString &appendPathPart(const StringPiece &s, UErrorCode &errorCode);
+
 private:
     MaybeStackArray<char, 40> buffer;
     int32_t len;

diff --git a/icu4c/source/common/putil.cpp b/icu4c/source/common/putil.cpp
index e517a81561ee7887dd3348788f92e62cdb0028f5..f65eb1b366085d6f5f04389d714d50a22bc6b286 100644 (file)
--- a/icu4c/source/common/putil.cpp
+++ b/icu4c/source/common/putil.cpp
@@ -2033,7 +2033,7 @@ u_versionFromUString(UVersionInfo versionArray, const UChar *versionString) {
 }
 
 U_CAPI void U_EXPORT2
-u_versionToString(UVersionInfo versionArray, char *versionString) {
+u_versionToString(const UVersionInfo versionArray, char *versionString) {
     uint16_t count, part;
     uint8_t field;
 

diff --git a/icu4c/source/common/unicode/uversion.h b/icu4c/source/common/unicode/uversion.h
index 2f259f1c13a4d58484773e07ed1739bf0ace606f..74e309105514adf38237a3b7a409801a815a5d56 100644 (file)
--- a/icu4c/source/common/unicode/uversion.h
+++ b/icu4c/source/common/unicode/uversion.h
@@ -178,7 +178,7 @@ u_versionFromUString(UVersionInfo versionArray, const UChar *versionString);
  * @stable ICU 2.4
  */
 U_STABLE void U_EXPORT2
-u_versionToString(UVersionInfo versionArray, char *versionString);
+u_versionToString(const UVersionInfo versionArray, char *versionString);
 
 /**
  * Gets the ICU release version.  The version array stores the version information

diff --git a/icu4c/source/data/unidata/ppucd.txt b/icu4c/source/data/unidata/ppucd.txt
index 65d033c1f4413e1b5c5600fad253fdf86916c9c3..e2dd3ed32e0eeef0a45c66fbc1f0072931f7b9c9 100644 (file)
--- a/icu4c/source/data/unidata/ppucd.txt
+++ b/icu4c/source/data/unidata/ppucd.txt
@@ -17195,7 +17195,7 @@ cp;ABF8;-Alpha;gc=Nd;-IDS;InSC=Other;lb=NU;na=MEETEI MAYEK DIGIT EIGHT;nt=De;nv=
 cp;ABF9;-Alpha;gc=Nd;-IDS;InSC=Other;lb=NU;na=MEETEI MAYEK DIGIT NINE;nt=De;nv=9;SB=NU;WB=NU;-XIDS
 cp;ABFA..ABFF;age=NA;-Alpha;gc=Cn;-Gr_Base;-IDC;-IDS;InSC=Other;lb=XX;SB=XX;sc=Zzzz;WB=XX;-XIDC;-XIDS
 
-block;AC00..D7AF;age=2.0;Alpha;blk=Hangul;ea=W;gc=Lo;GCB=LVT;Gr_Base;IDC;IDS;lb=H3;NFD_QC=N;NFKD_QC=N;SB=LE;sc=Hang;WB=LE;XIDC;XIDS
+block;AC00..D7AF;age=2.0;Alpha;blk=Hangul;dt=Can;ea=W;gc=Lo;GCB=LVT;Gr_Base;IDC;IDS;lb=H3;NFD_QC=N;NFKD_QC=N;SB=LE;sc=Hang;WB=LE;XIDC;XIDS
 # AC00..D7A3 Hangul Syllables
 algnamesrange;AC00..D7A3;hangul
 cp;AC00;GCB=LV;lb=H2
@@ -17597,7 +17597,7 @@ cp;D734;GCB=LV;lb=H2
 cp;D750;GCB=LV;lb=H2
 cp;D76C;GCB=LV;lb=H2
 cp;D788;GCB=LV;lb=H2
-cp;D7A4..D7AF;age=NA;-Alpha;ea=N;gc=Cn;GCB=XX;-Gr_Base;-IDC;-IDS;lb=XX;NFD_QC=Y;NFKD_QC=Y;SB=XX;sc=Zzzz;WB=XX;-XIDC;-XIDS
+cp;D7A4..D7AF;age=NA;-Alpha;dt=None;ea=N;gc=Cn;GCB=XX;-Gr_Base;-IDC;-IDS;lb=XX;NFD_QC=Y;NFKD_QC=Y;SB=XX;sc=Zzzz;WB=XX;-XIDC;-XIDS
 
 block;D7B0..D7FF;age=5.2;Alpha;blk=Jamo_Ext_B;ea=W;gc=Lo;GCB=T;Gr_Base;IDC;IDS;lb=JT;SB=LE;sc=Hang;WB=LE;XIDC;XIDS
 # D7B0..D7FF Hangul Jamo Extended-B

diff --git a/icu4c/source/tools/toolutil/ppucd.cpp b/icu4c/source/tools/toolutil/ppucd.cpp
index e193ec722cebef9d061f05c90fbb1ac55f8fd680..dd59afdbd4e9b201b9b6391f3baea9ece641cce3 100644 (file)
--- a/icu4c/source/tools/toolutil/ppucd.cpp
+++ b/icu4c/source/tools/toolutil/ppucd.cpp
@@ -42,8 +42,8 @@ PropertyNames::getPropertyValueEnum(int32_t property, const char *name) const {
 UniProps::UniProps()
         : start(U_SENTINEL), end(U_SENTINEL),
           bmg(U_SENTINEL),
-          numericValue(NULL),
-          name(NULL), uni1Name(NULL), nameAlias(NULL) {
+          digitValue(-1), numericValue(NULL),
+          name(NULL), nameAlias(NULL) {
     memset(binProps, 0, sizeof(binProps));
     memset(intProps, 0, sizeof(intProps));
     memset(age, 0, 4);
@@ -94,7 +94,7 @@ static const char *lineTypeStrings[]={
     "property",
     "binary",
     "value",
-    "default",
+    "defaults",
     "block",
     "cp",
     "algnamesrange"
@@ -200,7 +200,7 @@ PreparsedUCD::getProps(UnicodeSet &newValues, UErrorCode &errorCode) {
     if(!parseCodePointRange(field, start, end, errorCode)) { return NULL; }
     UniProps *props;
     switch(lineType) {
-    case DEFAULT_LINE:
+    case DEFAULTS_LINE:
         if(defaultLineIndex>=0) {
             fprintf(stderr,
                     "error in preparsed UCD: second line with default properties on line %ld\n",
@@ -322,26 +322,41 @@ PreparsedUCD::parseProperty(UniProps &props, const char *field, UnicodeSet &newV
                 field, (long)lineNumber);
         errorCode=U_PARSE_ERROR;
     } else if(prop<UCHAR_INT_LIMIT) {
-        int32_t value;
-        if((value=pnames->getPropertyValueEnum(prop, v))==UCHAR_INVALID_CODE) {
+        int32_t value=pnames->getPropertyValueEnum(prop, v);
+        if(value==UCHAR_INVALID_CODE && prop==UCHAR_CANONICAL_COMBINING_CLASS) {
+            // TODO: Make getPropertyValueEnum(UCHAR_CANONICAL_COMBINING_CLASS, v) work.
+            char *end;
+            unsigned long ccc=uprv_strtoul(v, &end, 10);
+            if(v<end && *end==0 && ccc<=254) {
+                value=(int32_t)ccc;
+            }
+        }
+        if(value==UCHAR_INVALID_CODE) {
             fprintf(stderr,
                     "error in preparsed UCD: '%s' is not a valid value on line %ld\n",
                     field, (long)lineNumber);
             errorCode=U_PARSE_ERROR;
         } else {
-            props.intProps[prop]=value;
+            props.intProps[prop-UCHAR_INT_START]=value;
         }
+    } else if(*v=='<' && lineType==DEFAULTS_LINE) {
+        // Ignore default values like <code point>.
+        return TRUE;
     } else {
+        char c;
         switch(prop) {
         case UCHAR_NUMERIC_VALUE:
             props.numericValue=v;
+            c=*v;
+            if('0'<=c && c<='9' && v[1]==0) {
+                props.digitValue=c-'0';
+            } else {
+                props.digitValue=-1;
+            }
             break;
         case UCHAR_NAME:
             props.name=v;
             break;
-        case UCHAR_UNICODE_1_NAME:
-            props.uni1Name=v;
-            break;
         case UCHAR_AGE:
             u_versionFromString(props.age, v);  // Writes 0.0.0.0 if v is not numeric.
             break;

diff --git a/icu4c/source/tools/toolutil/ppucd.h b/icu4c/source/tools/toolutil/ppucd.h
index 652a5e0ac1c4d6434d40926922c443fcce7a4d1f..dbf95c31cfb63ab78bb87da3fe5712c21f967ece 100644 (file)
--- a/icu4c/source/tools/toolutil/ppucd.h
+++ b/icu4c/source/tools/toolutil/ppucd.h
@@ -40,14 +40,16 @@ struct U_TOOLUTIL_API UniProps {
     UniProps();
     ~UniProps();
 
+    int32_t getIntProp(int32_t prop) const { return intProps[prop-UCHAR_INT_START]; }
+
     UChar32 start, end;
     UBool binProps[UCHAR_BINARY_LIMIT];
     int32_t intProps[UCHAR_INT_LIMIT-UCHAR_INT_START];
     UVersionInfo age;
     UChar32 bmg;
+    int32_t digitValue;
     const char *numericValue;
     const char *name;
-    const char *uni1Name;
     const char *nameAlias;
     UnicodeString cf;
 };
@@ -71,7 +73,7 @@ public:
         VALUE_LINE,
 
         /** defaults;0000..10FFFF;age=NA;bc=L;... */
-        DEFAULT_LINE,
+        DEFAULTS_LINE,
         /** block;0000..007F;age=1.1;blk=ASCII;ea=Na;... */
         BLOCK_LINE,
         /** cp;0030;AHex;bc=EN;gc=Nd;na=DIGIT ZERO;... */
@@ -108,7 +110,7 @@ public:
     const UVersionInfo &getUnicodeVersion() const { return ucdVersion; }
 
     /** Returns TRUE if the current line has property values. */
-    UBool lineHasPropertyValues() const { return DEFAULT_LINE<=lineType && lineType<=CP_LINE; }
+    UBool lineHasPropertyValues() const { return DEFAULTS_LINE<=lineType && lineType<=CP_LINE; }
 
     /**
      * Parses properties from the current line.