ICU-8972 genprops: use ppucd.txt for the age property; some cleanup

X-SVN-Rev: 31141

diff --git a/tools/unicode/c/genprops/corepropswriter.cpp b/tools/unicode/c/genprops/corepropswriter.cpp
index 0341fdf24a4fd6d58d5f31f8fe35e4de6f872f8d..3b6784ca6b32218d2699bdd6db6b3f18693d23cc 100644 (file)
--- a/tools/unicode/c/genprops/corepropswriter.cpp
+++ b/tools/unicode/c/genprops/corepropswriter.cpp
@@ -251,22 +251,20 @@ static UTrie2 *pTrie=NULL;
 
 /* -------------------------------------------------------------------------- */
 
-U_CFUNC void
+static void
 initStore() {
     UErrorCode errorCode=U_ZERO_ERROR;
     pTrie=utrie2_open(0, 0, &errorCode);
     if(U_FAILURE(errorCode)) {
-        fprintf(stderr, "error: utrie2_open() failed - %s\n", u_errorName(errorCode));
+        fprintf(stderr, "genprops error: corepropswriter utrie2_open() failed - %s\n",
+                u_errorName(errorCode));
         exit(errorCode);
     }
-
-    initAdditionalProperties();
 }
 
-U_CFUNC void
+static void
 exitStore() {
     utrie2_close(pTrie);
-    exitAdditionalProperties();
 }
 
 /* store a character's properties ------------------------------------------- */
@@ -419,7 +417,8 @@ generateData(const char *dataDir, UBool csource) {
 
     if(csource) {
         /* write .c file for hardcoded data */
-        FILE *f=usrc_create(dataDir, "uchar_props_data.h");
+        FILE *f=usrc_createFromGenerator(dataDir, "uchar_props_data.h",
+                                         "icu/tools/src/unicode/c/genprops/corepropswriter.cpp");
         if(f!=NULL) {
             fputs("#ifndef INCLUDED_FROM_UCHAR_C\n"
                   "#   error This file must be #included from uchar.c only.\n"
@@ -491,6 +490,9 @@ generateData(const char *dataDir, UBool csource) {
 
 class CorePropsWriter : public PropsWriter {
 public:
+    CorePropsWriter() { initStore(); }
+    virtual ~CorePropsWriter() { exitStore(); }
+
     virtual void setUnicodeVersion(const UVersionInfo version);
     virtual void setProps(const UniProps &, const UnicodeSet &newValues, UErrorCode &errorCode);
 };

diff --git a/tools/unicode/c/genprops/genprops.cpp b/tools/unicode/c/genprops/genprops.cpp
index dfe841f8c1e118e2cad9284965bdeeddceaae344..930f9bdaf409e1d9bca77d3aebe33d4411bb255d 100644 (file)
--- a/tools/unicode/c/genprops/genprops.cpp
+++ b/tools/unicode/c/genprops/genprops.cpp
@@ -50,6 +50,7 @@ U_NAMESPACE_USE
 
 UBool beVerbose=FALSE, haveCopyright=TRUE;
 
+PropsWriter::~PropsWriter() {}
 void PropsWriter::setUnicodeVersion(const UVersionInfo version) {}
 void PropsWriter::setProps(const UniProps &, const UnicodeSet &, UErrorCode &) {}
 
@@ -142,8 +143,6 @@ main(int argc, char* argv[]) {
     destDir=options[DESTDIR].value;
 
     /* initialize */
-    initStore();
-
     IcuToolErrorCode errorCode("genprops");
     LocalPointer<PropsWriter> corePropsWriter(createCorePropsWriter(errorCode));
     LocalPointer<PropsWriter> props2Writer(createProps2Writer(errorCode));
@@ -163,7 +162,6 @@ main(int argc, char* argv[]) {
     }
     PreparsedUCD::LineType lineType;
     UnicodeSet newValues;
-    int i=0;
     while((lineType=ppucd.readLine(errorCode))!=PreparsedUCD::NO_LINE) {
         if(ppucd.lineHasPropertyValues()) {
             const UniProps *props=ppucd.getProps(newValues, errorCode);
@@ -172,9 +170,13 @@ main(int argc, char* argv[]) {
             const UVersionInfo &version=ppucd.getUnicodeVersion();
             corePropsWriter->setUnicodeVersion(version);
         }
-        ++i;
+        if(errorCode.isFailure()) {
+            fprintf(stderr,
+                    "genprops: error parsing or setting values from ppucd.txt line %ld - %s\n",
+                    (long)ppucd.getLineNumber(), errorCode.errorName());
+            return errorCode.reset();
+        }
     }
-    printf("*** parsed %d lines from ppucd.txt\n", i);
 
     if(argc>=2) {
         suffix=argv[1];
@@ -207,8 +209,6 @@ main(int argc, char* argv[]) {
         generateData(destDir, options[CSOURCE].doesOccur);
     }
 
-    exitStore();
-    u_cleanup();
     return errorCode;
 }
 

diff --git a/tools/unicode/c/genprops/genprops.h b/tools/unicode/c/genprops/genprops.h
index d450606ec64d9a7c22ce28f168ec63dbe97b0dae..630dd5e1665d788afdb49c45793045f150ad653a 100644 (file)
--- a/tools/unicode/c/genprops/genprops.h
+++ b/tools/unicode/c/genprops/genprops.h
@@ -28,6 +28,7 @@
 
 class PropsWriter {
 public:
+    virtual ~PropsWriter();
     virtual void setUnicodeVersion(const UVersionInfo version);
     virtual void setProps(const UniProps &props, const UnicodeSet &newValues, UErrorCode &errorCode);
     // virtual writeCSourceFile(icusrcroot);
@@ -51,9 +52,6 @@ U_CFUNC UBool beVerbose, haveCopyright;
 U_CFUNC const char *const
 genCategoryNames[];
 
-/* properties vectors in props2.cpp */
-U_CFUNC UPropsVectors *pv;
-
 /* prototypes */
 U_CFUNC void
 writeUCDFilename(char *basename, const char *filename, const char *suffix);
@@ -64,12 +62,6 @@ isToken(const char *token, const char *s);
 U_CFUNC int32_t
 getTokenIndex(const char *const tokens[], int32_t countTokens, const char *s);
 
-U_CFUNC void
-initStore(void);
-
-U_CFUNC void
-exitStore(void);
-
 U_CFUNC uint32_t
 makeProps(Props *p);
 
@@ -85,13 +77,6 @@ repeatProps(uint32_t first, uint32_t last, uint32_t props);
 U_CFUNC void
 generateData(const char *dataDir, UBool csource);
 
-/* props2.c */
-U_CFUNC void
-initAdditionalProperties(void);
-
-U_CFUNC void
-exitAdditionalProperties(void);
-
 U_CFUNC void
 generateAdditionalProperties(char *filename, const char *suffix, UErrorCode *pErrorCode);
 

diff --git a/tools/unicode/c/genprops/props2writer.cpp b/tools/unicode/c/genprops/props2writer.cpp
index b776fcb0c4ed9344f753c08794aa34d7899bea4b..c6e498d36a940a4735c2ddad62d226f718259d30 100644 (file)
--- a/tools/unicode/c/genprops/props2writer.cpp
+++ b/tools/unicode/c/genprops/props2writer.cpp
@@ -38,7 +38,7 @@ U_NAMESPACE_USE
 /* data --------------------------------------------------------------------- */
 
 static UTrie2 *newTrie;
-UPropsVectors *pv;
+static UPropsVectors *pv;
 
 static UnicodeString *scriptExtensions;
 
@@ -77,11 +77,6 @@ parseTwoFieldFile(char *filename, char *basename,
     }
 }
 
-static void U_CALLCONV
-ageLineFn(void *context,
-          char *fields[][2], int32_t fieldCount,
-          UErrorCode *pErrorCode);
-
 static void U_CALLCONV
 scriptExtensionsLineFn(void *context,
                        char *fields[][2], int32_t fieldCount,
@@ -417,18 +412,19 @@ parseBinariesFile(char *filename, char *basename, const char *suffix,
 
 /* -------------------------------------------------------------------------- */
 
-U_CFUNC void
+static void
 initAdditionalProperties() {
     UErrorCode errorCode=U_ZERO_ERROR;
     pv=upvec_open(UPROPS_VECTOR_WORDS, &errorCode);
     if(U_FAILURE(errorCode)) {
-        fprintf(stderr, "error: upvec_open() failed - %s\n", u_errorName(errorCode));
+        fprintf(stderr, "genprops error: props2writer upvec_open() failed - %s\n",
+                u_errorName(errorCode));
         exit(errorCode);
     }
     scriptExtensions=new UnicodeString;
 }
 
-U_CFUNC void
+static void
 exitAdditionalProperties() {
     utrie2_close(newTrie);
     upvec_close(pv);
@@ -446,8 +442,6 @@ generateAdditionalProperties(char *filename, const char *suffix, UErrorCode *pEr
     /* add Han numeric types & values */
     parseMultiFieldFile(filename, basename, "DerivedNumericValues", suffix, 2, numericLineFn, pErrorCode);
 
-    parseTwoFieldFile(filename, basename, "DerivedAge", suffix, ageLineFn, pErrorCode);
-
     parseSingleEnumFile(filename, basename, suffix, &scriptSingleEnum, pErrorCode);
 
     parseTwoFieldFile(filename, basename, "ScriptExtensions", suffix, scriptExtensionsLineFn, pErrorCode);
@@ -524,59 +518,6 @@ for(int32_t c=0; c<=0x10ffff; ++c) {
     }
 }
 
-/* DerivedAge.txt ----------------------------------------------------------- */
-
-static void U_CALLCONV
-ageLineFn(void *context,
-          char *fields[][2], int32_t fieldCount,
-          UErrorCode *pErrorCode) {
-    char *s, *numberLimit;
-    uint32_t value, start, end, version;
-
-    u_parseCodePointRange(fields[0][0], &start, &end, pErrorCode);
-    if(U_FAILURE(*pErrorCode)) {
-        fprintf(stderr, "genprops: syntax error in DerivedAge.txt field 0 at %s\n", fields[0][0]);
-        exit(*pErrorCode);
-    }
-
-    /* ignore "unassigned" (the default is already set to 0.0) */
-    s=(char *)u_skipWhitespace(fields[1][0]);
-    if(0==uprv_strncmp(s, "unassigned", 10) || 0==uprv_strncmp(s, "Unassigned", 10)) {
-        return;
-    }
-
-    /* parse version number */
-    value=(uint32_t)uprv_strtoul(s, &numberLimit, 10);
-    if(s==numberLimit || value==0 || value>15 || (*numberLimit!='.' && *numberLimit!=' ' && *numberLimit!='\t' && *numberLimit!=0)) {
-        fprintf(stderr, "genprops: syntax error in DerivedAge.txt field 1 at %s\n", fields[1][0]);
-        *pErrorCode=U_PARSE_ERROR;
-        exit(U_PARSE_ERROR);
-    }
-    version=value<<4;
-
-    /* parse minor version number */
-    if(*numberLimit=='.') {
-        s=(char *)u_skipWhitespace(numberLimit+1);
-        value=(uint32_t)uprv_strtoul(s, &numberLimit, 10);
-        if(s==numberLimit || value>15 || (*numberLimit!=' ' && *numberLimit!='\t' && *numberLimit!=0)) {
-            fprintf(stderr, "genprops: syntax error in DerivedAge.txt field 1 at %s\n", fields[1][0]);
-            *pErrorCode=U_PARSE_ERROR;
-            exit(U_PARSE_ERROR);
-        }
-        version|=value;
-    }
-
-    if(start==0 && end==0x10ffff) {
-        /* Also set bits for initialValue and errorValue. */
-        end=UPVEC_MAX_CP;
-    }
-    upvec_setValue(pv, start, end, 0, version<<UPROPS_AGE_SHIFT, UPROPS_AGE_MASK, pErrorCode);
-    if(U_FAILURE(*pErrorCode)) {
-        fprintf(stderr, "genprops error: unable to set character age: %s\n", u_errorName(*pErrorCode));
-        exit(*pErrorCode);
-    }
-}
-
 /* ScriptExtensions.txt ----------------------------------------------------- */
 
 static void U_CALLCONV
@@ -940,21 +881,42 @@ writeAdditionalData(FILE *f, uint8_t *p, int32_t capacity, int32_t indexes[UPROP
 
 class Props2Writer : public PropsWriter {
 public:
+    Props2Writer() { initAdditionalProperties(); }
+    virtual ~Props2Writer() { exitAdditionalProperties(); }
+
     virtual void setProps(const UniProps &, const UnicodeSet &newValues, UErrorCode &errorCode);
 };
 
 void
 Props2Writer::setProps(const UniProps &props, const UnicodeSet &newValues, UErrorCode &errorCode) {
     if(U_FAILURE(errorCode)) { return; }
+    UChar32 start=props.start;
+    UChar32 end=props.end;
+    if(start==0 && end==0x10ffff) {
+        // Also set bits for initialValue and errorValue.
+        end=UPVEC_MAX_CP;
+    }
     if(newValues.contains(UCHAR_DECOMPOSITION_TYPE)) {
-        upvec_setValue(pv, props.start, props.end,
+        upvec_setValue(pv, start, end,
                        2, (uint32_t)props.getIntProp(UCHAR_DECOMPOSITION_TYPE), UPROPS_DT_MASK,
                        &errorCode);
-        if(U_FAILURE(errorCode)) {
-            fprintf(stderr, "genprops error: unable to set decomposition type: %s\n",
-                    u_errorName(errorCode));
-            exit(errorCode);
+    }
+    if(newValues.contains(UCHAR_AGE)) {
+        if(props.age[0]>15 || props.age[1]>15 || props.age[2]!=0 || props.age[3]!=0) {
+            char buffer[U_MAX_VERSION_STRING_LENGTH];
+            u_versionToString(props.age, buffer);
+            fprintf(stderr, "genprops error: age %s cannot be encoded\n", buffer);
+            errorCode=U_ILLEGAL_ARGUMENT_ERROR;
+            return;
         }
+        uint32_t version=(props.age[0]<<4)|props.age[1];
+        upvec_setValue(pv, start, end,
+                       0, version<<UPROPS_AGE_SHIFT, UPROPS_AGE_MASK,
+                       &errorCode);
+    }
+    if(U_FAILURE(errorCode)) {
+        fprintf(stderr, "genprops error: unable to set values for %04lX..%04lX: %s\n",
+                (long)start, (long)end, u_errorName(errorCode));
     }
 }