ICU-11058 support nested collation rule imports; make the importer stateless

X-SVN-Rev: 36157

icu4c/source/data/coll/es.txt

@@ -14,117 +14,7 @@ es{
     collations{
         search{
             Sequence{
-                "[normalization on][suppressContractions [เ-ไ ເ-ໄ ꪵ ꪶ ꪹ ꪻ ꪼ]]"
+                "[import und-u-co-search]"
-                "&'='<'≠'"
-                "&ا"
-                "<<<ﺎ<<<ﺍ"
-                "<<آ"
-                "<<<ﺂ<<<ﺁ"
-                "<<أ"
-                "<<<ﺄ<<<ﺃ"
-                "<<إ"
-                "<<<ﺈ<<<ﺇ"
-                "&و"
-                "<<<ۥ"
-                "<<<ﻮ<<<ﻭ"
-                "<<ؤ"
-                "<<<ﺆ<<<ﺅ"
-                "&ي"
-                "<<<ۦ"
-                "<<<ﻳ<<<ﻴ<<<ﻲ<<<ﻱ"
-                "<<ئ"
-                "<<<ﺋ<<<ﺌ<<<ﺊ<<<ﺉ"
-                "<<ى"
-                "<<<ﯨ<<<ﯩ"
-                "<<<ﻰ<<<ﻯ"
-                "&ه"
-                "<<<ﻫ<<<ﻬ<<<ﻪ<<<ﻩ"
-                "<<ة"
-                "<<<ﺔ<<<ﺓ"
-                "&[last primary ignorable]<<׳"
-                "<<״"
-                "<<ـ"
-                "<<ฺ"
-                "&ᄀ"
-                "=ᆨ"
-                "&ᄀᄀ"
-                "=ᄁ=ᆩ"
-                "&ᄀᄉ"
-                "=ᆪ"
-                "&ᄂ"
-                "=ᆫ"
-                "&ᄂᄌ"
-                "=ᆬ"
-                "&ᄂᄒ"
-                "=ᆭ"
-                "&ᄃ"
-                "=ᆮ"
-                "&ᄃᄃ"
-                "=ᄄ"
-                "&ᄅ"
-                "=ᆯ"
-                "&ᄅᄀ"
-                "=ᆰ"
-                "&ᄅᄆ"
-                "=ᆱ"
-                "&ᄅᄇ"
-                "=ᆲ"
-                "&ᄅᄉ"
-                "=ᆳ"
-                "&ᄅᄐ"
-                "=ᆴ"
-                "&ᄅᄑ"
-                "=ᆵ"
-                "&ᄅᄒ"
-                "=ᆶ"
-                "&ᄆ"
-                "=ᆷ"
-                "&ᄇ"
-                "=ᆸ"
-                "&ᄇᄇ"
-                "=ᄈ"
-                "&ᄇᄉ"
-                "=ᆹ"
-                "&ᄉ"
-                "=ᆺ"
-                "&ᄉᄉ"
-                "=ᄊ=ᆻ"
-                "&ᄋ"
-                "=ᆼ"
-                "&ᄌ"
-                "=ᆽ"
-                "&ᄌᄌ"
-                "=ᄍ"
-                "&ᄎ"
-                "=ᆾ"
-                "&ᄏ"
-                "=ᆿ"
-                "&ᄐ"
-                "=ᇀ"
-                "&ᄑ"
-                "=ᇁ"
-                "&ᄒ"
-                "=ᇂ"
-                "&ᅡᅵ"
-                "=ᅢ"
-                "&ᅣᅵ"
-                "=ᅤ"
-                "&ᅥᅵ"
-                "=ᅦ"
-                "&ᅧᅵ"
-                "=ᅨ"
-                "&ᅩᅡ"
-                "=ᅪ"
-                "&ᅩᅡᅵ"
-                "=ᅫ"
-                "&ᅩᅵ"
-                "=ᅬ"
-                "&ᅮᅴ"
-                "=ᅯ"
-                "&ᅮᅴᅵ"
-                "=ᅰ"
-                "&ᅮᅵ"
-                "=ᅱ"
                 "&N<ñ<<<Ñ"
             }
             Version{"25"}

icu4c/source/i18n/collationbuilder.cpp

@@ -54,26 +54,22 @@ namespace {
 
 class BundleImporter : public CollationRuleParser::Importer {
 public:
-    BundleImporter() : rules(NULL) {}
+    BundleImporter() {}
     virtual ~BundleImporter();
-    virtual const UnicodeString *getRules(
+    virtual void getRules(
             const char *localeID, const char *collationType,
+            UnicodeString &rules,
             const char *&errorReason, UErrorCode &errorCode);
-
-private:
-    UnicodeString *rules;
 };
 
-BundleImporter::~BundleImporter() {
+BundleImporter::~BundleImporter() {}
-    delete rules;
-}
 
-const UnicodeString *
+void
 BundleImporter::getRules(
         const char *localeID, const char *collationType,
+        UnicodeString &rules,
         const char *& /*errorReason*/, UErrorCode &errorCode) {
-    delete rules;
+    CollationLoader::loadRules(localeID, collationType, rules, errorCode);
-    return rules = CollationLoader::loadRules(localeID, collationType, errorCode);
 }
 
 }  // namespace

icu4c/source/i18n/collationruleparser.cpp

@@ -638,10 +638,9 @@ CollationRuleParser::parseSetting(UErrorCode &errorCode) {
             if(importer == NULL) {
                 setParseError("[import langTag] is not supported", errorCode);
             } else {
-                const UnicodeString *importedRules =
+                UnicodeString importedRules;
-                    importer->getRules(baseID,
+                importer->getRules(baseID, length > 0 ? collationType : "standard",
-                                       length > 0 ? collationType : "standard",
+                                   importedRules, errorReason, errorCode);
-                                       errorReason, errorCode);
                 if(U_FAILURE(errorCode)) {
                     if(errorReason == NULL) {
                         errorReason = "[import langTag] failed";
@@ -651,7 +650,7 @@ CollationRuleParser::parseSetting(UErrorCode &errorCode) {
                 }
                 const UnicodeString *outerRules = rules;
                 int32_t outerRuleIndex = ruleIndex;
-                parse(*importedRules, errorCode);
+                parse(importedRules, errorCode);
                 if(U_FAILURE(errorCode)) {
                     if(parseError != NULL) {
                         parseError->offset = outerRuleIndex;

icu4c/source/i18n/collationruleparser.h

@@ -93,8 +93,9 @@ public:
     class U_I18N_API Importer : public UObject {
     public:
         virtual ~Importer();
-        virtual const UnicodeString *getRules(
+        virtual void getRules(
                 const char *localeID, const char *collationType,
+                UnicodeString &rules,
                 const char *&errorReason, UErrorCode &errorCode) = 0;
     };
 

icu4c/source/i18n/ucol_imp.h

@@ -61,8 +61,8 @@ class UnicodeString;
 class CollationLoader {
 public:
     static void appendRootRules(UnicodeString &s);
-    static UnicodeString *loadRules(const char *localeID, const char *collationType,
+    static void loadRules(const char *localeID, const char *collationType,
-                                    UErrorCode &errorCode);
+                          UnicodeString &rules, UErrorCode &errorCode);
     static const CollationTailoring *loadTailoring(const Locale &locale, Locale &validLocale,
                                                    UErrorCode &errorCode);
 

icu4c/source/i18n/ucol_res.cpp

@@ -100,16 +100,17 @@ CollationLoader::appendRootRules(UnicodeString &s) {
     }
 }
 
-UnicodeString *
+void
-CollationLoader::loadRules(const char *localeID, const char *collationType, UErrorCode &errorCode) {
+CollationLoader::loadRules(const char *localeID, const char *collationType,
-    if(U_FAILURE(errorCode)) { return NULL; }
+                           UnicodeString &rules, UErrorCode &errorCode) {
+    if(U_FAILURE(errorCode)) { return; }
     U_ASSERT(collationType != NULL && *collationType != 0);
     // Copy the type for lowercasing.
     char type[16];
     int32_t typeLength = uprv_strlen(collationType);
     if(typeLength >= LENGTHOF(type)) {
         errorCode = U_ILLEGAL_ARGUMENT_ERROR;
-        return NULL;
+        return;
     }
     uprv_memcpy(type, collationType, typeLength + 1);
     T_CString_toLowerCase(type);
@@ -121,15 +122,13 @@ CollationLoader::loadRules(const char *localeID, const char *collationType, UErr
             ures_getByKeyWithFallback(collations.getAlias(), type, NULL, &errorCode));
     int32_t length;
     const UChar *s =  ures_getStringByKey(data.getAlias(), "Sequence", &length, &errorCode);
-    if(U_FAILURE(errorCode)) { return NULL; }
+    if(U_FAILURE(errorCode)) { return; }
 
     // No string pointer aliasing so that we need not hold onto the resource bundle.
-    UnicodeString *rules = new UnicodeString(s, length);
+    rules.setTo(s, length);
-    if(rules == NULL) {
+    if(rules.isBogus()) {
         errorCode = U_MEMORY_ALLOCATION_ERROR;
-        return NULL;
     }
-    return rules;
 }
 
 const CollationTailoring *

icu4c/source/tools/genrb/parse.cpp

@@ -673,21 +673,22 @@ class GenrbImporter : public icu::CollationRuleParser::Importer {
 public:
     GenrbImporter(const char *in, const char *out) : inputDir(in), outputDir(out) {}
     virtual ~GenrbImporter();
-    virtual const UnicodeString *getRules(
+    virtual void getRules(
             const char *localeID, const char *collationType,
+            UnicodeString &rules,
             const char *&errorReason, UErrorCode &errorCode);
 
 private:
     const char *inputDir;
     const char *outputDir;
-    UnicodeString rules;
 };
 
 GenrbImporter::~GenrbImporter() {}
 
-const UnicodeString *
+void
 GenrbImporter::getRules(
         const char *localeID, const char *collationType,
+        UnicodeString &rules,
         const char *& /*errorReason*/, UErrorCode &errorCode) {
     struct SRBRoot *data         = NULL;
     UCHARBUF       *ucbuf        = NULL;
@@ -718,11 +719,11 @@ GenrbImporter::getRules(
 
 
     if (U_FAILURE(errorCode)) {
-        return NULL;
+        return;
     }
     if(filename==NULL){
         errorCode=U_ILLEGAL_ARGUMENT_ERROR;
-        return NULL;
+        return;
     }else{
         filelen = (int32_t)uprv_strlen(filename);
     }
@@ -810,6 +811,9 @@ GenrbImporter::getRules(
 
     /* Parse the data into an SRBRoot */
     data = parse(ucbuf, inputDir, outputDir, filename, FALSE, FALSE, &errorCode);
+    if (U_FAILURE(errorCode)) {
+        goto finish;
+    }
 
     root = data->fRoot;
     collations = resLookup(root, "collations");
@@ -818,7 +822,8 @@ GenrbImporter::getRules(
       if (collation != NULL) {
         sequence = resLookup(collation, "Sequence");
         if (sequence != NULL) {
-          rules.setTo(FALSE, sequence->u.fString.fChars, sequence->u.fString.fLength);
+          // No string pointer aliasing so that we need not hold onto the resource bundle.
+          rules.setTo(sequence->u.fString.fChars, sequence->u.fString.fLength);
         }
       }
     }
@@ -835,8 +840,6 @@ finish:
     if(ucbuf) {
         ucbuf_close(ucbuf);
     }
-
-    return &rules;
 }
 
 // Quick-and-dirty escaping function.