ICU-11761 share equal strings even if they need explicit length; more generic root resource for flexibility (experiment with more work on pool bundles)

X-SVN-Rev: 37649

icu4c/source/tools/genrb/genrb.cpp

@@ -565,10 +565,9 @@ processFile(const char *filename, const char *cp,
         goto finish;
     }
     if(options[WRITE_POOL_BUNDLE].doesOccur) {
-        int32_t newKeysLength;
-        const char *newKeys, *newKeysLimit;
         data->compactKeys(*status);
-        newKeys = data->getKeyBytes(&newKeysLength);
+        int32_t newKeysLength;
+        const char *newKeys = data->getKeyBytes(&newKeysLength);
         newPoolBundle->addKeyBytes(newKeys, newKeysLength, *status);
         if(U_FAILURE(*status)) {
             fprintf(stderr, "bundle_compactKeys(%s) or bundle_getKeyBytes() failed: %s\n",
@@ -576,7 +575,7 @@ processFile(const char *filename, const char *cp,
             goto finish;
         }
         /* count the number of just-added key strings */
-        for(newKeysLimit = newKeys + newKeysLength; newKeys < newKeysLimit; ++newKeys) {
+        for(const char *newKeysLimit = newKeys + newKeysLength; newKeys < newKeysLimit; ++newKeys) {
             if(*newKeys == 0) {
                 ++newPoolBundle->fKeysCount;
             }

icu4c/source/tools/genrb/parse.cpp

@@ -2046,9 +2046,12 @@ parse(UCHARBUF *buf, const char *inputDir, const char *outputDir, const char *fi
         state.bundle->fNoFallback=TRUE;
     }
     /* top-level tables need not handle special table names like "collations" */
-    realParseTable(&state, state.bundle->fRoot, NULL, line, status);
+    assert(!state.bundle->fIsPoolBundle);
+    assert(state.bundle->fRoot->fType == URES_TABLE);
+    TableResource *rootTable = static_cast<TableResource *>(state.bundle->fRoot);
+    realParseTable(&state, rootTable, NULL, line, status);
     if(dependencyArray!=NULL){
-        state.bundle->fRoot->add(dependencyArray, 0, *status);
+        rootTable->add(dependencyArray, 0, *status);
         dependencyArray = NULL;
     }
    if (U_FAILURE(*status))

icu4c/source/tools/genrb/reslist.cpp

@@ -1002,7 +1002,7 @@ struct SResource *bin_open(struct SRBRoot *bundle, const char *tag, uint32_t len
 
 SRBRoot::SRBRoot(const UString *comment, UBool isPoolBundle, UErrorCode &errorCode)
         : fRoot(NULL), fLocale(NULL), fIndexLength(0), fMaxTableLength(0), fNoFallback(FALSE),
-          fStringsForm(0), fIsPoolBundle(FALSE),
+          fStringsForm(0), fIsPoolBundle(isPoolBundle),
           fKeys(NULL), fKeyMap(NULL),
           fKeysBottom(0), fKeysTop(0), fKeysCapacity(0), fKeysCount(0), fLocalKeyLimit(0),
           f16BitUnits(), f16BitStringsLength(0),
@@ -1022,7 +1022,6 @@ SRBRoot::SRBRoot(const UString *comment, UBool isPoolBundle, UErrorCode &errorCo
 
     fKeysCapacity = KEY_SPACE_SIZE;
     /* formatVersion 1.1: start fKeysTop after the root item and indexes[] */
-    fIsPoolBundle = isPoolBundle;
     if (gUsePoolBundle || isPoolBundle) {
         fIndexLength = URES_INDEX_POOL_CHECKSUM + 1;
     } else if (gFormatVersion >= 2) {
@@ -1417,9 +1416,9 @@ SRBRoot::compactStringsV2(UHashtable *stringSet, UErrorCode &errorCode) {
         int32_t j;
         for (j = i + 1; j < count; ++j) {
             StringResource *suffixRes = array[j];
-            /* Is it a suffix of the earlier, longer key? */
+            /* Is it a suffix of the earlier, longer string? */
             if (res->fString.endsWith(suffixRes->fString)) {
-                if (suffixRes->fNumCharsForLength == 0) {
+                if (suffixRes->fNumCharsForLength == 0 || res->length() == suffixRes->length()) {
                     /* yes, point to the earlier string */
                     suffixRes->fSame = res;
                     suffixRes->fSuffixOffset = res->length() - suffixRes->length();
@@ -1449,12 +1448,17 @@ SRBRoot::compactStringsV2(UHashtable *stringSet, UErrorCode &errorCode) {
     }
     if (f16BitUnits.isBogus()) {
         errorCode = U_MEMORY_ALLOCATION_ERROR;
+        return;
     }
     /* Write the suffix strings. Make each point to the real string. */
     for (; i < count; ++i) {
         StringResource *res = array[i];
         StringResource *same = res->fSame;
-        res->fRes = same->fRes + same->fNumCharsForLength + res->fSuffixOffset;
+        if (res->length() == same->length()) {
+            res->fRes = same->fRes;
+        } else {
+            res->fRes = same->fRes + same->fNumCharsForLength + res->fSuffixOffset;
+        }
         res->fSame = NULL;
         res->fWritten = TRUE;
     }

icu4c/source/tools/genrb/reslist.h

@@ -50,7 +50,7 @@ struct ResFile {
     int32_t fChecksum;
 };
 
-class TableResource;
+struct SResource;
 
 typedef struct KeyMapEntry {
     int32_t oldpos, newpos;
@@ -83,7 +83,7 @@ private:
 public:
     // TODO: private
 
-  TableResource *fRoot;
+  SResource *fRoot;  // Normally a TableResource.
   char *fLocale;
   int32_t fIndexLength;
   int32_t fMaxTableLength;
@@ -124,6 +124,7 @@ void bundle_write_xml(struct SRBRoot *bundle, const char *outputDir,const char*
 struct SResource* res_none(void);
 
 class ArrayResource;
+class TableResource;
 class IntVectorResource;
 
 TableResource *table_open(struct SRBRoot *bundle, const char *tag, const struct UString* comment, UErrorCode *status);