ICU-1126 complete implementation of u_strToTitle()

X-SVN-Rev: 7732

icu4c/source/common/uchar.c

@@ -1488,9 +1488,14 @@ single:
     return (result==c) ? -length : length;
 }
 
+/*
+ * Lowercases [srcStart..srcLimit[ but takes
+ * context [0..srcLength[ into account.
+ */
 U_CFUNC int32_t
 u_internalStrToLower(UChar *dest, int32_t destCapacity,
                      const UChar *src, int32_t srcLength,
+                     int32_t srcStart, int32_t srcLimit,
                      const char *locale,
                      UErrorCode *pErrorCode) {
     UCharIterator iter;
@@ -1504,6 +1509,9 @@ u_internalStrToLower(UChar *dest, int32_t destCapacity,
          * If we do not have real character properties data,
          * then we only do a fixed-length ASCII case mapping.
          */
+        src+=srcStart;
+        srcLength-=srcStart;
+
         if(srcLength<=destCapacity) {
             destIndex=srcLength;
             *pErrorCode=U_USING_DEFAULT_ERROR;
@@ -1528,9 +1536,10 @@ u_internalStrToLower(UChar *dest, int32_t destCapacity,
     uiter_setString(&iter, src, srcLength);
 
     /* case mapping loop */
-    srcIndex=destIndex=0;
-    while(srcIndex<srcLength) {
-        UTF_NEXT_CHAR(src, srcIndex, srcLength, c);
+    srcIndex=srcStart;
+    destIndex=0;
+    while(srcIndex<srcLimit) {
+        UTF_NEXT_CHAR(src, srcIndex, srcLimit, c);
         GET_PROPS_UNSAFE(c, props);
         if(!PROPS_VALUE_IS_EXCEPTION(props)) {
             if((1UL<<GET_CATEGORY(props))&(1UL<<U_UPPERCASE_LETTER|1UL<<U_TITLECASE_LETTER)) {

icu4c/source/common/ustr_imp.h

@@ -19,6 +19,9 @@
 #include "unicode/ucnv.h"
 #include "unicode/uiter.h"
 
+/** Simple declaration for u_strToTitle() to avoid including unicode/ubrk.h. */
+typedef void *UBreakIterator;
+
 /**
  * Are the Unicode properties loaded?
  * This must be used before internal functions are called that do
@@ -52,13 +55,17 @@ u_growBufferFromStatic(void *context,
                        int32_t length);
 
 /*
- * Internal string casing functions implementing ustring.c and UnicodeString
- * case mapping functions.
+ * Internal string casing functions implementing
+ * ustring.h/ustrcase.c and UnicodeString case mapping functions.
+ *
+ * Lowercases [srcStart..srcLimit[ but takes
+ * context [0..srcLength[ into account.
  * @internal
  */
 U_CFUNC int32_t
 u_internalStrToLower(UChar *dest, int32_t destCapacity,
                      const UChar *src, int32_t srcLength,
+                     int32_t srcStart, int32_t srcLimit,
                      const char *locale,
                      UErrorCode *pErrorCode);
 
@@ -71,6 +78,16 @@ u_internalStrToUpper(UChar *dest, int32_t destCapacity,
                      const char *locale,
                      UErrorCode *pErrorCode);
 
+/**
+ * @internal
+ */
+U_CFUNC int32_t
+u_internalStrToTitle(UChar *dest, int32_t destCapacity,
+                     const UChar *src, int32_t srcLength,
+                     UBreakIterator *titleIter,
+                     const char *locale,
+                     UErrorCode *pErrorCode);
+
 /**
  * Internal case folding function.
  * @internal

icu4c/source/common/ustrcase.c

@@ -32,7 +32,7 @@
  *
  * Must get titleIter!=NULL.
  */
-static int32_t
+U_CFUNC int32_t
 u_internalStrToTitle(UChar *dest, int32_t destCapacity,
                      const UChar *src, int32_t srcLength,
                      UBreakIterator *titleIter,
@@ -41,31 +41,39 @@ u_internalStrToTitle(UChar *dest, int32_t destCapacity,
     UCharIterator iter;
     UChar32 c;
     int32_t prev, index, destIndex, length;
+    UBool isFirstIndex;
 
     /* set up local variables */
     uiter_setString(&iter, src, srcLength);
     destIndex=0;
     prev=0;
+    isFirstIndex=TRUE;
 
     /* titlecasing loop */
     while(prev<srcLength) {
         /* find next index where to titlecase */
-        index=ubrk_next(titleIter);
-        if(index==UBRK_DONE) {
+        if(isFirstIndex) {
+            isFirstIndex=FALSE;
+            index=ubrk_first(titleIter);
+        } else {
+            index=ubrk_next(titleIter);
+        }
+        if(index==UBRK_DONE || index>srcLength) {
             index=srcLength;
         }
 
         /* lowercase [prev..index[ */
-        /* ### TODO this really needs to pass the full UCharIterator into the internal strToXXX functions so that they see the full context! */
         if(prev<index) {
             if(destIndex<destCapacity) {
                 length=u_internalStrToLower(dest+destIndex, destCapacity-destIndex,
-                                            src+prev, index-prev,
+                                            src, srcLength,
+                                            prev, index,
                                             locale,
                                             pErrorCode);
             } else {
                 length=u_internalStrToLower(NULL, 0,
-                                            src+prev, index-prev,
+                                            src, srcLength,
+                                            prev, index,
                                             locale,
                                             pErrorCode);
             }
@@ -163,7 +171,9 @@ u_strCaseMap(UChar *dest, int32_t destCapacity,
     ownTitleIter=FALSE;
 
     if(toWhichCase==TO_LOWER) {
-        destLength=u_internalStrToLower(temp, destCapacity, src, srcLength,
+        destLength=u_internalStrToLower(temp, destCapacity,
+                                        src, srcLength,
+                                        0, srcLength,
                                         locale, pErrorCode);
     } else if(toWhichCase==TO_UPPER) {
         destLength=u_internalStrToUpper(temp, destCapacity, src, srcLength,