ICU-2949 define and fix ucnv_getMaxCharSize() behavior

X-SVN-Rev: 13537

icu4c/source/common/ucnv.c

@@ -645,7 +645,7 @@ ucnv_resetFromUnicode(UConverter *converter)
 U_CAPI int8_t   U_EXPORT2
 ucnv_getMaxCharSize (const UConverter * converter)
 {
-    return converter->sharedData->staticData->maxBytesPerChar;
+    return converter->maxBytesPerUChar;
 }
 
 

icu4c/source/common/ucnv2022.c

@@ -435,10 +435,10 @@ _ISO2022Open(UConverter *cnv, const char *name, const char *locale,uint32_t opti
             uprv_strcpy(myConverterData->name,"ISO_2022");
         }
 
+        cnv->maxBytesPerUChar=cnv->sharedData->staticData->maxBytesPerChar;
     } else {
         *errorCode = U_MEMORY_ALLOCATION_ERROR;
     }
-
 }
 
 
@@ -2823,7 +2823,7 @@ static const UConverterStaticData _ISO2022StaticData={
     UCNV_IBM,
     UCNV_ISO_2022,
     1,
-    4,
+    3, /* max 3 bytes per UChar from UTF-8 (4 bytes from surrogate _pair_) */
     { 0x1a, 0, 0, 0 },
     1,
     FALSE,
@@ -2873,7 +2873,7 @@ static const UConverterStaticData _ISO2022JPStaticData={
     UCNV_IBM,
     UCNV_ISO_2022,
     1,
-    6,
+    6, /* max 6 bytes per UChar: 4-byte escape sequence + DBCS */
     { 0x1a, 0, 0, 0 },
     1,
     FALSE,
@@ -2923,7 +2923,7 @@ static const UConverterStaticData _ISO2022KRStaticData={
     UCNV_IBM,
     UCNV_ISO_2022,
     1,
-    3,
+    3, /* max 3 bytes per UChar: SO+DBCS */
     { 0x1a, 0, 0, 0 },
     1,
     FALSE,
@@ -2974,7 +2974,7 @@ static const UConverterStaticData _ISO2022CNStaticData={
     UCNV_IBM,
     UCNV_ISO_2022,
     2,
-    8,
+    8, /* max 8 bytes per UChar: 4-byte CNS designator + 2 bytes for SS2/SS3 + DBCS */
     { 0x1a, 0, 0, 0 },
     1,
     FALSE,

icu4c/source/common/ucnv_bld.c

@@ -775,6 +775,7 @@ ucnv_createConverterFromSharedData(UConverter *myUConverter,
     myUConverter->fromCharErrorBehaviour = (UConverterToUCallback) UCNV_TO_U_CALLBACK_SUBSTITUTE;
     myUConverter->fromUCharErrorBehaviour = (UConverterFromUCallback) UCNV_FROM_U_CALLBACK_SUBSTITUTE;
     myUConverter->toUnicodeStatus = myUConverter->sharedData->toUnicodeStatus;
+    myUConverter->maxBytesPerUChar = myUConverter->sharedData->staticData->maxBytesPerChar;
     myUConverter->subChar1 = myUConverter->sharedData->staticData->subChar1;
     myUConverter->subCharLen = myUConverter->sharedData->staticData->subCharLen;
     uprv_memcpy (myUConverter->subChar, myUConverter->sharedData->staticData->subChar, myUConverter->subCharLen);

icu4c/source/common/ucnv_bld.h

@@ -64,7 +64,7 @@ typedef struct UConverterStaticData {   /* +offset: size */
     int8_t conversionType;          /* +69: 1 conversion type */
 
     int8_t minBytesPerChar;         /* +70: 1 Minimum # bytes per char in this codepage */
-    int8_t maxBytesPerChar;         /* +71: 1 Maximum # bytes per char in this codepage */
+    int8_t maxBytesPerChar;         /* +71: 1 Maximum # bytes output per UChar in this codepage */
 
     uint8_t subChar[UCNV_MAX_SUBCHAR_LEN]; /* +72: 4  [note:  4 and 8 byte boundary] */
     int8_t subCharLen;              /* +76: 1 */
@@ -161,6 +161,14 @@ struct UConverter {
      */
     UChar32 fromUChar32;
 
+    /*
+     * value for ucnv_getMaxCharSize()
+     *
+     * usually simply copied from the static data, but ucnvmbcs.c modifies
+     * the value depending on the converter type and options
+     */
+    int8_t maxBytesPerUChar;
+
     int8_t subCharLen;                  /* length of the codepage specific character sequence */
     int8_t invalidCharLength;
     int8_t charErrorBufferLength;       /* number of valid bytes in charErrorBuffer */

icu4c/source/common/ucnv_ext.h

@@ -124,7 +124,19 @@
  *   [15] index of fromUStage3b[] (array of uint32_t like fromUTableValues[])
  *   [16] length of fromUStage3b[]
  *
- *   [17]..[30] reserved
+ *   [17] Bit field containing numbers of bytes:
+ *        31..24 reserved, 0
+ *        23..16 maximum input bytes
+ *        15.. 8 maximum output bytes
+ *         7.. 0 maximum bytes per UChar
+ *
+ *   [18] Bit field containing numbers of UChars:
+ *        31..24 reserved, 0
+ *        23..16 maximum input UChars
+ *        15.. 8 maximum output UChars
+ *         7.. 0 maximum UChars per byte
+ *
+ *   [19]..[30] reserved, 0
  *   [31] number of bytes for the entire extension structure
  *   [>31] reserved; there are indexes[0] indexes
  *
@@ -303,7 +315,10 @@ enum {
     UCNV_EXT_FROM_U_STAGE_3B_INDEX,
     UCNV_EXT_FROM_U_STAGE_3B_LENGTH,
 
-    UCNV_EXT_RESERVED_INDEX,            /* 17, moves with additional indexes */
+    UCNV_EXT_COUNT_BYTES,               /* 17 */
+    UCNV_EXT_COUNT_UCHARS,
+
+    UCNV_EXT_RESERVED_INDEX,            /* 19, moves with additional indexes */
 
     UCNV_EXT_SIZE=31,
     UCNV_EXT_INDEXES_MIN_LENGTH=32
@@ -313,6 +328,9 @@ enum {
 #define UCNV_EXT_ARRAY(indexes, index, itemType) \
     ((const itemType *)((const char *)(indexes)+(indexes)[index]))
 
+#define UCNV_GET_MAX_BYTES_PER_UCHAR(indexes) \
+    ((indexes)[UCNV_EXT_COUNT_BYTES]&0xff)
+
 /* internal API ------------------------------------------------------------- */
 
 U_CFUNC UBool

icu4c/source/common/ucnv_lmb.c

@@ -528,7 +528,7 @@ static const UConverterImpl _LMBCSImpl##n={\
 static const UConverterStaticData _LMBCSStaticData##n={\
   sizeof(UConverterStaticData),\
  "LMBCS-"  #n,\
-    0, UCNV_IBM, UCNV_LMBCS_##n, 1, 2,\
+    0, UCNV_IBM, UCNV_LMBCS_##n, 1, 3,\
     { 0x3f, 0, 0, 0 },1,FALSE,FALSE,0,0,{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0} \
 };\
 const UConverterSharedData _LMBCSData##n={\

icu4c/source/common/ucnv_u8.c

@@ -757,7 +757,8 @@ static const UConverterImpl _UTF8Impl={
 static const UConverterStaticData _UTF8StaticData={
     sizeof(UConverterStaticData),
     "UTF-8",
-    1208, UCNV_IBM, UCNV_UTF8, 1, 4,
+    1208, UCNV_IBM, UCNV_UTF8,
+    1, 3, /* max 3 bytes per UChar from UTF-8 (4 bytes from surrogate _pair_) */
     { 0xef, 0xbf, 0xbd, 0 },3,FALSE,FALSE,
     0,
     0,

icu4c/source/common/ucnvmbcs.c

@@ -898,6 +898,10 @@ _MBCSOpen(UConverter *cnv,
           const char *locale,
           uint32_t options,
           UErrorCode *pErrorCode) {
+    const int32_t *extIndexes;
+    uint8_t outputType;
+    int8_t maxBytesPerUChar;
+
     if((options&UCNV_OPTION_SWAP_LFNL)!=0) {
         /* do this because double-checked locking is broken */
         UBool isCached;
@@ -914,7 +918,6 @@ _MBCSOpen(UConverter *cnv,
         }
     }
 
-
     if(uprv_strstr(name, "18030")!=NULL) {
         if(uprv_strstr(name, "gb18030")!=NULL || uprv_strstr(name, "GB18030")!=NULL) {
             /* set a flag for GB 18030 mode, which changes the callback behavior */
@@ -922,6 +925,24 @@ _MBCSOpen(UConverter *cnv,
         }
     }
 
+    /* fix maxBytesPerUChar depending on outputType and options etc. */
+    outputType=cnv->sharedData->table->mbcs.outputType;
+    if(outputType==MBCS_OUTPUT_2_SISO) {
+        cnv->maxBytesPerUChar=3; /* SO+DBCS */
+    }
+
+    extIndexes=cnv->sharedData->table->mbcs.extIndexes;
+    if(extIndexes!=NULL) {
+        maxBytesPerUChar=(int8_t)UCNV_GET_MAX_BYTES_PER_UCHAR(extIndexes);
+        if(outputType==MBCS_OUTPUT_2_SISO) {
+            ++maxBytesPerUChar; /* SO + multiple DBCS */
+        }
+
+        if(maxBytesPerUChar>cnv->maxBytesPerUChar) {
+            cnv->maxBytesPerUChar=maxBytesPerUChar;
+        }
+    }
+
 #if 0
     /*
      * documentation of UConverter fields used for status