'used' should be set even if we are only computing the necessary

unicode/wctomb.c

@@ -76,33 +76,53 @@ static inline int is_valid_sbcs_mapping( const struct sbcs_table *table, int fla
 }
 
 /* query necessary dst length for src string */
-static inline int get_length_sbcs( const struct sbcs_table *table, int flags,
-                                   const WCHAR *src, unsigned int srclen )
+static int get_length_sbcs( const struct sbcs_table *table, int flags,
+                            const WCHAR *src, unsigned int srclen, int *used )
 {
-    unsigned int ret = srclen;
-
-    if (flags & WC_COMPOSITECHECK)
-    {
     const unsigned char  * const uni2cp_low = table->uni2cp_low;
     const unsigned short * const uni2cp_high = table->uni2cp_high;
+    int ret, tmp;
     WCHAR composed;
 
-        for (ret = 0; srclen > 1; ret++, srclen--, src++)
+    if (!used) used = &tmp;  /* avoid checking on every char */
+    *used = 0;
+
+    for (ret = 0; srclen; ret++, src++, srclen--)
     {
-            if (!(composed = compose(src))) continue;
-            /* check if we should skip the next char */
+        WCHAR wch = *src;
+        unsigned char ch;
 
-            /* in WC_DEFAULTCHAR and WC_DISCARDNS mode, we always skip */
-            /* the next char no matter if the composition is valid or not */
-            if (!(flags & (WC_DEFAULTCHAR|WC_DISCARDNS)))
+        if ((flags & WC_COMPOSITECHECK) && (srclen > 1) && (composed = compose(src)))
+        {
+            /* now check if we can use the composed char */
+            ch = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
+            if (is_valid_sbcs_mapping( table, flags, composed, ch ))
+            {
+                /* we have a good mapping, use it */
+                src++;
+                srclen--;
+                continue;
+            }
+            /* no mapping for the composed char, check the other flags */
+            if (flags & WC_DEFAULTCHAR) /* use the default char instead */
             {
-                unsigned char ch = uni2cp_low[uni2cp_high[composed >> 8] + (composed & 0xff)];
-                if (!is_valid_sbcs_mapping( table, flags, composed, ch )) continue;
+                *used = 1;
+                src++;  /* skip the non-spacing char */
+                srclen--;
+                continue;
             }
+            if (flags & WC_DISCARDNS) /* skip the second char of the composition */
+            {
                 src++;
                 srclen--;
             }
-        if (srclen) ret++;  /* last char */
+            /* WC_SEPCHARS is the default */
+        }
+        if (!*used)
+        {
+            ch = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
+            *used = !is_valid_sbcs_mapping( table, flags, wch, ch );
+        }
     }
     return ret;
 }
@@ -238,15 +258,15 @@ static inline int is_valid_dbcs_mapping( const struct dbcs_table *table, int fla
 /* query necessary dst length for src string */
 static int get_length_dbcs( const struct dbcs_table *table, int flags,
                             const WCHAR *src, unsigned int srclen,
-                            const char *defchar )
+                            const char *defchar, int *used )
 {
     const unsigned short * const uni2cp_low = table->uni2cp_low;
     const unsigned short * const uni2cp_high = table->uni2cp_high;
     WCHAR defchar_value = table->info.def_char;
     WCHAR composed;
-    int len;
+    int len, tmp;
 
-    if (!defchar && !(flags & WC_COMPOSITECHECK))
+    if (!defchar && !used && !(flags & WC_COMPOSITECHECK))
     {
         for (len = 0; srclen; srclen--, src++, len++)
         {
@@ -256,6 +276,8 @@ static int get_length_dbcs( const struct dbcs_table *table, int flags,
     }
 
     if (defchar) defchar_value = defchar[1] ? ((defchar[0] << 8) | defchar[1]) : defchar[0];
+    if (!used) used = &tmp;  /* avoid checking on every char */
+    *used = 0;
     for (len = 0; srclen; len++, srclen--, src++)
     {
         unsigned short res;
@@ -278,6 +300,7 @@ static int get_length_dbcs( const struct dbcs_table *table, int flags,
             if (flags & WC_DEFAULTCHAR) /* use the default char instead */
             {
                 if (defchar_value & 0xff00) len++;
+                *used = 1;
                 src++;  /* skip the non-spacing char */
                 srclen--;
                 continue;
@@ -291,7 +314,11 @@ static int get_length_dbcs( const struct dbcs_table *table, int flags,
         }
 
         res = uni2cp_low[uni2cp_high[wch >> 8] + (wch & 0xff)];
-        if (!is_valid_dbcs_mapping( table, flags, wch, res )) res = defchar_value;
+        if (!is_valid_dbcs_mapping( table, flags, wch, res ))
+        {
+            res = defchar_value;
+            *used = 1;
+        }
         if (res & 0xff00) len++;
     }
     return len;
@@ -399,15 +426,18 @@ int cp_wcstombs( const union cptable *table, int flags,
 {
     if (table->info.char_size == 1)
     {
-        if (!dstlen) return get_length_sbcs( &table->sbcs, flags, src, srclen );
         if (flags || defchar || used)
+        {
+            if (!dstlen) return get_length_sbcs( &table->sbcs, flags, src, srclen, used );
             return wcstombs_sbcs_slow( &table->sbcs, flags, src, srclen,
                                        dst, dstlen, defchar, used );
+        }
+        if (!dstlen) return srclen;
         return wcstombs_sbcs( &table->sbcs, src, srclen, dst, dstlen );
     }
     else /* mbcs */
     {
-        if (!dstlen) return get_length_dbcs( &table->dbcs, flags, src, srclen, defchar );
+        if (!dstlen) return get_length_dbcs( &table->dbcs, flags, src, srclen, defchar, used );
         if (flags || defchar || used)
             return wcstombs_dbcs_slow( &table->dbcs, flags, src, srclen,
                                        dst, dstlen, defchar, used );