Use ints rather than unsigned shorts in unknown encoding support

diff --git a/expat/xmlparse/xmlparse.c b/expat/xmlparse/xmlparse.c
index b8ec1e9106f4901bf4e985b1dcffd455b6258f6f..57906d427250eed08e9c3c75b702cbca4e739019 100755 (executable)
--- a/expat/xmlparse/xmlparse.c
+++ b/expat/xmlparse/xmlparse.c
@@ -1350,7 +1350,7 @@ handleUnknownEncoding(XML_Parser parser, const XML_Char *encodingName)
     XML_Encoding info;
     int i;
     for (i = 0; i < 256; i++)
-      info.map[i] = 0;
+      info.map[i] = -1;
     info.convert = 0;
     info.data = 0;
     info.release = 0;

diff --git a/expat/xmlparse/xmlparse.h b/expat/xmlparse/xmlparse.h
index edd16c9efa6ca9c9e3bce638e2672cadf82e7bf2..680f8141fdecd4433a8a431a9d8242db67e3445d 100755 (executable)
--- a/expat/xmlparse/xmlparse.h
+++ b/expat/xmlparse/xmlparse.h
@@ -111,9 +111,9 @@ typedef int (*XML_ExternalEntityRefHandler)(XML_Parser parser,
                                            const XML_Char *publicId);
 
 typedef struct {
-  unsigned short map[256];
+  int map[256];
   void *data;
-  unsigned short (*convert)(void *data, const char *s);
+  int (*convert)(void *data, const char *s);
   void (*release)(void *data);
 } XML_Encoding;
 

diff --git a/expat/xmltok/xmltok.c b/expat/xmltok/xmltok.c
index 94eaa109128f4311286e4e4f181a11f6185e5922..2c7129e3516349152912b43b7c87fe77d2ffdd6a 100755 (executable)
--- a/expat/xmltok/xmltok.c
+++ b/expat/xmltok/xmltok.c
@@ -962,7 +962,7 @@ int XmlUtf16Encode(int charNum, unsigned short *buf)
 
 struct unknown_encoding {
   struct normal_encoding normal;
-  unsigned short (*convert)(void *userData, const char *p);
+  int (*convert)(void *userData, const char *p);
   void *userData;
   unsigned short utf16[256];
   unsigned char utf8[256][4];
@@ -976,24 +976,29 @@ int XmlSizeOfUnknownEncoding()
 static
 int unknown_isName(const ENCODING *enc, const char *p)
 {
-  unsigned short c = ((const struct unknown_encoding *)enc)
-                     ->convert(((const struct unknown_encoding *)enc)->userData, p);
+  int c = ((const struct unknown_encoding *)enc)
+         ->convert(((const struct unknown_encoding *)enc)->userData, p);
+  if (c & ~0xFFFF)
+    return 0;
   return UCS2_GET_NAMING(namePages, c >> 8, c & 0xFF);
 }
 
 static
 int unknown_isNmstrt(const ENCODING *enc, const char *p)
 {
-  unsigned short c = ((const struct unknown_encoding *)enc)
-                     ->convert(((const struct unknown_encoding *)enc)->userData, p);
+  int c = ((const struct unknown_encoding *)enc)
+         ->convert(((const struct unknown_encoding *)enc)->userData, p);
+  if (c & ~0xFFFF)
+    return 0;
   return UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xFF);
 }
 
 static
 int unknown_isInvalid(const ENCODING *enc, const char *p)
 {
-  return ((const struct unknown_encoding *)enc)
-        ->convert(((const struct unknown_encoding *)enc)->userData, p) == 0;
+  int c = ((const struct unknown_encoding *)enc)
+          ->convert(((const struct unknown_encoding *)enc)->userData, p);
+  return (c & ~0xFFFF) || checkCharRefNumber(c) < 0;
 }
 
 static
@@ -1010,9 +1015,8 @@ void unknown_toUtf8(const ENCODING *enc,
     utf8 = ((const struct unknown_encoding *)enc)->utf8[(unsigned char)**fromP];
     n = *utf8++;
     if (n == 0) {
-      unsigned short c
-       = ((const struct unknown_encoding *)enc)
-         ->convert(((const struct unknown_encoding *)enc)->userData, *fromP);
+      int c = ((const struct unknown_encoding *)enc)
+             ->convert(((const struct unknown_encoding *)enc)->userData, *fromP);
       n = XmlUtf8Encode(c, buf);
       if (n > toLim - *toP)
        break;
@@ -1040,7 +1044,7 @@ void unknown_toUtf16(const ENCODING *enc,
     unsigned short c
       = ((const struct unknown_encoding *)enc)->utf16[(unsigned char)**fromP];
     if (c == 0) {
-      c = ((const struct unknown_encoding *)enc)
+      c = (unsigned short)((const struct unknown_encoding *)enc)
           ->convert(((const struct unknown_encoding *)enc)->userData, *fromP);
       *fromP += ((const struct normal_encoding *)enc)->type[(unsigned char)**fromP]
                 - (BT_LEAD2 - 2);
@@ -1053,8 +1057,8 @@ void unknown_toUtf16(const ENCODING *enc,
 
 ENCODING *
 XmlInitUnknownEncoding(void *mem,
-                      unsigned short *table,
-                      unsigned short (*convert)(void *userData, const char *p),
+                      int *table,
+                      int (*convert)(void *userData, const char *p),
                       void *userData)
 {
   int i;
@@ -1067,23 +1071,25 @@ XmlInitUnknownEncoding(void *mem,
        && table[i] != i)
       return 0;
   for (i = 0; i < 256; i++) {
-    unsigned short c = table[i];
-    if (c < 0x80) {
+    int c = table[i];
+    if (c == -1)
+      c = 0xFFFF;
+    if (c < 0) {
+      if (c < -4)
+       return 0;
+      e->normal.type[i] = BT_LEAD2 - (c + 2);
+      e->utf8[i][0] = 0;
+      e->utf16[i] = 0;
+    }
+    else if (c < 0x80) {
       if (latin1_encoding.type[c] != BT_OTHER
          && latin1_encoding.type[c] != BT_NONXML
          && c != i)
-        return 0;
-      if (c >= 2 && c <= 4) {
-       e->normal.type[i] = BT_LEAD2 + (c - 2);
-       e->utf8[i][0] = 0;
-       e->utf16[i] = 0;
-      }
-      else {
-       e->normal.type[i] = latin1_encoding.type[c];
-       e->utf8[i][0] = 1;
-       e->utf8[i][1] = (char)c;
-       e->utf16[i] = c == 0 ? 0xFFFF : c;
-      }
+       return 0;
+      e->normal.type[i] = latin1_encoding.type[c];
+      e->utf8[i][0] = 1;
+      e->utf8[i][1] = (char)c;
+      e->utf16[i] = c == 0 ? 0xFFFF : c;
     }
     else if (checkCharRefNumber(c) < 0) {
       e->normal.type[i] = BT_NONXML;
@@ -1092,6 +1098,8 @@ XmlInitUnknownEncoding(void *mem,
       e->utf8[i][1] = 0;
     }
     else {
+      if (c > 0xFFFF)
+       return 0;
       if (UCS2_GET_NAMING(nmstrtPages, c >> 8, c & 0xff))
        e->normal.type[i] = BT_NMSTRT;
       else if (UCS2_GET_NAMING(namePages, c >> 8, c & 0xff))

diff --git a/expat/xmltok/xmltok.h b/expat/xmltok/xmltok.h
index f28414a919788c6196218600a420228787fe4da3..6d0b91dff2ce69d45d11f48bee3db578aa03a7ed 100755 (executable)
--- a/expat/xmltok/xmltok.h
+++ b/expat/xmltok/xmltok.h
@@ -265,8 +265,8 @@ int XMLTOKAPI XmlUtf16Encode(int charNumber, unsigned short *buf);
 int XMLTOKAPI XmlSizeOfUnknownEncoding();
 ENCODING XMLTOKAPI *
 XmlInitUnknownEncoding(void *mem,
-                      unsigned short *table,
-                      unsigned short (*convert)(void *userData, const char *p),
+                      int *table,
+                      int (*convert)(void *userData, const char *p),
                       void *userData);
 
 #ifdef __cplusplus

diff --git a/expat/xmlwf/codepage.c b/expat/xmlwf/codepage.c
index 8144e954553d8e36ed16c024a999071d767aa529..d4791c36a99cbe896b8e9ccc7ce538198f5a2645 100755 (executable)
--- a/expat/xmlwf/codepage.c
+++ b/expat/xmlwf/codepage.c
@@ -23,14 +23,14 @@ Contributor(s):
 #ifdef WIN32
 #include <windows.h>
 
-int codepageMap(int cp, unsigned short *map)
+int codepageMap(int cp, int *map)
 {
   int i;
   CPINFO info;
   if (!GetCPInfo(cp, &info) || info.MaxCharSize > 2)
     return 0;
   for (i = 0; i < 256; i++)
-    map[i] = 0;
+    map[i] = -1;
   if (info.MaxCharSize > 1) {
     for (i = 0; i < MAX_LEADBYTES; i++) {
       int j, lim;
@@ -38,27 +38,28 @@ int codepageMap(int cp, unsigned short *map)
         break;
       lim = info.LeadByte[i + 1];
       for (j = info.LeadByte[i]; j < lim; j++)
-       map[j] = 2;
+       map[j] = -2;
     }
   }
   for (i = 0; i < 256; i++) {
-   if (map[i] == 0) {
+   if (map[i] == -1) {
      char c = i;
+     unsigned short n;
      if (MultiByteToWideChar(cp, MB_PRECOMPOSED|MB_ERR_INVALID_CHARS,
-                            &c, 1, map + i, 1) == 0)
-       map[i] = 0;
+                            &c, 1, &n, 1) == 1)
+       map[i] = n;
    }
   }
   return 1;
 }
 
-unsigned short codepageConvert(int cp, const char *p)
+int codepageConvert(int cp, const char *p)
 {
   unsigned short c;
   if (MultiByteToWideChar(cp, MB_PRECOMPOSED|MB_ERR_INVALID_CHARS,
                          p, 2, &c, 1) == 1)
     return c;
-  return 0;
+  return -1;
 }
 
 #else /* not WIN32 */
@@ -70,7 +71,7 @@ int codepageMap(int cp, unsigned short *map)
 
 unsigned short codepageConvert(int cp, const char *p)
 {
-  return 0;
+  return -1;
 }
 
 #endif /* not WIN32 */

diff --git a/expat/xmlwf/codepage.h b/expat/xmlwf/codepage.h
index a85632766854a60f1fb2570244c22e8823cbddda..94c66f563f7a1be1c5abc6a2dd7540a70120df69 100755 (executable)
--- a/expat/xmlwf/codepage.h
+++ b/expat/xmlwf/codepage.h
@@ -18,5 +18,5 @@ James Clark. All Rights Reserved.
 Contributor(s):
 */
 
-int codepageMap(int cp, unsigned short *map);
-unsigned short codepageConvert(int cp, const char *p);
+int codepageMap(int cp, int *map);
+int codepageConvert(int cp, const char *p);

diff --git a/expat/xmlwf/xmlwf.c b/expat/xmlwf/xmlwf.c
index 80a69d78935930d2a79fd081748259e79adaa39b..72ce770ab4ae33c0c14e2a88444e4879cd958120 100755 (executable)
--- a/expat/xmlwf/xmlwf.c
+++ b/expat/xmlwf/xmlwf.c
@@ -405,7 +405,7 @@ int externalEntityRefStream(XML_Parser parser,
 }
 
 static
-unsigned short unknownEncodingConvert(void *data, const char *p)
+int unknownEncodingConvert(void *data, const char *p)
 {
   return codepageConvert(*(int *)data, p);
 }