Deal with externally specified encodings

expat/xmlparse/xmlparse.c

@@ -84,6 +84,7 @@ typedef enum XML_Error Processor(XML_Parser parser,
 static Processor prologProcessor;
 static Processor contentProcessor;
 static Processor epilogProcessor;
+static Processor errorProcessor;
 
 static enum XML_Error
 doContent(XML_Parser parser, int startTagLevel, const ENCODING *enc,
@@ -210,7 +211,6 @@ XML_Parser XML_ParserCreate(const char *encodingName)
   if (!parser)
     return parser;
   processor = prologProcessor;
-  XmlInitEncoding(&initEncoding, &encoding);
   XmlPrologStateInit(&prologState);
   userData = 0;
   startElementHandler = 0;
@@ -245,6 +245,10 @@ XML_Parser XML_ParserCreate(const char *encodingName)
     return 0;
   }
   dataBufEnd = dataBuf + INIT_DATA_BUF_SIZE;
+  if (!XmlInitEncoding(&initEncoding, &encoding, encodingName)) {
+    errorCode = XML_ERROR_UNKNOWN_ENCODING;
+    processor = errorProcessor;
+  }
   return parser;
 }
 
@@ -1147,6 +1151,15 @@ enum XML_Error epilogProcessor(XML_Parser parser,
   }
 }
 
+static
+enum XML_Error errorProcessor(XML_Parser parser,
+			      const char *s,
+			      const char *end,
+			      const char **nextPtr)
+{
+  return errorCode;
+}
+
 static enum XML_Error
 storeAttributeValue(XML_Parser parser, const ENCODING *enc, int isCdata,
 		    const char *ptr, const char *end,

expat/xmltok/xmltok.c

@@ -393,6 +393,24 @@ static const struct encoding big2_encoding = { VTABLE, 2 };
 
 #undef PREFIX
 
+static
+int streqci(const char *s1, const char *s2)
+{
+  for (;;) {
+    char c1 = *s1++;
+    char c2 = *s2++;
+    if ('a' <= c1 && c1 <= 'z')
+      c1 += 'A' - 'a';
+    if ('a' <= c2 && c2 <= 'z')
+      c2 += 'A' - 'a';
+    if (c1 != c2)
+      return 0;
+    if (!c1)
+      break;
+  }
+  return 1;
+}
+
 static
 int initScan(const ENCODING *enc, int state, const char *ptr, const char *end,
 	     const char **nextTokPtr)
@@ -463,14 +481,27 @@ const ENCODING *XmlGetInternalEncoding(int e)
   return 0;
 }
 
-void XmlInitEncoding(INIT_ENCODING *p, const ENCODING **encPtr)
+int XmlInitEncoding(INIT_ENCODING *p, const ENCODING **encPtr, const char *name)
 {
+  if (name) {
+    if (streqci(name, "ISO-8859-1")) {
+      *encPtr = &latin1_encoding.enc;
+      return 1;
+    }
+    if (streqci(name, "UTF-8")) {
+      *encPtr = &utf8_encoding.enc;
+      return 1;
+    }
+    if (!streqci(name, "UTF-16"))
+      return 0;
+  }
   p->initEnc.scanners[XML_PROLOG_STATE] = initScanProlog;
   p->initEnc.scanners[XML_CONTENT_STATE] = initScanContent;
   p->initEnc.updatePosition = initUpdatePosition;
   p->initEnc.minBytesPerChar = 1;
   p->encPtr = encPtr;
   *encPtr = &(p->initEnc);
+  return 1;
 }
 
 static
@@ -581,15 +612,6 @@ int parsePseudoAttribute(const ENCODING *enc,
   return 1;
 }
 
-static
-int streq(const char *s1, const char *s2)
-{
-  for (; *s1 == *s2; s1++, s2++)
-    if (!*s1)
-      return 1;
-  return 0;
-}
-
 static
 const ENCODING *findEncoding(const ENCODING *enc, const char *ptr, const char *end)
 {
@@ -605,18 +627,15 @@ const ENCODING *findEncoding(const ENCODING *enc, const char *ptr, const char *e
     if ('a' <= buf[i] && buf[i] <= 'z')
       buf[i] +=  'A' - 'a';
   }
-  if (streq(buf, "UTF-8"))
+  if (streqci(buf, "UTF-8"))
     return &utf8_encoding.enc;
-  if (streq(buf, "ISO-8859-1"))
+  if (streqci(buf, "ISO-8859-1"))
     return &latin1_encoding.enc;
-  if (streq(buf, "UTF-16")) {
+  if (streqci(buf, "UTF-16")) {
     static const unsigned short n = 1;
     if (enc->minBytesPerChar == 2)
       return enc;
-    if (*(const char *)&n)
+    return &big2_encoding;
-      return &little2_encoding;
-    else
-      return &big2_encoding;
   }
   return 0;  
 }

expat/xmltok/xmltok.h

@@ -222,7 +222,7 @@ int XMLTOKAPI XmlParseXmlDecl(int isGeneralTextEntity,
 			      const ENCODING **namedEncodingPtr,
 			      int *standalonePtr);
 
-void XMLTOKAPI XmlInitEncoding(INIT_ENCODING *, const ENCODING **);
+int XMLTOKAPI XmlInitEncoding(INIT_ENCODING *, const ENCODING **, const char *name);
 const ENCODING XMLTOKAPI *XmlGetInternalEncoding(int);
 
 #ifdef __cplusplus

expat/xmlwf/xmlwf.c

@@ -184,6 +184,7 @@ int main(int argc, char **argv)
 {
   int i;
   const char *outputDir = 0;
+  const char *encoding = 0;
   int useFilemap = 1;
 
 #ifdef _MSC_VER
@@ -212,6 +213,16 @@ int main(int argc, char **argv)
 	outputDir = argv[i] + j + 1;
       i++;
     }
+    if (argv[i][j] == 'e') {
+      if (argv[i][j + 1] == '\0') {
+	if (++i == argc)
+	  usage(argv[0]);
+	encoding = argv[i];
+      }
+      else
+	encoding = argv[i] + j + 1;
+      i++;
+    }
     else if (argv[i][j] == '\0' && j > 1)
       i++;
     else
@@ -223,7 +234,7 @@ int main(int argc, char **argv)
     FILE *fp = 0;
     char *outName = 0;
     int result;
-    XML_Parser parser = XML_ParserCreate(0);
+    XML_Parser parser = XML_ParserCreate(encoding);
     if (outputDir) {
       outName = malloc(strlen(outputDir) + strlen(argv[i]) + 2);
       strcpy(outName, outputDir);