Update expat to 2.2.0 to fix CVE vulnerability.

third_party/expat/files/lib/xmltok_impl.c

 /* Copyright (c) 1998, 1999 Thai Open Source Software Center Ltd
    See the file COPYING for copying permission.
 */
 
 /* This file is included! */
 #ifdef XML_TOK_IMPL_C
 
 #ifndef IS_INVALID_CHAR
 #define IS_INVALID_CHAR(enc, ptr, n) (0)
 #endif
@@ skipping 69 matching lines @@
     ptr += MINBPC(enc); \
     break; \
   CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
   CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
   CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
 
 #ifndef PREFIX
 #define PREFIX(ident) ident
 #endif
 
+
+#define HAS_CHARS(enc, ptr, end, count) \
+    (end - ptr >= count * MINBPC(enc))
+
+#define HAS_CHAR(enc, ptr, end) \
+    HAS_CHARS(enc, ptr, end, 1)
+
+#define REQUIRE_CHARS(enc, ptr, end, count) \
+    { \
+      if (! HAS_CHARS(enc, ptr, end, count)) { \
+        return XML_TOK_PARTIAL; \
+      } \
+    }
+
+#define REQUIRE_CHAR(enc, ptr, end) \
+    REQUIRE_CHARS(enc, ptr, end, 1)
+
+
 /* ptr points to character following "<!-" */
 
 static int PTRCALL
 PREFIX(scanComment)(const ENCODING *enc, const char *ptr,
                     const char *end, const char **nextTokPtr)
 {
-  if (ptr != end) {
+  if (HAS_CHAR(enc, ptr, end)) {
     if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
     ptr += MINBPC(enc);
-    while (ptr != end) {
+    while (HAS_CHAR(enc, ptr, end)) {
       switch (BYTE_TYPE(enc, ptr)) {
       INVALID_CASES(ptr, nextTokPtr)
       case BT_MINUS:
-        if ((ptr += MINBPC(enc)) == end)
-          return XML_TOK_PARTIAL;
+        ptr += MINBPC(enc);
+        REQUIRE_CHAR(enc, ptr, end);
         if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
-          if ((ptr += MINBPC(enc)) == end)
-            return XML_TOK_PARTIAL;
+          ptr += MINBPC(enc);
+          REQUIRE_CHAR(enc, ptr, end);
           if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
             *nextTokPtr = ptr;
             return XML_TOK_INVALID;
           }
           *nextTokPtr = ptr + MINBPC(enc);
           return XML_TOK_COMMENT;
         }
         break;
       default:
         ptr += MINBPC(enc);
         break;
       }
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 /* ptr points to character following "<!" */
 
 static int PTRCALL
 PREFIX(scanDecl)(const ENCODING *enc, const char *ptr,
                  const char *end, const char **nextTokPtr)
 {
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   case BT_MINUS:
     return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_LSQB:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_COND_SECT_OPEN;
   case BT_NMSTRT:
   case BT_HEX:
     ptr += MINBPC(enc);
     break;
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_PERCNT:
-      if (ptr + MINBPC(enc) == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHARS(enc, ptr, end, 2);
       /* don't allow <!ENTITY% foo "whatever"> */
       switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
       case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       /* fall through */
     case BT_S: case BT_CR: case BT_LF:
       *nextTokPtr = ptr;
       return XML_TOK_DECL_OPEN;
     case BT_NMSTRT:
     case BT_HEX:
       ptr += MINBPC(enc);
       break;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 static int PTRCALL
-PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr,
+PREFIX(checkPiTarget)(const ENCODING *UNUSED_P(enc), const char *ptr,
                       const char *end, int *tokPtr)
 {
   int upper = 0;
   *tokPtr = XML_TOK_PI;
   if (end - ptr != MINBPC(enc)*3)
     return 1;
   switch (BYTE_TO_ASCII(enc, ptr)) {
   case ASCII_x:
     break;
   case ASCII_X:
@@ skipping 29 matching lines @@
 }
 
 /* ptr points to character following "<?" */
 
 static int PTRCALL
 PREFIX(scanPi)(const ENCODING *enc, const char *ptr,
                const char *end, const char **nextTokPtr)
 {
   int tok;
   const char *target = ptr;
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_S: case BT_CR: case BT_LF:
       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       ptr += MINBPC(enc);
-      while (ptr != end) {
+      while (HAS_CHAR(enc, ptr, end)) {
         switch (BYTE_TYPE(enc, ptr)) {
         INVALID_CASES(ptr, nextTokPtr)
         case BT_QUEST:
           ptr += MINBPC(enc);
-          if (ptr == end)
-            return XML_TOK_PARTIAL;
+          REQUIRE_CHAR(enc, ptr, end);
           if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
             *nextTokPtr = ptr + MINBPC(enc);
             return tok;
           }
           break;
         default:
           ptr += MINBPC(enc);
           break;
         }
       }
       return XML_TOK_PARTIAL;
     case BT_QUEST:
       if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       ptr += MINBPC(enc);
-      if (ptr == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHAR(enc, ptr, end);
       if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
         *nextTokPtr = ptr + MINBPC(enc);
         return tok;
       }
       /* fall through */
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 static int PTRCALL
-PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr,
+PREFIX(scanCdataSection)(const ENCODING *UNUSED_P(enc), const char *ptr,
                          const char *end, const char **nextTokPtr)
 {
   static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A,
                                      ASCII_T, ASCII_A, ASCII_LSQB };
   int i;
   /* CDATA[ */
-  if (end - ptr < 6 * MINBPC(enc))
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHARS(enc, ptr, end, 6);
   for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
     if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   *nextTokPtr = ptr;
   return XML_TOK_CDATA_SECT_OPEN;
 }
 
 static int PTRCALL
 PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr,
                         const char *end, const char **nextTokPtr)
 {
-  if (ptr == end)
+  if (ptr >= end)
     return XML_TOK_NONE;
   if (MINBPC(enc) > 1) {
     size_t n = end - ptr;
     if (n & (MINBPC(enc) - 1)) {
       n &= ~(MINBPC(enc) - 1);
       if (n == 0)
         return XML_TOK_PARTIAL;
       end = ptr + n;
     }
   }
   switch (BYTE_TYPE(enc, ptr)) {
   case BT_RSQB:
     ptr += MINBPC(enc);
-    if (ptr == end)
-      return XML_TOK_PARTIAL;
+    REQUIRE_CHAR(enc, ptr, end);
     if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
       break;
     ptr += MINBPC(enc);
-    if (ptr == end)
-      return XML_TOK_PARTIAL;
+    REQUIRE_CHAR(enc, ptr, end);
     if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
       ptr -= MINBPC(enc);
       break;
     }
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_CDATA_SECT_CLOSE;
   case BT_CR:
     ptr += MINBPC(enc);
-    if (ptr == end)
-      return XML_TOK_PARTIAL;
+    REQUIRE_CHAR(enc, ptr, end);
     if (BYTE_TYPE(enc, ptr) == BT_LF)
       ptr += MINBPC(enc);
     *nextTokPtr = ptr;
     return XML_TOK_DATA_NEWLINE;
   case BT_LF:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_DATA_NEWLINE;
   INVALID_CASES(ptr, nextTokPtr)
   default:
     ptr += MINBPC(enc);
     break;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
 #define LEAD_CASE(n) \
     case BT_LEAD ## n: \
       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
         *nextTokPtr = ptr; \
         return XML_TOK_DATA_CHARS; \
       } \
       ptr += n; \
       break;
     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
@@ skipping 14 matching lines @@
   *nextTokPtr = ptr;
   return XML_TOK_DATA_CHARS;
 }
 
 /* ptr points to character following "</" */
 
 static int PTRCALL
 PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr,
                    const char *end, const char **nextTokPtr)
 {
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_S: case BT_CR: case BT_LF:
-      for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
+      for (ptr += MINBPC(enc); HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
         switch (BYTE_TYPE(enc, ptr)) {
         case BT_S: case BT_CR: case BT_LF:
           break;
         case BT_GT:
           *nextTokPtr = ptr + MINBPC(enc);
           return XML_TOK_END_TAG;
         default:
           *nextTokPtr = ptr;
           return XML_TOK_INVALID;
         }
@@ skipping 16 matching lines @@
   }
   return XML_TOK_PARTIAL;
 }
 
 /* ptr points to character following "&#X" */
 
 static int PTRCALL
 PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr,
                        const char *end, const char **nextTokPtr)
 {
-  if (ptr != end) {
+  if (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_DIGIT:
     case BT_HEX:
       break;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
-    for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
+    for (ptr += MINBPC(enc); HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
       switch (BYTE_TYPE(enc, ptr)) {
       case BT_DIGIT:
       case BT_HEX:
         break;
       case BT_SEMI:
         *nextTokPtr = ptr + MINBPC(enc);
         return XML_TOK_CHAR_REF;
       default:
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 /* ptr points to character following "&#" */
 
 static int PTRCALL
 PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr,
                     const char *end, const char **nextTokPtr)
 {
-  if (ptr != end) {
+  if (HAS_CHAR(enc, ptr, end)) {
     if (CHAR_MATCHES(enc, ptr, ASCII_x))
       return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_DIGIT:
       break;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
-    for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
+    for (ptr += MINBPC(enc); HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
       switch (BYTE_TYPE(enc, ptr)) {
       case BT_DIGIT:
         break;
       case BT_SEMI:
         *nextTokPtr = ptr + MINBPC(enc);
         return XML_TOK_CHAR_REF;
       default:
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 /* ptr points to character following "&" */
 
 static int PTRCALL
 PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
                 const char **nextTokPtr)
 {
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   case BT_NUM:
     return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_SEMI:
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_ENTITY_REF;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 /* ptr points to character following first character of attribute name */
 
 static int PTRCALL
 PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
                  const char **nextTokPtr)
 {
 #ifdef XML_NS
   int hadColon = 0;
 #endif
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 #ifdef XML_NS
     case BT_COLON:
       if (hadColon) {
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       hadColon = 1;
       ptr += MINBPC(enc);
-      if (ptr == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHAR(enc, ptr, end);
       switch (BYTE_TYPE(enc, ptr)) {
       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
       default:
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       break;
 #endif
     case BT_S: case BT_CR: case BT_LF:
       for (;;) {
         int t;
 
         ptr += MINBPC(enc);
-        if (ptr == end)
-          return XML_TOK_PARTIAL;
+        REQUIRE_CHAR(enc, ptr, end);
         t = BYTE_TYPE(enc, ptr);
         if (t == BT_EQUALS)
           break;
         switch (t) {
         case BT_S:
         case BT_LF:
         case BT_CR:
           break;
         default:
           *nextTokPtr = ptr;
           return XML_TOK_INVALID;
         }
       }
     /* fall through */
     case BT_EQUALS:
       {
         int open;
 #ifdef XML_NS
         hadColon = 0;
 #endif
         for (;;) {
           ptr += MINBPC(enc);
-          if (ptr == end)
-            return XML_TOK_PARTIAL;
+          REQUIRE_CHAR(enc, ptr, end);
           open = BYTE_TYPE(enc, ptr);
           if (open == BT_QUOT || open == BT_APOS)
             break;
           switch (open) {
           case BT_S:
           case BT_LF:
           case BT_CR:
             break;
           default:
             *nextTokPtr = ptr;
             return XML_TOK_INVALID;
           }
         }
         ptr += MINBPC(enc);
         /* in attribute value */
         for (;;) {
           int t;
-          if (ptr == end)
-            return XML_TOK_PARTIAL;
+          REQUIRE_CHAR(enc, ptr, end);
           t = BYTE_TYPE(enc, ptr);
           if (t == open)
             break;
           switch (t) {
           INVALID_CASES(ptr, nextTokPtr)
           case BT_AMP:
             {
               int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
               if (tok <= 0) {
                 if (tok == XML_TOK_INVALID)
                   *nextTokPtr = ptr;
                 return tok;
               }
               break;
             }
           case BT_LT:
             *nextTokPtr = ptr;
             return XML_TOK_INVALID;
           default:
             ptr += MINBPC(enc);
             break;
           }
         }
         ptr += MINBPC(enc);
-        if (ptr == end)
-          return XML_TOK_PARTIAL;
+        REQUIRE_CHAR(enc, ptr, end);
         switch (BYTE_TYPE(enc, ptr)) {
         case BT_S:
         case BT_CR:
         case BT_LF:
           break;
         case BT_SOL:
           goto sol;
         case BT_GT:
           goto gt;
         default:
           *nextTokPtr = ptr;
           return XML_TOK_INVALID;
         }
         /* ptr points to closing quote */
         for (;;) {
           ptr += MINBPC(enc);
-          if (ptr == end)
-            return XML_TOK_PARTIAL;
+          REQUIRE_CHAR(enc, ptr, end);
           switch (BYTE_TYPE(enc, ptr)) {
           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
           case BT_S: case BT_CR: case BT_LF:
             continue;
           case BT_GT:
           gt:
             *nextTokPtr = ptr + MINBPC(enc);
             return XML_TOK_START_TAG_WITH_ATTS;
           case BT_SOL:
           sol:
             ptr += MINBPC(enc);
-            if (ptr == end)
-              return XML_TOK_PARTIAL;
+            REQUIRE_CHAR(enc, ptr, end);
             if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
               *nextTokPtr = ptr;
               return XML_TOK_INVALID;
             }
             *nextTokPtr = ptr + MINBPC(enc);
             return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
           default:
             *nextTokPtr = ptr;
             return XML_TOK_INVALID;
           }
@@ skipping 11 matching lines @@
 
 /* ptr points to character following "<" */
 
 static int PTRCALL
 PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
                const char **nextTokPtr)
 {
 #ifdef XML_NS
   int hadColon;
 #endif
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   case BT_EXCL:
-    if ((ptr += MINBPC(enc)) == end)
-      return XML_TOK_PARTIAL;
+    ptr += MINBPC(enc);
+    REQUIRE_CHAR(enc, ptr, end);
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_MINUS:
       return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
     case BT_LSQB:
       return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc),
                                       end, nextTokPtr);
     }
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   case BT_QUEST:
     return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_SOL:
     return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
 #ifdef XML_NS
   hadColon = 0;
 #endif
   /* we have a start-tag */
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
 #ifdef XML_NS
     case BT_COLON:
       if (hadColon) {
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       hadColon = 1;
       ptr += MINBPC(enc);
-      if (ptr == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHAR(enc, ptr, end);
       switch (BYTE_TYPE(enc, ptr)) {
       CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
       default:
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       break;
 #endif
     case BT_S: case BT_CR: case BT_LF:
       {
         ptr += MINBPC(enc);
-        while (ptr != end) {
+        while (HAS_CHAR(enc, ptr, end)) {
           switch (BYTE_TYPE(enc, ptr)) {
           CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
           case BT_GT:
             goto gt;
           case BT_SOL:
             goto sol;
           case BT_S: case BT_CR: case BT_LF:
             ptr += MINBPC(enc);
             continue;
           default:
             *nextTokPtr = ptr;
             return XML_TOK_INVALID;
           }
           return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
         }
         return XML_TOK_PARTIAL;
       }
     case BT_GT:
     gt:
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_START_TAG_NO_ATTS;
     case BT_SOL:
     sol:
       ptr += MINBPC(enc);
-      if (ptr == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHAR(enc, ptr, end);
       if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
         *nextTokPtr = ptr;
         return XML_TOK_INVALID;
       }
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_EMPTY_ELEMENT_NO_ATTS;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 static int PTRCALL
 PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end,
                    const char **nextTokPtr)
 {
-  if (ptr == end)
+  if (ptr >= end)
     return XML_TOK_NONE;
   if (MINBPC(enc) > 1) {
     size_t n = end - ptr;
     if (n & (MINBPC(enc) - 1)) {
       n &= ~(MINBPC(enc) - 1);
       if (n == 0)
         return XML_TOK_PARTIAL;
       end = ptr + n;
     }
   }
   switch (BYTE_TYPE(enc, ptr)) {
   case BT_LT:
     return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_AMP:
     return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_CR:
     ptr += MINBPC(enc);
-    if (ptr == end)
+    if (! HAS_CHAR(enc, ptr, end))
       return XML_TOK_TRAILING_CR;
     if (BYTE_TYPE(enc, ptr) == BT_LF)
       ptr += MINBPC(enc);
     *nextTokPtr = ptr;
     return XML_TOK_DATA_NEWLINE;
   case BT_LF:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_DATA_NEWLINE;
   case BT_RSQB:
     ptr += MINBPC(enc);
-    if (ptr == end)
+    if (! HAS_CHAR(enc, ptr, end))
       return XML_TOK_TRAILING_RSQB;
     if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
       break;
     ptr += MINBPC(enc);
-    if (ptr == end)
+    if (! HAS_CHAR(enc, ptr, end))
       return XML_TOK_TRAILING_RSQB;
     if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
       ptr -= MINBPC(enc);
       break;
     }
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   INVALID_CASES(ptr, nextTokPtr)
   default:
     ptr += MINBPC(enc);
     break;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
 #define LEAD_CASE(n) \
     case BT_LEAD ## n: \
       if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
         *nextTokPtr = ptr; \
         return XML_TOK_DATA_CHARS; \
       } \
       ptr += n; \
       break;
     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 #undef LEAD_CASE
     case BT_RSQB:
-      if (ptr + MINBPC(enc) != end) {
+      if (HAS_CHARS(enc, ptr, end, 2)) {
          if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
            ptr += MINBPC(enc);
            break;
          }
-         if (ptr + 2*MINBPC(enc) != end) {
+         if (HAS_CHARS(enc, ptr, end, 3)) {
            if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) {
              ptr += MINBPC(enc);
              break;
            }
            *nextTokPtr = ptr + 2*MINBPC(enc);
            return XML_TOK_INVALID;
          }
       }
       /* fall through */
     case BT_AMP:
@@ skipping 13 matching lines @@
   *nextTokPtr = ptr;
   return XML_TOK_DATA_CHARS;
 }
 
 /* ptr points to character following "%" */
 
 static int PTRCALL
 PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end,
                     const char **nextTokPtr)
 {
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   case BT_S: case BT_LF: case BT_CR: case BT_PERCNT:
     *nextTokPtr = ptr;
     return XML_TOK_PERCENT;
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_SEMI:
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_PARAM_ENTITY_REF;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 static int PTRCALL
 PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end,
                       const char **nextTokPtr)
 {
-  if (ptr == end)
-    return XML_TOK_PARTIAL;
+  REQUIRE_CHAR(enc, ptr, end);
   switch (BYTE_TYPE(enc, ptr)) {
   CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_CR: case BT_LF: case BT_S:
     case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
       *nextTokPtr = ptr;
       return XML_TOK_POUND_NAME;
     default:
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   }
   return -XML_TOK_POUND_NAME;
 }
 
 static int PTRCALL
 PREFIX(scanLit)(int open, const ENCODING *enc,
                 const char *ptr, const char *end,
                 const char **nextTokPtr)
 {
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     int t = BYTE_TYPE(enc, ptr);
     switch (t) {
     INVALID_CASES(ptr, nextTokPtr)
     case BT_QUOT:
     case BT_APOS:
       ptr += MINBPC(enc);
       if (t != open)
         break;
-      if (ptr == end)
+      if (! HAS_CHAR(enc, ptr, end))
         return -XML_TOK_LITERAL;
       *nextTokPtr = ptr;
       switch (BYTE_TYPE(enc, ptr)) {
       case BT_S: case BT_CR: case BT_LF:
       case BT_GT: case BT_PERCNT: case BT_LSQB:
         return XML_TOK_LITERAL;
       default:
         return XML_TOK_INVALID;
       }
     default:
       ptr += MINBPC(enc);
       break;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 static int PTRCALL
 PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
                   const char **nextTokPtr)
 {
   int tok;
-  if (ptr == end)
+  if (ptr >= end)
     return XML_TOK_NONE;
   if (MINBPC(enc) > 1) {
     size_t n = end - ptr;
     if (n & (MINBPC(enc) - 1)) {
       n &= ~(MINBPC(enc) - 1);
       if (n == 0)
         return XML_TOK_PARTIAL;
       end = ptr + n;
     }
   }
   switch (BYTE_TYPE(enc, ptr)) {
   case BT_QUOT:
     return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_APOS:
     return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_LT:
     {
       ptr += MINBPC(enc);
-      if (ptr == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHAR(enc, ptr, end);
       switch (BYTE_TYPE(enc, ptr)) {
       case BT_EXCL:
         return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
       case BT_QUEST:
         return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
       case BT_NMSTRT:
       case BT_HEX:
       case BT_NONASCII:
       case BT_LEAD2:
       case BT_LEAD3:
       case BT_LEAD4:
         *nextTokPtr = ptr - MINBPC(enc);
         return XML_TOK_INSTANCE_START;
       }
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     }
   case BT_CR:
     if (ptr + MINBPC(enc) == end) {
       *nextTokPtr = end;
       /* indicate that this might be part of a CR/LF pair */
       return -XML_TOK_PROLOG_S;
     }
     /* fall through */
   case BT_S: case BT_LF:
     for (;;) {
       ptr += MINBPC(enc);
-      if (ptr == end)
+      if (! HAS_CHAR(enc, ptr, end))
         break;
       switch (BYTE_TYPE(enc, ptr)) {
       case BT_S: case BT_LF:
         break;
       case BT_CR:
         /* don't split CR/LF pair */
         if (ptr + MINBPC(enc) != end)
           break;
         /* fall through */
       default:
         *nextTokPtr = ptr;
         return XML_TOK_PROLOG_S;
       }
     }
     *nextTokPtr = ptr;
     return XML_TOK_PROLOG_S;
   case BT_PERCNT:
     return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
   case BT_COMMA:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_COMMA;
   case BT_LSQB:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_OPEN_BRACKET;
   case BT_RSQB:
     ptr += MINBPC(enc);
-    if (ptr == end)
+    if (! HAS_CHAR(enc, ptr, end))
       return -XML_TOK_CLOSE_BRACKET;
     if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
-      if (ptr + MINBPC(enc) == end)
-        return XML_TOK_PARTIAL;
+      REQUIRE_CHARS(enc, ptr, end, 2);
       if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) {
         *nextTokPtr = ptr + 2*MINBPC(enc);
         return XML_TOK_COND_SECT_CLOSE;
       }
     }
     *nextTokPtr = ptr;
     return XML_TOK_CLOSE_BRACKET;
   case BT_LPAR:
     *nextTokPtr = ptr + MINBPC(enc);
     return XML_TOK_OPEN_PAREN;
   case BT_RPAR:
     ptr += MINBPC(enc);
-    if (ptr == end)
+    if (! HAS_CHAR(enc, ptr, end))
       return -XML_TOK_CLOSE_PAREN;
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_AST:
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_CLOSE_PAREN_ASTERISK;
     case BT_QUEST:
       *nextTokPtr = ptr + MINBPC(enc);
       return XML_TOK_CLOSE_PAREN_QUESTION;
     case BT_PLUS:
       *nextTokPtr = ptr + MINBPC(enc);
@@ skipping 55 matching lines @@
     if (IS_NAME_CHAR_MINBPC(enc, ptr)) {
       ptr += MINBPC(enc);
       tok = XML_TOK_NMTOKEN;
       break;
     }
     /* fall through */
   default:
     *nextTokPtr = ptr;
     return XML_TOK_INVALID;
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
     case BT_GT: case BT_RPAR: case BT_COMMA:
     case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
     case BT_S: case BT_CR: case BT_LF:
       *nextTokPtr = ptr;
       return tok;
 #ifdef XML_NS
     case BT_COLON:
       ptr += MINBPC(enc);
       switch (tok) {
       case XML_TOK_NAME:
-        if (ptr == end)
-          return XML_TOK_PARTIAL;
+        REQUIRE_CHAR(enc, ptr, end);
         tok = XML_TOK_PREFIXED_NAME;
         switch (BYTE_TYPE(enc, ptr)) {
         CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
         default:
           tok = XML_TOK_NMTOKEN;
           break;
         }
         break;
       case XML_TOK_PREFIXED_NAME:
         tok = XML_TOK_NMTOKEN;
@@ skipping 28 matching lines @@
     }
   }
   return -tok;
 }
 
 static int PTRCALL
 PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr,
                           const char *end, const char **nextTokPtr)
 {
   const char *start;
-  if (ptr == end)
+  if (ptr >= end)
     return XML_TOK_NONE;
+  else if (! HAS_CHAR(enc, ptr, end))
+    return XML_TOK_PARTIAL;
   start = ptr;
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
 #define LEAD_CASE(n) \
     case BT_LEAD ## n: ptr += n; break;
     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 #undef LEAD_CASE
     case BT_AMP:
       if (ptr == start)
         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_LT:
       /* this is for inside entity references */
       *nextTokPtr = ptr;
       return XML_TOK_INVALID;
     case BT_LF:
       if (ptr == start) {
         *nextTokPtr = ptr + MINBPC(enc);
         return XML_TOK_DATA_NEWLINE;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_CR:
       if (ptr == start) {
         ptr += MINBPC(enc);
-        if (ptr == end)
+        if (! HAS_CHAR(enc, ptr, end))
           return XML_TOK_TRAILING_CR;
         if (BYTE_TYPE(enc, ptr) == BT_LF)
           ptr += MINBPC(enc);
         *nextTokPtr = ptr;
         return XML_TOK_DATA_NEWLINE;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_S:
       if (ptr == start) {
         *nextTokPtr = ptr + MINBPC(enc);
         return XML_TOK_ATTRIBUTE_VALUE_S;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     default:
       ptr += MINBPC(enc);
       break;
     }
   }
   *nextTokPtr = ptr;
   return XML_TOK_DATA_CHARS;
 }
 
 static int PTRCALL
 PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr,
                        const char *end, const char **nextTokPtr)
 {
   const char *start;
-  if (ptr == end)
+  if (ptr >= end)
     return XML_TOK_NONE;
+  else if (! HAS_CHAR(enc, ptr, end))
+    return XML_TOK_PARTIAL;
   start = ptr;
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
 #define LEAD_CASE(n) \
     case BT_LEAD ## n: ptr += n; break;
     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 #undef LEAD_CASE
     case BT_AMP:
       if (ptr == start)
         return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_PERCNT:
       if (ptr == start) {
         int tok =  PREFIX(scanPercent)(enc, ptr + MINBPC(enc),
                                        end, nextTokPtr);
         return (tok == XML_TOK_PERCENT) ? XML_TOK_INVALID : tok;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_LF:
       if (ptr == start) {
         *nextTokPtr = ptr + MINBPC(enc);
         return XML_TOK_DATA_NEWLINE;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     case BT_CR:
       if (ptr == start) {
         ptr += MINBPC(enc);
-        if (ptr == end)
+        if (! HAS_CHAR(enc, ptr, end))
           return XML_TOK_TRAILING_CR;
         if (BYTE_TYPE(enc, ptr) == BT_LF)
           ptr += MINBPC(enc);
         *nextTokPtr = ptr;
         return XML_TOK_DATA_NEWLINE;
       }
       *nextTokPtr = ptr;
       return XML_TOK_DATA_CHARS;
     default:
       ptr += MINBPC(enc);
@@ skipping 11 matching lines @@
                          const char *end, const char **nextTokPtr)
 {
   int level = 0;
   if (MINBPC(enc) > 1) {
     size_t n = end - ptr;
     if (n & (MINBPC(enc) - 1)) {
       n &= ~(MINBPC(enc) - 1);
       end = ptr + n;
     }
   }
-  while (ptr != end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
     INVALID_CASES(ptr, nextTokPtr)
     case BT_LT:
-      if ((ptr += MINBPC(enc)) == end)
-        return XML_TOK_PARTIAL;
+      ptr += MINBPC(enc);
+      REQUIRE_CHAR(enc, ptr, end);
       if (CHAR_MATCHES(enc, ptr, ASCII_EXCL)) {
-        if ((ptr += MINBPC(enc)) == end)
-          return XML_TOK_PARTIAL;
+        ptr += MINBPC(enc);
+        REQUIRE_CHAR(enc, ptr, end);
         if (CHAR_MATCHES(enc, ptr, ASCII_LSQB)) {
           ++level;
           ptr += MINBPC(enc);
         }
       }
       break;
     case BT_RSQB:
-      if ((ptr += MINBPC(enc)) == end)
-        return XML_TOK_PARTIAL;
+      ptr += MINBPC(enc);
+      REQUIRE_CHAR(enc, ptr, end);
       if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
-        if ((ptr += MINBPC(enc)) == end)
-          return XML_TOK_PARTIAL;
+        ptr += MINBPC(enc);
+        REQUIRE_CHAR(enc, ptr, end);
         if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
           ptr += MINBPC(enc);
           if (level == 0) {
             *nextTokPtr = ptr;
             return XML_TOK_IGNORE_SECT;
           }
           --level;
         }
       }
       break;
     default:
       ptr += MINBPC(enc);
       break;
     }
   }
   return XML_TOK_PARTIAL;
 }
 
 #endif /* XML_DTD */
 
 static int PTRCALL
 PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
                    const char **badPtr)
 {
   ptr += MINBPC(enc);
   end -= MINBPC(enc);
-  for (; ptr != end; ptr += MINBPC(enc)) {
+  for (; HAS_CHAR(enc, ptr, end); ptr += MINBPC(enc)) {
     switch (BYTE_TYPE(enc, ptr)) {
     case BT_DIGIT:
     case BT_HEX:
     case BT_MINUS:
     case BT_APOS:
     case BT_LPAR:
     case BT_RPAR:
     case BT_PLUS:
     case BT_COMMA:
     case BT_SOL:
@@ skipping 127 matching lines @@
         return nAtts;
       break;
     default:
       break;
     }
   }
   /* not reached */
 }
 
 static int PTRFASTCALL
-PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr)
+PREFIX(charRefNumber)(const ENCODING *UNUSED_P(enc), const char *ptr)
 {
   int result = 0;
   /* skip &# */
   ptr += 2*MINBPC(enc);
   if (CHAR_MATCHES(enc, ptr, ASCII_x)) {
     for (ptr += MINBPC(enc);
          !CHAR_MATCHES(enc, ptr, ASCII_SEMI);
          ptr += MINBPC(enc)) {
       int c = BYTE_TO_ASCII(enc, ptr);
       switch (c) {
@@ skipping 23 matching lines @@
       result *= 10;
       result += (c - ASCII_0);
       if (result >= 0x110000)
         return -1;
     }
   }
   return checkCharRefNumber(result);
 }
 
 static int PTRCALL
-PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr,
+PREFIX(predefinedEntityName)(const ENCODING *UNUSED_P(enc), const char *ptr,
                              const char *end)
 {
   switch ((end - ptr)/MINBPC(enc)) {
   case 2:
     if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_t)) {
       switch (BYTE_TO_ASCII(enc, ptr)) {
       case ASCII_l:
         return ASCII_LT;
       case ASCII_g:
         return ASCII_GT;
@@ skipping 97 matching lines @@
         return 0;
       default:
         return 1;
       }
     }
   }
   /* not reached */
 }
 
 static int PTRCALL
-PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1,
+PREFIX(nameMatchesAscii)(const ENCODING *UNUSED_P(enc), const char *ptr1,
                          const char *end1, const char *ptr2)
 {
   for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
-    if (ptr1 == end1)
+    if (end1 - ptr1 < MINBPC(enc))
       return 0;
     if (!CHAR_MATCHES(enc, ptr1, *ptr2))
       return 0;
   }
   return ptr1 == end1;
 }
 
 static int PTRFASTCALL
 PREFIX(nameLength)(const ENCODING *enc, const char *ptr)
 {
@@ skipping 36 matching lines @@
     }
   }
 }
 
 static void PTRCALL
 PREFIX(updatePosition)(const ENCODING *enc,
                        const char *ptr,
                        const char *end,
                        POSITION *pos)
 {
-  while (ptr < end) {
+  while (HAS_CHAR(enc, ptr, end)) {
     switch (BYTE_TYPE(enc, ptr)) {
 #define LEAD_CASE(n) \
     case BT_LEAD ## n: \
       ptr += n; \
       break;
     LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
 #undef LEAD_CASE
     case BT_LF:
       pos->columnNumber = (XML_Size)-1;
       pos->lineNumber++;
       ptr += MINBPC(enc);
       break;
     case BT_CR:
       pos->lineNumber++;
       ptr += MINBPC(enc);
-      if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF)
+      if (HAS_CHAR(enc, ptr, end) && BYTE_TYPE(enc, ptr) == BT_LF)
         ptr += MINBPC(enc);
       pos->columnNumber = (XML_Size)-1;
       break;
     default:
       ptr += MINBPC(enc);
       break;
     }
     pos->columnNumber++;
   }
 }
 
 #undef DO_LEAD_CASE
 #undef MULTIBYTE_CASES
 #undef INVALID_CASES
 #undef CHECK_NAME_CASE
 #undef CHECK_NAME_CASES
 #undef CHECK_NMSTRT_CASE
 #undef CHECK_NMSTRT_CASES
 
 #endif /* XML_TOK_IMPL_C */