Function Html_parse_numeric_charref

src/html.cc:876–953

* Parse a numeric character reference (e.g., "/" or "/"). * The "&#" has already been consumed. */

 * The "&#" has already been consumed.
 */
static const char *Html_parse_numeric_charref(DilloHtml *html, char *tok,
                                              bool_t is_attr, int *entsize)
{
   static char buf[5];
   char *s = tok;
   int n, codepoint = -1;

   errno = 0;

   if (*s == 'x' || *s == 'X') {
      if (isxdigit(*++s)) {
         /* strtol with base 16 accepts leading "0x" - we don't */
         if (*s == '0' && s[1] == 'x') {
            s++;
            codepoint = 0;
         } else {
            codepoint = strtol(s, &s, 16);
         }
      }
   } else if (isdigit(*s)) {
      codepoint = strtol(s, &s, 10);
   }
   if (errno)
      codepoint = -1;

   if (*s == ';')
      s++;
   else {
      if (prefs.show_extra_warnings && (html->DocType == DT_XHTML ||
          (html->DocType == DT_HTML && html->DocTypeVersion <= 4.01f))) {
         char c = *s;
         *s = '\0';
         BUG_MSG("Character reference '&#%s' lacks ';'.", tok);
         *s = c;
      }
      /* Don't require ';' for old HTML, except that our current heuristic
       * is to require it in attributes to avoid cases like "&copy=1" found
       * in URLs.
       */
      if (is_attr || html->DocType == DT_XHTML ||
          (html->DocType == DT_HTML && html->DocTypeVersion >= 5.0f)) {
         return NULL;
      }

   }
   if ((codepoint < 0x20 && codepoint != '\t' && codepoint != '\n' &&
        codepoint != '\f') ||
       (codepoint >= 0x7f && codepoint <= 0x9f) ||
       (codepoint >= 0xd800 && codepoint <= 0xdfff) || codepoint > 0x10ffff ||
       ((codepoint & 0xfffe) == 0xfffe) ||
       (!(html->DocType == DT_HTML && html->DocTypeVersion >= 5.0f) &&
        codepoint > 0xffff)) {
      /* this catches null bytes, errors, codes out of range, disallowed
       * control chars, permanently undefined chars, and surrogates.
       */
      char c = *s;
      *s = '\0';
      BUG_MSG("Numeric character reference '&#%s' is not valid.", tok);