Attachment #4635 for bug #8196

View | Details | Raw Unified | Return to bug 8196
Collapse All | Expand All

Lines 83-88 Link Here

(-)htmldoc-1.8.23.orig/htmldoc/htmldoc.cxx (-3 / +30 lines)
83	static void term_handler(int signum);	83	static void term_handler(int signum);
84	static void usage(void);	84	static void usage(void);
85		85
		86	/*
		87	* Local variables...
		88	*/
		89
		90	int autocharset = 0;
		91
86		92
87	/*	93	/*
88	* 'main()' - Main entry for HTMLDOC.	94	* 'main()' - Main entry for HTMLDOC.
Lines 247-253 Link Here
247	{	253	{
248	i ++;	254	i ++;
249	if (i < argc)	255	if (i < argc)
250	htmlSetCharSet(argv[i]);	256	{
		257	if (strcmp(argv[i], "auto") == 0)
		258	{
		259	autocharset = 1;
		260	}
		261	else htmlSetCharSet(argv[i]);
		262	}
251	else	263	else
252	usage();	264	usage();
253	}	265	}
Lines 1034-1039 Link Here
1034		1046
1035	while (document->prev != NULL)	1047	while (document->prev != NULL)
1036	document = document->prev;	1048	document = document->prev;
		1049
		1050	if (autocharset)
		1051	htmlSetMetaCharSet(document);
1037		1052
1038	htmlDebugStats("Document Tree", document);	1053	htmlDebugStats("Document Tree", document);
1039		1054
Lines 1269-1275 Link Here
1269	else if (strncasecmp(line, "XRXCOMMENTS=", 12) == 0)	1284	else if (strncasecmp(line, "XRXCOMMENTS=", 12) == 0)
1270	XRXComments = atoi(line + 12);	1285	XRXComments = atoi(line + 12);
1271	else if (strncasecmp(line, "CHARSET=", 8) == 0)	1286	else if (strncasecmp(line, "CHARSET=", 8) == 0)
1272	htmlSetCharSet(line + 8);	1287	{
		1288	if (strcmp(line + 8, "auto") == 0)
		1289	{
		1290	autocharset = 1;
		1291	}
		1292	else htmlSetCharSet(line + 8);
		1293	}
1273	else if (strncasecmp(line, "PAGEMODE=", 9) == 0)	1294	else if (strncasecmp(line, "PAGEMODE=", 9) == 0)
1274	PDFPageMode = atoi(line + 9);	1295	PDFPageMode = atoi(line + 9);
1275	else if (strncasecmp(line, "PAGELAYOUT=", 11) == 0)	1296	else if (strncasecmp(line, "PAGELAYOUT=", 11) == 0)
Lines 1994-2000 Link Here
1994	}	2015	}
1995	}	2016	}
1996	else if (strcmp(temp, "--charset") == 0)	2017	else if (strcmp(temp, "--charset") == 0)
1997	htmlSetCharSet(temp2);	2018	{
		2019	if (strcmp(temp2, "auto") == 0)
		2020	{
		2021	autocharset = 1;
		2022	}
		2023	else htmlSetCharSet(temp2);
		2024	}
1998	else if (strcmp(temp, "--pagemode") == 0)	2025	else if (strcmp(temp, "--pagemode") == 0)
1999	{	2026	{
2000	for (i = 0; i < (int)(sizeof(PDFModes) / sizeof(PDFModes[0])); i ++)	2027	for (i = 0; i < (int)(sizeof(PDFModes) / sizeof(PDFModes[0])); i ++)

Lines 309-314 Link Here

(-)htmldoc-1.8.23.orig/htmldoc/html.h (+1 lines)
309		309
310	extern void htmlSetBaseSize(float p, float s);	310	extern void htmlSetBaseSize(float p, float s);
311	extern void htmlSetCharSet(const char *cs);	311	extern void htmlSetCharSet(const char *cs);
		312	extern int htmlSetMetaCharSet(tree_t *tree);
312	extern void htmlSetTextColor(uchar *color);	313	extern void htmlSetTextColor(uchar *color);
313		314
314	extern void htmlDebugStats(const char title, tree_t t);	315	extern void htmlDebugStats(const char title, tree_t t);




  return (NULL);
}

/*
 * 'htmlGetMetaCharSet()' - Get document charset from "meta" data...
 */

int
htmlSetMetaCharSet(tree_t *tree)	/* I - Document tree */
{
  uchar	*tname,			/* Name value from tree entry */
	*tcontent,		/* Content value from tree entry */
	*tchar, *tchar2;
  uchar charset[256];

  while (tree != NULL)
  {
   /*
    * Check this tree entry...
    */
    if (tree->markup == MARKUP_META)
    {
      if ((tname = htmlGetVariable(tree, (uchar *)"HTTP-EQUIV")) != NULL &&
          (tcontent = htmlGetVariable(tree, (uchar *)"CONTENT")) != NULL)
	if (strcasecmp((char *)tname, "Content-Type") == 0)
	  if ((tchar = (uchar *)strstr((char *)tcontent, "charset=")) != NULL ||
	      (tchar = (uchar *)strstr((char *)tcontent, "CHARSET=")) != NULL)
	  {

	    tchar += 8;
	    for (tchar2 = charset; (*tchar >= (uchar)'a' && *tchar <= (uchar)'z') ||
	                           (*tchar >= (uchar)'A' && *tchar <= (uchar)'Z') ||
	                           (*tchar >= (uchar)'0' && *tchar <= (uchar)'9') ||
			           *tchar == (uchar)'_' || *tchar == (uchar)'-';)
	      *tchar2++ = *tchar++;
	    *tchar2 = (uchar)'\0';
	    for (tchar2 = charset; *tchar2 != (uchar)'\0'; tchar2++)
	      *tchar2 = (uchar)tolower((char)*tchar2);
	    htmlSetCharSet((char *)charset);
	    return 1;
	  }
      if ((tname = htmlGetVariable(tree, (uchar *)"CHARSET")) != NULL)
      {
        tchar = tname;
	for (tchar2 = charset; *tchar != (uchar)'\0';)
	  *tchar2++ = (uchar)tolower((char)*tchar++);
	*tchar2 = (uchar)'\0';
	htmlSetCharSet((char *)charset);
	return 1;
      }
    }

   /*
    * Check child entries...
    */

    if (tree->child != NULL)
      if (htmlSetMetaCharSet(tree->child))
        return 1;

   /*
    * Next tree entry...
    */

    tree = tree->next;
  }

  return 0;
}


/*
 * 'htmlGetStyle()' - Get a style value from a node's STYLE attribute.

Return to bug 8196

Lines 1854-1859 Link Here

(-)htmldoc-1.8.23.orig/htmldoc/htmllib.cxx (+67 lines)
1854	return (NULL);	1854	return (NULL);
1855	}	1855	}
1856		1856
		1857	/*
		1858	* 'htmlGetMetaCharSet()' - Get document charset from "meta" data...
		1859	*/
		1860
		1861	int
		1862	htmlSetMetaCharSet(tree_t tree) / I - Document tree */
		1863	{
		1864	uchar tname, / Name value from tree entry */
		1865	tcontent, / Content value from tree entry */
		1866	tchar, tchar2;
		1867	uchar charset[256];
		1868
		1869	while (tree != NULL)
		1870	{
		1871	/*
		1872	* Check this tree entry...
		1873	*/
		1874	if (tree->markup == MARKUP_META)
		1875	{
		1876	if ((tname = htmlGetVariable(tree, (uchar *)"HTTP-EQUIV")) != NULL &&
		1877	(tcontent = htmlGetVariable(tree, (uchar *)"CONTENT")) != NULL)
		1878	if (strcasecmp((char *)tname, "Content-Type") == 0)
		1879	if ((tchar = (uchar )strstr((char )tcontent, "charset=")) != NULL \|\|
		1880	(tchar = (uchar )strstr((char )tcontent, "CHARSET=")) != NULL)
		1881	{
		1882
		1883	tchar += 8;
		1884	for (tchar2 = charset; (tchar >= (uchar)'a' && tchar <= (uchar)'z') \|\|
		1885	(tchar >= (uchar)'A' && tchar <= (uchar)'Z') \|\|
		1886	(tchar >= (uchar)'0' && tchar <= (uchar)'9') \|\|
		1887	tchar == (uchar)'_' \|\| tchar == (uchar)'-';)
		1888	tchar2++ = tchar++;
		1889	*tchar2 = (uchar)'\0';
		1890	for (tchar2 = charset; *tchar2 != (uchar)'\0'; tchar2++)
		1891	tchar2 = (uchar)tolower((char)tchar2);
		1892	htmlSetCharSet((char *)charset);
		1893	return 1;
		1894	}
		1895	if ((tname = htmlGetVariable(tree, (uchar *)"CHARSET")) != NULL)
		1896	{
		1897	tchar = tname;
		1898	for (tchar2 = charset; *tchar != (uchar)'\0';)
		1899	tchar2++ = (uchar)tolower((char)tchar++);
		1900	*tchar2 = (uchar)'\0';
		1901	htmlSetCharSet((char *)charset);
		1902	return 1;
		1903	}
		1904	}
		1905
		1906	/*
		1907	* Check child entries...
		1908	*/
		1909
		1910	if (tree->child != NULL)
		1911	if (htmlSetMetaCharSet(tree->child))
		1912	return 1;
		1913
		1914	/*
		1915	* Next tree entry...
		1916	*/
		1917
		1918	tree = tree->next;
		1919	}
		1920
1921	return 0;
1922	}
1923
1857		1924
1858	/*	1925	/*
1859	* 'htmlGetStyle()' - Get a style value from a node's STYLE attribute.	1926	* 'htmlGetStyle()' - Get a style value from a node's STYLE attribute.