ignore malformed html, handle funny encodings

fizx · Jan 4, 2009 · 87d5ae6 · 87d5ae6
1 parent 08ca013
commit 87d5ae6
Show file tree

Hide file tree

Showing 2 changed files with 5 additions and 5 deletions.
diff --git a/dexter.c b/dexter.c
@@ -28,15 +28,15 @@ int yywrap(void){
 xmlDocPtr dex_parse_file(dexPtr dex, char* file, boolean html) {
 	if(html) {
 		htmlParserCtxtPtr htmlCtxt = htmlNewParserCtxt();
-  	htmlDocPtr html = htmlCtxtReadFile(htmlCtxt, file, "UTF-8", 3);
+  	htmlDocPtr html = htmlCtxtReadFile(htmlCtxt, file, NULL, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
 		if(html == NULL) {
 			asprintf(&dex->error, "Couldn't parse file: %s\n", file);
 			return NULL;
 		}
 		return dex_parse_doc(dex, html);
 	} else {
 		xmlParserCtxtPtr ctxt = xmlNewParserCtxt();
-		xmlDocPtr xml = xmlCtxtReadFile(ctxt, file, "UTF-8", 3);
+		xmlDocPtr xml = xmlCtxtReadFile(ctxt, file, NULL, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
 		if(xml == NULL) {
 			asprintf(&dex->error, "Couldn't parse file: %s\n", file);
 			return NULL;
@@ -48,15 +48,15 @@ xmlDocPtr dex_parse_file(dexPtr dex, char* file, boolean html) {
 xmlDocPtr dex_parse_string(dexPtr dex, char* string, size_t size, boolean html) {
 	if(html) {
 		htmlParserCtxtPtr htmlCtxt = htmlNewParserCtxt();
-  	htmlDocPtr html = htmlCtxtReadMemory(htmlCtxt, string, size, "http://kylemaxwell.com/dexter/memory", "UTF-8", 3);
+  	htmlDocPtr html = htmlCtxtReadMemory(htmlCtxt, string, size, "http://kylemaxwell.com/dexter/memory", NULL, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
 		if(html == NULL) {
 			asprintf(&dex->error, "Couldn't parse string\n");
 			return NULL;
 		}
 		return dex_parse_doc(dex, html);
 	} else {
 		xmlParserCtxtPtr ctxt = xmlNewParserCtxt();
- 		xmlDocPtr xml = xmlCtxtReadMemory(ctxt, string, size, "http://kylemaxwell.com/dexter/memory", "UTF-8", 3);
+ 		xmlDocPtr xml = xmlCtxtReadMemory(ctxt, string, size, "http://kylemaxwell.com/dexter/memory", NULL, HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
 		if(xml == NULL) {
 			asprintf(&dex->error, "Couldn't parse string\n");
 			return NULL;

diff --git a/functions.c b/functions.c
@@ -287,7 +287,7 @@ xsltLoadHtmlDocument(xsltTransformContextPtr ctxt, const xmlChar *URI) {
 	ret = ret->next;
     }
 
-    doc = htmlReadFile(URI, NULL, ctxt->parserOptions | HTML_PARSE_RECOVER);
+    doc = htmlReadFile(URI, NULL, ctxt->parserOptions | HTML_PARSE_RECOVER | HTML_PARSE_NOERROR |HTML_PARSE_NOWARNING);
 
     if (doc == NULL)
 	return(NULL);