std::string parse(const std::string& xml, const std::string& name, const std::string& url, const std::string& html, const std::string& htmlheader, std::vector<std::pair<std::string, std::string> >& attaches) {
std::string ret("");
std::string pp = getParserPath(url);
xsltStylesheetPtr xslt = xsltParseStylesheetFile(BAD_CAST pp.c_str());
htmlDocPtr doc = NULL;
static std::string encoding("gb18030");
std::string mimetype = getMIMEType(htmlheader, html);
if (!mimetype.empty() && mimetype == "text/xml") {
doc = html.empty() ? NULL : xmlReadDoc(BAD_CAST html.c_str(), NULL, encoding.c_str(), XML_PARSE_RECOVER);
} else {
doc = html.empty() ? NULL : htmlParseDoc(BAD_CAST html.c_str(), encoding.c_str());
}
if (doc != NULL) {
const char *params[7] = {0};
size_t n_param = 0;
params[n_param] = NULL;
xmlDocPtr res = xsltApplyStylesheet(xslt, doc, params);
//free_xslt_params(params, n_param);
if (res != NULL) {
xmlChar *s = NULL;
int len = 0;
if (xsltSaveResultToString(&s, &len, res, xslt) >= 0) {
ret.assign((const char *)s, len);
xmlFree(s);
}
xmlFreeDoc(res);
}
xmlFreeDoc(doc);
}
return ret;
}
int main(int argc, char **argv)
{
assert(argv[1]);
std::ifstream t(argv[1]);
std::string str((std::istreambuf_iterator<char>(t)),
std::istreambuf_iterator<char>());
//tidy::tidy(str);
//xmlDocPtr doc = htmlParseFile(f_doc, NULL);
xmlDocPtr doc = htmlParseDoc(BAD_CAST str.c_str(), NULL);
if (!doc) {
printf("libxml failed to parse file <%s>\n", argv[1]);
return 1;
}
xmlNode *doc_head = xmlDocGetRootElement(doc);
if (argc < 3) {
print_element_names(doc_head, 0);
} else {
run_xpath(doc, argv[2]);
}
//xmlFree(doc_head);
return 0;
}
Pvoid_t html_tokenize_into_features(const char * html, Pvoid_t features) {
xmlSubstituteEntitiesDefault(0);
htmlDocPtr doc = htmlParseDoc(BAD_CAST html, "UTF-8");
if (doc) {
Buffer *buf = extractText(doc);
features = tokenize_text(buf->buf, buf->length, features);
features = tokenize_uris(doc, features);
free_buffer(buf);
xmlFreeDoc(doc);
}
return features;
}