xmlNode* htmlParseFragment(void *doc, void *buffer, int buffer_len, void *url, int options, void *error_buffer, int error_buffer_len) {
xmlNode* root_element = NULL;
xmlParserErrors errCode;
errCode = xmlParseInNodeContext((xmlNodePtr)doc, buffer, buffer_len, options, &root_element);
if (errCode != XML_ERR_OK) {
char *c_error_buffer = (char*)error_buffer;
snprintf(c_error_buffer, error_buffer_len, "xml fragemnt parsing error (xmlParserErrors):%d", errCode);
return NULL;
}
return root_element;
}
str
BATXMLcontent(bat *ret, const bat *bid)
{
BAT *b, *bn;
BUN p, q;
BATiter bi;
xmlDocPtr doc;
xmlNodePtr root;
size_t size = BUFSIZ;
str buf = GDKmalloc(size);
const char *err = OPERATION_FAILED;
xmlBufferPtr xbuf;
if (buf == NULL)
throw(MAL,"xml.content",MAL_MALLOC_FAIL);
if ((b = BATdescriptor(*bid)) == NULL) {
GDKfree(buf);
throw(MAL, "xml.content", INTERNAL_BAT_ACCESS);
}
doc = xmlParseMemory("<doc/>", 6);
root = xmlDocGetRootElement(doc);
prepareResult(bn, b, TYPE_xml, "content", GDKfree(buf));
bi = bat_iterator(b);
xbuf = xmlBufferCreate();
BATloop(b, p, q) {
const char *t = (const char *) BUNtail(bi, p);
size_t len;
xmlNodePtr elem;
xmlParserErrors xerr;
const xmlChar *s;
if (strNil(t)) {
bunfastapp(bn, str_nil);
bn->T->nonil = 0;
continue;
}
len = strlen(t);
xerr = xmlParseInNodeContext(root, t, (int) len, 0, &elem);
if (xerr != XML_ERR_OK) {
err = XML_PARSE_ERROR;
goto bunins_failed;
}
xmlNodeDump(xbuf, doc, elem, 0, 0);
s = xmlBufferContent(xbuf);
len = strlen((const char *) s);
if (len + 2 >= size) {
GDKfree(buf);
size = len + 128;
buf = GDKmalloc(size);
if (buf == NULL) {
err = MAL_MALLOC_FAIL;
goto bunins_failed;
}
}
buf[0] = 'C';
strcpy(buf + 1, (const char *) s);
bunfastapp(bn, buf);
xmlBufferEmpty(xbuf);
xmlFreeNodeList(elem);
}
xmlBufferFree(xbuf);
xmlFreeDoc(doc);
GDKfree(buf);
finalizeResult(ret, bn, b);
return MAL_SUCCEED;
bunins_failed:
xmlBufferFree(xbuf);
xmlFreeDoc(doc);
if (buf != NULL)
GDKfree(buf);
BBPunfix(b->batCacheid);
BBPunfix(bn->batCacheid);
throw(MAL, "xml.document", "%s", err);
}
str
BATXMLxmltext(bat *ret, const bat *bid)
{
BAT *b, *bn;
BUN p, q;
BATiter bi;
size_t size = 0;
str buf = NULL;
xmlDocPtr doc = NULL;
xmlNodePtr elem;
str content = NULL;
const char *err = OPERATION_FAILED;
if ((b = BATdescriptor(*bid)) == NULL)
throw(MAL, "xml.text", INTERNAL_BAT_ACCESS);
prepareResult(bn, b, TYPE_str, "text", (void) 0);
bi = bat_iterator(b);
BATloop(b, p, q) {
const char *t = (const char *) BUNtail(bi, p);
size_t len;
if (strNil(t)) {
bunfastapp(bn, t);
bn->T->nonil = 0;
continue;
}
len = strlen(t);
switch (*t) {
case 'D': {
xmlDocPtr d = xmlParseMemory(t + 1, (int) (len - 1));
elem = xmlDocGetRootElement(d);
content = (str) xmlNodeGetContent(elem);
xmlFreeDoc(d);
if (content == NULL) {
err = MAL_MALLOC_FAIL;
goto bunins_failed;
}
break;
}
case 'C':
if (doc == NULL)
doc = xmlParseMemory("<doc/>", 6);
xmlParseInNodeContext(xmlDocGetRootElement(doc), t + 1, (int) (len - 1), 0, &elem);
content = (str) xmlNodeGetContent(elem);
xmlFreeNodeList(elem);
if (content == NULL) {
err = MAL_MALLOC_FAIL;
goto bunins_failed;
}
break;
case 'A': {
str s;
if (buf == NULL || size < len) {
size = len + 128;
if (buf != NULL)
GDKfree(buf);
buf = GDKmalloc(size);
if (buf == NULL) {
err = MAL_MALLOC_FAIL;
goto bunins_failed;
}
}
s = buf;
t++;
while (*t) {
if (*t == '"' || *t == '\'') {
char q = *t++;
s += XMLunquotestring(&t, q, s);
}
t++;
}
*s = 0;
break;
}
default:
assert(*t == 'A' || *t == 'C' || *t == 'D');
bunfastapp(bn, str_nil);
bn->T->nonil = 0;
continue;
}
assert(content != NULL || buf != NULL);
bunfastapp(bn, content != NULL ? content : buf);
if (content != NULL)
GDKfree(content);
content = NULL;
}
finalizeResult(ret, bn, b);
if (buf != NULL)
GDKfree(buf);
if (doc != NULL)
xmlFreeDoc(doc);
return MAL_SUCCEED;
bunins_failed:
BBPunfix(b->batCacheid);
BBPunfix(bn->batCacheid);
if (buf != NULL)
GDKfree(buf);
if (doc != NULL)
xmlFreeDoc(doc);
if (content != NULL)
GDKfree(content);
throw(MAL, "xml.text", "%s", err);
}
