mirror of
https://github.com/privacore/open-source-search-engine.git
synced 2025-07-15 02:36:08 -04:00
Fix get_titlerec
This commit is contained in:
@ -93,65 +93,71 @@ int main(int argc, char **argv) {
|
||||
key96_t startKey = Titledb::makeFirstKey(docId);
|
||||
key96_t endKey = Titledb::makeLastKey(docId);
|
||||
|
||||
msg5.getList(RDB_TITLEDB, 0, &list, startKey, endKey, 500000000, true, 0, 0, -1, NULL, NULL, 0, true, NULL, 0, -1, -1LL, false, true);
|
||||
|
||||
if (list.getNumRecs() != 1) {
|
||||
if (!msg5.getList(RDB_TITLEDB, 0, &list, startKey, endKey, 500000000, true, 0, 0, -1, NULL, NULL, 0, true, NULL, 0, -1, -1LL, false, true)) {
|
||||
logf(LOG_TRACE, "msg5.getlist didn't block");
|
||||
cleanup();
|
||||
exit(1);
|
||||
}
|
||||
|
||||
if (list.isEmpty()) {
|
||||
logf(LOG_TRACE, "Unable to find titlerec for docId=%" PRIu64, docId);
|
||||
cleanup();
|
||||
exit(1);
|
||||
}
|
||||
|
||||
XmlDoc xmlDoc;
|
||||
if (!xmlDoc.set2(list.getCurrentRec(), list.getCurrentRecSize(), "main", NULL, 0)) {
|
||||
logf(LOG_TRACE, "Unable to set XmlDoc for docId=%" PRIu64, docId);
|
||||
cleanup();
|
||||
exit(1);
|
||||
}
|
||||
for (list.resetListPtr(); !list.isExhausted(); list.skipCurrentRecord()) {
|
||||
XmlDoc xmlDoc;
|
||||
if (!xmlDoc.set2(list.getCurrentRec(), list.getCurrentRecSize(), "main", NULL, 0)) {
|
||||
logf(LOG_TRACE, "Unable to set XmlDoc for docId=%" PRIu64, docId);
|
||||
cleanup();
|
||||
exit(1);
|
||||
}
|
||||
|
||||
logf(LOG_TRACE, "XmlDoc info");
|
||||
logf(LOG_TRACE, "\tfirstUrl : %.*s", xmlDoc.size_firstUrl, xmlDoc.ptr_firstUrl);
|
||||
logf(LOG_TRACE, "\tredirUrl : %.*s", xmlDoc.size_redirUrl, xmlDoc.ptr_redirUrl);
|
||||
logf(LOG_TRACE, "\trootTitle : %.*s", xmlDoc.size_rootTitleBuf, xmlDoc.ptr_rootTitleBuf);
|
||||
// logf(LOG_TRACE, "\timageData :");
|
||||
logf(LOG_TRACE, "\t");
|
||||
loghex(LOG_TRACE, xmlDoc.ptr_utf8Content, xmlDoc.size_utf8Content, "\tutf8Content:");
|
||||
logf(LOG_TRACE, "\tsite : %.*s", xmlDoc.size_site, xmlDoc.ptr_site);
|
||||
logf(LOG_TRACE, "XmlDoc info");
|
||||
logf(LOG_TRACE, "\tfirstUrl : %.*s", xmlDoc.size_firstUrl, xmlDoc.ptr_firstUrl);
|
||||
logf(LOG_TRACE, "\tredirUrl : %.*s", xmlDoc.size_redirUrl, xmlDoc.ptr_redirUrl);
|
||||
logf(LOG_TRACE, "\trootTitle : %.*s", xmlDoc.size_rootTitleBuf, xmlDoc.ptr_rootTitleBuf);
|
||||
// logf(LOG_TRACE, "\timageData :");
|
||||
logf(LOG_TRACE, "\t");
|
||||
loghex(LOG_TRACE, xmlDoc.ptr_utf8Content, xmlDoc.size_utf8Content, "\tutf8Content:");
|
||||
logf(LOG_TRACE, "\tsite : %.*s", xmlDoc.size_site, xmlDoc.ptr_site);
|
||||
|
||||
logf(LOG_TRACE, "\tlinkInfo");
|
||||
LinkInfo* linkInfo = xmlDoc.getLinkInfo1();
|
||||
logf(LOG_TRACE, "\t\tm_numGoodInlinks : %d", linkInfo->m_numGoodInlinks);
|
||||
logf(LOG_TRACE, "\t\tm_numInlinksInternal : %d", linkInfo->m_numInlinksInternal);
|
||||
logf(LOG_TRACE, "\t\tm_numStoredInlinks : %d", linkInfo->m_numStoredInlinks);
|
||||
logf(LOG_TRACE, "\tlinkInfo");
|
||||
LinkInfo *linkInfo = xmlDoc.getLinkInfo1();
|
||||
logf(LOG_TRACE, "\t\tm_numGoodInlinks : %d", linkInfo->m_numGoodInlinks);
|
||||
logf(LOG_TRACE, "\t\tm_numInlinksInternal : %d", linkInfo->m_numInlinksInternal);
|
||||
logf(LOG_TRACE, "\t\tm_numStoredInlinks : %d", linkInfo->m_numStoredInlinks);
|
||||
|
||||
int i = 0;
|
||||
for (Inlink *inlink = linkInfo->getNextInlink(NULL); inlink; inlink = linkInfo->getNextInlink(inlink)) {
|
||||
logf(LOG_TRACE, "\t\tinlink #%d", i++);
|
||||
logf(LOG_TRACE, "\t\t\tdocId : %" PRIu64, inlink->m_docId);
|
||||
logf(LOG_TRACE, "\t\t\turl : %s", inlink->getUrl());
|
||||
logf(LOG_TRACE, "\t\t\tlinktext : %s", inlink->getLinkText());
|
||||
logf(LOG_TRACE, "\t\t\tcountry : %s", getCountryCode(inlink->m_country));
|
||||
logf(LOG_TRACE, "\t\t\tlanguage : %s", getLanguageAbbr(inlink->m_language));
|
||||
}
|
||||
int i = 0;
|
||||
for (Inlink *inlink = linkInfo->getNextInlink(NULL); inlink; inlink = linkInfo->getNextInlink(inlink)) {
|
||||
logf(LOG_TRACE, "\t\tinlink #%d", i++);
|
||||
logf(LOG_TRACE, "\t\t\tdocId : %" PRIu64, inlink->m_docId);
|
||||
logf(LOG_TRACE, "\t\t\turl : %s", inlink->getUrl());
|
||||
logf(LOG_TRACE, "\t\t\tlinktext : %s", inlink->getLinkText());
|
||||
logf(LOG_TRACE, "\t\t\tcountry : %s", getCountryCode(inlink->m_country));
|
||||
logf(LOG_TRACE, "\t\t\tlanguage : %s", getLanguageAbbr(inlink->m_language));
|
||||
}
|
||||
|
||||
loghex(LOG_TRACE, xmlDoc.ptr_linkdbData, xmlDoc.size_linkdbData, "\tlinkdbData");
|
||||
loghex(LOG_TRACE, xmlDoc.ptr_linkdbData, xmlDoc.size_linkdbData, "\tlinkdbData");
|
||||
|
||||
logf(LOG_TRACE, "\ttagRec");
|
||||
TagRec *tagRec = xmlDoc.getTagRec();
|
||||
for (Tag *tag = tagRec->getFirstTag(); tag; tag = tagRec->getNextTag(tag)) {
|
||||
SafeBuf sb;
|
||||
tag->printDataToBuf(&sb);
|
||||
logf(LOG_TRACE, "\t\t%-12s: %s", getTagStrFromType(tag->m_type), sb.getBufStart());
|
||||
}
|
||||
logf(LOG_TRACE, "\ttagRec");
|
||||
TagRec *tagRec = xmlDoc.getTagRec();
|
||||
for (Tag *tag = tagRec->getFirstTag(); tag; tag = tagRec->getNextTag(tag)) {
|
||||
SafeBuf sb;
|
||||
tag->printDataToBuf(&sb);
|
||||
logf(LOG_TRACE, "\t\t%-12s: %s", getTagStrFromType(tag->m_type), sb.getBufStart());
|
||||
}
|
||||
|
||||
logf(LOG_TRACE, "\t");
|
||||
|
||||
logf(LOG_TRACE, "Links info");
|
||||
g_log.m_disabled = true;
|
||||
Links *links = xmlDoc.getLinks();
|
||||
g_log.m_disabled = false;
|
||||
for (int i = 0; i < links->getNumLinks(); ++i) {
|
||||
logf(LOG_TRACE, "\tlink : %.*s", links->getLinkLen(i), links->getLinkPtr(i));
|
||||
logf(LOG_TRACE, "\t");
|
||||
|
||||
logf(LOG_TRACE, "Links info");
|
||||
g_log.m_disabled = true;
|
||||
Links *links = xmlDoc.getLinks();
|
||||
g_log.m_disabled = false;
|
||||
for (int i = 0; i < links->getNumLinks(); ++i) {
|
||||
logf(LOG_TRACE, "\tlink : %.*s", links->getLinkLen(i), links->getLinkPtr(i));
|
||||
}
|
||||
}
|
||||
cleanup();
|
||||
|
||||
|
Reference in New Issue
Block a user