tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

snippets: need a multimap for the group container

Browse Source
This commit is contained in:
"Jean-Francois Dockes ext:(%22) 2012-09-27 13:12:07 +02:00
parent be7aed5659
commit eeb014077f
2 changed files with 10 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

26
src/rcldb/rclabstract.cpp
View File

@ -112,7 +112,7 @@ void Query::Native::setDbWideQTermsFreqs()
// aggregated frequency. // aggregated frequency.
double Query::Native::qualityTerms(Xapian::docid docid, double Query::Native::qualityTerms(Xapian::docid docid,
const vector<string>& terms, const vector<string>& terms,
map<double, vector<string> >& byQ) multimap<double, vector<string> >& byQ)
{ {
LOGABS(("qualityTerms\n")); LOGABS(("qualityTerms\n"));
setDbWideQTermsFreqs(); setDbWideQTermsFreqs();
@ -129,6 +129,7 @@ double Query::Native::qualityTerms(Xapian::docid docid,
m_q->m_sd->getTerms(hld); m_q->m_sd->getTerms(hld);
} }
// Group the input terms by the user term they were possibly expanded from
map<string, vector<string> > byRoot; map<string, vector<string> > byRoot;
for (vector<string>::const_iterator qit = terms.begin(); for (vector<string>::const_iterator qit = terms.begin();
qit != terms.end(); qit++) { qit != terms.end(); qit++) {
@ -143,14 +144,14 @@ double Query::Native::qualityTerms(Xapian::docid docid,
} }
} }
if (!found) { if (!found) {
LOGABS(("qualityTerms: [%s] not found\n", (*qit).c_str())); LOGDEB0(("qualityTerms: [%s] not found in hld\n", (*qit).c_str()));
byRoot[*qit].push_back(*qit); byRoot[*qit].push_back(*qit);
} }
} }
// Compute in-document and global frequencies for the groups.
map<string, double> grpwdfs; map<string, double> grpwdfs;
map<string, double> grptfreqs; map<string, double> grptfreqs;
for (map<string, vector<string> >::const_iterator git = byRoot.begin(); for (map<string, vector<string> >::const_iterator git = byRoot.begin();
git != byRoot.end(); git++) { git != byRoot.end(); git++) {
for (vector<string>::const_iterator qit = git->second.begin(); for (vector<string>::const_iterator qit = git->second.begin();
@ -169,6 +170,7 @@ double Query::Native::qualityTerms(Xapian::docid docid,
} }
} }
// Build a sorted by quality container for the groups
for (map<string, vector<string> >::const_iterator git = byRoot.begin(); for (map<string, vector<string> >::const_iterator git = byRoot.begin();
git != byRoot.end(); git++) { git != byRoot.end(); git++) {
double q = (grpwdfs[git->first]) * grptfreqs[git->first]; double q = (grpwdfs[git->first]) * grptfreqs[git->first];
@ -185,21 +187,11 @@ double Query::Native::qualityTerms(Xapian::docid docid,
q = 1; q = 1;
} }
totalweight += q; totalweight += q;
for (vector<string>::const_iterator qit = git->second.begin(); byQ.insert(pair<double, vector<string> >(q, git->second));
qit != git->second.end(); qit++) {
termQcoefs[*qit] = q;
}
}
// Build a sorted by quality term list.
for (vector<string>::const_iterator qit = terms.begin();
qit != terms.end(); qit++) {
if (termQcoefs.find(*qit) != termQcoefs.end())
byQ[termQcoefs[*qit]].push_back(*qit);
} }
#ifdef DEBUGABSTRACT #ifdef DEBUGABSTRACT
for (map<double, vector<string> >::reverse_iterator mit = byQ.rbegin(); for (multimap<double, vector<string> >::reverse_iterator mit= byQ.rbegin();
mit != byQ.rend(); mit++) { mit != byQ.rend(); mit++) {
LOGABS(("qualityTerms: group\n")); LOGABS(("qualityTerms: group\n"));
for (vector<string>::const_iterator qit = mit->second.begin(); for (vector<string>::const_iterator qit = mit->second.begin();
@ -240,7 +232,7 @@ int Query::Native::getFirstMatchPage(Xapian::docid docid)
setDbWideQTermsFreqs(); setDbWideQTermsFreqs();
// We try to use a page which matches the "best" term. Get a sorted list // We try to use a page which matches the "best" term. Get a sorted list
map<double, vector<string> > byQ; multimap<double, vector<string> > byQ;
double totalweight = qualityTerms(docid, terms, byQ); double totalweight = qualityTerms(docid, terms, byQ);
for (map<double, vector<string> >::reverse_iterator mit = byQ.rbegin(); for (map<double, vector<string> >::reverse_iterator mit = byQ.rbegin();
@ -303,7 +295,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
// such a group prevents displaying matches for other terms (by // such a group prevents displaying matches for other terms (by
// removing its meaning from the maximum occurrences per term test // removing its meaning from the maximum occurrences per term test
// used while walking the list below) // used while walking the list below)
map<double, vector<string> > byQ; multimap<double, vector<string> > byQ;
double totalweight = qualityTerms(docid, matchedTerms, byQ); double totalweight = qualityTerms(docid, matchedTerms, byQ);
LOGABS(("makeAbstract:%d: computed Qcoefs.\n", chron.ms())); LOGABS(("makeAbstract:%d: computed Qcoefs.\n", chron.ms()));
// This can't happen, but would crash us // This can't happen, but would crash us

2
src/rcldb/rclquery_p.h
View File

@ -60,7 +60,7 @@ public:
void setDbWideQTermsFreqs(); void setDbWideQTermsFreqs();
double qualityTerms(Xapian::docid docid, double qualityTerms(Xapian::docid docid,
const std::vector<std::string>& terms, const std::vector<std::string>& terms,
std::map<double, std::vector<std::string> >& byQ); std::multimap<double, std::vector<std::string> >& byQ);
}; };
} }