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

adjust makeAbstract for prefixed terms

Browse Source
This commit is contained in:
dockes 2007-01-19 10:23:26 +00:00
parent 9b2db79f97
commit 238cd56d6c
1 changed files with 30 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
src/rcldb/rcldb.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.102 2007-01-17 13:53:41 dockes Exp $ (C) 2004 J.F.Dockes"; static char rcsid[] = "@(#$Id: rcldb.cpp,v 1.103 2007-01-19 10:23:26 dockes Exp $ (C) 2004 J.F.Dockes";
#endif #endif
/* /*
* This program is free software; you can redistribute it and/or modify * This program is free software; you can redistribute it and/or modify
@ -214,29 +214,49 @@ bool Native::dbDataToRclDoc(Xapian::docid docid, std::string &data, Doc &doc)
return true; return true;
} }
static list<string> noPrefixList(const list<string>& in)
{
list<string> out;
for (list<string>::const_iterator qit = in.begin();
qit != in.end(); qit++) {
if ('A' <= qit->at(0) && qit->at(0) <= 'Z') {
string term = *qit;
while (term.length() && 'A' <= term.at(0) && term.at(0) <= 'Z')
term.erase(0, 1);
if (term.length())
out.push_back(term);
continue;
} else {
out.push_back(*qit);
}
}
return out;
}
// Build a document abstract by extracting text chunks around the query terms // Build a document abstract by extracting text chunks around the query terms
// This uses the db termlists, not the original document. // This uses the db termlists, not the original document.
string Native::makeAbstract(Xapian::docid docid, const list<string>& terms) string Native::makeAbstract(Xapian::docid docid, const list<string>& iterms)
{ {
Chrono chron; Chrono chron;
LOGDEB(("makeAbstract:%d: maxlen %d wWidth %d\n", chron.ms(), LOGDEB2(("makeAbstract:%d: maxlen %d wWidth %d\n", chron.ms(),
m_db->m_synthAbsLen, m_db->m_synthAbsWordCtxLen)); m_db->m_synthAbsLen, m_db->m_synthAbsWordCtxLen));
list<string> terms = noPrefixList(iterms);
if (terms.empty()) { if (terms.empty()) {
return ""; return "";
} }
// We may want to use the db-wide freqs to tune the abstracts one // We may want to use the db-wide freqs to tune the abstracts one
// day but we currently don't // day but we currently don't
#if 0 #if 0
if (m_termfreqs.empty()) { if (m_termfreqs.empty()) {
for (list<string>::const_iterator qit = terms.begin(); for (list<string>::const_iterator qit = terms.begin();
qit != terms.end(); qit++) { qit != terms.end(); qit++) {
m_termfreqs[*qit] = db.get_termfreq(*qit); m_termfreqs[*qit] = db.get_termfreq(*qit);
LOGDEB2(("makeAbstract: [%s] db freq %d\n", qit->c_str(), LOGDEB(("makeAbstract: [%s] db freq %d\n", qit->c_str(),
m_termfreqs[*qit])); m_termfreqs[*qit]));
} }
LOGDEB2(("makeAbstract:%d: got termfreqs\n", chron.ms())); LOGDEB(("makeAbstract:%d: got termfreqs\n", chron.ms()));
} }
#endif #endif
@ -360,6 +380,9 @@ string Native::makeAbstract(Xapian::docid docid, const list<string>& terms)
for (term = db.termlist_begin(docid); for (term = db.termlist_begin(docid);
term != db.termlist_end(docid); term++) { term != db.termlist_end(docid); term++) {
// Ignore prefixed terms
if ('A' <= (*term).at(0) && (*term).at(0) <= 'Z')
continue;
if (cutoff-- < 0) { if (cutoff-- < 0) {
LOGDEB(("makeAbstract: max term count cutoff\n")); LOGDEB(("makeAbstract: max term count cutoff\n"));
break; break;