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

make sure that both the user term and the stem are in the expanded list

Browse Source
This commit is contained in:
dockes 2007-01-19 15:19:51 +00:00
parent b05230b00b
commit 6b394537b2
1 changed files with 35 additions and 27 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

62
src/rcldb/stemdb.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.6 2006-12-19 12:11:21 dockes Exp $ (C) 2005 J.F.Dockes"; static char rcsid[] = "@(#$Id: stemdb.cpp,v 1.7 2007-01-19 15:19:51 dockes Exp $ (C) 2005 J.F.Dockes";
#endif #endif
/** /**
@ -228,48 +228,56 @@ bool stemExpand(const std::string& dbdir,
Xapian::Stem stemmer(lang); Xapian::Stem stemmer(lang);
string stem = stemmer.stem_word(term); string stem = stemmer.stem_word(term);
LOGDEB(("stemExpand: [%s] stem-> [%s]\n", term.c_str(), stem.c_str())); LOGDEB(("stemExpand: [%s] stem-> [%s]\n", term.c_str(), stem.c_str()));
// Try to fetch the doc from the stem db
// Open stem database
string stemdbdir = stemdbname(dbdir, lang); string stemdbdir = stemdbname(dbdir, lang);
Xapian::Database sdb(stemdbdir); Xapian::Database sdb(stemdbdir);
LOGDEB1(("stemExpand: %s lastdocid: %d\n", LOGDEB0(("stemExpand: %s lastdocid: %d\n",
stemdbdir.c_str(), sdb.get_lastdocid())); stemdbdir.c_str(), sdb.get_lastdocid()));
// Try to fetch the doc from the stem db
if (!sdb.term_exists(stem)) { if (!sdb.term_exists(stem)) {
LOGDEB1(("Db::stemExpand: no term for %s\n", stem.c_str())); LOGDEB0(("Db::stemExpand: no term for %s\n", stem.c_str()));
result.push_back(term); } else {
return true; Xapian::PostingIterator did = sdb.postlist_begin(stem);
} if (did == sdb.postlist_end(stem)) {
Xapian::PostingIterator did = sdb.postlist_begin(stem); LOGDEB0(("stemExpand: no term(1) for %s\n",stem.c_str()));
if (did == sdb.postlist_end(stem)) { } else {
LOGDEB1(("stemExpand: no term(1) for %s\n",stem.c_str())); Xapian::Document doc = sdb.get_document(*did);
result.push_back(term); string data = doc.get_data();
return true;
}
Xapian::Document doc = sdb.get_document(*did);
string data = doc.get_data();
// Build expansion list from database data // Build expansion list from database data No need for
// No need for a conftree, but we need to massage the data a little // a conftree, but we need to massage the data a
string::size_type pos = data.find_first_of("="); // little
++pos; string::size_type pos = data.find_first_of("=");
string::size_type pos1 = data.find_last_of("\n"); ++pos;
if (pos == string::npos || pos1 == string::npos ||pos1 <= pos) { // ?? string::size_type pos1 = data.find_last_of("\n");
result.push_back(term); if (pos == string::npos || pos1 == string::npos ||
return true; pos1 <= pos) {
// ??
} else {
stringToStrings(data.substr(pos, pos1-pos), result);
}
}
} }
stringToStrings(data.substr(pos, pos1-pos), result);
// If the user term itself is not in the list, add it. // If the user term or stem are not in the list, add them
if (find(result.begin(), result.end(), term) == result.end()) { if (find(result.begin(), result.end(), term) == result.end()) {
result.push_back(term); result.push_back(term);
} }
LOGDEB(("stemExpand: %s -> %s\n", stem.c_str(), if (find(result.begin(), result.end(), stem) == result.end()) {
stringlistdisp(result).c_str())); result.push_back(stem);
}
LOGDEB0(("stemExpand: %s -> %s\n", stem.c_str(),
stringlistdisp(result).c_str()));
} catch (...) { } catch (...) {
LOGERR(("stemExpand: error accessing stem db. dbdir [%s] lang [%s]\n", LOGERR(("stemExpand: error accessing stem db. dbdir [%s] lang [%s]\n",
dbdir.c_str(), lang.c_str())); dbdir.c_str(), lang.c_str()));
result.push_back(term); result.push_back(term);
return false; return false;
} }
return true; return true;
} }