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

comments and justification

Browse Source
This commit is contained in:
Jean-Francois Dockes
2022-01-07 11:41:54 +01:00
parent 8c122d7de8
commit 25d6d78902
6 changed files with 34 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/rcldb/rclabstract.cpp
+4 -7
View File
@@ -254,7 +254,7 @@ double Query::Native::qualityTerms(Xapian::docid docid,
}
// Return page number for first match of "significant" term.
// Choose most interesting term and return the page number for its first match
int Query::Native::getFirstMatchPage(Xapian::docid docid, string& term)
{
LOGDEB("Query::Native::getFirstMatchPage\n");
@@ -286,9 +286,7 @@ int Query::Native::getFirstMatchPage(Xapian::docid docid, string& term)
qualityTerms(docid, terms, byQ);
for (auto mit = byQ.rbegin(); mit != byQ.rend(); mit++) {
for (vector<string>::const_iterator qit = mit->second.begin();
qit != mit->second.end(); qit++) {
string qterm = *qit;
for (const auto& qterm : mit->second) {
Xapian::PositionIterator pos;
string emptys;
try {
@@ -619,9 +617,8 @@ int Query::Native::abstractFromIndex(
// possibly retried by our caller.
//
// @param[out] vabs the abstract is returned as a vector of snippets.
int Query::Native::makeAbstract(Xapian::docid docid,
vector<Snippet>& vabs,
int imaxoccs, int ictxwords, bool sortbypage)
int Query::Native::makeAbstract(
Xapian::docid docid, vector<Snippet>& vabs, int imaxoccs, int ictxwords, bool sortbypage)
{
chron.restart();
LOGDEB("makeAbstract: docid " << docid << " imaxoccs " <<
src/rcldb/rcldb.cpp
+8 -7
View File
@@ -96,10 +96,13 @@ const string pathelt_prefix = "XP";
static const string udi_prefix("Q");
static const string parent_prefix("F");
// Special terms to mark begin/end of field (for anchored searches), and
// page breaks
// Special terms to mark begin/end of field (for anchored searches).
string start_of_field_term;
string end_of_field_term;
// Special term for page breaks. Note that we use a complicated mechanism for multiple page
// breaks at the same position, when it would have been probably simpler to use XXPG/n terms
// instead (did not try to implement though). A change would force users to reindex.
const string page_break_term = "XXPG/";
// Special term to mark documents with children.
@@ -1846,16 +1849,14 @@ bool Db::addOrUpdate(const string &udi, const string &parent_udi, Doc &doc)
}
}
// If empty pages (multiple break at same pos) were recorded, save
// them (this is because we have no way to record them in the
// Xapian list
// If empty pages (multiple break at same pos) were recorded, save them (this is
// because we have no way to record them in the Xapian list)
if (!tpidx.m_pageincrvec.empty()) {
ostringstream multibreaks;
for (unsigned int i = 0; i < tpidx.m_pageincrvec.size(); i++) {
if (i != 0)
multibreaks << ",";
multibreaks << tpidx.m_pageincrvec[i].first << "," <<
tpidx.m_pageincrvec[i].second;
multibreaks << tpidx.m_pageincrvec[i].first << "," << tpidx.m_pageincrvec[i].second;
}
RECORD_APPEND(record, string(cstr_mbreaks), multibreaks.str());
}
src/rcldb/rclquery.cpp
+1 -3
View File
@@ -360,7 +360,6 @@ int Query::getFirstMatchPage(const Doc &doc, string& term)
return m_reason.empty() ? pagenum : -1;
}
// Mset size
// Note: times for retrieving (multiple times)all docs from a sample
// 25k docs db (q: mime:*)
@@ -511,8 +510,7 @@ vector<string> Query::expand(const Doc &doc)
Xapian::ESet eset = m_nq->xenquire->get_eset(20, rset, false);
LOGDEB("ESet terms:\n");
// We filter out the special terms
for (Xapian::ESetIterator it = eset.begin();
it != eset.end(); it++) {
for (Xapian::ESetIterator it = eset.begin(); it != eset.end(); it++) {
LOGDEB(" [" << (*it) << "]\n");
if ((*it).empty() || has_prefix(*it))
continue;