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

Fixed the autophrase fix: need a full tree flatten to work

Browse Source
This commit is contained in:
Jean-Francois Dockes 2015-06-14 09:22:34 +02:00
parent 3e9581345d
commit 2a69d30701
2 changed files with 206 additions and 71 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

167
src/rcldb/searchdata.cpp
View File

@ -25,6 +25,7 @@
#include <vector>
#include <algorithm>
#include <sstream>
#include <iostream>
using namespace std;
#include "xapian.h"
@ -79,6 +80,10 @@ SearchData::~SearchData()
bool SearchData::maybeAddAutoPhrase(Rcl::Db& db, double freqThreshold)
{
LOGDEB0(("SearchData::maybeAddAutoPhrase()\n"));
// cerr << "BEFORE SIMPLIFY\n"; dump(cerr);
simplify();
// cerr << "AFTER SIMPLIFY\n"; dump(cerr);
if (!m_query.size()) {
LOGDEB2(("SearchData::maybeAddAutoPhrase: empty query\n"));
return false;
@ -90,28 +95,12 @@ bool SearchData::maybeAddAutoPhrase(Rcl::Db& db, double freqThreshold)
// field names, bail out.
for (qlist_it_t it = m_query.begin(); it != m_query.end(); it++) {
SClType tp = (*it)->m_tp;
SearchDataClauseSimple *clp = 0;
if (tp == SCLT_SUB) {
// The query language parser produces subqueries for simple terms
SearchDataClauseSub *subclp =
dynamic_cast<SearchDataClauseSub*>(*it);
if (subclp == 0) {
LOGDEB2(("SearchData::maybeAddAutoPhrase: "
"dyncast to clauseSub failed\n"));
return false;
}
if (!subclp->getSub()->singleSimple()) {
LOGDEB2(("SearchData::maybeAddAutoPhrase: !pureSingle\n"));
return false;
}
clp = dynamic_cast<SearchDataClauseSimple*>(
*(subclp->getSub()->m_query.begin()));
} else if (tp != SCLT_AND && tp != SCLT_OR) {
if (tp != SCLT_AND && tp != SCLT_OR) {
LOGDEB2(("SearchData::maybeAddAutoPhrase: wrong tp %d\n", tp));
return false;
} else {
clp = dynamic_cast<SearchDataClauseSimple*>(*it);
}
SearchDataClauseSimple *clp =
dynamic_cast<SearchDataClauseSimple*>(*it);
if (clp == 0) {
LOGDEB2(("SearchData::maybeAddAutoPhrase: dyncast failed\n"));
return false;
@ -130,9 +119,11 @@ bool SearchData::maybeAddAutoPhrase(Rcl::Db& db, double freqThreshold)
LOGDEB2(("SearchData::maybeAddAutoPhrase: wildcards\n"));
return false;
}
// Do a simple word-split here, don't bother with the full-blown
// textsplit. The autophrase thing is just "best effort", it's
// normal that it won't work in strange cases.
// Do a simple word-split here, not the full-blown
// textsplit. Spans of stopwords should not be trimmed later
// in this function, they will be properly split when the
// phrase gets processed by toNativeQuery() later on.
vector<string> wl;
stringToStrings(clp->gettext(), wl);
words.insert(words.end(), wl.begin(), wl.end());
@ -154,7 +145,7 @@ bool SearchData::maybeAddAutoPhrase(Rcl::Db& db, double freqThreshold)
swords.append(1, ' ');
swords += *it;
} else {
LOGDEB0(("Autophrase: [%s] too frequent (%.2f %%)\n",
LOGDEB0(("SearchData::Autophrase: [%s] too frequent (%.2f %%)\n",
it->c_str(), 100 * freq));
slack++;
}
@ -199,6 +190,65 @@ bool SearchData::fileNameOnly()
return true;
}
void SearchData::simplify()
{
for (unsigned int i = 0; i < m_query.size(); i++) {
if (m_query[i]->m_tp != SCLT_SUB)
continue;
//C[est ce dyncast qui crashe??
SearchDataClauseSub *clsubp =
dynamic_cast<SearchDataClauseSub*>(m_query[i]);
if (clsubp == 0) {
// ??
continue;
}
if (clsubp->getSub()->m_tp != m_tp)
continue;
clsubp->getSub()->simplify();
// If this subquery has special attributes, it's not a
// candidate for collapsing
if (!clsubp->getSub()->m_filetypes.empty() ||
!clsubp->getSub()->m_nfiletypes.empty() ||
clsubp->getSub()->m_haveDates ||
clsubp->getSub()->m_maxSize != size_t(-1) ||
clsubp->getSub()->m_minSize != size_t(-1) ||
clsubp->getSub()->m_haveWildCards)
continue;
bool allsametp = true;
for (qlist_it_t it1 = clsubp->getSub()->m_query.begin();
it1 != clsubp->getSub()->m_query.end(); it1++) {
// We want all AND or OR clause, and same as our conjunction
if (((*it1)->getTp() != SCLT_AND && (*it1)->getTp() != SCLT_OR) ||
(*it1)->getTp() != m_tp) {
allsametp = false;
break;
}
}
if (!allsametp)
continue;
// All ok: delete the clause_sub, and insert the queries from
// its searchdata in its place
m_query.erase(m_query.begin() + i);
m_query.insert(m_query.begin() + i,
clsubp->getSub()->m_query.begin(),
clsubp->getSub()->m_query.end());
for (unsigned int j = i;
j < i + clsubp->getSub()->m_query.size(); j++) {
m_query[j]->setParent(this);
}
i += clsubp->getSub()->m_query.size() - 1;
// We don't want the clauses to be deleted when the parent is, as we
// know own them.
clsubp->getSub()->m_query.clear();
delete clsubp;
}
}
bool SearchData::singleSimple()
{
if (m_query.size() != 1 || !m_filetypes.empty() || !m_nfiletypes.empty() ||
@ -206,8 +256,9 @@ bool SearchData::singleSimple()
m_haveWildCards)
return false;
SearchDataClause *clp = *m_query.begin();
if (clp->getTp() != SCLT_AND && clp->getTp() != SCLT_OR)
if (clp->getTp() != SCLT_AND && clp->getTp() != SCLT_OR) {
return false;
}
return true;
}
@ -219,4 +270,72 @@ void SearchData::getTerms(HighlightData &hld) const
return;
}
void SearchData::dump(ostream& o) const
{
o << "SearchData: " << " qs " << int(m_query.size()) <<
" ft " << m_filetypes.size() << " nft " << m_nfiletypes.size() <<
" hd " << m_haveDates << " maxs " << int(m_maxSize) << " mins " <<
int(m_minSize) << " wc " << m_haveWildCards << "\n";
for (std::vector<SearchDataClause*>::const_iterator it =
m_query.begin(); it != m_query.end(); it++) {
(*it)->dump(o);
o << "\n";
}
o << "\n";
}
void SearchDataClause::dump(ostream& o) const
{
o << "SearchDataClause??";
}
void SearchDataClauseSimple::dump(ostream& o) const
{
o << "ClauseSimple: ";
if (m_exclude)
o << "- ";
o << "[" ;
if (!m_field.empty())
o << m_field << " : ";
o << m_text << "]";
}
void SearchDataClauseFilename::dump(ostream& o) const
{
o << "ClauseFN: ";
if (m_exclude)
o << " - ";
o << "[" << m_text << "]";
}
void SearchDataClausePath::dump(ostream& o) const
{
o << "ClausePath: ";
if (m_exclude)
o << " - ";
o << "[" << m_text << "]";
}
void SearchDataClauseDist::dump(ostream& o) const
{
if (m_tp == SCLT_NEAR)
o << "ClauseDist: NEAR: ";
else
o << "ClauseDist: PHRA: ";
if (m_exclude)
o << " - ";
o << "[";
if (!m_field.empty())
o << m_field << " : ";
o << m_text << "]";
}
void SearchDataClauseSub::dump(ostream& o) const
{
o << "ClauseSub {\n";
m_sub.getconstptr()->dump(o);
o << "}";
}
} // Namespace Rcl

18
src/rcldb/searchdata.h
View File

@ -26,6 +26,7 @@
*/
#include <string>
#include <vector>
#include <ostream>
#include "rcldb.h"
#include "refcntr.h"
@ -161,6 +162,7 @@ public:
int getMaxExp() {return m_maxexp;}
int getMaxCl() {return m_maxcl;}
int getSoftMaxExp() {return m_softmaxexpand;}
void dump(ostream& o) const;
friend class ::AdvSearch;
@ -207,6 +209,11 @@ private:
// value during "find-as-you-type" operations from the GUI
int m_softmaxexpand;
// Collapse bogus subqueries generated by the query parser, mostly
// so that we can check if this is an autophrase candidate (else
// Xapian will do it anyway)
void simplify();
bool expandFileTypes(Rcl::Db &db, std::vector<std::string>& exptps);
bool clausesToQuery(Rcl::Db &db, SClType tp,
std::vector<SearchDataClause*>& query,
@ -297,6 +304,8 @@ public:
virtual Relation getrel() {
return m_rel;
}
virtual void dump(ostream& o) const;
friend class SearchData;
protected:
std::string m_reason;
@ -361,6 +370,8 @@ public:
virtual void setfield(const string& field) {
m_field = field;
}
virtual void dump(ostream& o) const;
protected:
std::string m_text; // Raw user entry text.
std::string m_field; // Field specification if any
@ -382,6 +393,7 @@ protected:
int mods, void *pq, bool useNear, int slack);
};
/**
* Filename search clause. This is special because term expansion is only
* performed against the unsplit file name terms.
@ -404,6 +416,7 @@ public:
}
virtual bool toNativeQuery(Rcl::Db &, void *);
virtual void dump(ostream& o) const;
};
@ -441,7 +454,7 @@ public:
}
virtual bool toNativeQuery(Rcl::Db &, void *);
virtual void dump(ostream& o) const;
};
/**
@ -468,6 +481,7 @@ public:
virtual void setslack(int slack) {
m_slack = slack;
}
virtual void dump(ostream& o) const;
private:
int m_slack;
};
@ -494,6 +508,8 @@ public:
virtual RefCntr<SearchData> getSub() {
return m_sub;
}
virtual void dump(ostream& o) const;
protected:
RefCntr<SearchData> m_sub;
};