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

propagate wasa nostem modifier

Browse Source
This commit is contained in:
dockes 2007-02-13 10:58:32 +00:00
parent b6d4fd34bd
commit 9084a6516b
3 changed files with 78 additions and 44 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

70
src/query/wasatorcl.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: wasatorcl.cpp,v 1.7 2007-02-07 12:00:17 dockes Exp $ (C) 2006 J.F.Dockes"; static char rcsid[] = "@(#$Id: wasatorcl.cpp,v 1.8 2007-02-13 10:58:31 dockes Exp $ (C) 2006 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
@ -21,6 +21,7 @@ static char rcsid[] = "@(#$Id: wasatorcl.cpp,v 1.7 2007-02-07 12:00:17 dockes Ex
#include "rcldb.h" #include "rcldb.h"
#include "searchdata.h" #include "searchdata.h"
#include "wasatorcl.h" #include "wasatorcl.h"
#include "debuglog.h"
Rcl::SearchData *wasaStringToRcl(const string &qs, string &reason) Rcl::SearchData *wasaStringToRcl(const string &qs, string &reason)
{ {
@ -44,6 +45,8 @@ Rcl::SearchData *wasaQueryToRcl(WasaQuery *wasa)
Rcl::SearchData *sdata = new Rcl::SearchData(Rcl::SCLT_AND); Rcl::SearchData *sdata = new Rcl::SearchData(Rcl::SCLT_AND);
WasaQuery::subqlist_t::iterator it; WasaQuery::subqlist_t::iterator it;
Rcl::SearchDataClause *nclause;
for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) { for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) {
switch ((*it)->m_op) { switch ((*it)->m_op) {
case WasaQuery::OP_NULL: case WasaQuery::OP_NULL:
@ -53,16 +56,23 @@ Rcl::SearchData *wasaQueryToRcl(WasaQuery *wasa)
continue; continue;
case WasaQuery::OP_LEAF: case WasaQuery::OP_LEAF:
if ((*it)->m_value.find_first_of(" \t\n\r") != string::npos) { if ((*it)->m_value.find_first_of(" \t\n\r") != string::npos) {
sdata->addClause nclause = new Rcl::SearchDataClauseDist(Rcl::SCLT_PHRASE,
(new Rcl::SearchDataClauseDist(Rcl::SCLT_PHRASE, (*it)->m_value, 0,
(*it)->m_value, 0, (*it)->m_fieldspec);
(*it)->m_fieldspec));
} else { } else {
sdata->addClause nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_AND,
(new Rcl::SearchDataClauseSimple(Rcl::SCLT_AND, (*it)->m_value,
(*it)->m_value, (*it)->m_fieldspec);
(*it)->m_fieldspec));
} }
if (nclause == 0) {
LOGERR(("wasaQueryToRcl: out of memory\n"));
return 0;
}
if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM) {
fprintf(stderr, "Setting NOSTEM\n");
nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
}
sdata->addClause(nclause);
break; break;
case WasaQuery::OP_EXCL: case WasaQuery::OP_EXCL:
// Note: have to add dquotes which will be translated to // Note: have to add dquotes which will be translated to
@ -70,15 +80,24 @@ Rcl::SearchData *wasaQueryToRcl(WasaQuery *wasa)
// but should work. If there is actually a single // but should work. If there is actually a single
// word, it will not be taken as a phrase, and // word, it will not be taken as a phrase, and
// stem-expansion will work normally // stem-expansion will work normally
sdata->addClause nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL,
(new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL, string("\"") +
string("\"") + (*it)->m_value + "\"",
(*it)->m_value + "\"", (*it)->m_fieldspec);
(*it)->m_fieldspec));
if (nclause == 0) {
LOGERR(("wasaQueryToRcl: out of memory\n"));
return 0;
}
if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
sdata->addClause(nclause);
break; break;
case WasaQuery::OP_OR: case WasaQuery::OP_OR:
// Concatenate all OR values as phrases. Hope there are no // Concatenate all OR values as phrases. Hope there are no
// stray dquotes in there // stray dquotes in there. Note that single terms won't be
// handled as real phrases inside searchdata.cpp (so the
// added dquotes won't interfer with stemming)
{ {
string orvalue; string orvalue;
WasaQuery::subqlist_t::iterator orit; WasaQuery::subqlist_t::iterator orit;
@ -86,20 +105,19 @@ Rcl::SearchData *wasaQueryToRcl(WasaQuery *wasa)
orit != (*it)->m_subs.end(); orit++) { orit != (*it)->m_subs.end(); orit++) {
orvalue += string("\"") + (*orit)->m_value + "\" "; orvalue += string("\"") + (*orit)->m_value + "\" ";
} }
sdata->addClause nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_OR,
(new Rcl::SearchDataClauseSimple(Rcl::SCLT_OR, orvalue,
orvalue, (*it)->m_fieldspec);
(*it)->m_fieldspec)); if (nclause == 0) {
LOGERR(("wasaQueryToRcl: out of memory\n"));
return 0;
}
if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
sdata->addClause(nclause);
} }
} }
} }
// File type and sort specs. We only know about mime types for now.
if (wasa->m_typeKind == WasaQuery::WQTK_MIME) {
for (vector<string>::const_iterator it = wasa->m_types.begin();
it != wasa->m_types.end(); it++) {
sdata->addFiletype(*it);
}
}
return sdata; return sdata;
} }

27
src/rcldb/searchdata.cpp
View File

@ -1,5 +1,5 @@
#ifndef lint #ifndef lint
static char rcsid[] = "@(#$Id: searchdata.cpp,v 1.12 2007-01-29 13:51:08 dockes Exp $ (C) 2006 J.F.Dockes"; static char rcsid[] = "@(#$Id: searchdata.cpp,v 1.13 2007-02-13 10:58:31 dockes Exp $ (C) 2006 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
@ -310,10 +310,10 @@ static void addPrefix(list<string>& terms, const string& prefix)
* count) * count)
*/ */
bool StringToXapianQ::processUserString(const string &iq, bool StringToXapianQ::processUserString(const string &iq,
const string &prefix, const string &prefix,
string &ermsg, string &ermsg,
list<Xapian::Query> &pqueries, list<Xapian::Query> &pqueries,
int slack, bool useNear) int slack, bool useNear)
{ {
LOGDEB(("StringToXapianQ:: query string: [%s]\n", iq.c_str())); LOGDEB(("StringToXapianQ:: query string: [%s]\n", iq.c_str()));
ermsg.erase(); ermsg.erase();
@ -400,8 +400,8 @@ bool StringToXapianQ::processUserString(const string &iq,
// Some version of xapian will accept only one OR clause // Some version of xapian will accept only one OR clause
// inside NEAR, all others must be leafs // inside NEAR, all others must be leafs
bool nostemexp = bool nostemexp =
(op == Xapian::Query::OP_PHRASE || hadmultiple) ? op == Xapian::Query::OP_PHRASE || hadmultiple;
true : false;
string sterm; string sterm;
list<string>exp; list<string>exp;
stripExpandTerm(nostemexp, *it, exp, sterm); stripExpandTerm(nostemexp, *it, exp, sterm);
@ -472,10 +472,15 @@ static string fieldToPrefix(const string& i_field)
return ""; return "";
} }
static const string nullstemlang;
// Translate a simple OR, AND, or EXCL search clause. // Translate a simple OR, AND, or EXCL search clause.
bool SearchDataClauseSimple::toNativeQuery(Rcl::Db &db, void *p, bool SearchDataClauseSimple::toNativeQuery(Rcl::Db &db, void *p,
const string& stemlang) const string& stemlang)
{ {
const string& l_stemlang = (m_modifiers&SDCM_NOSTEMMING)? nullstemlang:
stemlang;
m_terms.clear(); m_terms.clear();
m_groups.clear(); m_groups.clear();
Xapian::Query *qp = (Xapian::Query *)p; Xapian::Query *qp = (Xapian::Query *)p;
@ -502,7 +507,7 @@ bool SearchDataClauseSimple::toNativeQuery(Rcl::Db &db, void *p,
(m_parentSearch && !m_parentSearch->haveWildCards()) || (m_parentSearch && !m_parentSearch->haveWildCards()) ||
(m_parentSearch == 0 && !m_haveWildCards); (m_parentSearch == 0 && !m_haveWildCards);
StringToXapianQ tr(db, stemlang, doBoostUserTerm); StringToXapianQ tr(db, l_stemlang, doBoostUserTerm);
if (!tr.processUserString(m_text, prefix, m_reason, pqueries)) if (!tr.processUserString(m_text, prefix, m_reason, pqueries))
return false; return false;
if (pqueries.empty()) { if (pqueries.empty()) {
@ -516,7 +521,7 @@ bool SearchDataClauseSimple::toNativeQuery(Rcl::Db &db, void *p,
// Translate a FILENAME search clause. // Translate a FILENAME search clause.
bool SearchDataClauseFilename::toNativeQuery(Rcl::Db &db, void *p, bool SearchDataClauseFilename::toNativeQuery(Rcl::Db &db, void *p,
const string& stemlang) const string&)
{ {
Xapian::Query *qp = (Xapian::Query *)p; Xapian::Query *qp = (Xapian::Query *)p;
*qp = Xapian::Query(); *qp = Xapian::Query();
@ -532,6 +537,8 @@ bool SearchDataClauseFilename::toNativeQuery(Rcl::Db &db, void *p,
bool SearchDataClauseDist::toNativeQuery(Rcl::Db &db, void *p, bool SearchDataClauseDist::toNativeQuery(Rcl::Db &db, void *p,
const string& stemlang) const string& stemlang)
{ {
const string& l_stemlang = (m_modifiers&SDCM_NOSTEMMING)? nullstemlang:
stemlang;
LOGDEB(("SearchDataClauseDist::toNativeQuery\n")); LOGDEB(("SearchDataClauseDist::toNativeQuery\n"));
m_terms.clear(); m_terms.clear();
m_groups.clear(); m_groups.clear();
@ -562,7 +569,7 @@ bool SearchDataClauseDist::toNativeQuery(Rcl::Db &db, void *p,
} }
string s = string("\"") + m_text + string("\""); string s = string("\"") + m_text + string("\"");
bool useNear = (m_tp == SCLT_NEAR); bool useNear = (m_tp == SCLT_NEAR);
StringToXapianQ tr(db, stemlang, doBoostUserTerm); StringToXapianQ tr(db, l_stemlang, doBoostUserTerm);
if (!tr.processUserString(s, prefix, m_reason, pqueries, m_slack, useNear)) if (!tr.processUserString(s, prefix, m_reason, pqueries, m_slack, useNear))
return false; return false;
if (pqueries.empty()) { if (pqueries.empty()) {

25
src/rcldb/searchdata.h
View File

@ -16,7 +16,7 @@
*/ */
#ifndef _SEARCHDATA_H_INCLUDED_ #ifndef _SEARCHDATA_H_INCLUDED_
#define _SEARCHDATA_H_INCLUDED_ #define _SEARCHDATA_H_INCLUDED_
/* @(#$Id: searchdata.h,v 1.9 2007-01-25 15:50:54 dockes Exp $ (C) 2004 J.F.Dockes */ /* @(#$Id: searchdata.h,v 1.10 2007-02-13 10:58:32 dockes Exp $ (C) 2004 J.F.Dockes */
/** /**
* Structures to hold data coming almost directly from the gui * Structures to hold data coming almost directly from the gui
@ -121,28 +121,37 @@ private:
class SearchDataClause { class SearchDataClause {
public: public:
SearchDataClause(SClType tp) : m_tp(tp), m_parentSearch(0) {} SearchDataClause(SClType tp)
: m_tp(tp), m_parentSearch(0), m_modifiers(SDCM_NONE)
{}
virtual ~SearchDataClause() {} virtual ~SearchDataClause() {}
virtual bool toNativeQuery(Rcl::Db &db, void *, const string&) = 0; virtual bool toNativeQuery(Rcl::Db &db, void *, const string&) = 0;
virtual bool isFileName() const {return m_tp==SCLT_FILENAME ? true: false;} bool isFileName() const {return m_tp==SCLT_FILENAME ? true: false;}
virtual string getReason() const {return m_reason;} virtual string getReason() const {return m_reason;}
virtual bool getTerms(vector<string>&, vector<vector<string> >&, virtual bool getTerms(vector<string>&, vector<vector<string> >&,
vector<int>&) const vector<int>&) const
{return true;} {return true;}
virtual SClType getTp() {return m_tp;}
virtual void setParent(SearchData *p) {m_parentSearch = p;} SClType getTp() {return m_tp;}
void setParent(SearchData *p) {m_parentSearch = p;}
friend class SearchData; friend class SearchData;
enum Modifier {SDCM_NONE=0, SDCM_NOSTEMMING=1};
virtual void setModifiers(Modifier mod) {m_modifiers = mod;}
protected: protected:
string m_reason; string m_reason;
SClType m_tp; SClType m_tp;
SearchData *m_parentSearch; SearchData *m_parentSearch;
bool m_haveWildCards; bool m_haveWildCards;
Modifier m_modifiers;
}; };
/** /**