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

GUI snippets window: add options for the max list length and for sorting the snippets by page number

Browse Source
This commit is contained in:
Jean-Francois Dockes 2019-06-28 14:20:47 +02:00
parent eb83ebfec6
commit 3ccd2364b9
13 changed files with 290 additions and 224 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
src/qtgui/guiutils.cpp
View File

@ -164,10 +164,6 @@ void rwSettings(bool writing)
} }
} }
// Abstract snippet separator
SETTING_RW(prefs.abssep, "/Recoll/prefs/reslist/abssep", String,"…");
if (!writing && prefs.abssep == "")
prefs.abssep = "…";
SETTING_RW(prefs.reslistdateformat, "/Recoll/prefs/reslist/dateformat", SETTING_RW(prefs.reslistdateformat, "/Recoll/prefs/reslist/dateformat",
String," %Y-%m-%d %H:%M:%S %z"); String," %Y-%m-%d %H:%M:%S %z");
if (!writing && prefs.reslistdateformat == "") if (!writing && prefs.reslistdateformat == "")
@ -224,6 +220,13 @@ void rwSettings(bool writing)
Int, 250); Int, 250);
SETTING_RW(prefs.syntAbsCtx, "/Recoll/prefs/query/syntAbsCtx", SETTING_RW(prefs.syntAbsCtx, "/Recoll/prefs/query/syntAbsCtx",
Int, 4); Int, 4);
// Abstract snippet separator
SETTING_RW(prefs.abssep, "/Recoll/prefs/reslist/abssep", String,"…");
if (!writing && prefs.abssep == "")
prefs.abssep = "…";
SETTING_RW(prefs.snipwMaxLength, "/Recoll/prefs/snipwin/maxlen", Int, 1000);
SETTING_RW(prefs.snipwSortByPage,"/Recoll/prefs/snipwin/bypage", Bool, false);
SETTING_RW(prefs.autoSuffs, "/Recoll/prefs/query/autoSuffs", String, ""); SETTING_RW(prefs.autoSuffs, "/Recoll/prefs/query/autoSuffs", String, "");
SETTING_RW(prefs.autoSuffsEnable, SETTING_RW(prefs.autoSuffsEnable,
"/Recoll/prefs/query/autoSuffsEnable", Bool, false); "/Recoll/prefs/query/autoSuffsEnable", Bool, false);

15
src/qtgui/guiutils.h
View File

@ -57,8 +57,6 @@ class PrefsPack {
QString reslistformat; QString reslistformat;
string creslistformat; string creslistformat;
QString reslistheadertext; QString reslistheadertext;
// Abstract snippet separator
QString abssep;
// Date strftime format // Date strftime format
QString reslistdateformat; QString reslistdateformat;
string creslistdateformat; string creslistdateformat;
@ -86,6 +84,15 @@ class PrefsPack {
// Abstract preferences. Building abstracts can slow result display // Abstract preferences. Building abstracts can slow result display
bool queryBuildAbstract; bool queryBuildAbstract;
bool queryReplaceAbstract; bool queryReplaceAbstract;
// Synthetized abstract length (chars) and word context size (words)
int syntAbsLen;
int syntAbsCtx;
// Abstract snippet separator
QString abssep;
// Snippets window max list size
int snipwMaxLength;
// Snippets window sort by page (dflt: by weight)
bool snipwSortByPage;
bool startWithAdvSearchOpen; bool startWithAdvSearchOpen;
// Try to display html if it exists in the internfile stack. // Try to display html if it exists in the internfile stack.
bool previewHtml; bool previewHtml;
@ -122,10 +129,6 @@ class PrefsPack {
QStringList restableFields; QStringList restableFields;
vector<int> restableColWidths; vector<int> restableColWidths;
// Synthetized abstract length and word context size
int syntAbsLen;
int syntAbsCtx;
// Remembered term match mode // Remembered term match mode
int termMatchType; int termMatchType;

4
src/qtgui/snippets_w.cpp
View File

@ -107,6 +107,7 @@ void SnippetsW::init()
connect(nextPB, SIGNAL(clicked()), this, SLOT(slotEditFindNext())); connect(nextPB, SIGNAL(clicked()), this, SLOT(slotEditFindNext()));
connect(prevPB, SIGNAL(clicked()), this, SLOT(slotEditFindPrevious())); connect(prevPB, SIGNAL(clicked()), this, SLOT(slotEditFindPrevious()));
delete browserw;
#if defined(USING_WEBKIT) #if defined(USING_WEBKIT)
browserw = new QWebView(this); browserw = new QWebView(this);
verticalLayout->insertWidget(0, browserw); verticalLayout->insertWidget(0, browserw);
@ -165,7 +166,8 @@ void SnippetsW::init()
setWindowTitle(title); setWindowTitle(title);
vector<Rcl::Snippet> vpabs; vector<Rcl::Snippet> vpabs;
m_source->getAbstract(m_doc, vpabs); m_source->getAbstract(m_doc, vpabs,
prefs.snipwMaxLength, prefs.snipwSortByPage);
HighlightData hdata; HighlightData hdata;
m_source->getTerms(hdata); m_source->getTerms(hdata);

64
src/qtgui/uiprefs.ui
View File

@ -45,15 +45,15 @@
</item> </item>
<item> <item>
<widget class="QLineEdit" name="qtermStyleLE"> <widget class="QLineEdit" name="qtermStyleLE">
<property name="toolTip">
<string>Query terms highlighting in results. &lt;br&gt;Maybe try something like "color:red;background:yellow" for something more lively than the default blue...</string>
</property>
<property name="minimumSize"> <property name="minimumSize">
<size> <size>
<width>50</width> <width>50</width>
<height>0</height> <height>0</height>
</size> </size>
</property> </property>
<property name="toolTip">
<string>Query terms highlighting in results. &lt;br&gt;Maybe try something like &quot;color:red;background:yellow&quot; for something more lively than the default blue...</string>
</property>
</widget> </widget>
</item> </item>
</layout> </layout>
@ -137,12 +137,12 @@
</item> </item>
<item> <item>
<widget class="QCheckBox" name="previewActiveLinksCB"> <widget class="QCheckBox" name="previewActiveLinksCB">
<property name="text">
<string>Activate links in preview.</string>
</property>
<property name="toolTip"> <property name="toolTip">
<string>Make links inside the preview window clickable, and start an external browser when they are clicked.</string> <string>Make links inside the preview window clickable, and start an external browser when they are clicked.</string>
</property> </property>
<property name="text">
<string>Activate links in preview.</string>
</property>
<property name="checked"> <property name="checked">
<bool>false</bool> <bool>false</bool>
</property> </property>
@ -527,6 +527,52 @@
</item> </item>
</layout> </layout>
</item> </item>
<item>
<layout class="QHBoxLayout">
<item>
<widget class="QLabel" name="snipwMaxLenLBL">
<property name="sizePolicy">
<sizepolicy hsizetype="Preferred" vsizetype="Preferred">
<horstretch>1</horstretch>
<verstretch>0</verstretch>
</sizepolicy>
</property>
<property name="text">
<string>Maximum number of snippets displayed in the snippets window</string>
</property>
<property name="wordWrap">
<bool>false</bool>
</property>
</widget>
</item>
<item>
<widget class="QSpinBox" name="snipwMaxLenSB">
<property name="minimum">
<number>1</number>
</property>
<property name="maximum">
<number>10000000</number>
</property>
<property name="singleStep">
<number>10</number>
</property>
<property name="value">
<number>1000</number>
</property>
</widget>
</item>
</layout>
</item>
<item>
<widget class="QCheckBox" name="snipwByPageCB">
<property name="text">
<string>Sort snippets by page number (default: by weigth).</string>
</property>
<property name="checked">
<bool>false</bool>
</property>
</widget>
</item>
</layout> </layout>
</item> </item>
<item> <item>
@ -903,12 +949,12 @@ May be slow for big documents.</string>
</item> </item>
<item> <item>
<widget class="QPushButton" name="ptransPB"> <widget class="QPushButton" name="ptransPB">
<property name="text">
<string>Paths translations</string>
</property>
<property name="toolTip"> <property name="toolTip">
<string>Set path translations for the selected index or for the main one if no selection exists.</string> <string>Set path translations for the selected index or for the main one if no selection exists.</string>
</property> </property>
<property name="text">
<string>Paths translations</string>
</property>
</widget> </widget>
</item> </item>
</layout> </layout>

5
src/qtgui/uiprefs_w.cpp
View File

@ -184,7 +184,8 @@ void UIPrefsDialog::setFromPrefs()
string nm = path_getsimple((const char *)snipCssFile.toLocal8Bit()); string nm = path_getsimple((const char *)snipCssFile.toLocal8Bit());
snipCssPB->setText(QString::fromLocal8Bit(nm.c_str())); snipCssPB->setText(QString::fromLocal8Bit(nm.c_str()));
} }
snipwMaxLenSB->setValue(prefs.snipwMaxLength);
snipwByPageCB->setChecked(prefs.snipwSortByPage);
paraFormat = prefs.reslistformat; paraFormat = prefs.reslistformat;
headerText = prefs.reslistheadertext; headerText = prefs.reslistheadertext;
@ -301,6 +302,8 @@ void UIPrefsDialog::accept()
prefs.reslistformat = prefs.dfltResListFormat; prefs.reslistformat = prefs.dfltResListFormat;
paraFormat = prefs.reslistformat; paraFormat = prefs.reslistformat;
} }
prefs.snipwMaxLength = snipwMaxLenSB->value();
prefs.snipwSortByPage = snipwByPageCB->isChecked();
prefs.creslistformat = (const char*)prefs.reslistformat.toUtf8(); prefs.creslistformat = (const char*)prefs.reslistformat.toUtf8();

68
src/query/docseq.h
View File

@ -39,7 +39,7 @@ struct ResListEntry {
/** Sort specification. */ /** Sort specification. */
class DocSeqSortSpec { class DocSeqSortSpec {
public: public:
DocSeqSortSpec() : desc(false) {} DocSeqSortSpec() : desc(false) {}
bool isNotNull() const {return !field.empty();} bool isNotNull() const {return !field.empty();}
void reset() {field.erase();} void reset() {field.erase();}
@ -50,7 +50,7 @@ class DocSeqSortSpec {
/** Filtering spec. This is only used to filter by doc category for now, hence /** Filtering spec. This is only used to filter by doc category for now, hence
the rather specialized interface */ the rather specialized interface */
class DocSeqFiltSpec { class DocSeqFiltSpec {
public: public:
DocSeqFiltSpec() {} DocSeqFiltSpec() {}
enum Crit {DSFS_MIMETYPE, DSFS_QLANG, DSFS_PASSALL}; enum Crit {DSFS_MIMETYPE, DSFS_QLANG, DSFS_PASSALL};
void orCrit(Crit crit, const std::string& value) { void orCrit(Crit crit, const std::string& value) {
@ -75,7 +75,7 @@ class DocSeqFiltSpec {
the current one will have to do for now. the current one will have to do for now.
*/ */
class DocSequence { class DocSequence {
public: public:
DocSequence(const std::string &t) : m_title(t) {} DocSequence(const std::string &t) : m_title(t) {}
virtual ~DocSequence() {} virtual ~DocSequence() {}
@ -101,19 +101,16 @@ class DocSequence {
abs.push_back(doc.meta[Rcl::Doc::keyabs]); abs.push_back(doc.meta[Rcl::Doc::keyabs]);
return true; return true;
} }
virtual bool getAbstract(Rcl::Doc& doc, virtual bool getAbstract(Rcl::Doc& doc, std::vector<Rcl::Snippet>& abs,
std::vector<Rcl::Snippet>& abs) int, bool) {
{
abs.push_back(Rcl::Snippet(0, doc.meta[Rcl::Doc::keyabs])); abs.push_back(Rcl::Snippet(0, doc.meta[Rcl::Doc::keyabs]));
return true; return true;
} }
virtual int getFirstMatchPage(Rcl::Doc&, std::string&) virtual int getFirstMatchPage(Rcl::Doc&, std::string&) {
{
return -1; return -1;
} }
/** Get duplicates. */ /** Get duplicates. */
virtual bool docDups(const Rcl::Doc&, std::vector<Rcl::Doc>&) virtual bool docDups(const Rcl::Doc&, std::vector<Rcl::Doc>&) {
{
return false; return false;
} }
@ -155,10 +152,11 @@ class DocSequence {
virtual bool canSort() {return false;} virtual bool canSort() {return false;}
virtual bool setFiltSpec(const DocSeqFiltSpec &) {return false;} virtual bool setFiltSpec(const DocSeqFiltSpec &) {return false;}
virtual bool setSortSpec(const DocSeqSortSpec &) {return false;} virtual bool setSortSpec(const DocSeqSortSpec &) {return false;}
virtual std::shared_ptr<DocSequence> getSourceSeq() {return std::shared_ptr<DocSequence>();} virtual std::shared_ptr<DocSequence> getSourceSeq() {
return std::shared_ptr<DocSequence>();}
static void set_translations(const std::string& sort, const std::string& filt) static void set_translations(const std::string& sort,
{ const std::string& filt) {
o_sort_trans = sort; o_sort_trans = sort;
o_filt_trans = filt; o_filt_trans = filt;
} }
@ -172,7 +170,7 @@ protected:
static std::string o_filt_trans; static std::string o_filt_trans;
std::string m_reason; std::string m_reason;
private: private:
std::string m_title; std::string m_title;
}; };
@ -186,63 +184,53 @@ public:
{} {}
virtual ~DocSeqModifier() {} virtual ~DocSeqModifier() {}
virtual bool getAbstract(Rcl::Doc& doc, std::vector<std::string>& abs) virtual bool getAbstract(Rcl::Doc& doc, std::vector<std::string>& abs) {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->getAbstract(doc, abs); return m_seq->getAbstract(doc, abs);
} }
virtual bool getAbstract(Rcl::Doc& doc, virtual bool getAbstract(Rcl::Doc& doc, std::vector<Rcl::Snippet>& abs,
std::vector<Rcl::Snippet>& abs) int maxlen, bool bypage) override {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->getAbstract(doc, abs); return m_seq->getAbstract(doc, abs, maxlen, bypage);
} }
/** Get duplicates. */ /** Get duplicates. */
virtual bool docDups(const Rcl::Doc& doc, std::vector<Rcl::Doc>& dups) virtual bool docDups(const Rcl::Doc& doc, std::vector<Rcl::Doc>& dups) {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->docDups(doc, dups); return m_seq->docDups(doc, dups);
} }
virtual bool snippetsCapable() virtual bool snippetsCapable() {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->snippetsCapable(); return m_seq->snippetsCapable();
} }
virtual std::string getDescription() virtual std::string getDescription() {
{
if (!m_seq) if (!m_seq)
return ""; return "";
return m_seq->getDescription(); return m_seq->getDescription();
} }
virtual void getTerms(HighlightData& hld) virtual void getTerms(HighlightData& hld) {
{
if (!m_seq) if (!m_seq)
return; return;
m_seq->getTerms(hld); m_seq->getTerms(hld);
} }
virtual bool getEnclosing(Rcl::Doc& doc, Rcl::Doc& pdoc) virtual bool getEnclosing(Rcl::Doc& doc, Rcl::Doc& pdoc) {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->getEnclosing(doc, pdoc); return m_seq->getEnclosing(doc, pdoc);
} }
virtual std::string getReason() virtual std::string getReason() {
{
if (!m_seq) if (!m_seq)
return string(); return string();
return m_seq->getReason(); return m_seq->getReason();
} }
virtual std::string title() virtual std::string title() {
{
return m_seq->title(); return m_seq->title();
} }
virtual std::shared_ptr<DocSequence> getSourceSeq() virtual std::shared_ptr<DocSequence> getSourceSeq() {
{
return m_seq; return m_seq;
} }
@ -269,14 +257,12 @@ public:
virtual bool canSort() {return true;} virtual bool canSort() {return true;}
virtual bool setFiltSpec(const DocSeqFiltSpec &); virtual bool setFiltSpec(const DocSeqFiltSpec &);
virtual bool setSortSpec(const DocSeqSortSpec &); virtual bool setSortSpec(const DocSeqSortSpec &);
virtual bool getDoc(int num, Rcl::Doc &doc, std::string *sh = 0) virtual bool getDoc(int num, Rcl::Doc &doc, std::string *sh = 0) {
{
if (!m_seq) if (!m_seq)
return false; return false;
return m_seq->getDoc(num, doc, sh); return m_seq->getDoc(num, doc, sh);
} }
virtual int getResCnt() virtual int getResCnt() {
{
if (!m_seq) if (!m_seq)
return 0; return 0;
return m_seq->getResCnt(); return m_seq->getResCnt();
@ -290,4 +276,4 @@ private:
DocSeqSortSpec m_sspec; DocSeqSortSpec m_sspec;
}; };
#endif /* _DOCSEQ_H_INCLUDED_ */ #endif /* _DOCSEQ_H_ */

21
src/query/docseqdb.cpp
View File

@ -76,21 +76,22 @@ static const string cstr_mre("[...]");
// This one only gets called to fill-up the snippets window // This one only gets called to fill-up the snippets window
// We ignore most abstract/snippets preferences. // We ignore most abstract/snippets preferences.
bool DocSequenceDb::getAbstract(Rcl::Doc &doc, vector<Rcl::Snippet>& vpabs) bool DocSequenceDb::getAbstract(Rcl::Doc &doc, vector<Rcl::Snippet>& vpabs,
int maxlen, bool sortbypage)
{ {
LOGDEB("DocSequenceDb::getAbstract/pair\n" ); LOGDEB("DocSequenceDb::getAbstract/pair\n");
std::unique_lock<std::mutex> locker(o_dblock); std::unique_lock<std::mutex> locker(o_dblock);
if (!setQuery()) if (!setQuery())
return false; return false;
// Have to put the limit somewhere. // Have to put the limit somewhere.
int maxoccs = 1000;
int ret = Rcl::ABSRES_ERROR; int ret = Rcl::ABSRES_ERROR;
if (m_q->whatDb()) { if (m_q->whatDb()) {
ret = m_q->makeDocAbstract(doc, vpabs, maxoccs, ret = m_q->makeDocAbstract(
m_q->whatDb()->getAbsCtxLen()+ 2); doc, vpabs, maxlen, m_q->whatDb()->getAbsCtxLen() + 2, sortbypage);
} }
LOGDEB("DocSequenceDb::getAbstract: got ret " << (ret) << " vpabs len " << (vpabs.size()) << "\n" ); LOGDEB("DocSequenceDb::getAbstract: got ret " << ret << " vpabs len " <<
vpabs.size() << "\n");
if (vpabs.empty()) { if (vpabs.empty()) {
return true; return true;
} }
@ -156,7 +157,7 @@ string DocSequenceDb::title()
bool DocSequenceDb::setFiltSpec(const DocSeqFiltSpec &fs) bool DocSequenceDb::setFiltSpec(const DocSeqFiltSpec &fs)
{ {
LOGDEB("DocSequenceDb::setFiltSpec\n" ); LOGDEB("DocSequenceDb::setFiltSpec\n");
std::unique_lock<std::mutex> locker(o_dblock); std::unique_lock<std::mutex> locker(o_dblock);
if (fs.isNotNull()) { if (fs.isNotNull()) {
// We build a search spec by adding a filtering layer to the base one. // We build a search spec by adding a filtering layer to the base one.
@ -204,7 +205,8 @@ bool DocSequenceDb::setFiltSpec(const DocSeqFiltSpec &fs)
bool DocSequenceDb::setSortSpec(const DocSeqSortSpec &spec) bool DocSequenceDb::setSortSpec(const DocSeqSortSpec &spec)
{ {
LOGDEB("DocSequenceDb::setSortSpec: fld [" << (spec.field) << "] " << (spec.desc ? "desc" : "asc") << "\n" ); LOGDEB("DocSequenceDb::setSortSpec: fld [" << spec.field << "] " <<
(spec.desc ? "desc" : "asc") << "\n");
std::unique_lock<std::mutex> locker(o_dblock); std::unique_lock<std::mutex> locker(o_dblock);
if (spec.isNotNull()) { if (spec.isNotNull()) {
m_q->setSortBy(spec.field, !spec.desc); m_q->setSortBy(spec.field, !spec.desc);
@ -227,7 +229,8 @@ bool DocSequenceDb::setQuery()
m_lastSQStatus = m_q->setQuery(m_fsdata); m_lastSQStatus = m_q->setQuery(m_fsdata);
if (!m_lastSQStatus) { if (!m_lastSQStatus) {
m_reason = m_q->getReason(); m_reason = m_q->getReason();
LOGERR("DocSequenceDb::setQuery: rclquery::setQuery failed: " << (m_reason) << "\n" ); LOGERR("DocSequenceDb::setQuery: rclquery::setQuery failed: " <<
m_reason << "\n");
} }
return m_lastSQStatus; return m_lastSQStatus;
} }

16
src/query/docseqdb.h
View File

@ -16,9 +16,10 @@
*/ */
#ifndef _DOCSEQDB_H_INCLUDED_ #ifndef _DOCSEQDB_H_INCLUDED_
#define _DOCSEQDB_H_INCLUDED_ #define _DOCSEQDB_H_INCLUDED_
#include "docseq.h"
#include <memory> #include <memory>
#include "docseq.h"
#include "searchdata.h" #include "searchdata.h"
#include "rclquery.h" #include "rclquery.h"
@ -26,21 +27,22 @@
class DocSequenceDb : public DocSequence { class DocSequenceDb : public DocSequence {
public: public:
DocSequenceDb(std::shared_ptr<Rcl::Db> db, DocSequenceDb(std::shared_ptr<Rcl::Db> db,
std::shared_ptr<Rcl::Query> q, const string &t, std::shared_ptr<Rcl::Query> q, const std::string &t,
std::shared_ptr<Rcl::SearchData> sdata); std::shared_ptr<Rcl::SearchData> sdata);
virtual ~DocSequenceDb() {} virtual ~DocSequenceDb() {}
virtual bool getDoc(int num, Rcl::Doc &doc, string * = 0); virtual bool getDoc(int num, Rcl::Doc &doc, std::string * = 0);
virtual int getResCnt(); virtual int getResCnt();
virtual void getTerms(HighlightData& hld); virtual void getTerms(HighlightData& hld);
// Called to fill-up the snippets window. Ignoers // Called to fill-up the snippets window. Ignoers
// buildabstract/replaceabstract and syntabslen // buildabstract/replaceabstract and syntabslen
virtual bool getAbstract(Rcl::Doc &doc, vector<Rcl::Snippet>&); virtual bool getAbstract(Rcl::Doc &doc, std::vector<Rcl::Snippet>&,
int maxlen, bool sortbypage) override;
virtual bool getAbstract(Rcl::Doc &doc, vector<string>&); virtual bool getAbstract(Rcl::Doc &doc, std::vector<std::string>&);
virtual int getFirstMatchPage(Rcl::Doc&, std::string& term); virtual int getFirstMatchPage(Rcl::Doc&, std::string& term);
virtual bool docDups(const Rcl::Doc& doc, std::vector<Rcl::Doc>& dups); virtual bool docDups(const Rcl::Doc& doc, std::vector<Rcl::Doc>& dups);
virtual string getDescription(); virtual std::string getDescription();
virtual std::list<std::string> expand(Rcl::Doc &doc); virtual std::list<std::string> expand(Rcl::Doc &doc);
virtual bool canFilter() {return true;} virtual bool canFilter() {return true;}
virtual bool setFiltSpec(const DocSeqFiltSpec &filtspec); virtual bool setFiltSpec(const DocSeqFiltSpec &filtspec);
@ -56,7 +58,7 @@ class DocSequenceDb : public DocSequence {
{ {
return true; return true;
} }
virtual string title(); virtual std::string title();
protected: protected:
virtual std::shared_ptr<Rcl::Db> getDb() { virtual std::shared_ptr<Rcl::Db> getDb() {

22
src/rcldb/rclabsfromtext.cpp
View File

@ -138,12 +138,14 @@ public:
if (maxtermcount && termcount++ > maxtermcount) { if (maxtermcount && termcount++ > maxtermcount) {
LOGINF("Rclabsfromtext: stopping because maxtermcount reached: "<< LOGINF("Rclabsfromtext: stopping because maxtermcount reached: "<<
maxtermcount << endl); maxtermcount << endl);
retflags |= ABSRES_TRUNC;
return false; return false;
} }
// Also limit the number of fragments (just in case safety) // Also limit the number of fragments (just in case safety)
if (m_fragments.size() > maxtermcount / 100) { if (m_fragments.size() > maxtermcount / 100) {
LOGINF("Rclabsfromtext: stopping because maxfragments reached: "<< LOGINF("Rclabsfromtext: stopping because maxfragments reached: "<<
maxtermcount/100 << endl); maxtermcount/100 << endl);
retflags |= ABSRES_TRUNC;
return false; return false;
} }
// Remember recent past // Remember recent past
@ -327,6 +329,10 @@ public:
return; return;
} }
int getretflags() {
return retflags;
}
private: private:
// Past terms because we need to go back for context before a hit // Past terms because we need to go back for context before a hit
deque<pair<int,int>> m_prevterms; deque<pair<int,int>> m_prevterms;
@ -364,6 +370,7 @@ private:
unsigned int termcount{0}; unsigned int termcount{0};
unsigned int maxtermcount{0}; unsigned int maxtermcount{0};
int retflags{0};
}; };
int Query::Native::abstractFromText( int Query::Native::abstractFromText(
@ -375,7 +382,8 @@ int Query::Native::abstractFromText(
int ctxwords, int ctxwords,
unsigned int maxtotaloccs, unsigned int maxtotaloccs,
vector<Snippet>& vabs, vector<Snippet>& vabs,
Chrono& chron Chrono& chron,
bool sortbypage
) )
{ {
(void)chron; (void)chron;
@ -423,13 +431,21 @@ int Query::Native::abstractFromText(
// Sort the fragments by decreasing weight // Sort the fragments by decreasing weight
const vector<MatchFragment>& res1 = splitter.getFragments(); const vector<MatchFragment>& res1 = splitter.getFragments();
vector<MatchFragment> result(res1.begin(), res1.end()); vector<MatchFragment> result(res1.begin(), res1.end());
if (sortbypage) {
std::sort(result.begin(), result.end(),
[](const MatchFragment& a,
const MatchFragment& b) -> bool {
return a.hitpos < b.hitpos;
}
);
} else {
std::sort(result.begin(), result.end(), std::sort(result.begin(), result.end(),
[](const MatchFragment& a, [](const MatchFragment& a,
const MatchFragment& b) -> bool { const MatchFragment& b) -> bool {
return a.coef > b.coef; return a.coef > b.coef;
} }
); );
}
vector<int> vpbreaks; vector<int> vpbreaks;
ndb->getPagePositions(docid, vpbreaks); ndb->getPagePositions(docid, vpbreaks);
@ -464,7 +480,7 @@ int Query::Native::abstractFromText(
if (count++ >= maxtotaloccs) if (count++ >= maxtotaloccs)
break; break;
} }
return ABSRES_OK; return ABSRES_OK | splitter.getretflags();
} }
} }

9
src/rcldb/rclabstract.cpp
View File

@ -623,11 +623,12 @@ int Query::Native::abstractFromIndex(
// @param[out] vabs the abstract is returned as a vector of snippets. // @param[out] vabs the abstract is returned as a vector of snippets.
int Query::Native::makeAbstract(Xapian::docid docid, int Query::Native::makeAbstract(Xapian::docid docid,
vector<Snippet>& vabs, vector<Snippet>& vabs,
int imaxoccs, int ictxwords) int imaxoccs, int ictxwords, bool sortbypage)
{ {
chron.restart(); chron.restart();
LOGABS("makeAbstract: docid " << docid << " imaxoccs " << LOGDEB("makeAbstract: docid " << docid << " imaxoccs " <<
imaxoccs << " ictxwords " << ictxwords << "\n"); imaxoccs << " ictxwords " << ictxwords << " sort by page " <<
sortbypage << "\n");
// The (unprefixed) terms matched by this document // The (unprefixed) terms matched by this document
vector<string> matchedTerms; vector<string> matchedTerms;
@ -675,7 +676,7 @@ int Query::Native::makeAbstract(Xapian::docid docid,
if (ndb->m_storetext) { if (ndb->m_storetext) {
return abstractFromText(ndb, docid, matchedTerms, byQ, return abstractFromText(ndb, docid, matchedTerms, byQ,
totalweight, ctxwords, maxtotaloccs, vabs, totalweight, ctxwords, maxtotaloccs, vabs,
chron); chron, sortbypage);
} else { } else {
return abstractFromIndex(ndb, docid, matchedTerms, byQ, return abstractFromIndex(ndb, docid, matchedTerms, byQ,
totalweight, ctxwords, maxtotaloccs, vabs, totalweight, ctxwords, maxtotaloccs, vabs,

4
src/rcldb/rclquery.cpp
View File

@ -267,7 +267,7 @@ bool Query::getQueryTerms(vector<string>& terms)
} }
int Query::makeDocAbstract(const Doc &doc, vector<Snippet>& abstract, int Query::makeDocAbstract(const Doc &doc, vector<Snippet>& abstract,
int maxoccs, int ctxwords) int maxoccs, int ctxwords, bool sortbypage)
{ {
LOGDEB("makeDocAbstract: maxoccs " << maxoccs << " ctxwords " << LOGDEB("makeDocAbstract: maxoccs " << maxoccs << " ctxwords " <<
ctxwords << "\n"); ctxwords << "\n");
@ -276,7 +276,7 @@ int Query::makeDocAbstract(const Doc &doc, vector<Snippet>& abstract,
return ABSRES_ERROR; return ABSRES_ERROR;
} }
int ret = ABSRES_ERROR; int ret = ABSRES_ERROR;
XAPTRY(ret = m_nq->makeAbstract(doc.xdocid, abstract, maxoccs, ctxwords), XAPTRY(ret = m_nq->makeAbstract(doc.xdocid, abstract, maxoccs, ctxwords, sortbypage),
m_db->m_ndb->xrdb, m_reason); m_db->m_ndb->xrdb, m_reason);
if (!m_reason.empty()) { if (!m_reason.empty()) {
LOGDEB("makeDocAbstract: makeAbstract: reason: " << m_reason << "\n"); LOGDEB("makeDocAbstract: makeAbstract: reason: " << m_reason << "\n");

2
src/rcldb/rclquery.h
View File

@ -107,7 +107,7 @@ public:
bool makeDocAbstract(const Doc &doc, std::vector<std::string>& abstract); bool makeDocAbstract(const Doc &doc, std::vector<std::string>& abstract);
// Returned as a vector of pair<page,snippet> page is 0 if unknown // Returned as a vector of pair<page,snippet> page is 0 if unknown
int makeDocAbstract(const Doc &doc, std::vector<Snippet>& abst, int makeDocAbstract(const Doc &doc, std::vector<Snippet>& abst,
int maxoccs= -1, int ctxwords = -1); int maxoccs= -1, int ctxwords = -1, bool sortbypage=false);
/** Retrieve page number for first match for "significant" query term /** Retrieve page number for first match for "significant" query term
* @param term returns the chosen term */ * @param term returns the chosen term */
int getFirstMatchPage(const Doc &doc, std::string& term); int getFirstMatchPage(const Doc &doc, std::string& term);

5
src/rcldb/rclquery_p.h
View File

@ -55,7 +55,7 @@ public:
/** Return a list of terms which matched for a specific result document */ /** Return a list of terms which matched for a specific result document */
bool getMatchTerms(unsigned long xdocid, std::vector<std::string>& terms); bool getMatchTerms(unsigned long xdocid, std::vector<std::string>& terms);
int makeAbstract(Xapian::docid id, std::vector<Snippet>&, int makeAbstract(Xapian::docid id, std::vector<Snippet>&,
int maxoccs = -1, int ctxwords = -1); int maxoccs, int ctxwords, bool sortbypage);
int getFirstMatchPage(Xapian::docid docid, std::string& term); int getFirstMatchPage(Xapian::docid docid, std::string& term);
void setDbWideQTermsFreqs(); void setDbWideQTermsFreqs();
double qualityTerms(Xapian::docid docid, double qualityTerms(Xapian::docid docid,
@ -109,7 +109,8 @@ public:
int ctxwords, int ctxwords,
unsigned int maxtotaloccs, unsigned int maxtotaloccs,
vector<Snippet>& vabs, vector<Snippet>& vabs,
Chrono& chron Chrono& chron,
bool sortbypage
); );
}; };