yetangitu/recoll
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

arrange to enable passing a match term as an argument when opening a document with the native editor (only / mostly pdf for now)

Browse source
This commit is contained in:
Jean-Francois Dockes 2012-10-02 14:40:09 +02:00
parent 0b65142149
commit bf6bbfbf14
13 changed files with 115 additions and 56 deletions
Download patch file Download diff file Expand all files Collapse all files

8
src/doc/user/usermanual.sgml
View file

@ -4563,6 +4563,14 @@ x-my-tag = mailmytag
snippet.</para></formalpara>
</listitem>
<listitem><formalpara><title>%s</title>
<para>Search term. The value will only be set for documents
with indexed page numbers (ie: PDF). The value will be one of
the matched search terms. It would allow pre-setting the
value in the "Find" entry inside Evince for example, for easy
highlighting of the term.</para></formalpara>
</listitem>
<listitem><formalpara><title>%U, %u</title>
<para>Url.</para></formalpara>
</listitem>

8
src/qtgui/rclmain_w.cpp
View file

@ -1498,7 +1498,7 @@ static bool lookForHtmlBrowser(string &exefile)
return false;
}
void RclMain::startNativeViewer(Rcl::Doc doc, int pagenum)
void RclMain::startNativeViewer(Rcl::Doc doc, int pagenum, QString term)
{
LOGDEB(("RclMain::startNativeViewer: page %d\n", pagenum));
// Look for appropriate viewer
@ -1520,10 +1520,13 @@ void RclMain::startNativeViewer(Rcl::Doc doc, int pagenum)
if (pagenum == -1) {
pagenum = 1;
string lterm;
if (m_source.isNotNull())
pagenum = m_source->getFirstMatchPage(doc);
pagenum = m_source->getFirstMatchPage(doc, lterm);
if (pagenum == -1)
pagenum = 1;
else
term = QString::fromUtf8(lterm.c_str());
}
char cpagenum[20];
sprintf(cpagenum, "%d", pagenum);
@ -1666,6 +1669,7 @@ void RclMain::startNativeViewer(Rcl::Doc doc, int pagenum)
subs["i"] = doc.ipath;
subs["M"] = doc.mimetype;
subs["p"] = cpagenum;
subs["s"] = (const char*)term.toLocal8Bit();
subs["U"] = url;
subs["u"] = url;
// Let %(xx) access all metadata.

3
src/qtgui/rclmain_w.h
View file

@ -119,7 +119,8 @@ public slots:
virtual void docExpand(Rcl::Doc);
virtual void startPreview(int docnum, Rcl::Doc doc, int keymods);
virtual void startPreview(Rcl::Doc);
virtual void startNativeViewer(Rcl::Doc, int pagenum = -1);
virtual void startNativeViewer(Rcl::Doc, int pagenum = -1,
QString term=QString());
virtual void saveDocToFile(Rcl::Doc);
virtual void previewNextInTab(Preview *, int sid, int docnum);
virtual void previewPrevInTab(Preview *, int sid, int docnum);

14
src/qtgui/reslist.cpp
View file

@ -34,6 +34,7 @@
#include <qclipboard.h>
#include <qscrollbar.h>
#include <QTextBlock>
#include <QShortcut>
#ifndef __APPLE__
#include <qx11info_x11.h>
#endif
@ -59,9 +60,8 @@
#include "rclaspell.h"
#endif
#ifndef MIN
#define MIN(A,B) ((A) < (B) ? (A) : (B))
#endif
static const QKeySequence quitKeySeq("Ctrl+q");
static const QKeySequence closeKeySeq("Ctrl+w");
#ifndef RESLIST_TEXTBROWSER
#include <QWebFrame>
@ -805,9 +805,13 @@ void ResList::newSnippetsW(const Rcl::Doc& doc)
{
SnippetsW *sp = new SnippetsW(doc, m_source);
if (m_parent) {
connect(sp, SIGNAL(startNativeViewer(Rcl::Doc, int)),
m_parent, SLOT(startNativeViewer(Rcl::Doc, int)));
connect(sp, SIGNAL(startNativeViewer(Rcl::Doc, int, QString)),
m_parent, SLOT(startNativeViewer(Rcl::Doc, int, QString)));
connect(new QShortcut(quitKeySeq, sp), SIGNAL (activated()),
m_parent, SLOT (fileExit()));
}
connect(new QShortcut(closeKeySeq, sp), SIGNAL (activated()),
sp, SLOT (close()));
sp->show();
}

25
src/qtgui/snippets_w.cpp
View file

@ -61,7 +61,7 @@ void SnippetsW::init()
setWindowTitle(QString::fromUtf8(titleOrFilename.c_str()));
vector<pair<int, string> > vpabs;
vector<Rcl::Snippet> vpabs;
m_source->getAbstract(m_doc, vpabs);
HighlightData hdata;
@ -77,21 +77,23 @@ void SnippetsW::init()
g_hiliter.set_inputhtml(false);
for (vector<pair<int, string> >::const_iterator it = vpabs.begin();
for (vector<Rcl::Snippet>::const_iterator it = vpabs.begin();
it != vpabs.end(); it++) {
html += "<tr><td>";
if (it->first > 0) {
char buf[100];
sprintf(buf, "P.&nbsp;%d", it->first);
if (it->page > 0) {
char txt[100];
sprintf(txt, "P.&nbsp;%d", it->page);
char url[100];
sprintf(url, "P%dT%s", it->page, it->term.c_str());
html += "<a href=\"";
html += buf;
html += url;
html += "\">";
html += buf;
html += txt;
html += "</a>";
}
html += "</td><td>";
list<string> lr;
g_hiliter.plaintorich(it->second, lr, hdata);
g_hiliter.plaintorich(it->snippet, lr, hdata);
html.append(QString::fromUtf8(lr.front().c_str()));
html.append("</td></tr>\n");
}
@ -117,7 +119,12 @@ void SnippetsW::linkWasClicked(const QUrl &url)
if (numpos == string::npos)
return;
int page = atoi(ascurl.c_str() + numpos);
emit startNativeViewer(m_doc, page);
string::size_type termpos = ascurl.find_first_of("T");
string term;
if (termpos != string::npos)
term = ascurl.substr(termpos+1);
emit startNativeViewer(m_doc, page,
QString::fromUtf8(term.c_str()));
return;
}
}

3
src/qtgui/snippets_w.h
View file

@ -16,6 +16,7 @@
*/
#ifndef _SNIPPETS_W_H_INCLUDED_
#define _SNIPPETS_W_H_INCLUDED_
#include <QString>
#include "rcldoc.h"
#include "refcntr.h"
@ -39,7 +40,7 @@ protected slots:
virtual void linkWasClicked(const QUrl &);
signals:
void startNativeViewer(Rcl::Doc, int pagenum);
void startNativeViewer(Rcl::Doc, int pagenum, QString term);
private:
void init();

12
src/query/docseq.h
View file

@ -25,6 +25,9 @@
#include "refcntr.h"
#include "hldata.h"
// Need this for the "Snippet" class def.
#include "rclquery.h"
// A result list entry.
struct ResListEntry {
Rcl::Doc doc;
@ -96,13 +99,12 @@ class DocSequence {
return true;
}
virtual bool getAbstract(Rcl::Doc& doc,
std::vector<std::pair<int, std::string> >& abs)
std::vector<Rcl::Snippet>& abs)
{
abs.push_back(std::pair<int, std::string>(0,
doc.meta[Rcl::Doc::keyabs]));
abs.push_back(Rcl::Snippet(0, doc.meta[Rcl::Doc::keyabs]));
return true;
}
virtual int getFirstMatchPage(Rcl::Doc&)
virtual int getFirstMatchPage(Rcl::Doc&, std::string&)
{
return -1;
}
@ -173,7 +175,7 @@ public:
return m_seq->getAbstract(doc, abs);
}
virtual bool getAbstract(Rcl::Doc& doc,
std::vector<std::pair<int, std::string> >& abs)
std::vector<Rcl::Snippet>& abs)
{
if (m_seq.isNull())
return false;

17
src/query/docseqdb.cpp
View file

@ -64,11 +64,11 @@ int DocSequenceDb::getResCnt()
}
return m_rescnt;
}
static const string cstr_mre("[...]");
// This one only gets called to fill-up the snippets window
// We ignore most abstract/snippets preferences.
bool DocSequenceDb::getAbstract(Rcl::Doc &doc,
vector<pair<int, string> >& vpabs)
bool DocSequenceDb::getAbstract(Rcl::Doc &doc, vector<Rcl::Snippet>& vpabs)
{
LOGDEB(("DocSequenceDb::getAbstract/pair\n"));
setQuery();
@ -77,15 +77,16 @@ bool DocSequenceDb::getAbstract(Rcl::Doc &doc,
int maxoccs = 500;
Rcl::abstract_result ret = Rcl::ABSRES_ERROR;
if (m_q->whatDb()) {
ret = m_q->makeDocAbstract(doc,vpabs, maxoccs,
ret = m_q->makeDocAbstract(doc, vpabs, maxoccs,
m_q->whatDb()->getAbsCtxLen()+ 2);
}
if (vpabs.empty())
vpabs.push_back(pair<int, string>(0, doc.meta[Rcl::Doc::keyabs]));
vpabs.push_back(Rcl::Snippet(0, doc.meta[Rcl::Doc::keyabs]));
// If the list was probably truncated, indicate it.
if (ret == Rcl::ABSRES_TRUNC)
vpabs.push_back(pair<int, string>(-1, "[...]"));
if (ret == Rcl::ABSRES_TRUNC) {
vpabs.push_back(Rcl::Snippet(-1, cstr_mre));
}
return true;
}
@ -102,11 +103,11 @@ bool DocSequenceDb::getAbstract(Rcl::Doc &doc, vector<string>& vabs)
return true;
}
int DocSequenceDb::getFirstMatchPage(Rcl::Doc &doc)
int DocSequenceDb::getFirstMatchPage(Rcl::Doc &doc, string& term)
{
setQuery();
if (m_q->whatDb()) {
return m_q->getFirstMatchPage(doc);
return m_q->getFirstMatchPage(doc, term);
}
return -1;
}

4
src/query/docseqdb.h
View file

@ -34,10 +34,10 @@ class DocSequenceDb : public DocSequence {
// Called to fill-up the snippets window. Ignoers
// buildabstract/replaceabstract and syntabslen
virtual bool getAbstract(Rcl::Doc &doc, vector<pair<int, string> >&);
virtual bool getAbstract(Rcl::Doc &doc, vector<Rcl::Snippet>&);
virtual bool getAbstract(Rcl::Doc &doc, vector<string>&);
virtual int getFirstMatchPage(Rcl::Doc&);
virtual int getFirstMatchPage(Rcl::Doc&, std::string& term);
virtual bool getEnclosing(Rcl::Doc& doc, Rcl::Doc& pdoc);
virtual string getDescription();
virtual list<string> expand(Rcl::Doc &doc);

27
src/rcldb/rclabstract.cpp
View file

@ -19,6 +19,9 @@
#include <math.h>
#include <map>
#include <tr1/unordered_set>
using std::tr1::unordered_set;
using namespace std;
#include "debuglog.h"
@ -204,11 +207,11 @@ double Query::Native::qualityTerms(Xapian::docid docid,
}
// Return page number for first match of "significant" term.
int Query::Native::getFirstMatchPage(Xapian::docid docid)
int Query::Native::getFirstMatchPage(Xapian::docid docid, string& term)
{
if (!m_q|| !m_q->m_db || !m_q->m_db->m_ndb || !m_q->m_db->m_ndb->m_isopen) {
LOGERR(("Query::getFirstMatchPage: no db\n"));
return false;
return -1;
}
Rcl::Db::Native *ndb(m_q->m_db->m_ndb);
Xapian::Database& xrdb(ndb->xrdb);
@ -246,8 +249,10 @@ int Query::Native::getFirstMatchPage(Xapian::docid docid)
for (pos = xrdb.positionlist_begin(docid, qterm);
pos != xrdb.positionlist_end(docid, qterm); pos++) {
int pagenum = ndb->getPageNumberForPosition(pagepos, *pos);
if (pagenum > 0)
if (pagenum > 0) {
term = qterm;
return pagenum;
}
}
} catch (...) {
// Term does not occur. No problem.
@ -263,7 +268,7 @@ int Query::Native::getFirstMatchPage(Xapian::docid docid)
// DatabaseModified and other general exceptions are catched and
// possibly retried by our caller
abstract_result Query::Native::makeAbstract(Xapian::docid docid,
vector<pair<int, string> >& vabs,
vector<Snippet>& vabs,
int imaxoccs, int ictxwords)
{
Chrono chron;
@ -315,6 +320,9 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
// The terms 'array' that we partially populate with the document
// terms, at their positions around the search terms positions:
map<unsigned int, string> sparseDoc;
// Also remember apart the search term positions so that we can list
// them with their snippets.
unordered_set<unsigned int> searchTermPositions;
// Total number of occurences for all terms. We stop when we have too much
unsigned int totaloccs = 0;
@ -392,6 +400,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
for (unsigned int ii = sta; ii <= sto; ii++) {
if (ii == (unsigned int)ipos) {
sparseDoc[ii] = qterm;
searchTermPositions.insert(ii);
} else if (ii > (unsigned int)ipos &&
ii < (unsigned int)ipos + qtrmwrdcnt) {
sparseDoc[ii] = occupiedmarker;
@ -470,7 +479,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
break;
}
map<unsigned int, string>::iterator vit;
if ((vit=sparseDoc.find(*pos)) != sparseDoc.end()) {
if ((vit = sparseDoc.find(*pos)) != sparseDoc.end()) {
// Don't replace a term: the terms list is in
// alphabetic order, and we may have several terms
// at the same position, we want to keep only the
@ -513,6 +522,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
string chunk;
bool incjk = false;
int page = 0;
string term;
for (map<unsigned int, string>::const_iterator it = sparseDoc.begin();
it != sparseDoc.end(); it++) {
LOGDEB2(("Abtract:output %u -> [%s]\n", it->first,it->second.c_str()));
@ -522,6 +532,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
page = ndb->getPageNumberForPosition(vpbreaks, it->first);
if (page < 0)
page = 0;
term.clear();
}
Utf8Iter uit(it->second);
bool newcjk = false;
@ -530,8 +541,10 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
if (!incjk || (incjk && !newcjk))
chunk += " ";
incjk = newcjk;
if (searchTermPositions.find(it->first) != searchTermPositions.end())
term = it->second;
if (it->second == cstr_ellipsis) {
vabs.push_back(pair<int,string>(page, chunk));
vabs.push_back(Snippet(page, chunk).setTerm(term));
chunk.clear();
} else {
if (it->second.compare(end_of_field_term) &&
@ -540,7 +553,7 @@ abstract_result Query::Native::makeAbstract(Xapian::docid docid,
}
}
if (!chunk.empty())
vabs.push_back(pair<int, string>(page, chunk));
vabs.push_back(Snippet(page, chunk).setTerm(term));
LOGDEB2(("makeAbtract: done in %d mS\n", chron.millis()));
return ret;

22
src/rcldb/rclquery.cpp
View file

@ -300,7 +300,7 @@ bool Query::getMatchTerms(unsigned long xdocid, vector<string>& terms)
}
abstract_result Query::makeDocAbstract(Doc &doc,
vector<pair<int, string> >& abstract,
vector<Snippet>& abstract,
int maxoccs, int ctxwords)
{
LOGDEB(("makeDocAbstract: maxoccs %d ctxwords %d\n", maxoccs, ctxwords));
@ -318,19 +318,19 @@ abstract_result Query::makeDocAbstract(Doc &doc,
bool Query::makeDocAbstract(Doc &doc, vector<string>& abstract)
{
vector<pair<int, string> > vpabs;
vector<Snippet> vpabs;
if (!makeDocAbstract(doc, vpabs))
return false;
for (vector<pair<int, string> >::const_iterator it = vpabs.begin();
for (vector<Snippet>::const_iterator it = vpabs.begin();
it != vpabs.end(); it++) {
string chunk;
if (it->first > 0) {
if (it->page > 0) {
doc.haspages = true;
ostringstream ss;
ss << it->first;
ss << it->page;
chunk += string(" [p ") + ss.str() + "] ";
}
chunk += it->second;
chunk += it->snippet;
abstract.push_back(chunk);
}
return true;
@ -338,18 +338,18 @@ bool Query::makeDocAbstract(Doc &doc, vector<string>& abstract)
bool Query::makeDocAbstract(Doc &doc, string& abstract)
{
vector<pair<int, string> > vpabs;
vector<Snippet> vpabs;
if (!makeDocAbstract(doc, vpabs))
return false;
for (vector<pair<int, string> >::const_iterator it = vpabs.begin();
for (vector<Snippet>::const_iterator it = vpabs.begin();
it != vpabs.end(); it++) {
abstract.append(it->second);
abstract.append(it->snippet);
abstract.append(cstr_ellipsis);
}
return m_reason.empty() ? true : false;
}
int Query::getFirstMatchPage(Doc &doc)
int Query::getFirstMatchPage(Doc &doc, string& term)
{
LOGDEB1(("Db::getFirstMatchPages\n"));;
if (!m_nq) {
@ -357,7 +357,7 @@ int Query::getFirstMatchPage(Doc &doc)
return false;
}
int pagenum = -1;
XAPTRY(pagenum = m_nq->getFirstMatchPage(Xapian::docid(doc.xdocid)),
XAPTRY(pagenum = m_nq->getFirstMatchPage(Xapian::docid(doc.xdocid), term),
m_db->m_ndb->xrdb, m_reason);
return m_reason.empty() ? pagenum : -1;
}

24
src/rcldb/rclquery.h
View file

@ -25,12 +25,12 @@ using std::vector;
#endif
#include "refcntr.h"
#include "searchdata.h"
#ifndef NO_NAMESPACES
namespace Rcl {
#endif
class SearchData;
class Db;
class Doc;
@ -40,6 +40,24 @@ enum abstract_result {
ABSRES_TRUNC = 2
};
// Snippet entry for makeDocAbstract
class Snippet {
public:
Snippet(int page, const string& snip)
: page(page), snippet(snip)
{
}
Snippet& setTerm(const string& trm)
{
term = trm;
return *this;
}
int page;
string term;
string snippet;
};
/**
* An Rcl::Query is a question (SearchData) applied to a
* database. Handles access to the results. Somewhat equivalent to a
@ -89,10 +107,10 @@ class Query {
// Returned as a snippets vector
bool makeDocAbstract(Doc &doc, vector<string>& abstract);
// Returned as a vector of pair<page,snippet> page is 0 if unknown
abstract_result makeDocAbstract(Doc &doc, vector<pair<int, string> >& abst,
abstract_result makeDocAbstract(Doc &doc, vector<Snippet>& abst,
int maxoccs= -1, int ctxwords = -1);
/** Retrieve detected page breaks positions */
int getFirstMatchPage(Doc &doc);
int getFirstMatchPage(Doc &doc, std::string& term);
/** Expand query to look for documents like the one passed in */
vector<string> expand(const Doc &doc);

4
src/rcldb/rclquery_p.h
View file

@ -54,9 +54,9 @@ public:
delete xenquire; xenquire = 0;
termfreqs.clear();
}
abstract_result makeAbstract(Xapian::docid id, vector<pair<int, string> >&,
abstract_result makeAbstract(Xapian::docid id, vector<Snippet>&,
int maxoccs = -1, int ctxwords = -1);
int getFirstMatchPage(Xapian::docid docid);
int getFirstMatchPage(Xapian::docid docid, std::string& term);
void setDbWideQTermsFreqs();
double qualityTerms(Xapian::docid docid,
const std::vector<std::string>& terms,