tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects 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 44ab1963f9
commit f025ff33dd
13 changed files with 115 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/rcldb/rclabstract.cpp
+20 -7
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;
src/rcldb/rclquery.cpp
+11 -11
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;
}
src/rcldb/rclquery.h
+21 -3
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);
src/rcldb/rclquery_p.h
+2 -2
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,