[Opentrep-svn] SF.net SVN: opentrep:[141] trunk/opentrep/opentrep/bom
Status: Beta
Brought to you by:
denis_arnaud
Menu
▾
▴
[Opentrep-svn] SF.net SVN: opentrep:[141] trunk/opentrep/opentrep/bom
|
From: <den...@us...> - 2009-07-19 10:47:09
|
Revision: 141
http://opentrep.svn.sourceforge.net/opentrep/?rev=141&view=rev
Author: denis_arnaud
Date: 2009-07-19 10:47:08 +0000 (Sun, 19 Jul 2009)
Log Message:
-----------
[Dev] Improved the notification scheme when there are several matches with the same percentage.
Modified Paths:
--------------
trunk/opentrep/opentrep/bom/Document.cpp
trunk/opentrep/opentrep/bom/Document.hpp
trunk/opentrep/opentrep/bom/ResultHolder.cpp
trunk/opentrep/opentrep/bom/StringMatcher.cpp
Modified: trunk/opentrep/opentrep/bom/Document.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/Document.cpp 2009-07-18 23:30:37 UTC (rev 140)
+++ trunk/opentrep/opentrep/bom/Document.cpp 2009-07-19 10:47:08 UTC (rev 141)
@@ -5,6 +5,7 @@
#include <cassert>
// OpenTREP
#include <opentrep/bom/Document.hpp>
+#include <opentrep/service/Logger.hpp>
namespace OPENTREP {
@@ -22,20 +23,20 @@
// //////////////////////////////////////////////////////////////////////
std::string Document::describeKey() const {
std::ostringstream oStr;
- oStr << describeShortKey();
+ oStr << "`" << describeShortKey() << "´";
if (_correctedQueryString.empty() == false) {
- oStr << " (corrected into " << _correctedQueryString << ")";
+ oStr << " (corrected into `" << _correctedQueryString << "´)";
}
return oStr.str();
}
// //////////////////////////////////////////////////////////////////////
- std::string Document::toString() const {
+ std::string Document::toShortString() const {
std::ostringstream oStr;
- oStr << describeKey() << std::endl;
+ oStr << describeShortKey();
const Xapian::docid& lDocID = _document.get_docid();
- oStr << "Document ID " << lDocID << "\t" << _percentage
+ oStr << " => Document ID " << lDocID << " matching at " << _percentage
<< "% [" << _document.get_data() << "]";
if (_documentList.empty() == false) {
@@ -62,6 +63,38 @@
}
// //////////////////////////////////////////////////////////////////////
+ std::string Document::toString() const {
+ std::ostringstream oStr;
+ oStr << describeKey();
+
+ const Xapian::docid& lDocID = _document.get_docid();
+ oStr << " => Document ID " << lDocID << " matching at " << _percentage
+ << "% [" << _document.get_data() << "]";
+
+ if (_documentList.empty() == false) {
+ oStr << " along with " << _documentList.size()
+ << " other matching document(s) { ";
+
+ unsigned short idx = 0;
+ for (XapianDocumentList_T::const_iterator itDoc = _documentList.begin();
+ itDoc != _documentList.end(); ++itDoc, ++idx) {
+ const Xapian::Document& lXapianDoc = *itDoc;
+ const Xapian::docid& lDocID = lXapianDoc.get_docid();
+ if (idx != 0) {
+ oStr << ", ";
+ }
+ oStr << "Doc ID " << lDocID << " [" << lXapianDoc.get_data() << "]";
+ }
+ oStr << " }." << std::endl;
+
+ } else {
+ oStr << std::endl;
+ }
+
+ return oStr.str();
+ }
+
+ // //////////////////////////////////////////////////////////////////////
void Document::toStream (std::ostream& ioOut) const {
ioOut << toString();
}
@@ -70,4 +103,22 @@
void Document::fromStream (std::istream& ioIn) {
}
+ // //////////////////////////////////////////////////////////////////////
+ NbOfMatches_T Document::notifyIfExtraMatch () const {
+ NbOfMatches_T oNbOfMatches = _documentList.size();
+
+ // DEBUG
+ if (oNbOfMatches != 0) {
+ OPENTREP_LOG_NOTIFICATION ("NOTE: the following document gets several "
+ << "matches with the same matching percentage."
+ << " You may want to alter the SQL database "
+ << "and re-index the Xapian database, so as "
+ << "to allow a more specific match:"
+ << std::endl << toString());
+ }
+
+ // Return the total number of matches (main plus extras)
+ return (1 + oNbOfMatches);
+ }
+
}
Modified: trunk/opentrep/opentrep/bom/Document.hpp
===================================================================
--- trunk/opentrep/opentrep/bom/Document.hpp 2009-07-18 23:30:37 UTC (rev 140)
+++ trunk/opentrep/opentrep/bom/Document.hpp 2009-07-19 10:47:08 UTC (rev 141)
@@ -80,20 +80,35 @@
_documentList.push_back (iMatchingDocument);
}
+
+ public:
+ // /////////// Business methods /////////
+ /** Retrieve the number of extra matches for the given query string,
+ and print a notification in the logs, so that the SQL and Xapian
+ databases can be altered in order to get more specific matches.
+ @param const TravelQuery_T& Query string having yielded that matching
+ document.
+ @return NbOfMatches_T The number of matches (i.e., one chosen match
+ added to the extra matches). */
+ NbOfMatches_T notifyIfExtraMatch() const;
+
public:
// /////////// Display support methods /////////
- /** Dump a Business Object into an output stream.
+ /** Dump the structure into an output stream.
@param ostream& the output stream. */
void toStream (std::ostream& ioOut) const;
- /** Read a Business Object from an input stream.
+ /** Read a structure from an input stream.
@param istream& the input stream. */
void fromStream (std::istream& ioIn);
- /** Get the serialised version of the Business Object. */
+ /** Get the serialised version of the structure. */
std::string toString() const;
+ /** Get a shorter serialised version of the structure. */
+ std::string toShortString() const;
+
/** Get a string describing the whole key (differentiating two objects
at any level). */
std::string describeKey() const;
Modified: trunk/opentrep/opentrep/bom/ResultHolder.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/ResultHolder.cpp 2009-07-18 23:30:37 UTC (rev 140)
+++ trunk/opentrep/opentrep/bom/ResultHolder.cpp 2009-07-19 10:47:08 UTC (rev 141)
@@ -156,9 +156,8 @@
ioDocumentList.push_back (lMatchingDocument);
// DEBUG
- const XapianDocumentList_T& lXapianDocumentList =
- lMatchingDocument.getExtraDocumentList();
- const NbOfMatches_T lNbOfMatches = 1 + lXapianDocumentList.size();
+ const NbOfMatches_T lNbOfMatches =
+ lMatchingDocument.notifyIfExtraMatch();
OPENTREP_LOG_DEBUG ("==> " << lNbOfMatches
<< " matches for the query string: `"
<< lMatchedString << "´ (from `"
Modified: trunk/opentrep/opentrep/bom/StringMatcher.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/StringMatcher.cpp 2009-07-18 23:30:37 UTC (rev 140)
+++ trunk/opentrep/opentrep/bom/StringMatcher.cpp 2009-07-19 10:47:08 UTC (rev 141)
@@ -295,21 +295,41 @@
*/
Xapian::MSetIterator itDoc = iMatchingSet.begin();
assert (itDoc != iMatchingSet.end());
-
+
+ // Store the percentage
const Xapian::percent& lBestPercentage = itDoc.get_percent();
ioMatchingDocument.setXapianPercentage (lBestPercentage);
- ioMatchingDocument.setXapianDocument (itDoc.get_document());
+ // Store the (Xapian) document itself
+ const Xapian::Document& lBestDocument = itDoc.get_document();
+ ioMatchingDocument.setXapianDocument (lBestDocument);
+
+ // Go on in the list of matches, if any
+ ++itDoc;
+
+ /*
+ if (itDoc != iMatchingSet.end()) {
+ // DEBUG
+ OPENTREP_LOG_DEBUG ("There are extra matches for Document ID "
+ << lBestDocument.get_docid()
+ << ", itself matching at " << lBestPercentage
+ << "%.");
+ }
+ */
+
/** Add all the Xapian documents having reached the same matching
percentage. */
NbOfMatches_T idx = 1;
- for (++itDoc ; itDoc != iMatchingSet.end(); ++itDoc, ++idx) {
+ for ( ; itDoc != iMatchingSet.end(); ++itDoc, ++idx) {
const Xapian::percent& lPercentage = itDoc.get_percent();
+ const Xapian::Document& lDocument = itDoc.get_document();
// DEBUG
- OPENTREP_LOG_DEBUG ("[" << idx << "] extra. Best " << lBestPercentage
- << "%, other " << lPercentage << "% ("
- << itDoc.get_document().get_docid() << ")");
+ /*
+ OPENTREP_LOG_DEBUG ("Extra #" << idx << ": Doc ID "
+ << lDocument.get_docid() << " matching at "
+ << lPercentage << "%.");
+ */
if (lPercentage == lBestPercentage) {
ioMatchingDocument.addExtraDocument (itDoc.get_document());
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
