[Opentrep-svn] SF.net SVN: opentrep:[170] trunk/opentrep/opentrep
Status: Beta
Brought to you by:
denis_arnaud
Menu
▾
▴
[Opentrep-svn] SF.net SVN: opentrep:[170] trunk/opentrep/opentrep
|
From: <den...@us...> - 2009-07-27 05:56:55
|
Revision: 170
http://opentrep.svn.sourceforge.net/opentrep/?rev=170&view=rev
Author: denis_arnaud
Date: 2009-07-27 05:56:43 +0000 (Mon, 27 Jul 2009)
Log Message:
-----------
[Dev] Prepared the code to dig out the edit distance and extra and alternate locations.
Modified Paths:
--------------
trunk/opentrep/opentrep/bom/Document.cpp
trunk/opentrep/opentrep/bom/Document.hpp
trunk/opentrep/opentrep/bom/Place.cpp
trunk/opentrep/opentrep/bom/Place.hpp
trunk/opentrep/opentrep/bom/ResultHolder.cpp
trunk/opentrep/opentrep/bom/StringMatcher.cpp
trunk/opentrep/opentrep/bom/StringMatcher.hpp
trunk/opentrep/opentrep/command/RequestInterpreter.cpp
trunk/opentrep/opentrep/factory/FacPlace.cpp
trunk/opentrep/opentrep/factory/FacPlace.hpp
Modified: trunk/opentrep/opentrep/bom/Document.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/Document.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/Document.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -25,7 +25,8 @@
std::ostringstream oStr;
oStr << "`" << describeShortKey() << "'";
if (_correctedQueryString.empty() == false) {
- oStr << " (corrected into `" << _correctedQueryString << "')";
+ oStr << " (corrected into `" << _correctedQueryString
+ << "' with an edit distance/error of " << _editDistance << ")";
}
return oStr.str();
}
@@ -37,11 +38,12 @@
const Xapian::docid& lDocID = _document.get_docid();
oStr << " => Document ID " << lDocID << " matching at " << _percentage
- << "% [" << _document.get_data() << "]";
+ << "% (edit distance of " << _editDistance << ") ["
+ << _document.get_data() << "]";
if (_documentList.empty() == false) {
oStr << " along with " << _documentList.size()
- << " other matching document(s) (";
+ << " other equivalent matching document(s) (";
unsigned short idx = 0;
for (XapianDocumentList_T::const_iterator itDoc = _documentList.begin();
@@ -53,6 +55,25 @@
}
oStr << lDocID;
}
+ oStr << ")";
+ }
+
+ if (_alternateDocumentList.empty() == false) {
+ oStr << " and with still " << _alternateDocumentList.size()
+ << " other less matching document(s) (";
+
+ unsigned short idx = 0;
+ for (XapianAlternateDocumentList_T::const_iterator itDoc =
+ _alternateDocumentList.begin();
+ itDoc != _alternateDocumentList.end(); ++itDoc, ++idx) {
+ const Xapian::percent& lPercentage = itDoc->first;
+ const Xapian::Document& lXapianDoc = itDoc->second;
+ const Xapian::docid& lDocID = lXapianDoc.get_docid();
+ if (idx != 0) {
+ oStr << ", ";
+ }
+ oStr << lDocID << " / " << lPercentage << "%";
+ }
oStr << ")." << std::endl;
} else {
@@ -73,7 +94,7 @@
if (_documentList.empty() == false) {
oStr << " along with " << _documentList.size()
- << " other matching document(s) { ";
+ << " other equivalent matching document(s) { ";
unsigned short idx = 0;
for (XapianDocumentList_T::const_iterator itDoc = _documentList.begin();
@@ -85,6 +106,26 @@
}
oStr << "Doc ID " << lDocID << " [" << lXapianDoc.get_data() << "]";
}
+ oStr << " }";
+ }
+
+ if (_alternateDocumentList.empty() == false) {
+ oStr << " and with still " << _alternateDocumentList.size()
+ << " other less matching document(s) { ";
+
+ unsigned short idx = 0;
+ for (XapianAlternateDocumentList_T::const_iterator itDoc =
+ _alternateDocumentList.begin();
+ itDoc != _alternateDocumentList.end(); ++itDoc, ++idx) {
+ const Xapian::percent& lPercentage = itDoc->first;
+ const Xapian::Document& lXapianDoc = itDoc->second;
+ const Xapian::docid& lDocID = lXapianDoc.get_docid();
+ if (idx != 0) {
+ oStr << ", ";
+ }
+ oStr << lDocID << " / " << lPercentage << "% ["
+ << lXapianDoc.get_data() << "]";
+ }
oStr << " }." << std::endl;
} else {
Modified: trunk/opentrep/opentrep/bom/Document.hpp
===================================================================
--- trunk/opentrep/opentrep/bom/Document.hpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/Document.hpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -17,6 +17,12 @@
// //////////////// Type definitions /////////////////
/** List of Xapian documents. */
typedef std::list<Xapian::Document> XapianDocumentList_T;
+
+ /** Pair of a Xapian document and its associated matching percentage. */
+ typedef std::pair<Xapian::percent, Xapian::Document> XapianDocumentPair_T;
+
+ /** List of Xapian documents. */
+ typedef std::list<XapianDocumentPair_T> XapianAlternateDocumentList_T;
// //////////////// Main Class /////////////////
@@ -48,12 +54,19 @@
return _percentage;
}
- /** Get the extra list of matching Xapian documents. */
+ /** Get the extra list of matching Xapian documents (i.e., those
+ having matched with the same weight as the main one). */
const XapianDocumentList_T& getExtraDocumentList() const {
return _documentList;
}
+ /** Get the alternate list of matching Xapian documents (i.e., those
+ having matched with a lower weight than the main one). */
+ const XapianAlternateDocumentList_T& getAlternateDocumentList() const {
+ return _alternateDocumentList;
+ }
+
// ////////////////// Setters ////////////////
/** Set the query string. */
void setQueryString (const TravelQuery_T& iQueryString) {
@@ -75,12 +88,25 @@
_percentage = iPercentage;
}
+ /** Set the edit distance/error, with which the matching has been made. */
+ void setEditDistance (const NbOfErrors_T& iEditDistance) {
+ _editDistance = iEditDistance;
+ }
+
/** Add a matching Xapian document (having the same matching percentage). */
void addExtraDocument (const Xapian::Document& iMatchingDocument) {
_documentList.push_back (iMatchingDocument);
}
+ /** Add a matching Xapian document (having a lower matching percentage). */
+ void addAlternateDocument (const Xapian::percent& iMatchingPercentage,
+ const Xapian::Document& iMatchingDocument) {
+ _alternateDocumentList.
+ push_back (XapianDocumentPair_T (iMatchingPercentage,
+ iMatchingDocument));
+ }
+
public:
// /////////// Business methods /////////
/** Retrieve the number of extra matches for the given query string,
@@ -143,10 +169,18 @@
/** Matching document, as returned by the Xapian full text search. */
Xapian::Document _document;
+ /** Edit distance/error, with which the matching has been made. */
+ NbOfErrors_T _editDistance;
+
/** List of Xapian documents having the same matching percentage.
<br>Hence, any of those other Xapian documents could have been
chosen, instead of the main one. */
XapianDocumentList_T _documentList;
+
+ /** List of Xapian documents having the a lower matching percentage.
+ <br>Those alternate matches can be suggested (in the famous
+ "Did you mean Xxx?" question) to the end user. */
+ XapianAlternateDocumentList_T _alternateDocumentList;
};
}
Modified: trunk/opentrep/opentrep/bom/Place.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/Place.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/Place.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -10,12 +10,13 @@
namespace OPENTREP {
// //////////////////////////////////////////////////////////////////////
- Place::Place () : _world (NULL), _placeHolder (NULL) {
+ Place::Place () : _world (NULL), _placeHolder (NULL), _mainPlace (NULL) {
}
// //////////////////////////////////////////////////////////////////////
Place::Place (const Place& iPlace) :
_world (iPlace._world), _placeHolder (iPlace._placeHolder),
+ _mainPlace (iPlace._mainPlace),
_placeCode (iPlace._placeCode), _cityCode (iPlace._cityCode),
_stateCode (iPlace._stateCode), _countryCode (iPlace._countryCode),
_regionCode (iPlace._regionCode), _continentCode (iPlace._continentCode),
Modified: trunk/opentrep/opentrep/bom/Place.hpp
===================================================================
--- trunk/opentrep/opentrep/bom/Place.hpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/Place.hpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -14,6 +14,7 @@
#include <opentrep/Location.hpp>
#include <opentrep/bom/BomAbstract.hpp>
#include <opentrep/bom/Names.hpp>
+#include <opentrep/bom/PlaceList.hpp>
namespace OPENTREP {
@@ -93,6 +94,16 @@
language. */
bool getNameList (const Language::EN_Language&, NameList_T&) const;
+ /** Get the list of extra matching (similar) places. */
+ const PlaceOrderedList_T& getExtraPlaceList() const {
+ return _extraPlaceList;
+ }
+
+ /** Get the list of alternate matching (less similar) places. */
+ const PlaceOrderedList_T& getAlternatePlaceList() const {
+ return _alternatePlaceList;
+ }
+
// ///////// Setters ////////
/** Set the Place code. */
@@ -206,9 +217,14 @@
/** Parent World. */
World* _world;
- /** Parent PlaceHolder. */
+ /** Parent PlaceHolder (not always defined,for instance if the
+ current Place object is an extra or alternate one). */
PlaceHolder* _placeHolder;
+ /** Parent (main) Place (not always defined,for instance if the
+ current Place object is itself a main one). */
+ Place* _mainPlace;
+
private:
// /////// Attributes /////////
/** Place code. */
@@ -233,6 +249,12 @@
NameMatrix_T _nameMatrix;
/** Xapian document ID. */
XapianDocID_T _docID;
+
+ /** List of extra matching (similar) places. */
+ PlaceOrderedList_T _extraPlaceList;
+
+ /** List of alternate matching (less similar) places. */
+ PlaceOrderedList_T _alternatePlaceList;
};
}
Modified: trunk/opentrep/opentrep/bom/ResultHolder.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/ResultHolder.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/ResultHolder.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -110,26 +110,26 @@
<< lMaxEditDistance << ".");
// Retrieve the list of Xapian documents matching the query string
+ NbOfErrors_T lCalculatedEditDistance = 0;
oMatchedString =
- StringMatcher::searchString (ioMatchingSet, ioPartialQueryString,
- lMaxEditDistance,
- hasReachedMaximalAllowableEditDistance,
- _database);
+ StringMatcher::searchString(ioMatchingSet, ioPartialQueryString,
+ lCalculatedEditDistance, lMaxEditDistance,
+ hasReachedMaximalAllowableEditDistance,
+ _database);
// DEBUG
OPENTREP_LOG_DEBUG ("---- Current query string: `"
<< ioPartialQueryString << "' --- Kept query: `"
<< oMatchedString
- << "', with a maximal edit distance of "
- << lMaxEditDistance << ", for "
+ << "', with an edit distance of a maximum of "
+ << lCalculatedEditDistance << " (over "
+ << lMaxEditDistance << "), for "
<< ioMatchingSet.size() << " matches.");
if (ioMatchingSet.empty() == false) {
- // Create the corresponding list of documents
- StringMatcher::
- extractBestMatchingDocumentFromMSet (ioMatchingSet,
- ioMatchingDocument);
-
+ // Store the calculated (and applied) edit distance/erro
+ ioMatchingDocument.setEditDistance (lCalculatedEditDistance);
+
// Since a result has been found, the search can be stopped
// for that part of the query.
shouldStop = true;
Modified: trunk/opentrep/opentrep/bom/StringMatcher.cpp
===================================================================
--- trunk/opentrep/opentrep/bom/StringMatcher.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/StringMatcher.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -114,12 +114,13 @@
// ///////////////////////////////////////////////////////////////////
void checkAndAlterIfNeeded (TravelQuery_T& ioSuggestedString,
const TravelQuery_T& iOriginalString,
+ NbOfErrors_T& ioCalculatedEditDistance,
const NbOfErrors_T& iMaxEditDistance,
const Xapian::Database& iDatabase) {
/**
- Store a copy of the suggested string, as it will me altered by
- the below method.
+ Store a copy of the suggested string, as it will be altered by
+ the below method, i.e., removeFurthestLeftWord().
*/
TravelQuery_T lOriginalStringCopy (iOriginalString);
StringMatcher::removeFurthestLeftWord (lOriginalStringCopy);
@@ -128,15 +129,14 @@
Get a spell-corrected suggestion for the reduced original string.
<br>Limit the edit distance to the given maximal one.
*/
- NbOfErrors_T lCalculatedEditDistance =
- calculateEditDistance (lOriginalStringCopy);
+ ioCalculatedEditDistance = calculateEditDistance (lOriginalStringCopy);
- lCalculatedEditDistance = std::min (lCalculatedEditDistance,
- iMaxEditDistance);
+ ioCalculatedEditDistance = std::min (ioCalculatedEditDistance,
+ iMaxEditDistance);
std::string lSuggestionForReducedOriginalString =
iDatabase.get_spelling_suggestion (lOriginalStringCopy,
- lCalculatedEditDistance);
+ ioCalculatedEditDistance);
/**
Note that if the suggestion on the reduced-original string is
@@ -154,7 +154,7 @@
OPENTREP_LOG_DEBUG ("The suggestion (`" << ioSuggestedString
<< "') for `" << iOriginalString
<< "', with an edit distance/error of "
- << lCalculatedEditDistance
+ << ioCalculatedEditDistance
<< " over " << iMaxEditDistance << " allowable"
<< ", is the same as the suggestion for the reduced "
<< "original string (`" << lOriginalStringCopy
@@ -184,14 +184,14 @@
Get a spell-corrected suggestion for the reduced original string.
<br>Limit the edit distance to the given maximal one.
*/
- lCalculatedEditDistance = calculateEditDistance (lOriginalStringCopy);
+ ioCalculatedEditDistance = calculateEditDistance (lOriginalStringCopy);
- lCalculatedEditDistance = std::min (lCalculatedEditDistance,
- iMaxEditDistance);
+ ioCalculatedEditDistance = std::min (ioCalculatedEditDistance,
+ iMaxEditDistance);
lSuggestionForReducedOriginalString =
iDatabase.get_spelling_suggestion (lOriginalStringCopy,
- lCalculatedEditDistance);
+ ioCalculatedEditDistance);
/**
Note that if the suggestion on the reduced-original string is
@@ -209,7 +209,7 @@
OPENTREP_LOG_DEBUG ("The suggestion (`" << ioSuggestedString
<< "') for `" << iOriginalString
<< "', with an edit distance/error of "
- << lCalculatedEditDistance
+ << ioCalculatedEditDistance
<< " over " << iMaxEditDistance << " allowable"
<< ", is the same as the suggestion for the reduced "
<< "original string (`" << lOriginalStringCopy
@@ -228,6 +228,7 @@
std::string StringMatcher::
searchString (Xapian::MSet& ioMatchingSet,
const TravelQuery_T& iSearchString,
+ NbOfErrors_T& ioCalculatedEditDistance,
NbOfErrors_T& ioMaxEditDistance,
bool& ioHasReachedMaximalAllowableEditDistance,
const Xapian::Database& iDatabase) {
@@ -407,26 +408,26 @@
phrase/string. With the above example, 'sna francisco' yields the
suggestion 'san francisco'.
*/
- NbOfErrors_T lCalculatedEditDistance =
- calculateEditDistance (lOriginalQueryString);
+ ioCalculatedEditDistance = calculateEditDistance (lOriginalQueryString);
// Store the greatest edit distance/error
- lMaxEditDistance = std::max (lMaxEditDistance, lCalculatedEditDistance);
+ lMaxEditDistance = std::max (lMaxEditDistance, ioCalculatedEditDistance);
// Limit the edit distance to the given maximal one
- lCalculatedEditDistance = std::min (lCalculatedEditDistance,
- ioMaxEditDistance);
+ ioCalculatedEditDistance = std::min (ioCalculatedEditDistance,
+ ioMaxEditDistance);
std::string lFullWordCorrectedString =
iDatabase.get_spelling_suggestion (lOriginalQueryString,
- lCalculatedEditDistance);
+ ioCalculatedEditDistance);
/**
Check that the suggestion does not encompass extra words, which
will be otherwise/rather recognised in another step.
*/
checkAndAlterIfNeeded (lFullWordCorrectedString, lOriginalQueryString,
- ioMaxEditDistance, iDatabase);
+ ioCalculatedEditDistance, ioMaxEditDistance,
+ iDatabase);
/**
Since there is still no match, we search on the string
@@ -528,7 +529,7 @@
NbOfMatches_T idx = 1;
for ( ; itDoc != iMatchingSet.end(); ++itDoc, ++idx) {
const Xapian::percent& lPercentage = itDoc.get_percent();
- // const Xapian::Document& lDocument = itDoc.get_document();
+ const Xapian::Document& lDocument = itDoc.get_document();
// DEBUG
/*
@@ -536,12 +537,15 @@
<< lDocument.get_docid() << " matching at "
<< lPercentage << "%.");
*/
-
+
+ /** If the matching percentage is the same as for the main
+ (chosen) Xapian document, then add it to the dedicated
+ list. Otherwise, add it to the alternative choices. */
if (lPercentage == lBestPercentage) {
- ioMatchingDocument.addExtraDocument (itDoc.get_document());
+ ioMatchingDocument.addExtraDocument (lDocument);
} else {
- break;
+ ioMatchingDocument.addAlternateDocument (lPercentage, lDocument);
}
}
}
Modified: trunk/opentrep/opentrep/bom/StringMatcher.hpp
===================================================================
--- trunk/opentrep/opentrep/bom/StringMatcher.hpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/bom/StringMatcher.hpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -28,6 +28,7 @@
words of the search string.
@param Xapian::MSet& The Xapian matching set. It can be empty.
@param const std::string& The query string.
+ @param NbOfErrors_T& The calculated (and applied) edit distance/error.
@param NbOfErrors_T& The maximal allowable edit distance/error.
@param bool& Whether or not the maximal allowable edit distance/error
has become greater than the maximum of the edit distance/errors
@@ -37,6 +38,7 @@
which has yielded matches. */
static std::string searchString (Xapian::MSet&,
const std::string& iSearchString,
+ NbOfErrors_T& ioCalculatedEditDistance,
NbOfErrors_T& ioMaxEditDistance,
bool& ioHasReachedMaximalAllowableEditDistance,
const Xapian::Database&);
Modified: trunk/opentrep/opentrep/command/RequestInterpreter.cpp
===================================================================
--- trunk/opentrep/opentrep/command/RequestInterpreter.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/command/RequestInterpreter.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -72,6 +72,13 @@
const Result* lResult_ptr = *itResult;
assert (lResult_ptr != NULL);
+ /**
+ TODO: Add a loop for retrieving both extra and alternate Documents
+ Use FacPlace::initLinkWithExtraPlace() and
+ FacPlace::initLinkWithAlternatePlace()
+ */
+
+
// Retrieve the parameters of the best matching document
const Xapian::Document& lDocument = lResult_ptr->getXapianDocument();
const Xapian::percent& lDocPercentage =
Modified: trunk/opentrep/opentrep/factory/FacPlace.cpp
===================================================================
--- trunk/opentrep/opentrep/factory/FacPlace.cpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/factory/FacPlace.cpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -63,4 +63,26 @@
return *oPlace_ptr;
}
+ // //////////////////////////////////////////////////////////////////////
+ void FacPlace::initLinkWithExtraPlace (Place& ioMainPlace,
+ Place& ioExtraPlace) {
+ // Link the main Place to the extra Place, and vice versa
+ ioExtraPlace._mainPlace = &ioMainPlace;
+
+ // Add the extra Place to the main Place internal map (of extra
+ // Place objects)
+ ioMainPlace._extraPlaceList.push_back (&ioExtraPlace);
+ }
+
+ // //////////////////////////////////////////////////////////////////////
+ void FacPlace::initLinkWithAlternatePlace (Place& ioMainPlace,
+ Place& ioAlternatePlace) {
+ // Link the main Place to the alternate Place, and vice versa
+ ioAlternatePlace._mainPlace = &ioMainPlace;
+
+ // Add the alternate Place to the main Place internal map (of
+ // alternate Place objects)
+ ioMainPlace._extraPlaceList.push_back (&ioAlternatePlace);
+ }
+
}
Modified: trunk/opentrep/opentrep/factory/FacPlace.hpp
===================================================================
--- trunk/opentrep/opentrep/factory/FacPlace.hpp 2009-07-25 22:27:45 UTC (rev 169)
+++ trunk/opentrep/opentrep/factory/FacPlace.hpp 2009-07-27 05:56:43 UTC (rev 170)
@@ -36,6 +36,20 @@
@return Place& The newly created object. */
Place& clone (const Place&);
+ /** Initialise the link between a Place and an extra Place.
+ @param Place& Main Place object.
+ @param Place& Extra Place object.
+ @exception FacExceptionNullPointer
+ @exception FacException.*/
+ static void initLinkWithExtraPlace (Place&, Place&);
+
+ /** Initialise the link between a Place and an alternate Place.
+ @param Place& Main Place object.
+ @param Place& Alternate Place object.
+ @exception FacExceptionNullPointer
+ @exception FacException.*/
+ static void initLinkWithAlternatePlace (Place&, Place&);
+
private:
/** Default Constructor.
<br>This constructor is private in order to ensure the singleton
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
