From kde-commits  Wed Apr 30 22:22:11 2008
From: Sebastian Trueg <sebastian () trueg ! de>
Date: Wed, 30 Apr 2008 22:22:11 +0000
To: kde-commits
Subject: playground/base/nepomuk-kde/search-api
Message-Id: <1209594131.731193.18739.nullmailer () svn ! kde ! org>
X-MARC-Message: https://marc.info/?l=kde-commits&m=120959413901032

SVN commit 802869 by trueg:

Support full property and resource URIs in query strings. Fixed values in quotes. In the end the problem was that either QRegExp's \b is broken or I did not understand it.

 M  +53 -7     queryparser.cpp  


--- trunk/playground/base/nepomuk-kde/search-api/queryparser.cpp #802868:802869
@@ -32,21 +32,36 @@
  */
 
 namespace {
-    QString s_plainTermPattern( "([^\\s\"':=\\<\\>]+|(?:([\"'])[^\"':=\\<\\>]+\\2))" );
+    QString s_plainTermPattern( "([^\\s\"':=\\<\\>]+|(?:([\"'])[^\"']+\\%1))" );
     QString s_inExclusionPattern( "([\\+\\-]?)" );
+    QString s_uriPattern( "<([^<>]+)>" );
+    QString s_comparatorPattern( "(:|=|\\<|\\>|\\<=|\\>=)" );
 
     // match a simple search text
     // captures: 1 - The optional + or - sign (may be empty)
     //           2 - the search text (including optional paranthesis)
-    QRegExp s_plainTermRx( "\\b" + s_inExclusionPattern + s_plainTermPattern + "\\b" );
+    QRegExp s_plainTermRx( QString( s_inExclusionPattern + s_plainTermPattern ).arg( 3 ) );
 
     // match a field search term: fieldname + relation (:, =, etc) + search text with optional paranthesis
     // captures: 1 - The optional + or - sign (may be empty)
     //           2 - fieldname
     //           3 - relation
     //           4 - search text (including optional paranthesis)
-    QRegExp s_fieldRx( "\\b" + s_inExclusionPattern + "(\\S+)([:=\\<\\>])" + s_plainTermPattern + "\\b" );
+    QRegExp s_fieldRx( QString( s_inExclusionPattern + "(\\S+)" + s_comparatorPattern + s_plainTermPattern ).arg( 5 ) );
 
+    // match a property URI search term: property URI + relation (:, =, etc) + search text with optional paranthesis
+    // captures: 1 - The optional + or - sign (may be empty)
+    //           2 - property URI
+    //           3 - relation
+    //           4 - search text (including optional paranthesis)
+    QRegExp s_propertyRx( QString( s_inExclusionPattern + s_uriPattern + s_comparatorPattern + s_plainTermPattern ).arg( 5 ) );
+
+    // match a property URI search term: property URI + relation (:, =, etc) + resource URI
+    // captures: 1 - The optional + or - sign (may be empty)
+    //           2 - property URI
+    //           3 - resource URI
+    QRegExp s_resourceRx( QString( s_inExclusionPattern + s_uriPattern + "(?::|=)" + s_uriPattern ) );
+
     // FIXME: allow comparators in field and values when we have ' or " around them
 
     Nepomuk::Search::Term::Comparator fieldTypeRelationFromString( const QString& s ) {
@@ -70,6 +85,16 @@
             return Nepomuk::Search::Term::Equal;
         }
     }
+
+    QString stripQuotes( const QString& s ) {
+        if ( s[0] == '\'' ||
+             s[0] == '\"' ) {
+            return s.mid( 1 ).left( s.length()-2 );
+        }
+        else {
+            return s;
+        }
+    }
 }
 
 
@@ -123,11 +148,32 @@
         Term term;
 
         if ( pos < query.length() ) {
-            if ( s_fieldRx.indexIn( query, pos ) == pos ) {
+            if ( s_resourceRx.indexIn( query, pos ) == pos ) {
                 // FIXME: honour the +-
-                kDebug() << "matched field term at" << pos << s_fieldRx.cap( 2 );
+                kDebug() << "matched resource term at" << pos << s_resourceRx.cap( 0 );
+                term = Term( QUrl( s_resourceRx.cap( 2 ) ), QUrl( s_resourceRx.cap( 3 ) ) );
+                pos += s_resourceRx.matchedLength();
+            }
+            else if ( s_propertyRx.indexIn( query, pos ) == pos ) {
+                // FIXME: honour the +-
+                kDebug() << "matched property term at" << pos << s_propertyRx.cap( 0 );
+                term.setProperty( QUrl( s_propertyRx.cap( 2 ) ) );
+                term.setValue( stripQuotes( s_propertyRx.cap( 4 ) ) );
+                QString comparator = s_propertyRx.cap( 3 );
+                if( comparator == QChar(':') ) {
+                    term.setType( Term::ContainsTerm );
+                }
+                else {
+                    term.setType( Term::ComparisonTerm );
+                    term.setComparator( fieldTypeRelationFromString( comparator ) );
+                }
+                pos += s_propertyRx.matchedLength();
+            }
+            else if ( s_fieldRx.indexIn( query, pos ) == pos ) {
+                // FIXME: honour the +-
+                kDebug() << "matched field term at" << pos << s_fieldRx.cap( 0 );
                 term.setField( s_fieldRx.cap( 2 ) );
-                term.setValue( s_fieldRx.cap( 4 ) );
+                term.setValue( stripQuotes( s_fieldRx.cap( 4 ) ) );
                 QString comparator = s_fieldRx.cap( 3 );
                 if( comparator == QChar(':') ) {
                     term.setType( Term::ContainsTerm );
@@ -140,7 +186,7 @@
             }
             else if ( s_plainTermRx.indexIn( query, pos ) == pos ) {
                 // FIXME: honour the +-
-                QString value = s_plainTermRx.cap( 2 );
+                QString value = stripQuotes( s_plainTermRx.cap( 2 ) );
                 if ( value.length() == d->orKeyword.length() && value.toLower() == d->orKeyword ) {
                     inOrBlock = true;
                 }