matching enhancements -- allow matching rule to be changed on the fly
[koha_gimpoz] / C4 / Search.pm
index 02a4438..9d345d4 100644 (file)
@@ -21,13 +21,12 @@ use C4::Context;
 use C4::Biblio;    # GetMarcFromKohaField
 use C4::Koha;      # getFacets
 use Lingua::Stem;
+use C4::Date;
 
 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
 
 # set the version for version checking
-$VERSION = do { my @v = '$Revision$' =~ /\d+/g;
-    shift(@v) . "." . join( "_", map { sprintf "%03d", $_ } @v );
-};
+$VERSION = 3.00;
 
 =head1 NAME
 
@@ -100,6 +99,11 @@ sub FindDuplicate {
     my ( $biblionumber, $title );
 
     # search duplicate on ISBN, easy and fast..
+    # ... normalize first
+    if ( $result->{isbn} ) {
+        $result->{isbn} =~ s/\(.*$//;
+        $result->{isbn} =~ s/\s+$//; 
+    }
     #$search->{'avoidquerylog'}=1;
     if ( $result->{isbn} ) {
         $query = "isbn=$result->{isbn}";
@@ -109,6 +113,8 @@ sub FindDuplicate {
         $result->{title} =~ s /\"//g;
         $result->{title} =~ s /\(//g;
         $result->{title} =~ s /\)//g;
+               # remove valid operators
+               $result->{title} =~ s/(and|or|not)//g;
         $query = "ti,ext=$result->{title}";
         $query .= " and mt=$result->{itemtype}" if ($result->{itemtype});    
         if ($result->{author}){
@@ -116,6 +122,8 @@ sub FindDuplicate {
           $result->{author} =~ s /\"//g;
           $result->{author} =~ s /\(//g;
           $result->{author} =~ s /\)//g;
+                 # remove valid operators
+                 $result->{author} =~ s/(and|or|not)//g;
           $query .= " and au,ext=$result->{author}";
         }     
     }
@@ -327,15 +335,54 @@ sub getRecords {
             }
         };
         if ($@) {
-            warn "prob with query  toto $query_to_use " . $@;
+            warn "WARNING: query problem with $query_to_use " . $@;
         }
 
         # concatenate the sort_by limits and pass them to the results object
         my $sort_by;
         foreach my $sort (@sort_by) {
-            $sort_by .= $sort . " ";    # used to be $sort,
+            if ($sort eq "author_az") {
+                $sort_by.="1=1003 <i ";
+            }
+            elsif ($sort eq "author_za") {
+                $sort_by.="1=1003 >i ";
+            }
+                       elsif ($sort eq "popularity_asc") {
+                               $sort_by.="1=9003 <i ";
+                       }
+                       elsif ($sort eq "popularity_dsc") {
+                $sort_by.="1=9003 >i ";
+            }
+                       elsif ($sort eq "call_number_asc") {
+                $sort_by.="1=20  <i ";
+            }
+                       elsif ($sort eq "call_number_dsc") {
+                $sort_by.="1=20 >i ";
+            }
+                       elsif ($sort eq "pubdate_asc") {
+                $sort_by.="1=31 <i ";
+            }
+                       elsif ($sort eq "pubdate_dsc") {
+                $sort_by.="1=31 >i ";
+            }
+                       elsif ($sort eq "acqdate_asc") {
+                $sort_by.="1=32 <i ";
+            }
+                       elsif ($sort eq "acqdate_dsc") {
+                $sort_by.="1=32 >i ";
+            }
+                       elsif ($sort eq "title_az") {
+                $sort_by.="1=4 <i ";
+            }
+                       elsif ($sort eq "title_za") {
+                $sort_by.="1=4 >i ";
+            }
         }
-        $results[$i]->sort( "yaz", $sort_by ) if $sort_by;
+               if ($sort_by) {
+                       if ( $results[$i]->sort( "yaz", $sort_by ) < 0) {
+                       warn "WARNING sort $sort_by failed";
+                       }
+               }
     }
     while ( ( my $i = ZOOM::event( \@zconns ) ) != 0 ) {
         my $ev = $zconns[ $i - 1 ]->last_event();
@@ -508,9 +555,154 @@ sub getRecords {
     return ( undef, $results_hashref, \@facets_loop );
 }
 
+sub _remove_stopwords {
+       my ($operand,$index) = @_;
+       # if the index contains more than one qualifier, but not phrase:    
+       if ($index!~m/phr|ext/){
+       # operand may be a wordlist deleting stopwords
+       # remove stopwords from operand : parse all stopwords & remove them (case insensitive)
+       #       we use IsAlpha unicode definition, to deal correctly with diacritics.
+       #       otherwise, a french word like "leçon" is splitted in "le" "çon", le is an empty word, we get "çon"
+       #       and don't find anything...
+               foreach (keys %{C4::Context->stopwords}) {
+      next if ($_ =~/(and|or|not)/); # don't remove operators 
+                       $operand=~ s/\P{IsAlpha}$_\P{IsAlpha}/ /i;
+                       $operand=~ s/^$_\P{IsAlpha}/ /i;
+                       $operand=~ s/\P{IsAlpha}$_$/ /i;
+               }
+       }
+       return $operand;
+}
+
+sub _add_truncation {
+       my ($operand,$index) = @_;
+       my (@nontruncated,@righttruncated,@lefttruncated,@rightlefttruncated,@regexpr);
+       # if the index contains more than one qualifier, but not phrase, add truncation qualifiers
+       #if (index($index,"phr")<0 && index($index,",")>0){
+       # warn "ADDING TRUNCATION QUALIFIERS";
+               $operand =~s/^ //g;
+               my @wordlist= split (/\s/,$operand);
+               foreach my $word (@wordlist){
+                       if ($word=~s/^\*([^\*]+)\*$/$1/){
+        push @rightlefttruncated,$word;
+                       } 
+                       elsif($word=~s/^\*([^\*]+)$/$1/){
+                               push @lefttruncated,$word;
+                        
+                       } 
+                       elsif ($word=~s/^([^\*]+)\*$/$1/){
+                               push @righttruncated,$word;
+                       } 
+                       elsif (index($word,"*")<0){
+                               push @nontruncated,$word;
+                       }
+                       else {
+                               push @regexpr,$word;
+                        
+                       }
+               }
+       #}
+       return (\@nontruncated,\@righttruncated,\@lefttruncated,\@rightlefttruncated,\@regexpr);
+}
+
+sub _build_stemmed_operand {
+       my ($operand) = @_;
+       my $stemmed_operand;
+       #$operand =~ s/^(and |or |not )//i;
+       # STEMMING FIXME: may need to refine the field weighting so stemmed operands don't 
+       # disrupt the query ranking, this needs more testing
+       # FIXME: the locale should be set based on the user's language and/or search choice
+       my $stemmer = Lingua::Stem->new( -locale => 'EN-US' );
+       # FIXME: these should be stored in the db so the librarian can modify the behavior
+       $stemmer->add_exceptions(
+                       {
+                               'and' => 'and',
+                'or'  => 'or',
+                'not' => 'not',
+                       }
+                    
+               );
+       my @words = split( / /, $operand );
+       my $stems = $stemmer->stem(@words);
+       foreach my $stem (@$stems) {
+                       $stemmed_operand .= "$stem";
+                       $stemmed_operand .= "?" unless ( $stem =~ /(and$|or$|not$)/ ) || ( length($stem) < 3 );
+                       $stemmed_operand .= " ";
+       }
+       #warn "STEMMED OPERAND: $stemmed_operand";
+       return $stemmed_operand;
+}
+
+sub _build_weighted_query {
+       # FIELD WEIGHTING - This is largely experimental stuff. What I'm committing works
+       # pretty well but will work much better when we have an actual query parser
+       my ($operand,$stemmed_operand,$index) = @_;
+    my $stemming      = C4::Context->preference("QueryStemming")     || 0;
+    my $weight_fields = C4::Context->preference("QueryWeightFields") || 0;
+    my $fuzzy_enabled = C4::Context->preference("QueryFuzzy") || 0;
+
+    my $weighted_query .= " (rk=(";     # Specifies that we're applying rank
+       # keyword has different weight properties
+       if ( ( $index =~ /kw/ ) || ( !$index ) ) {
+       # a simple way to find out if this query uses an index
+               if ( $operand =~ /(\=|\:)/ ) {
+                       $weighted_query .= " $operand";
+               }
+               else {
+                       $weighted_query .=" Title-cover,ext,r1=\"$operand\"";   # title cover as exact
+                       $weighted_query .=" or ti,ext,r2=\"$operand\"";             # exact title elsewhere
+                       $weighted_query .= " or ti,phr,r3=\"$operand\"";          # index as phrase
+                       #$weighted_query .= " or any,ext,r4=$operand";         # index as exact
+                       #$weighted_query .=" or kw,wrdl,r5=\"$operand\"";            # all the words in the query (wordlist)
+                       $weighted_query .= " or wrd,fuzzy,r8=\"$operand\"" if $fuzzy_enabled; # add fuzzy
+                       $weighted_query .= " or wrd,right-Truncation,r9=\"$stemmed_operand\"" if ($stemming and $stemmed_operand); # add stemming
+                       # embedded sorting: 0 a-z; 1 z-a
+                       #$weighted_query .= ") or (sort1,aut=1";
+               }
+                    
+       }
+       #TODO: build better cases based on specific search indexes
+       #elsif ( $index =~ /au/ ) {
+       #       $weighted_query .=" $index,ext,r1=$operand";    # index label as exact
+       #       #$weighted_query .= " or (title-sort-az=0 or $index,startswithnt,st-word,r3=$operand #)";
+       #       $weighted_query .=" or $index,phr,r3=$operand";    # index as phrase
+       #       $weighted_query .= " or $index,rt,wrd,r3=$operand";
+       #}
+       #elsif ( $index =~ /ti/ ) {
+       #       $weighted_query .=" Title-cover,ext,r1=$operand"; # index label as exact
+       #       $weighted_query .= " or Title-series,ext,r2=$operand";
+       #       #$weighted_query .= " or ti,ext,r2=$operand";
+       #       #$weighted_query .= " or ti,phr,r3=$operand";
+       #       #$weighted_query .= " or ti,wrd,r3=$operand";
+       #       $weighted_query .=" or (title-sort-az=0 or Title-cover,startswithnt,st-word,r3=$operand #)";
+       #       $weighted_query .=" or (title-sort-az=0 or Title-cover,phr,r6=$operand)";
+               #$weighted_query .= " or Title-cover,wrd,r5=$operand";
+               #$weighted_query .= " or ti,ext,r6=$operand";
+               #$weighted_query .= " or ti,startswith,phr,r7=$operand";
+               #$weighted_query .= " or ti,phr,r8=$operand";
+               #$weighted_query .= " or ti,wrd,r9=$operand";
+               #$weighted_query .= " or ti,ext,r2=$operand";         # index as exact
+               #$weighted_query .= " or ti,phr,r3=$operand";              # index as  phrase
+               #$weighted_query .= " or any,ext,r4=$operand";         # index as exact
+               #$weighted_query .= " or kw,wrd,r5=$operand";         # index as exact
+       #}
+       else {
+               warn "WEIGHT GENERIC";
+               $weighted_query .=" $index=$operand";
+               #$weighted_query .=" $index,ext,r1=$operand";    # index label as exact
+               #$weighted_query .= " or $index,ext,r2=$operand";            # index as exact
+               #$weighted_query .=" or $index,phr,r3=$operand";    # index as phrase
+               #$weighted_query .= " or $index,rt,wrd,r3=$operand";
+               #$weighted_query .=" or $index,wrd,r5=$operand";    # index as word right-truncated
+               #$weighted_query .= " or $index,wrd,fuzzy,r8=$operand" if $fuzzy_enabled;
+       }
+       $weighted_query .= "))";    # close rank specification
+       return $weighted_query;
+}
+
 # build the query itself
 sub buildQuery {
-    my ( $query, $operators, $operands, $indexes, $limits, $sort_by ) = @_;
+    my ( $operators, $operands, $indexes, $limits, $sort_by ) = @_;
 
     my @operators = @$operators if $operators;
     my @indexes   = @$indexes   if $indexes;
@@ -518,26 +710,19 @@ sub buildQuery {
     my @limits    = @$limits    if $limits;
     my @sort_by   = @$sort_by   if $sort_by;
 
+       my $stemming      = C4::Context->preference("QueryStemming")     || 0;
+       my $weight_fields = C4::Context->preference("QueryWeightFields") || 0;
+       my $fuzzy_enabled = C4::Context->preference("QueryFuzzy") || 0;
+
     my $human_search_desc;      # a human-readable query
     my $machine_search_desc;    #a machine-readable query
-        # FIXME: the locale should be set based on the syspref
-    my $stemmer = Lingua::Stem->new( -locale => 'EN-US' );
-
-# FIXME: these should be stored in the db so the librarian can modify the behavior
-    $stemmer->add_exceptions(
-        {
-            'and' => 'and',
-            'or'  => 'or',
-            'not' => 'not',
-        }
-    );
-
-
+       #warn "OPERATORS: >@operators< INDEXES: >@indexes< OPERANDS: >@operands< LIMITS: >@limits< SORTS: >@sort_by<";
+       my $query = $operands[0];
 # STEP I: determine if this is a form-based / simple query or if it's complex (if complex,
 # we can't handle field weighting, stemming until a formal query parser is written
-# I'll work on this soon -- JF
-#if (!$query) { # form-based
-# check if this is a known query language query, if it is, return immediately:
+
+# check if this is a known query language query, if it is, return immediately,
+# the user is responsible for constructing valid syntax:
     if ( $query =~ /^ccl=/ ) {
         return ( undef, $', $', $', 'ccl' );
     }
@@ -547,162 +732,91 @@ sub buildQuery {
     if ( $query =~ /^pqf=/ ) {
         return ( undef, $', $', $', 'pqf' );
     }
-    if ( $query =~ /(\(|\))/ ) {    # sorry, too complex
+    if ( $query =~ /(\(|\))/ ) {    # sorry, too complex, assume CCL
         return ( undef, $query, $query, $query, 'ccl' );
     }
 
-# form-based queries are limited to non-nested a specific depth, so we can easily
+# form-based queries are limited to non-nested at a specific depth, so we can easily
 # modify the incoming query operands and indexes to do stemming and field weighting
 # Once we do so, we'll end up with a value in $query, just like if we had an
 # incoming $query from the user
     else {
-        $query = ""
-          ; # clear it out so we can populate properly with field-weighted stemmed query
-        my $previous_operand
-          ;    # a flag used to keep track if there was a previous query
-               # if there was, we can apply the current operator
+        $query = ""; # clear it out so we can populate properly with field-weighted stemmed query
+        my $previous_operand;    # a flag used to keep track if there was a previous query
+                                                       # if there was, we can apply the current operator
+               # for every operand
         for ( my $i = 0 ; $i <= @operands ; $i++ ) {
-            my $operand = $operands[$i];
-            # remove stopwords from operand : parse all stopwords & remove them (case insensitive)
-            foreach (keys %{C4::Context->stopwords}) {
-                $operand=~ s/\b$_\b//i;
-            }
-            my $index   = $indexes[$i];
-            my $stemmed_operand;
-            my $stemming      = C4::Context->parameters("Stemming")     || 0;
-            my $weight_fields = C4::Context->parameters("WeightFields") || 0;
-
-            if ( $operands[$i] ) {
-                       $operand =~ s/^(and |or |not )//i;
-
-# STEMMING FIXME: need to refine the field weighting so stemmed operands don't disrupt the query ranking
-                if ($stemming) {
-                    my @words = split( / /, $operands[$i] );
-                    my $stems = $stemmer->stem(@words);
-                    foreach my $stem (@$stems) {
-                        $stemmed_operand .= "$stem";
-                        $stemmed_operand .= "?"
-                          unless ( $stem =~ /(and$|or$|not$)/ )
-                          || ( length($stem) < 3 );
-                        $stemmed_operand .= " ";
-
-                        #warn "STEM: $stemmed_operand";
-                    }
 
-                    #$operand = $stemmed_operand;
-                }
+                       # COMBINE OPERANDS, INDEXES AND OPERATORS
+                       if ( $operands[$i] ) {
+               my $operand = $operands[$i];
+               my $index   = $indexes[$i];
 
-# FIELD WEIGHTING - This is largely experimental stuff. What I'm committing works
-# pretty well but will work much better when we have an actual query parser
-                my $weighted_query;
-                if ($weight_fields) {
-                    $weighted_query .=
-                      " rk=(";    # Specifies that we're applying rank
-                                  # keyword has different weight properties
-                    if ( ( $index =~ /kw/ ) || ( !$index ) )
-                    { # FIXME: do I need to add right-truncation in the case of stemming?
-                          # a simple way to find out if this query uses an index
-                        if ( $operand =~ /(\=|\:)/ ) {
-                            $weighted_query .= " $operand";
-                        }
-                        else {
-                            $weighted_query .=
-                              " Title-cover,ext,r1=\"$operand\""
-                              ;    # index label as exact
-                            $weighted_query .=
-                              " or ti,ext,r2=$operand";    # index as exact
-                             #$weighted_query .= " or ti,phr,r3=$operand";              # index as  phrase
-                             #$weighted_query .= " or any,ext,r4=$operand";         # index as exact
-                            $weighted_query .=
-                              " or kw,wrdl,r5=$operand";    # index as exact
-                            $weighted_query .= " or wrd,fuzzy,r9=$operand";
-                            $weighted_query .= " or wrd=$stemmed_operand"
-                              if $stemming;
-                        }
-                    }
-                    elsif ( $index =~ /au/ ) {
-                        $weighted_query .=
-                          " $index,ext,r1=$operand";    # index label as exact
-                         #$weighted_query .= " or (title-sort-az=0 or $index,startswithnt,st-word,r3=$operand #)";
-                        $weighted_query .=
-                          " or $index,phr,r3=$operand";    # index as phrase
-                        $weighted_query .= " or $index,rt,wrd,r3=$operand";
-                    }
-                    elsif ( $index =~ /ti/ ) {
-                        $weighted_query .=
-                          " Title-cover,ext,r1=$operand"; # index label as exact
-                        $weighted_query .= " or Title-series,ext,r2=$operand";
-
-                        #$weighted_query .= " or ti,ext,r2=$operand";
-                        #$weighted_query .= " or ti,phr,r3=$operand";
-                        #$weighted_query .= " or ti,wrd,r3=$operand";
-                        $weighted_query .=
-" or (title-sort-az=0 or Title-cover,startswithnt,st-word,r3=$operand #)";
-                        $weighted_query .=
-" or (title-sort-az=0 or Title-cover,phr,r6=$operand)";
-
-                        #$weighted_query .= " or Title-cover,wrd,r5=$operand";
-                        #$weighted_query .= " or ti,ext,r6=$operand";
-                        #$weighted_query .= " or ti,startswith,phr,r7=$operand";
-                        #$weighted_query .= " or ti,phr,r8=$operand";
-                        #$weighted_query .= " or ti,wrd,r9=$operand";
-
-   #$weighted_query .= " or ti,ext,r2=$operand";         # index as exact
-   #$weighted_query .= " or ti,phr,r3=$operand";              # index as  phrase
-   #$weighted_query .= " or any,ext,r4=$operand";         # index as exact
-   #$weighted_query .= " or kw,wrd,r5=$operand";         # index as exact
-                    }
-                    else {
-                        $weighted_query .=
-                          " $index,ext,r1=$operand";    # index label as exact
-                         #$weighted_query .= " or $index,ext,r2=$operand";            # index as exact
-                        $weighted_query .=
-                          " or $index,phr,r3=$operand";    # index as phrase
-                        $weighted_query .= " or $index,rt,wrd,r3=$operand";
-                        $weighted_query .=
-                          " or $index,wrd,r5=$operand"
-                          ;    # index as word right-truncated
-                        $weighted_query .= " or $index,wrd,fuzzy,r8=$operand";
-                    }
-                    $weighted_query .= ")";    # close rank specification
-                    $operand = $weighted_query;
-                }
+                               # if there's no index, don't use one, it will throw a CCL error
+                               my $index_plus;# $index_plus = "$index:" if $index;
+                               my $index_plus_comma;# $index_plus_comma="$index," if $index;
+
+                               # Remove Stopwords      
+                               $operand = _remove_stopwords($operand,$index);
+                               #warn "OP_SW: $operand";
+                               # Handle Truncation
+                               my ($nontruncated,$righttruncated,$lefttruncated,$rightlefttruncated,$regexpr);
+                               ($nontruncated,$righttruncated,$lefttruncated,$rightlefttruncated,$regexpr) = _add_truncation($operand,$index);
+                               #warn "TRUNCATION: NON:@$nontruncated RIGHT:@$righttruncated LEFT:@$lefttruncated RIGHTLEFT:@$rightlefttruncated REGEX:@$regexpr";
+
+                               # Handle Stemming
+                       my $stemmed_operand;
+                               $stemmed_operand = _build_stemmed_operand($operand) if $stemming;
+
+                               # Handle Field Weighting
+                               my $weighted_operand;
+                $weighted_operand = _build_weighted_query($operand,$stemmed_operand,$index) if $weight_fields;
+
+                               # proves we're operating in multi-leaf mode
+                               # $weighted_operand = "$weighted_operand and $weighted_operand";
+                               $operand = $weighted_operand if $weight_fields;
 
-                # only add an operator if there is a previous operand
+                # If there's a previous operand, we need to add an operator
                 if ($previous_operand) {
-                    if ( $operators[ $i - 1 ] ) {
-                        $query .= " $operators[$i-1] $index: $operand";
-                        if ( !$index ) {
-                            $human_search_desc .=
-                              "  $operators[$i-1] $operands[$i]";
-                        }
-                        else {
-                            $human_search_desc .=
-                              "  $operators[$i-1] $index: $operands[$i]";
-                        }
+                                       # user-specified operator
+                    if ( $operators[$i-1] ) {
+                                               $human_search_desc .="  $operators[$i-1] $index_plus $operands[$i]";
+                                               $query .= " $operators[$i-1] $index_plus $operand";
                     }
-
                     # the default operator is and
                     else {
-                        $query             .= " and $index: $operand";
-                        $human_search_desc .= "  and $index: $operands[$i]";
+                        $query             .= " and $index_plus $operand";
+                        $human_search_desc .= " and $index_plus $operands[$i]";
                     }
                 }
-                else {
-                    if ( !$index ) {
-                        $query             .= " $operand";
-                        $human_search_desc .= "  $operands[$i]";
-                    }
-                    else {
-                        $query             .= " $index: $operand";
-                        $human_search_desc .= "  $index: $operands[$i]";
-                    }
+                               # There's no previous operand - FIXME: completely ignoring our $query, no field weighting, no stemming
+                               # FIXME: also, doesn't preserve original order
+                else { 
+                                       # if there are terms to fit with truncation
+                                       if (scalar(@$righttruncated)+scalar(@$lefttruncated)+scalar(@$rightlefttruncated)>0){
+                                               # add the non-truncated ones first
+                                               $query.= "$index_plus @$nontruncated " if (scalar(@$nontruncated)>0);
+                                               if (scalar(@$righttruncated)>0){
+                                                       $query .= "and $index_plus_comma"."rtrn:@$righttruncated ";
+                                               }            
+                                               if (scalar(@$lefttruncated)>0){
+                                                       $query .= "and $index_plus_comma"."ltrn:@$lefttruncated ";
+                                               }            
+                                               if (scalar(@$rightlefttruncated)>0){
+                                                       $query .= "and $index_plus_comma"."rltrn:@$rightlefttruncated ";
+                                               }
+                        $query=~s/^and//; # FIXME: this is cheating :-)
+                        $human_search_desc .= $query;
+                                       } else {           
+                        $query             .= " $index_plus $operand";
+                        $human_search_desc .= " $index_plus $operands[$i]";
+                                       }            
                     $previous_operand = 1;
                 }
             }    #/if $operands
         }    # /for
     }
-
+       #warn "QUERY:".$query;
     # add limits
     my $limit_query;
     my $limit_search_desc;
@@ -711,9 +825,7 @@ sub buildQuery {
         # FIXME: not quite right yet ... will work on this soon -- JF
         my $type = $1 if $limit =~ m/([^:]+):([^:]*)/;
         if ( $limit =~ /available/ ) {
-            $limit_query .=
-" (($query and datedue=0000-00-00) or ($query and datedue=0000-00-00 not lost=1) or ($query and datedue=0000-00-00 not lost=2))";
-
+            $limit_query .= " (($query and datedue=0000-00-00) or ($query and datedue=0000-00-00 not lost=1) or ($query and datedue=0000-00-00 not lost=2))";
             #$limit_search_desc.=" and available";
         }
         elsif ( ($limit_query) && ( index( $limit_query, $type, 0 ) > 0 ) ) {
@@ -769,6 +881,7 @@ sub buildQuery {
         $limit_search_desc = " and ($limit_search_desc)" if $limit_search_desc;
 
     }
+       #warn "LIMIT: $limit_query";
     $query             .= $limit_query;
     $human_search_desc .= $limit_search_desc;
 
@@ -790,9 +903,9 @@ sub buildQuery {
     $human_search_desc =~ s/^ //g;
     my $koha_query = $query;
 
-#     warn "QUERY:".$koha_query;
-#     warn "SEARCHDESC:".$human_search_desc;
-#     warn "FEDERATED QUERY:".$federated_query;
+    #warn "QUERY:".$koha_query;
+    #warn "SEARCHDESC:".$human_search_desc;
+    #warn "FEDERATED QUERY:".$federated_query;
     return ( undef, $human_search_desc, $koha_query, $federated_query );
 }
 
@@ -827,13 +940,14 @@ sub searchResults {
     #find itemtype & itemtype image
     my %itemtypes;
     $bsth =
-      $dbh->prepare("SELECT itemtype,description,imageurl,summary FROM itemtypes");
+      $dbh->prepare("SELECT itemtype,description,imageurl,summary,notforloan FROM itemtypes");
     $bsth->execute();
     while ( my $bdata = $bsth->fetchrow_hashref ) {
         $itemtypes{ $bdata->{'itemtype'} }->{description} =
           $bdata->{'description'};
         $itemtypes{ $bdata->{'itemtype'} }->{imageurl} = $bdata->{'imageurl'};
         $itemtypes{ $bdata->{'itemtype'} }->{summary} = $bdata->{'summary'};
+        $itemtypes{ $bdata->{'itemtype'} }->{notforloan} = $bdata->{'notforloan'};
     }
 
     #search item field code
@@ -905,25 +1019,25 @@ sub searchResults {
             $summary =~ s/\n/<br>/g;
             $oldbiblio->{summary} = $summary;
         }
-        # add spans to search term in results
+        # add spans to search term in results for search term highlighting
         foreach my $term ( keys %$span_terms_hashref ) {
-
-            #warn "term: $term";
             my $old_term = $term;
             if ( length($term) > 3 ) {
-                $term =~ s/(.*=|\)|\(|\+|\.|\?)//g;
+                $term =~ s/(.*=|\)|\(|\+|\.|\?|\[|\])//g;
+                               $term =~ s/\\//g;
+                               $term =~ s/\*//g;
 
                 #FIXME: is there a better way to do this?
-                $oldbiblio->{'title'} =~ s/$term/<span class=term>$&<\/span>/gi;
+                $oldbiblio->{'title'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
                 $oldbiblio->{'subtitle'} =~
-                  s/$term/<span class=term>$&<\/span>/gi;
-
-                $oldbiblio->{'author'} =~ s/$term/<span class=term>$&<\/span>/gi;
-                $oldbiblio->{'publishercode'} =~ s/$term/<span class=term>$&<\/span>/gi;
-                $oldbiblio->{'place'} =~ s/$term/<span class=term>$&<\/span>/gi;
-                $oldbiblio->{'pages'} =~ s/$term/<span class=term>$&<\/span>/gi;
-                $oldbiblio->{'notes'} =~ s/$term/<span class=term>$&<\/span>/gi;
-                $oldbiblio->{'size'}  =~ s/$term/<span class=term>$&<\/span>/gi;
+                  s/$term/<span class=\"term\">$&<\/span>/gi;
+
+                $oldbiblio->{'author'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
+                $oldbiblio->{'publishercode'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
+                $oldbiblio->{'place'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
+                $oldbiblio->{'pages'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
+                $oldbiblio->{'notes'} =~ s/$term/<span class=\"term\">$&<\/span>/gi;
+                $oldbiblio->{'size'}  =~ s/$term/<span class=\"term\">$&<\/span>/gi;
             }
         }
 
@@ -943,113 +1057,85 @@ sub searchResults {
         my $itemlost_count    = 0;
         my $norequests        = 1;
 
+        #
+        # check the loan status of the item : 
+        # it is not stored in the MARC record, for pref (zebra reindexing)
+        # reason. Thus, we have to get the status from a specific SQL query
+        #
+        my $sth_issue = $dbh->prepare("
+            SELECT date_due,returndate 
+            FROM issues 
+            WHERE itemnumber=? AND returndate IS NULL");
+        my $items_count=scalar(@fields);
         foreach my $field (@fields) {
             my $item;
             foreach my $code ( keys %subfieldstosearch ) {
                 $item->{$code} = $field->subfield( $subfieldstosearch{$code} );
             }
+            $sth_issue->execute($item->{itemnumber});
+            $item->{due_date} = format_date($sth_issue->fetchrow);
+            $item->{onloan} = 1 if $item->{due_date};
+            # at least one item can be reserved : suppose no
+            $norequests = 1;
             if ( $item->{wthdrawn} ) {
                 $wthdrawn_count++;
-            }
-            elsif ( $item->{notforloan} == -1 ) {
-                $ordered_count++;
-                $norequests = 0;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{unavailable}=1;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{wthdrawn}=1;
             }
             elsif ( $item->{itemlost} ) {
                 $itemlost_count++;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{unavailable}=1;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{itemlost}=1;
             }
-            elsif ( ( $item->{onloan} ) && ( $item->{onloan} != '0000-00-00' ) )
+            unless ( $item->{notforloan}) {
+                # OK, this one can be issued, so at least one can be reserved
+                $norequests = 0;
+            }
+            if ( ( $item->{onloan} ) && ( $item->{onloan} != '0000-00-00' ) )
             {
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{unavailable}=1;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{onloancount} = 1;
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{due_date} = $item->{due_date};
                 $onloan_count++;
-                $norequests = 0;
             }
-            else {
-                $norequests = 0;
-                if ( $item->{'homebranch'} ) {
-                    $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{count}++;
-                }
+            if ( $item->{'homebranch'} ) {
+                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{count}++;
+            }
 
-                # Last resort
-                elsif ( $item->{'holdingbranch'} ) {
-                    $items->{ $item->{'holdingbranch'} }->{count}++;
-                }
-                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{itemcallnumber} =                $item->{itemcallnumber};
-                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{location} =                $item->{location};
-                $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{branchcode} =               $item->{homebranch};
+            # Last resort
+            elsif ( $item->{'holdingbranch'} ) {
+                $items->{ $item->{'holdingbranch'} }->{count}++;
             }
+            $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{itemcallnumber} =                $item->{itemcallnumber};
+            $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{location} =                $item->{location};
+            $items->{ $item->{'homebranch'}.'--'.$item->{'itemcallnumber'} }->{branchcode} =               $item->{homebranch};
         }    # notforloan, item level and biblioitem level
+
+        # last check for norequest : if itemtype is notforloan, it can't be reserved either, whatever the items
+        $norequests = 1 if $itemtypes{$oldbiblio->{itemtype}}->{notforloan};
+
         for my $key ( sort keys %$items ) {
             my $this_item = {
                 branchname     => $branches{$items->{$key}->{branchcode}},
                 branchcode     => $items->{$key}->{branchcode},
-                count          => $items->{$key}->{count}==1 ?"":$items->{$key}->{count},
+                count          => $items->{$key}->{count},
                 itemcallnumber => $items->{$key}->{itemcallnumber},
                 location => $items->{$key}->{location},
+                onloancount      => $items->{$key}->{onloancount},
+                due_date         => $items->{$key}->{due_date},
+                wthdrawn      => $items->{$key}->{wthdrawn},
+                lost         => $items->{$key}->{itemlost},
             };
             push @items_loop, $this_item;
         }
         $oldbiblio->{norequests}    = $norequests;
+        $oldbiblio->{items_count}    = $items_count;
         $oldbiblio->{items_loop}    = \@items_loop;
         $oldbiblio->{onloancount}   = $onloan_count;
         $oldbiblio->{wthdrawncount} = $wthdrawn_count;
         $oldbiblio->{itemlostcount} = $itemlost_count;
         $oldbiblio->{orderedcount}  = $ordered_count;
         $oldbiblio->{isbn}          =~ s/-//g; # deleting - in isbn to enable amazon content 
-        
-# FIXME
-#  Ugh ... this is ugly, I'll re-write it better above then delete it
-#     my $norequests = 1;
-#     my $noitems    = 1;
-#     if (@items) {
-#         $noitems = 0;
-#         foreach my $itm (@items) {
-#             $norequests = 0 unless $itm->{'itemnotforloan'};
-#         }
-#     }
-#     $oldbiblio->{'noitems'} = $noitems;
-#     $oldbiblio->{'norequests'} = $norequests;
-#     $oldbiblio->{'even'} = $even = not $even;
-#     $oldbiblio->{'itemcount'} = $counts{'total'};
-#     my $totalitemcounts = 0;
-#     foreach my $key (keys %counts){
-#         if ($key ne 'total'){
-#             $totalitemcounts+= $counts{$key};
-#             $oldbiblio->{'locationhash'}->{$key}=$counts{$key};
-#         }
-#     }
-#     my ($locationtext, $locationtextonly, $notavailabletext) = ('','','');
-#     foreach (sort keys %{$oldbiblio->{'locationhash'}}) {
-#         if ($_ eq 'notavailable') {
-#             $notavailabletext="Not available";
-#             my $c=$oldbiblio->{'locationhash'}->{$_};
-#             $oldbiblio->{'not-available-p'}=$c;
-#         } else {
-#             $locationtext.="$_";
-#             my $c=$oldbiblio->{'locationhash'}->{$_};
-#             if ($_ eq 'Item Lost') {
-#                 $oldbiblio->{'lost-p'} = $c;
-#             } elsif ($_ eq 'Withdrawn') {
-#                 $oldbiblio->{'withdrawn-p'} = $c;
-#             } elsif ($_ eq 'On Loan') {
-#                 $oldbiblio->{'on-loan-p'} = $c;
-#             } else {
-#                 $locationtextonly.= $_;
-#                 $locationtextonly.= " ($c)<br/> " if $totalitemcounts > 1;
-#             }
-#             if ($totalitemcounts>1) {
-#                 $locationtext.=" ($c)<br/> ";
-#             }
-#         }
-#     }
-#     if ($notavailabletext) {
-#         $locationtext.= $notavailabletext;
-#     } else {
-#         $locationtext=~s/, $//;
-#     }
-#     $oldbiblio->{'location'} = $locationtext;
-#     $oldbiblio->{'location-only'} = $locationtextonly;
-#     $oldbiblio->{'use-location-flags-p'} = 1;
-
         push( @newresults, $oldbiblio );
     }
     return @newresults;
@@ -1091,7 +1177,7 @@ sub NZgetRecords {
 sub NZanalyse {
     my ($string,$server) = @_;
     # $server contains biblioserver or authorities, depending on what we search on.
-    warn "querying : $string on $server";
+    #warn "querying : $string on $server";
     $server='biblioserver' unless $server;
     # if we have a ", replace the content to discard temporarily any and/or/not inside
     my $commacontent;
@@ -1152,44 +1238,58 @@ sub NZanalyse {
     # it's a leaf, do the real SQL query and return the result
     } else {
         $string =~  s/__X__/"$commacontent"/ if $commacontent;
-        $string =~ s/-|\.|\?|,|;|!|'|\(|\)|\[|\]|{|}|"|<|>|&|\+|\*|\// /g;
+        $string =~ s/-|\.|\?|,|;|!|'|\(|\)|\[|\]|{|}|"|&|\+|\*|\// /g;
 #         warn "leaf : $string\n";
         # parse the string in in operator/operand/value again
-        $string =~ /(.*)(=|>|>=|<|<=)(.*)/;
+        $string =~ /(.*)(>=|<=)(.*)/;
         my $left = $1;
         my $operator = $2;
         my $right = $3;
+        unless ($operator) {
+            $string =~ /(.*)(>|<|=)(.*)/;
+            $left = $1;
+            $operator = $2;
+            $right = $3;
+        }
         my $results;
         # automatic replace for short operators
-        $left='title' if $left eq 'ti';
-        $left='author' if $left eq 'au';
-        $left='publisher' if $left eq 'pb';
-        $left='subject' if $left eq 'su';
-        $left='koha-Auth-Number' if $left eq 'an';
-        $left='keyword' if $left eq 'kw';
+        $left='title' if $left =~ '^ti';
+        $left='author' if $left =~ '^au';
+        $left='publisher' if $left =~ '^pb';
+        $left='subject' if $left =~ '^su';
+        $left='koha-Auth-Number' if $left =~ '^an';
+        $left='keyword' if $left =~ '^kw';
         if ($operator) {
             #do a specific search
             my $dbh = C4::Context->dbh;
             $operator='LIKE' if $operator eq '=' and $right=~ /%/;
-            my $sth = $dbh->prepare("SELECT biblionumbers FROM nozebra WHERE server=? AND indexname=? AND value $operator ?");
+            my $sth = $dbh->prepare("SELECT biblionumbers,value FROM nozebra WHERE server=? AND indexname=? AND value $operator ?");
             warn "$left / $operator / $right\n";
             # split each word, query the DB and build the biblionumbers result
             foreach (split / /,$right) {
-                my $biblionumbers;
+                my ($biblionumbers,$value);
                 next unless $_;
-                warn "EXECUTE : $server, $left, $_";
+                warn "EXECUTE : $server, $left, $_";
                 $sth->execute($server, $left, $_);
-                while (my $line = $sth->fetchrow) {
-                    $biblionumbers .= $line;
-#                     warn "result : $line";
+                while (my ($line,$value) = $sth->fetchrow) {
+                    # if we are dealing with a numeric value, use only numeric results (in case of >=, <=, > or <)
+                    # otherwise, fill the result
+                    $biblionumbers .= $line unless ($right =~ /\d/ && $value =~ /\D/);
+                    warn "result : $value ". ($right =~ /\d/) . "==".(!$value =~ /\d/) ;#= $line";
                 }
                 # do a AND with existing list if there is one, otherwise, use the biblionumbers list as 1st result list
                 if ($results) {
                     my @leftresult = split /;/, $biblionumbers;
                     my $temp;
-                    foreach (@leftresult) {
-                        if ($results =~ "$_;") {
-                            $temp .= "$_;$_;";
+                    foreach my $entry (@leftresult) { # $_ contains biblionumber,title-weight
+                        # remove weight at the end
+                        my $cleaned = $entry;
+                        $cleaned =~ s/-\d*$//;
+                        # if the entry already in the hash, take it & increase weight
+#                         warn "===== $cleaned =====";
+                        if ($results =~ "$cleaned") {
+                            $temp .= "$entry;$entry;";
+#                             warn "INCLUDING $entry";
                         }
                     }
                     $results = $temp;
@@ -1213,15 +1313,23 @@ sub NZanalyse {
                 }
                 # do a AND with existing list if there is one, otherwise, use the biblionumbers list as 1st result list
                 if ($results) {
+#                 warn "RES for $_ = $biblionumbers";
                     my @leftresult = split /;/, $biblionumbers;
                     my $temp;
-                    foreach (@leftresult) {
-                        if ($results =~ "$_;") {
-                            $temp .= "$_;$_;";
+                    foreach my $entry (@leftresult) { # $_ contains biblionumber,title-weight
+                        # remove weight at the end
+                        my $cleaned = $entry;
+                        $cleaned =~ s/-\d*$//;
+                        # if the entry already in the hash, take it & increase weight
+#                         warn "===== $cleaned =====";
+                        if ($results =~ "$cleaned") {
+                            $temp .= "$entry;$entry;";
+#                             warn "INCLUDING $entry";
                         }
                     }
                     $results = $temp;
                 } else {
+#                 warn "NEW RES for $_ = $biblionumbers";
                     $results = $biblionumbers;
                 }
             }
@@ -1250,7 +1358,7 @@ sub NZorder {
     #
     # order by POPULARITY
     #
-    if ($ordering =~ /1=9523/) {
+    if ($ordering =~ /popularity/) {
         my %result;
         my %popularity;
         # popularity is not in MARC record, it's builded from a specific query
@@ -1269,7 +1377,7 @@ sub NZorder {
         # sort the hash and return the same structure as GetRecords (Zebra querying)
         my $result_hash;
         my $numbers=0;
-        if ($ordering eq '1=9523 >i') { # sort popularity DESC
+        if ($ordering eq 'popularity_dsc') { # sort popularity DESC
             foreach my $key (sort {$b cmp $a} (keys %popularity)) {
                 $result_hash->{'RECORDS'}[$numbers++] = $result{$popularity{$key}}->as_usmarc();
             }
@@ -1285,7 +1393,7 @@ sub NZorder {
     #
     # ORDER BY author
     #
-    } elsif ($ordering eq '1=1003 <i'){
+    } elsif ($ordering =~/author/){
         my %result;
         foreach (split /;/,$biblionumbers) {
             my ($biblionumber,$title) = split /,/,$_;
@@ -1304,7 +1412,7 @@ sub NZorder {
         # sort the hash and return the same structure as GetRecords (Zebra querying)
         my $result_hash;
         my $numbers=0;
-        if ($ordering eq '1=1003 <i') { # sort by author desc
+        if ($ordering eq 'author_za') { # sort by author desc
             foreach my $key (sort (keys %result)) {
                 $result_hash->{'RECORDS'}[$numbers++] = $result{$key}->as_usmarc();
             }
@@ -1320,7 +1428,7 @@ sub NZorder {
     #
     # ORDER BY callnumber
     #
-    } elsif ($ordering eq '1=20 <i'){
+    } elsif ($ordering =~/callnumber/){
         my %result;
         foreach (split /;/,$biblionumbers) {
             my ($biblionumber,$title) = split /,/,$_;
@@ -1340,7 +1448,7 @@ sub NZorder {
         # sort the hash and return the same structure as GetRecords (Zebra querying)
         my $result_hash;
         my $numbers=0;
-        if ($ordering eq '1=1003 <i') { # sort by title desc
+        if ($ordering eq 'call_number_dsc') { # sort by title desc
             foreach my $key (sort (keys %result)) {
                 $result_hash->{'RECORDS'}[$numbers++] = $result{$key}->as_usmarc();
             }
@@ -1353,7 +1461,7 @@ sub NZorder {
         $result_hash->{'hits'} = $numbers;
         $finalresult->{'biblioserver'} = $result_hash;
         return $finalresult;
-    } elsif ($ordering =~ /1=31/){ #pub year
+    } elsif ($ordering =~ /pubdate/){ #pub year
         my %result;
         foreach (split /;/,$biblionumbers) {
             my ($biblionumber,$title) = split /,/,$_;
@@ -1367,7 +1475,7 @@ sub NZorder {
         # sort the hash and return the same structure as GetRecords (Zebra querying)
         my $result_hash;
         my $numbers=0;
-        if ($ordering eq '1=31 <i') { # sort by pubyear desc
+        if ($ordering eq 'pubdate_dsc') { # sort by pubyear desc
             foreach my $key (sort (keys %result)) {
                 $result_hash->{'RECORDS'}[$numbers++] = $result{$key}->as_usmarc();
             }
@@ -1383,7 +1491,7 @@ sub NZorder {
     #
     # ORDER BY title
     #
-    } elsif ($ordering =~ /1=4/) { 
+    } elsif ($ordering =~ /title/) { 
         # the title is in the biblionumbers string, so we just need to build a hash, sort it and return
         my %result;
         foreach (split /;/,$biblionumbers) {
@@ -1397,7 +1505,7 @@ sub NZorder {
         # sort the hash and return the same structure as GetRecords (Zebra querying)
         my $result_hash;
         my $numbers=0;
-        if ($ordering eq '1=4 <i') { # sort by title desc
+        if ($ordering eq 'title_za') { # sort by title desc
             foreach my $key (sort (keys %result)) {
                 $result_hash->{'RECORDS'}[$numbers++] = $result{$key};
             }