Attachment #1142 for bug #3172

View | Details | Raw Unified | Return to bug 3172
Collapse All | Expand All




    require Exporter;
    @ISA    = qw(Exporter);
	@EXPORT_OK = qw(&output_ajax_with_http_headers &is_ajax); # More stuff should go here instead
	%EXPORT_TAGS = ( all =>[qw(&themelanguage &gettemplate setlanguagecookie getlanguagecookie pagination_bar
								&output_ajax_with_http_headers &output_html_with_http_headers)],
					ajax =>[qw(&output_ajax_with_http_headers is_ajax)],
					html =>[qw(&output_html_with_http_headers)]
				);
    push @EXPORT, qw(
        &themelanguage &gettemplate setlanguagecookie getlanguagecookie pagination_bar
    );
    push @EXPORT, qw(
        &output_html_with_http_headers

    );
}

sub getlanguagecookie {
    my ($query) = @_;
    my $lang;
    if ($query->cookie('KohaOpacLanguage')){
        $lang = $query->cookie('KohaOpacLanguage') ;
    }else{
        $lang = $ENV{HTTP_ACCEPT_LANGUAGE};
        
    }
    $lang = substr($lang, 0, 2);

    return $lang;
}

=item pagination_bar

   pagination_bar($base_url, $nb_pages, $current_page, $startfrom_name)




use C4::Context;
use C4::Biblio;    # GetMarcFromKohaField
use C4::Koha;      # getFacets
use Lingua::Stem;
use C4::Search::PazPar2;
use XML::Simple;
use C4::Dates qw(format_date);
use C4::XSLT;
use C4::Branch;
require Lingua::Stem::Snowball;

use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $DEBUG);



# STEMMING
sub _build_stemmed_operand {
    my ($operand, $lang) = @_;
    my $stemmed_operand;

    # If operand contains a digit, it is almost certainly an identifier, and should

    # "014100018X" to "x ", which for a MARC21 database would bring up irrelevant
    # results (e.g., "23 x 29 cm." from the 300$c).  Bug 2098.
    return $operand if $operand =~ /\d/;
    
    my $stemmer = Lingua::Stem::Snowball->new( lang => $lang,
                                               encoding => "UTF-8" );

# 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 );
    
    for my $word (@words) {
        my $stem = $stemmer->stem($word);
        $stemmed_operand .= "$stem";
        $stemmed_operand .= "?"
          unless ( $stem =~ /(and$|or$|not$)/ ) || ( length($stem) < 3 );

=cut

sub buildQuery {
    my ( $operators, $operands, $indexes, $limits, $sort_by, $scan, $lang ) = @_;

    warn "---------\nEnter buildQuery\n---------" if $DEBUG;



                # Handle Stemming
                my $stemmed_operand;
                $stemmed_operand = _build_stemmed_operand($operand, $lang)
                  if $stemming;
                warn "STEMMED OPERAND: >$stemmed_operand<" if $DEBUG;


Lines 562-568 WriteMakefile( Link Here

(-)a/Makefile.PL (-1 / +1 lines)
562	'JSON' => 2.07, # Needed by admin/item_circulation_alerts.pl	562	'JSON' => 2.07, # Needed by admin/item_circulation_alerts.pl
563	'LWP::Simple' => 1.41,	563	'LWP::Simple' => 1.41,
564	'LWP::UserAgent' => 2.033,	564	'LWP::UserAgent' => 2.033,
565	'Lingua::Stem' => 0.82,	565	'Lingua::Stem::Snowball' => 0.952, # optional
566	'List::Util' => 1.18,	566	'List::Util' => 1.18,
567	'List::MoreUtils' => 0.21,	567	'List::MoreUtils' => 0.21,
568	'Locale::Language' => 2.07,	568	'Locale::Language' => 2.07,

Lines 94-100 HTML::Scrubber Link Here

(-)a/about.pl (-1 / +1 lines)
94	JSON	94	JSON
95	LWP::Simple	95	LWP::Simple
96	LWP::UserAgent	96	LWP::UserAgent
97	Lingua::Stem	97	Lingua::Stem::Snowball
98	List::Util	98	List::Util
99	List::MoreUtils	99	List::MoreUtils
100	Locale::Language	100	Locale::Language

Lines 419-425 my ( $error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit Link Here

(-)a/catalogue/search.pl (-1 / +2 lines)
419	my @results;	419	my @results;
420		420
421	## I. BUILD THE QUERY	421	## I. BUILD THE QUERY
422	( $error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit_desc,$stopwords_removed,$query_type) = buildQuery(\@operators,\@operands,\@indexes,\@limits,\@sort_by,$scan);	422	my $lang = C4::Output::getlanguagecookie($cgi);
		423	( $error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit_desc,$stopwords_removed,$query_type) = buildQuery(\@operators,\@operands,\@indexes,\@limits,\@sort_by,$scan,$lang);
423		424
424	## parse the query_cgi string and put it into a form suitable for <input>s	425	## parse the query_cgi string and put it into a form suitable for <input>s
425	my @query_inputs;	426	my @query_inputs;

Lines 334-340 my ($error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit_ Link Here

(-)a/opac/opac-search.pl (-2 / +2 lines)
334	my @results;	334	my @results;
335		335
336	## I. BUILD THE QUERY	336	## I. BUILD THE QUERY
337	( $error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit_desc,$stopwords_removed,$query_type) = buildQuery(\@operators,\@operands,\@indexes,\@limits,\@sort_by);	337	my $lang = C4::Output::getlanguagecookie($cgi);
		338	( $error,$query,$simple_query,$query_cgi,$query_desc,$limit,$limit_cgi,$limit_desc,$stopwords_removed,$query_type) = buildQuery(\@operators,\@operands,\@indexes,\@limits,\@sort_by, 0, $lang);
338		339
339	sub _input_cgi_parse ($) {	340	sub _input_cgi_parse ($) {
340	my @elements;	341	my @elements;
341	-

Return to bug 3172

Lines 21-32 require Exporter; Link Here

(-)a/C4/Search.pm (-17 / +10 lines)
21	use C4::Context;	21	use C4::Context;
22	use C4::Biblio; # GetMarcFromKohaField	22	use C4::Biblio; # GetMarcFromKohaField
23	use C4::Koha; # getFacets	23	use C4::Koha; # getFacets
24	use Lingua::Stem;
25	use C4::Search::PazPar2;	24	use C4::Search::PazPar2;
26	use XML::Simple;	25	use XML::Simple;
27	use C4::Dates qw(format_date);	26	use C4::Dates qw(format_date);
28	use C4::XSLT;	27	use C4::XSLT;
29	use C4::Branch;	28	use C4::Branch;
		29	require Lingua::Stem::Snowball;
30		30
31	use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $DEBUG);	31	use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS $DEBUG);
32		32
Lines 736-742 sub _detect_truncation { Link Here
736		736
737	# STEMMING	737	# STEMMING
738	sub _build_stemmed_operand {	738	sub _build_stemmed_operand {
739	my ($operand) = @_;	739	my ($operand, $lang) = @_;
740	my $stemmed_operand;	740	my $stemmed_operand;
741		741
742	# If operand contains a digit, it is almost certainly an identifier, and should	742	# If operand contains a digit, it is almost certainly an identifier, and should
Lines 745-765 sub _build_stemmed_operand { Link Here
745	# "014100018X" to "x ", which for a MARC21 database would bring up irrelevant	745	# "014100018X" to "x ", which for a MARC21 database would bring up irrelevant
746	# results (e.g., "23 x 29 cm." from the 300$c). Bug 2098.	746	# results (e.g., "23 x 29 cm." from the 300$c). Bug 2098.
747	return $operand if $operand =~ /\d/;	747	return $operand if $operand =~ /\d/;
		748
		749	my $stemmer = Lingua::Stem::Snowball->new( lang => $lang,
		750	encoding => "UTF-8" );
748		751
749	# FIXME: the locale should be set based on the user's language and/or search choice
750	my $stemmer = Lingua::Stem->new( -locale => 'EN-US' );
751
752	# FIXME: these should be stored in the db so the librarian can modify the behavior
753	$stemmer->add_exceptions(
754	{
755	'and' => 'and',
756	'or' => 'or',
757	'not' => 'not',
758	}
759	);
760	my @words = split( / /, $operand );	752	my @words = split( / /, $operand );
761	my $stems = $stemmer->stem(@words);	753
762	for my $stem (@$stems) {	754	for my $word (@words) {
		755	my $stem = $stemmer->stem($word);
763	$stemmed_operand .= "$stem";	756	$stemmed_operand .= "$stem";
764	$stemmed_operand .= "?"	757	$stemmed_operand .= "?"
765	unless ( $stem =~ /(and$\|or$\|not$)/ ) \|\| ( length($stem) < 3 );	758	unless ( $stem =~ /(and$\|or$\|not$)/ ) \|\| ( length($stem) < 3 );
Lines 847-853 See verbose embedded documentation. Link Here
847	=cut	840	=cut
848		841
849	sub buildQuery {	842	sub buildQuery {
850	my ( $operators, $operands, $indexes, $limits, $sort_by, $scan ) = @_;	843	my ( $operators, $operands, $indexes, $limits, $sort_by, $scan, $lang ) = @_;
851		844
852	warn "---------\nEnter buildQuery\n---------" if $DEBUG;	845	warn "---------\nEnter buildQuery\n---------" if $DEBUG;
853		846
Lines 1038-1044 sub buildQuery { Link Here
1038		1031
1039	# Handle Stemming	1032	# Handle Stemming
1040	my $stemmed_operand;	1033	my $stemmed_operand;
1041	$stemmed_operand = _build_stemmed_operand($operand)	1034	$stemmed_operand = _build_stemmed_operand($operand, $lang)
1042	if $stemming;	1035	if $stemming;
1043	warn "STEMMED OPERAND: >$stemmed_operand<" if $DEBUG;	1036	warn "STEMMED OPERAND: >$stemmed_operand<" if $DEBUG;
1044		1037

Lines 39-51 BEGIN { Link Here

(-)a/C4/Output.pm (-2 / +16 lines)
39	require Exporter;	39	require Exporter;
40	@ISA = qw(Exporter);	40	@ISA = qw(Exporter);
41	@EXPORT_OK = qw(&output_ajax_with_http_headers &is_ajax); # More stuff should go here instead	41	@EXPORT_OK = qw(&output_ajax_with_http_headers &is_ajax); # More stuff should go here instead
42	%EXPORT_TAGS = ( all =>[qw(&themelanguage &gettemplate setlanguagecookie pagination_bar	42	%EXPORT_TAGS = ( all =>[qw(&themelanguage &gettemplate setlanguagecookie getlanguagecookie pagination_bar
43	&output_ajax_with_http_headers &output_html_with_http_headers)],	43	&output_ajax_with_http_headers &output_html_with_http_headers)],
44	ajax =>[qw(&output_ajax_with_http_headers is_ajax)],	44	ajax =>[qw(&output_ajax_with_http_headers is_ajax)],
45	html =>[qw(&output_html_with_http_headers)]	45	html =>[qw(&output_html_with_http_headers)]
46	);	46	);
47	push @EXPORT, qw(	47	push @EXPORT, qw(
48	&themelanguage &gettemplate setlanguagecookie pagination_bar	48	&themelanguage &gettemplate setlanguagecookie getlanguagecookie pagination_bar
49	);	49	);
50	push @EXPORT, qw(	50	push @EXPORT, qw(
51	&output_html_with_http_headers	51	&output_html_with_http_headers
Lines 208-213 sub setlanguagecookie { Link Here
208	);	208	);
209	}	209	}
210		210
		211	sub getlanguagecookie {
		212	my ($query) = @_;
		213	my $lang;
		214	if ($query->cookie('KohaOpacLanguage')){
		215	$lang = $query->cookie('KohaOpacLanguage') ;
		216	}else{
		217	$lang = $ENV{HTTP_ACCEPT_LANGUAGE};
		218
		219	}
		220	$lang = substr($lang, 0, 2);
		221
		222	return $lang;
		223	}
		224
211	=item pagination_bar	225	=item pagination_bar
212		226
213	pagination_bar($base_url, $nb_pages, $current_page, $startfrom_name)	227	pagination_bar($base_url, $nb_pages, $current_page, $startfrom_name)