swish-e/perl/test.pl

#!/usr/local/bin/perl -w

    use strict;

=pod
    Test script for the SWISHE library

    please see perldoc README-PERL for more information
    $Id: test.pl,v 1.13 2002/08/22 22:58:38 whmoseley Exp $
=cut

    # Import symbols from the SWISHE.pm module
    use SWISHE;


    # In this test we will use the same index twice
    # The results will seem odd since normally you would use
    # two different index files, but it demonstrates
    # how to process two index files.
    
    my $indexfilename1 = '../tests/test.index';
    my $indexfilename2 = $indexfilename1;

    die "Index file '$indexfilename1' not found!  Did you run make test from the top level directory?\n"
        unless -e $indexfilename1;

    my $indexfiles = $indexfilename1;        


    # To search for several indexes just put them together
    # Uncomment to test/demonstrate the use of two index files
    
    #my $indexfiles = "$indexfilename1 $indexfilename2";


    # First, open the index files. 
    # This reads in headers and prepares the index for searching
    # You can run more than one query once the index is opened.

    
    my $handle = SwishOpen( $indexfiles )
        or die "Failed to SwishOpen '$indexfiles'";

    # Get a few headers from the index files for display

    my @headers = ( qw/WordCharacters BeginCharacters EndCharacters IndexedOn FileCount FuzzyIndexingMode/ );
  
    for my $header ( @headers ) {
        print_header("Header '$header'");

        my @h =  SwishHeaderParameter( $handle, $header );

        print "$header for index $_ is $h[$_]\n" for 0..$#h;
    }


    # Now, let's run a few queries...

    # Define a few searches


    my @searches = (
        {
            title   => 'Normal search',
            query   => 'test',
            props   => '',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'MetaTag search 1',
            query   => 'meta1=metatest1',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'MetaTag search 2',
            query   => 'meta2=metatest2',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'XML Search',
            query   => 'meta3=metatest3',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'Phrase Search',
            query   => '"three little pigs"',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'Advanced search',
            query   => 'test or meta1=m* or meta2=m* or meta3=m*',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'Advanced search',
            query   => 'test or meta1=m* or meta2=m* or meta3=m*',
            props   => 'meta1 meta2 meta3',
            sort    => '',
            context => 1,   # Search the entire file
        },
        {
            title   => 'Limit to title',
            query   => 'test or meta1=m* or meta2=m* or meta3=m*',
            props   => 'meta1 meta2 meta3 swishrank swishdocpath swishlastmodified',
            sort    => '',
            context => 1,   # Search the entire file
            limit   => [ 'swishtitle', '<=', 'If you are seeing this, the test' ],
        },
        {
            title   => 'Limit to title - second test with same query',
            query   => 'test or meta1=m* or meta2=m* or meta3=m*',
            props   => 'meta1 meta2 meta3 swishrank swishdocpath swishlastmodified',
            sort    => '',
            context => 1,   # Search the entire file
            limit   => [ 'swishtitle', '<=', 'If you are seeing this, the test' ],
        },
    );

    # Need an array in perl to deliver the above hash contents to swish in
    # the correct order
    my @settings = qw/query context props sort/;


    print_header("*** Now searching ****");
    print "Note that some META names have embedded newlines.\n";


    # Use an array for a hash slice when reading results.  See SwishNext below.
    my @labels = qw/
        rank
        file_name
        title
        content_length
    /;


    for my $search ( @searches ) {
        print_header( "$search->{title} - Query: '$search->{query}'" );


        # Since we *might* use SetLimitParameter, make sure it's reset first
        ClearLimitParameter( $handle );
        
        if ( $search->{limit} ) {
            print "limiting to @{$search->{limit}}\n\n";
            SetLimitParameter( $handle, ,@{$search->{limit}});
        }

        # Here's the actual query
        
        my $num_results = SwishSearch( $handle, @{$search}{ @settings } );

        print "# Number of results = $num_results\n\n";

        if ( $num_results <= 0 ) {
            print ($num_results ? SwishErrorString( $num_results ) : 'No Results');

            my $error = SwishError( $handle );
            print "\nError number: $error\n" if $error;

            next;
        }

        my %result;
        my @properties = split /\s+/, $search->{props};
        my %props;

        while ( ( @result{ @labels }, @props{@properties} ) = SwishNext( $handle )) {

            for ( @labels ) {
                printf("  %20s -> '%s'\n", $_ ,$result{$_});
            }
            for ( @properties ) {
                printf("  %20s:(%-20s) -> '%s'\n", 'Property', $_, $props{$_} || '<blank>' );
            }
            print "-----------\n";
        }
    }

    print_header('Other Functions');


    # Now, demonstrate the use of SwishStem to find the stem of a word.

    my @stemwords = qw/parking libaries library librarians money monies running runs is/;
    print "\nStemming:\n";
    print "    '$_' => '" . ( SwishStem( $_ ) || 'returned undefined: Word not stemmed for some reason' ) . "'\n" for @stemwords;
    print "\n";


    # Grab the stop words from the header

    my @stopwords = SwishStopWords( $handle, $indexfilename1 );
    print 'Stopwords: ',
          ( @stopwords ? join(', ', @stopwords) : '** None **' ),
          "\n\n";


    # Let's see what words in the index begin with the letter "t".

    my $letter = 't';
    my @keywords = SwishWords( $handle, $indexfilename1, $letter);

    print "List of keywords that start with the letter '$letter':\n",
          join("\n", @keywords),
          "\n\n";


    # Free the memory.

    SwishClose( $handle );

    # If swish was built with memory debugging this will dump extra info
    SWISHE::MemSummary();


sub print_header {
    print "\n", '-' x length( $_[0] ),"\n",
          $_[0],
          "\n", '-' x length( $_[0] ),"\n";
}


1	adcroft	1.1	#!/usr/local/bin/perl -w
2
3			use strict;
4
5			=pod
6			Test script for the SWISHE library
7
8			please see perldoc README-PERL for more information
9			$Id: test.pl,v 1.13 2002/08/22 22:58:38 whmoseley Exp $
10			=cut
11
12			# Import symbols from the SWISHE.pm module
13			use SWISHE;
14
15
16			# In this test we will use the same index twice
17			# The results will seem odd since normally you would use
18			# two different index files, but it demonstrates
19			# how to process two index files.
20
21			my $indexfilename1 = '../tests/test.index';
22			my $indexfilename2 = $indexfilename1;
23
24			die "Index file '$indexfilename1' not found! Did you run make test from the top level directory?\n"
25			unless -e $indexfilename1;
26
27			my $indexfiles = $indexfilename1;
28
29
30			# To search for several indexes just put them together
31			# Uncomment to test/demonstrate the use of two index files
32
33			#my $indexfiles = "$indexfilename1 $indexfilename2";
34
35
36			# First, open the index files.
37			# This reads in headers and prepares the index for searching
38			# You can run more than one query once the index is opened.
39
40
41			my $handle = SwishOpen( $indexfiles )
42			or die "Failed to SwishOpen '$indexfiles'";
43
44			# Get a few headers from the index files for display
45
46			my @headers = ( qw/WordCharacters BeginCharacters EndCharacters IndexedOn FileCount FuzzyIndexingMode/ );
47
48			for my $header ( @headers ) {
49			print_header("Header '$header'");
50
51			my @h = SwishHeaderParameter( $handle, $header );
52
53			print "$header for index $_ is $h[$_]\n" for 0..$#h;
54			}
55
56
57			# Now, let's run a few queries...
58
59			# Define a few searches
60
61
62
63			my @searches = (
64			{
65			title => 'Normal search',
66			query => 'test',
67			props => '',
68			sort => '',
69			context => 1, # Search the entire file
70			},
71			{
72			title => 'MetaTag search 1',
73			query => 'meta1=metatest1',
74			props => 'meta1 meta2 meta3',
75			sort => '',
76			context => 1, # Search the entire file
77			},
78			{
79			title => 'MetaTag search 2',
80			query => 'meta2=metatest2',
81			props => 'meta1 meta2 meta3',
82			sort => '',
83			context => 1, # Search the entire file
84			},
85			{
86			title => 'XML Search',
87			query => 'meta3=metatest3',
88			props => 'meta1 meta2 meta3',
89			sort => '',
90			context => 1, # Search the entire file
91			},
92			{
93			title => 'Phrase Search',
94			query => '"three little pigs"',
95			props => 'meta1 meta2 meta3',
96			sort => '',
97			context => 1, # Search the entire file
98			},
99			{
100			title => 'Advanced search',
101			query => 'test or meta1=m* or meta2=m* or meta3=m*',
102			props => 'meta1 meta2 meta3',
103			sort => '',
104			context => 1, # Search the entire file
105			},
106			{
107			title => 'Advanced search',
108			query => 'test or meta1=m* or meta2=m* or meta3=m*',
109			props => 'meta1 meta2 meta3',
110			sort => '',
111			context => 1, # Search the entire file
112			},
113			{
114			title => 'Limit to title',
115			query => 'test or meta1=m* or meta2=m* or meta3=m*',
116			props => 'meta1 meta2 meta3 swishrank swishdocpath swishlastmodified',
117			sort => '',
118			context => 1, # Search the entire file
119			limit => [ 'swishtitle', '<=', 'If you are seeing this, the test' ],
120			},
121			{
122			title => 'Limit to title - second test with same query',
123			query => 'test or meta1=m* or meta2=m* or meta3=m*',
124			props => 'meta1 meta2 meta3 swishrank swishdocpath swishlastmodified',
125			sort => '',
126			context => 1, # Search the entire file
127			limit => [ 'swishtitle', '<=', 'If you are seeing this, the test' ],
128			},
129			);
130
131			# Need an array in perl to deliver the above hash contents to swish in
132			# the correct order
133			my @settings = qw/query context props sort/;
134
135
136
137			print_header("* Now searching **");
138			print "Note that some META names have embedded newlines.\n";
139
140
141			# Use an array for a hash slice when reading results. See SwishNext below.
142			my @labels = qw/
143			rank
144			file_name
145			title
146			content_length
147			/;
148
149
150			for my $search ( @searches ) {
151			print_header( "$search->{title} - Query: '$search->{query}'" );
152
153
154			# Since we might use SetLimitParameter, make sure it's reset first
155			ClearLimitParameter( $handle );
156
157			if ( $search->{limit} ) {
158			print "limiting to @{$search->{limit}}\n\n";
159			SetLimitParameter( $handle, ,@{$search->{limit}});
160			}
161
162			# Here's the actual query
163
164			my $num_results = SwishSearch( $handle, @{$search}{ @settings } );
165
166			print "# Number of results = $num_results\n\n";
167
168			if ( $num_results <= 0 ) {
169			print ($num_results ? SwishErrorString( $num_results ) : 'No Results');
170
171			my $error = SwishError( $handle );
172			print "\nError number: $error\n" if $error;
173
174			next;
175			}
176
177			my %result;
178			my @properties = split /\s+/, $search->{props};
179			my %props;
180
181			while ( ( @result{ @labels }, @props{@properties} ) = SwishNext( $handle )) {
182
183			for ( @labels ) {
184			printf(" %20s -> '%s'\n", $_ ,$result{$_});
185			}
186			for ( @properties ) {
187			printf(" %20s:(%-20s) -> '%s'\n", 'Property', $_, $props{$_} \|\| '<blank>' );
188			}
189			print "-----------\n";
190			}
191			}
192
193			print_header('Other Functions');
194
195
196			# Now, demonstrate the use of SwishStem to find the stem of a word.
197
198			my @stemwords = qw/parking libaries library librarians money monies running runs is/;
199			print "\nStemming:\n";
200			print " '$_' => '" . ( SwishStem( $_ ) \|\| 'returned undefined: Word not stemmed for some reason' ) . "'\n" for @stemwords;
201			print "\n";
202
203
204
205			# Grab the stop words from the header
206
207			my @stopwords = SwishStopWords( $handle, $indexfilename1 );
208			print 'Stopwords: ',
209			( @stopwords ? join(', ', @stopwords) : ' None ' ),
210			"\n\n";
211
212
213			# Let's see what words in the index begin with the letter "t".
214
215			my $letter = 't';
216			my @keywords = SwishWords( $handle, $indexfilename1, $letter);
217
218			print "List of keywords that start with the letter '$letter':\n",
219			join("\n", @keywords),
220			"\n\n";
221
222
223
224			# Free the memory.
225
226			SwishClose( $handle );
227
228			# If swish was built with memory debugging this will dump extra info
229			SWISHE::MemSummary();
230
231
232			sub print_header {
233			print "\n", '-' x length( $_[0] ),"\n",
234			$_[0],
235			"\n", '-' x length( $_[0] ),"\n";
236			}
237
238