trunk/t/10homer.t

#!/usr/bin/perl

use strict;
use warnings;

use Test::More tests => 5;
use blib;
use jsFind;
use Data::Dumper;

BEGIN { use_ok('jsFind'); }

my $t = new jsFind B => 100;

ok(-e "t/homer.txt", "Homer: The Odyssey preface");

my $p = 0;
my $text = '';
my %words;
my $word_count = 0;
my $max_words = 100;

my $res;

my $full_text;

ok(open(U, "t/homer.txt"), "open homer.txt");
while(<U>) {
        chomp;
        if (/^\s*$/) {
                $p++;

                $full_text .= "\n" if ($full_text);
                $full_text = "paragraph $p\n\n";

                diag "paragraph $p";

                my %usage;

                my @words = split(/\s+/,lc($text));

                foreach (@words) {
                        $usage{$_}++;
                }

                foreach my $word (@words) {

                        $words{$word} += $usage{$word};

                        $res->{$word}->{$p} = $usage{$word};

                        $t->B_search(
                                Key => $word,
                                Data => { "$p" => {
                                        t => "Ulyss paragraph $p",
                                        f => $usage{$word},
                                        },
                                },
                                Insert => 1,
                                Append => 1,
                        );

                        $word_count++;

                        $full_text .= "$word ";

                        last if ($max_words && $word_count >= $max_words);
                }

                $text = '';

                last if ($max_words && $word_count >= $max_words);
        } else {
                $text .= $_;
        }

}

my $test_data = Dumper($res);
$test_data =~ s/=>/:/gs;
$test_data =~ s/\$VAR1/var test_data/;
open(JS, "> html/test_data.js") || die "can't open test_data.js: $!";
print JS $test_data;
close(JS);

ok($test_data, "test_data saved");

#my $sum = 0;
#foreach my $w (keys %words) {
#       print STDERR "$w: $words{$w}\n";
#       $sum += $words{$w};
#}

if (open(T,"> homer_text.txt")) {
        print T $full_text;
        close(T);
}
diag "homer_text.txt created";

if (open(T,"> homer_words.txt")) {
        print T $t->to_string;
        close(T);
}
diag "homer_words.txt created";

my $total_words = scalar keys %words;

cmp_ok($t->to_jsfind("./html/homer"), '==', $total_words, " jsfind index with $total_words words");
1	dpavlin	20	#!/usr/bin/perl
2
3			use strict;
4			use warnings;
5
6			use Test::More tests => 5;
7			use blib;
8			use jsFind;
9			use Data::Dumper;
10
11			BEGIN { use_ok('jsFind'); }
12
13			my $t = new jsFind B => 100;
14
15	dpavlin	21	ok(-e "t/homer.txt", "Homer: The Odyssey preface");
16	dpavlin	20
17			my $p = 0;
18			my $text = '';
19			my %words;
20	dpavlin	23	my $word_count = 0;
21			my $max_words = 100;
22	dpavlin	20
23			my $res;
24
25	dpavlin	23	my $full_text;
26
27	dpavlin	21	ok(open(U, "t/homer.txt"), "open homer.txt");
28	dpavlin	20	while(<U>) {
29			chomp;
30			if (/^\s*$/) {
31			$p++;
32
33	dpavlin	23	$full_text .= "\n" if ($full_text);
34			$full_text = "paragraph $p\n\n";
35
36	dpavlin	20	diag "paragraph $p";
37
38			my %usage;
39	dpavlin	23
40			my @words = split(/\s+/,lc($text));
41
42			foreach (@words) {
43	dpavlin	20	$usage{$_}++;
44			}
45
46	dpavlin	23	foreach my $word (@words) {
47	dpavlin	20
48			$words{$word} += $usage{$word};
49
50			$res->{$word}->{$p} = $usage{$word};
51
52			$t->B_search(
53			Key => $word,
54			Data => { "$p" => {
55			t => "Ulyss paragraph $p",
56			f => $usage{$word},
57			},
58			},
59			Insert => 1,
60			Append => 1,
61			);
62	dpavlin	23
63			$word_count++;
64
65			$full_text .= "$word ";
66
67			last if ($max_words && $word_count >= $max_words);
68	dpavlin	20	}
69
70			$text = '';
71
72	dpavlin	23	last if ($max_words && $word_count >= $max_words);
73	dpavlin	20	} else {
74			$text .= $_;
75			}
76
77			}
78
79			my $test_data = Dumper($res);
80			$test_data =~ s/=>/:/gs;
81			$test_data =~ s/\$VAR1/var test_data/;
82			open(JS, "> html/test_data.js") \|\| die "can't open test_data.js: $!";
83			print JS $test_data;
84			close(JS);
85
86			ok($test_data, "test_data saved");
87
88			#my $sum = 0;
89			#foreach my $w (keys %words) {
90			# print STDERR "$w: $words{$w}\n";
91			# $sum += $words{$w};
92			#}
93
94	dpavlin	23	if (open(T,"> homer_text.txt")) {
95			print T $full_text;
96			close(T);
97			}
98			diag "homer_text.txt created";
99
100			if (open(T,"> homer_words.txt")) {
101			print T $t->to_string;
102			close(T);
103			}
104			diag "homer_words.txt created";
105
106	dpavlin	20	my $total_words = scalar keys %words;
107
108	dpavlin	21	cmp_ok($t->to_jsfind("./html/homer"), '==', $total_words, " jsfind index with $total_words words");