trunk/t/10homer.t

#!/usr/bin/perl

use strict;
use warnings;

use Test::More tests => 7;
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 $line = 0;
my $text = '';
my %words_usage;
my $word_count = 0;
my $max_words;
#$max_words = 100;

my $res;

my $full_text;

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

        $full_text = "$line: ";

        my %usage;

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

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

        foreach my $word (@words) {

                next if ($word eq '');

                $words_usage{"$word $line"} = $usage{$word};

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

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

                $word_count++;

                $full_text .= "$word ";

        }

        $full_text = "\n";

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

my $test_data = Dumper($res);
$test_data =~ s/=>/:/gs;
$test_data =~ s/\$VAR1/var test_data/;
ok(open(JS, "> html/test_data.js"), "test_data.js");
print JS $test_data;
close(JS);

ok($test_data, "test_data saved");

my $sum = 0;
ok(open(TD, "> homer_freq.txt"), "homer_freq.txt");
foreach my $w (keys %words_usage) {
        print TD "$w: $words_usage{$w}\n";
        $sum += $words_usage{$w};
}
close(TD);
diag "homer_freq.txt created";

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_usage;

cmp_ok($t->to_jsfind(dir=>"./html/homer"), '==', $total_words, " jsfind index with $total_words words");

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