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 => 200;

my $file = shift @ARGV || 't/homer.txt';

ok(-e $file, "reading input file $file");

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, $file), "open $file");
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	37	my $t = new jsFind B => 200;
14	dpavlin	20
15	dpavlin	37	my $file = shift @ARGV \|\| 't/homer.txt';
16	dpavlin	20
17	dpavlin	37	ok(-e $file, "reading input file $file");
18
19	dpavlin	25	my $line = 0;
20	dpavlin	20	my $text = '';
21	dpavlin	28	my %words_usage;
22	dpavlin	23	my $word_count = 0;
23	dpavlin	28	my $max_words;
24			#$max_words = 100;
25	dpavlin	20
26			my $res;
27
28	dpavlin	23	my $full_text;
29
30	dpavlin	37	ok(open(U, $file), "open $file");
31	dpavlin	20	while(<U>) {
32			chomp;
33	dpavlin	25	$line++;
34			next if (/^\s*$/);
35	dpavlin	20
36	dpavlin	25	$full_text = "$line: ";
37	dpavlin	23
38	dpavlin	25	my %usage;
39	dpavlin	20
40	dpavlin	25	my @words = split(/\s+/,lc($_));
41	dpavlin	23
42	dpavlin	25	foreach (@words) {
43			$usage{$_}++;
44			}
45	dpavlin	23
46	dpavlin	25	foreach my $word (@words) {
47	dpavlin	20
48	dpavlin	25	next if ($word eq '');
49	dpavlin	20
50	dpavlin	28	$words_usage{"$word $line"} = $usage{$word};
51	dpavlin	20
52	dpavlin	25	$res->{$word}->{$line} = $usage{$word};
53	dpavlin	20
54	dpavlin	25	$t->B_search(
55			Key => $word,
56			Data => { "$line" => {
57			t => "Odyssey line $line",
58			f => $usage{$word},
59	dpavlin	20	},
60	dpavlin	25	},
61			Insert => 1,
62			Append => 1,
63			);
64	dpavlin	23
65	dpavlin	25	$word_count++;
66	dpavlin	23
67	dpavlin	25	$full_text .= "$word ";
68	dpavlin	23
69	dpavlin	25	}
70	dpavlin	20
71	dpavlin	25	$full_text = "\n";
72	dpavlin	20
73	dpavlin	25	last if ($max_words && $word_count >= $max_words);
74	dpavlin	20	}
75
76			my $test_data = Dumper($res);
77			$test_data =~ s/=>/:/gs;
78			$test_data =~ s/\$VAR1/var test_data/;
79	dpavlin	28	ok(open(JS, "> html/test_data.js"), "test_data.js");
80	dpavlin	20	print JS $test_data;
81			close(JS);
82
83			ok($test_data, "test_data saved");
84
85	dpavlin	28	my $sum = 0;
86			ok(open(TD, "> homer_freq.txt"), "homer_freq.txt");
87			foreach my $w (keys %words_usage) {
88			print TD "$w: $words_usage{$w}\n";
89			$sum += $words_usage{$w};
90			}
91			close(TD);
92			diag "homer_freq.txt created";
93	dpavlin	20
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	28	my $total_words = scalar keys %words_usage;
107	dpavlin	20
108	dpavlin	35	cmp_ok($t->to_jsfind(dir=>"./html/homer"), '==', $total_words, " jsfind index with $total_words words");
109	dpavlin	25
110	dpavlin	28	#print Dumper($t);