File Coverage

blib/lib/WWW/Scraper/ISBN/AmazonUK_Driver.pm

Criterion	Covered	Total	%
statement	88	95	92.6
branch	32	44	72.7
condition	10	24	41.6
subroutine	9	9	100.0
pod	1	1	100.0
total	140	173	80.9

line	stmt	bran	cond	sub	pod	time	code
1							package WWW::Scraper::ISBN::AmazonUK_Driver;
2
3	6			6		122940	use strict;
	6					20
	6					342
4	6			6		42	use warnings;
	6					19
	6					272
5
6	6			6		72	use vars qw($VERSION);
	6					14
	6					548
7							$VERSION = '0.40';
8
9							#--------------------------------------------------------------------------
10
11							=head1 NAME
12
13							WWW::Scraper::ISBN::AmazonUK_Driver - Search driver for Amazon.co.uk
14
15							=head1 SYNOPSIS
16
17							See parent class documentation (L)
18
19							=head1 DESCRIPTION
20
21							Searches for book information from the (UK) Amazon online catalog.
22
23							=cut
24
25							#--------------------------------------------------------------------------
26
27							###########################################################################
28							# Inheritence
29
30	6			6		35	use base qw(WWW::Scraper::ISBN::Driver);
	6					21
	6					2016
31
32							###########################################################################
33							# Modules
34
35	6			6		4729	use WWW::Mechanize;
	6					498765
	6					200
36	6			6		1322	use JSON;
	6					13377
	6					69
37
38							###########################################################################
39							# Variables
40
41							my $AMA_SEARCH = 'http://www.amazon.co.uk/s/ref=nb_sb_noss?url=search-alias%3Daps&x=18&y=16&field-keywords=';
42							my $AMA_URL = 'http://www.amazon.co.uk/[^/]+/dp/[\dX]+/ref=sr_1_1/';
43							my $IN2MM = 0.0393700787; # number of inches in a millimetre (mm)
44							my $LB2G = 0.00220462; # number of pounds (lbs) in a gram
45							my $OZ2G = 0.035274; # number of ounces (oz) in a gram
46
47							#--------------------------------------------------------------------------
48
49							###########################################################################
50							# Public Interface
51
52							=head1 METHODS
53
54							=over 4
55
56							=item C
57
58							Creates a query string, then passes the appropriate form fields to the
59							Amazon (UK) server.
60
61							The returned page should be the correct catalog page for that ISBN. If not the
62							function returns zero and allows the next driver in the chain to have a go. If
63							a valid page is returned, the following fields are returned via the book hash:
64
65							isbn (now returns isbn13)
66							isbn10
67							isbn13
68							ean13 (industry name)
69							author
70							title
71							book_link
72							thumb_link
73							image_link
74							pubdate
75							publisher
76							binding (if known)
77							pages (if known)
78							weight (if known) (in grams)
79							width (if known) (in millimetres)
80							height (if known) (in millimetres)
81							depth (if known) (in millimetres)
82
83							The book_link, thumb_link and image_link refer back to the Amazon (UK) website.
84
85							=back
86
87							=cut
88
89							sub search {
90	3			3	1	23901	my $self = shift;
91	3					8	my $isbn = shift;
92	3					17	$self->found(0);
93	3					59	$self->book(undef);
94
95							# validate and convert into EAN13 format
96	3					37	my $ean = $self->convert_to_ean13($isbn);
97	3	50	66			157	return $self->handler("Invalid ISBN specified [$isbn]")
			33
			66
			33
98							if(!$ean \|\| (length $isbn == 13 && $isbn ne $ean)
99							\|\| (length $isbn == 10 && $isbn ne $self->convert_to_isbn10($ean)));
100
101	3					66	my $mech = WWW::Mechanize->new();
102	3					19729	$mech->agent_alias( 'Linux Mozilla' );
103
104	3					206	my $search = $AMA_SEARCH . $ean;
105
106	3					7	eval { $mech->get( $search ) };
	3					14
107	3	50	33			1487234	return $self->handler("Amazon UK website appears to be unavailable.")
			33
108							if($@ \|\| !$mech->success() \|\| !$mech->content());
109
110	3					213	my $content = $mech->content();
111							#print STDERR "\n# content=[$content]\n";
112	3					428	my ($link) = $content =~ m!($AMA_URL)!s;
113	3	50				13	return $self->handler("Failed to find that book on Amazon UK website.")
114							unless($link);
115
116	3					8	eval { $mech->get( $link ) };
	3					16
117	3	50	33			6396971	return $self->handler("Amazon UK website appears to be unavailable.")
			33
118							if($@ \|\| !$mech->success() \|\| !$mech->content());
119
120	3					583	return $self->_parse($mech);
121							}
122
123							sub _parse {
124	5			5		1503	my $self = shift;
125	5					11	my $mech = shift;
126
127							# The Book page
128	5					20	my $html = $mech->content;
129	5					538	my $data = {};
130
131							#print STDERR "\n# html=[$html]\n";
132
133	5			1		94454	my @size = $html =~ m! \sProduct Dimensions:\s\s([\d.]+) x ([\d.]+) x ([\d.]+) (cm)\s !si;
	1					21
	1					3
	1					27
134	5	100				31319	@size = $html =~ m! \sProduct Dimensions:\s\s([\d.]+) x ([\d.]+) x ([\d.]+) (inches)\s !si unless(@size);
135	5	100				24	if(@size) {
136	4					12	my $type = pop @size;
137	4					53	($data->{depth},$data->{width},$data->{height}) = sort @size;
138	4	50				23	if($type eq 'cm') {
		0
139	4					57	$data->{$_} = int($data->{$_} * 10) for(qw( height width depth ));
140							} elsif($type eq 'inches') {
141	0					0	$data->{$_} = int($data->{$_} / $IN2MM) for(qw( height width depth ));
142							}
143							}
144
145	5					23891	($data->{binding},$data->{pages}) = $html =~ m! (Paperback\|Hardcover):\s([\d.]+)\spages !si;
146	5					41319	($data->{weight}) = $html =~ m! Shipping Weight:\s([\d.]+)\sounces !si;
147	5					23814	($data->{published}) = $html =~ m! Publisher:\s(.?) !si;
148	5					23835	($data->{isbn10}) = $html =~ m! ISBN-10:\s(.?) !si;
149	5					23157	($data->{isbn13}) = $html =~ m! ISBN-13:\s(.?) !si;
150	5					6519	($data->{content}) = $html =~ m!
151	5					46891	($data->{description}) = $html =~ m!From the Back Cover\s* \s* (.*?)
152	5	100				25698	($data->{description}) = $html =~ m! ]>.?