Navigation

Changeset 70

Timestamp:

04/27/08 20:56:37 (2 months ago)

Author:

bkuhn

Message:

r76@hughes: bkuhn | 2008-04-27 20:49:20 -0400

Wrote basics for PDF::Extract::Text
Fixed various things in pdfdiff.in
Fixed version number of PDF::OCR required, and where version is
obtained for PDF::Extract::Text

Files:

r67	r70
	1	2008-04-27 Bradley M. Kuhn <bkuhn@ebb.org>
	2
	3	* Perl/lib/PDF/Extract/Text.pm (new): Wrote function.
	4	(PDF::OCR::Thorough::_pdftotext): Wrote override of Thorough
	5	function.
	6	(getText): Wrote function.
	7

r66	r70
29	29
30	30	use Getopt::Long qw(:config auto_help bundling);
31		use PDF::OCR::Thorough;
	31	use Pod::Usage;
	32	use PDF::Extract::Text;
32	33
33	34	=head1 SYNOPSIS
…	…
111	112	=cut
112	113
	114	my($saveTemplate, $pdiffOutputFile, $extraDiffOptions,$diffCmd, $useMeld,
	115	$help, $verbose);
	116
113	117	GetOptions("save-intermediates=s" => \$saveTemplate,
114	118	"pdiff=s" => \$pdiffOutputFile,
…	…
116	120	"diff=s" => \$diffCmd,
117	121	"meld" => \$useMeld,
118		"help" => \$help
	122	"help" => \$help,
119	123	"verbose" => \$verbose
120	124	) or pod2usage(2);
…	…
133	137	if ($saveTemplate) {
134	138	open(OLD_TEXT, ">${saveTemplate}-old.txt") or
135		pod2usage("$0: cannot open ${saveTemplate}-old.txt for writing: $!";
	139	pod2usage("$0: cannot open ${saveTemplate}-old.txt for writing: $!");
136	140	open(NEW_TEXT, ">${saveTemplate}-new.txt") or
137		pod2usage("$0: cannot open ${saveTemplate}-new.txt for writing: $!";
	141	pod2usage("$0: cannot open ${saveTemplate}-new.txt for writing: $!");
138	142	}
139		my $old~~Text = new PDF::OCR::Thorough(~~$oldFile)
	143	my $oldExtract = new PDF::Extract::Text(pdfFile => $oldFile)
140	144	or die("cannot do PDF::OCR::Thorough on $oldFile");
141		my $new~~Text = new PDF::OCR::Thorough(~~$newFile)
	145	my $newExtract = new PDF::Extract::Text(pdfFile => $newFile)
142	146	or die("cannot do PDF::OCR::Thorough on $newFile");
	147
	148	my $oldText = $oldExtract->getText();
	149	my $newText = $newExtract->getText();
143	150
144	151	print "OLD: $oldText\n";

r68	r70
20	20	use warnings;
21	21
	22	BEGIN {
	23	use PDF::OCR::Thorough;
	24	sub PDF::OCR::Thorough::_pdftotext {
	25	my $self = shift;
	26	$self->{pdftotextbin} \|\|= PDF::OCR::Thorough::which('pdftotext') or die("missing pdftotext?");
	27	return ($self->{pdftotextbin}, '-layout', '-nopgbrk');
	28	}
	29	}
	30
	31	package PDF::Extract::Text;
	32
22	33	require Exporter;
23	34	use AutoLoader qw(AUTOLOAD);
	35	use Carp qw(croak);
	36	use PDF::OCR::Thorough;
24	37
25	38	=head1 NAME
…	…
58	71	=head1 PUBLIC METHODS
59	72
	73	=head2 new PDF::Extract::Text
	74
	75	Creates a new PDF::Extract::Text object
	76
	77	new PDF::Extract::Text(pdfFile => $pdfFileName)
	78
60	79	=cut
61	80
	81	sub new {
	82	my $this = shift;
	83	my $class = ref($this) \|\| $this;
	84	my $self = {};
	85	bless $self, $class;
	86	my(%args) = @_;
	87
	88	$self->{pdfFile} = $args{pdfFile};
	89
	90	croak("usage: new PDF::Extract::Text(pdfFile => \$pdfFileName)")
	91	unless defined $self->{pdfFile};
	92	croak("cannot open $self->{pdfFile} for reading: $!")
	93	unless -r $self->{pdfFile};
	94
	95	$self->{ocrObj} = new PDF::OCR::Thorough($self->{pdfFile});
	96	return $self;
	97	}
	98
	99	=head2 getText
	100
	101	Returns the text, formatted as best as is possible.
	102
	103	=cut
	104
	105	sub getText {
	106	my $self = shift;
	107
	108	return $self->{ocrObj}->get_text;
	109	}
	110
	111	1;
	112
	113	__END__

r69	r70
27	27	PREREQ_FATAL => 1,
28	28	PREREQ_PM => {
29		'PDF::OCR' => '1.7'
	29	'PDF::OCR' => '1.07'
30	30	},
31	31	MAKEFILE => "Makefile.perl-generated",
32	32	NAME => 'PDF::Extract::Text',
33		VERSION_FROM => "lib/PDF/Extract/Text/Version.pm~~.in~~",
	33	VERSION_FROM => "lib/PDF/Extract/Text/Version.pm",
34	34
35	35	AUTHOR => 'Bradley M. Kuhn <bkuhn@ebb.org>',