From: Dobrica Pavlinusic <dpavlin@rot13.org>
Date: Tue, 20 Nov 2007 10:08:02 +0000 (+0000)
Subject:  r1650@llin:  dpavlin | 2007-11-20 11:07:57 +0100
X-Git-Url: http://git.rot13.org/?p=webpac2;a=commitdiff_plain;h=5d5fc808565b88bca0e2d10aa62bd89c7e748743

 r1650@llin:  dpavlin | 2007-11-20 11:07:57 +0100
 final tweaks for WebPAC::Input::PDF, emit fields A .. ZZ


git-svn-id: svn+ssh://mjesec/home/dpavlin/svn/webpac2/trunk@1057 07558da8-63fa-0310-ba24-9fe276d99e06
---

diff --git a/lib/WebPAC/Input/PDF.pm b/lib/WebPAC/Input/PDF.pm
index 91b9920..8cba888 100644
--- a/lib/WebPAC/Input/PDF.pm
+++ b/lib/WebPAC/Input/PDF.pm
@@ -69,7 +69,7 @@ sub new {
 
 	$log->info("opend $file with $pages pages");
 
-	my @lines;
+	my @lines = ();
 
 	foreach my $p ( 1 .. $pages ) {
 		my $tree = $doc->getPageContentTree($p);
@@ -123,7 +123,7 @@ sub new {
 		}
 	}
 
-	$self->size( $#lines );
+	$self->{_lines} = \@lines;
 
 	$log->debug("loaded ", $self->size, " records", sub { dump( @lines ) });
 
@@ -136,6 +136,10 @@ Return record with ID C<$mfn> from database
 
   my $rec = $input->fetch_rec( $mfn, $filter_coderef );
 
+Records are returned as field C<A>, C<B> and so on...
+
+Last supported column is C<ZZ>.
+
 =cut
 
 sub fetch_rec {
@@ -143,7 +147,36 @@ sub fetch_rec {
 
 	my ( $mfn, $filter_coderef ) = @_;
 
-	return $self->{_rec}->[$mfn-1];
+	my $rec = {
+		'000' => [ $mfn ],
+	};
+
+	my $line = $self->{_lines}->[ $mfn - 1 ] || return;
+	confess "expected ARRAY for _lines $mfn" unless ref($line) eq 'ARRAY';
+
+#	warn "## line = ",dump( $line );
+
+	my $col = 'A';
+	my $c = 0;
+	foreach my $e ( @$line ) {
+		$rec->{$col} = $e;
+		$c++;
+		# FIXME what about columns > ZZ
+		if ( $col eq 'Z' ) {
+			$col .= 'AA';
+		} elsif ( $col eq 'ZZ' ) {
+			$self->_get_logger()->logwarn("ignoring colums above ZZ (original ", $#$line + 1, " > $c max columns)");
+			last;
+		} elsif ( $col =~ m/([A-Z])Z$/ ) {
+			$col .= $1++ . 'A';
+		} else {
+			$col++;
+		}
+	}
+
+#	warn "## rec = ",dump( $rec );
+
+	return $rec;
 }
 
 
@@ -157,7 +190,7 @@ Return number of records in database
 
 sub size {
 	my $self = shift;
-	return $#{$self->{_rec}} + 1;
+	return $#{$self->{_lines}} + 1;
 }
 
 =head1 SEE ALSO
diff --git a/t/2-input-pdf.t b/t/2-input-pdf.t
index c0098ba..6a2151a 100755
--- a/t/2-input-pdf.t
+++ b/t/2-input-pdf.t
@@ -3,7 +3,7 @@
 use strict;
 use blib;
 
-use Test::More tests => 9;
+use Test::More tests => 11;
 
 BEGIN {
 use_ok( 'WebPAC::Test' );
@@ -24,6 +24,8 @@ ok(my $db = $input->open(
 ), "open");
 ok(my $size = $input->size, "size");
 
+$size = 3;
+
 foreach my $mfn ( 1 ... $size ) {
 	my $rec = $input->fetch;
 	if ($mfn <= 10 || $mfn == 20) {