X-Git-Url: http://git.rot13.org/?a=blobdiff_plain;f=Scraper.pm;h=692a4c6f4496c2a639a6f84852a73ed6ebff0744;hb=f26d4b82ad95b3975d8315a77928d3a1897090e4;hp=e1b13f497408aae5b0a8290a56b318ec7f0e9c46;hpb=6fe3f8a7ff856864af8366854b152ad15fb4506c;p=Biblio-Z3950.git

diff --git a/Scraper.pm b/Scraper.pm
index e1b13f4..692a4c6 100644
--- a/Scraper.pm
+++ b/Scraper.pm
@@ -5,14 +5,48 @@ use strict;
 
 use WWW::Mechanize;
 
-binmode STDOUT, ':utf8';
-
 sub new {
-    my ( $class ) = @_;
+    my ( $class, $database ) = @_;
+
+	$database ||= $class;
+
     my $self = {
 		mech => WWW::Mechanize->new(),
+		database => $database,
 	};
     bless $self, $class;
     return $self;
 }
 
+sub mech { $_[0]->{mech} }
+
+sub save_marc {
+	my ( $self, $id, $marc ) = @_;
+
+	my $database = $self->{database};
+	mkdir 'marc' unless -e 'marc';
+	mkdir "marc/$database" unless -e "marc/$database";
+
+	my $path = "marc/$database/$id";
+
+	open(my $out, '>:utf8', $path) || die "$path: $!";
+	print $out $marc;
+	close($out);
+
+	warn "# created $path ", -s $path, " bytes";
+
+}
+
+our $dump_nr = 1;
+
+sub save_content {
+	my $self = shift;
+	my $path = "/tmp/$dump_nr.html";
+	open(my $html, '>', $path);
+	print $html $self->{mech}->content;
+	close($html);
+	warn "# save_content $path ", -s $path, " bytes";
+	$dump_nr++;
+}
+
+1;