Fetch source by use of Mediawiki::API (not wget + custom mediawiki-extract decoding).

2 files changed, 34 insertions, 3 deletions

diff --git a/Makefile b/Makefile
index 29b20dd..774bd70 100644
--- a/Makefile
+++ b/Makefile
@@ -47,11 +47,10 @@ args_ebook = -V papersize=ebook -V fontsize=12pt -V classoption=oneside -V class
 all: $(flavors:%=$(stem)-%.pdf)
 
 download:
-	wget -O $(stem).raw '$(source_baseurl)w/index.php?title=$(source_basename)&action=edit'
+	./mediawiki-fetch $(source_baseurl) $(source_basename) $(stem).raw
 
 $(stem).mw: $(stem).raw
-	./mediawiki-extract $< $@
-	./mediawiki-trim $@
+	./mediawiki-trim $< $@
 	./mediawiki-blockquote $@
 	./mediawiki-uri-escape $@
 	./mediawiki-matter $@
diff --git a/mediawiki-fetch b/mediawiki-fetch
new file mode 100755
index 0000000..181c84e
--- /dev/null
+++ b/mediawiki-fetch
@@ -0,0 +1,32 @@
+#!/usr/bin/perl
+
+# fetch content of mediawiki page
+
+use warnings;
+use strict;
+
+use feature qw(say);
+
+use Path::Tiny;
+use MediaWiki::API;
+use DateTime::Format::ISO8601;
+
+my $base_uri = shift;
+my $page_title = shift;
+my $outfile = shift;
+
+my $mw = MediaWiki::API->new( { api_url => $base_uri . 'w/api.php' }  );
+
+my $page = $mw->get_page( { title => $page_title } )
+	|| die $mw->{error}->{code} . ': ' . $mw->{error}->{details};
+
+my $dt = DateTime::Format::ISO8601->parse_datetime( $page->{'timestamp'} );
+
+path($outfile)->spew_utf8( $page->{'*'} );
+path($outfile)->touch( $dt->epoch );
+
+say STDERR sprintf( 'Succesfully fetched %s (rev. %s)',
+	$page_title,
+	$page->{'revid'});
+
+1;