view release on metacpan or  search on metacpan

lib/Alvis/Wikipedia/XMLDump.pm  view on Meta::CPAN

package Alvis::Wikipedia::XMLDump;

use warnings;
use strict;

$Alvis::Wikipedia::XMLDump::VERSION = '0.1';

use Storable;
use Parse::MediaWikiDump;
use Digest::MD5;
use Encode;
use Data::Dumper;

use Alvis::Wikipedia::WikitextParser;
use Alvis::Wikipedia::Variables;
use Alvis::Wikipedia::CatGraph;
use Alvis::Canonical;

########################################################################
#
#  Exported constants
#
#######################################################################

# Record output formats
our ($OUTPUT_HTML,
     $OUTPUT_ALVIS
     )=(0..1);

############################################################################
#
#  Error message stuff
#
############################################################################

my ($ERR_OK,
    $ERR_VAR,
    $ERR_PARSER,
    $ERR_FIRST_PASS,
    $ERR_SECOND_PASS,
    $ERR_TEMPL_ADD,
    $ERR_EXPAND,
    $ERR_DUMP,
    $ERR_TABLE_PARSE,
    $ERR_REC_CB,
    $ERR_HTML,
    $ERR_CAN_DOC_CONV,
    $ERR_ALVIS,
    $ERR_BUILD_CAT_GRAPH,
    $ERR_CATEGORIES,
    $ERR_XML_PARSER,
    $ERR_CAN_DOC_CONVERSION,
    $ERR_ID,
    $ERR_TITLE,
    $ERR_CAT_PAGE_LINKS_ADD,
    $ERR_CAT_GRAPH,
    $ERR_LOAD_TEMPLATES,
    $ERR_CAT_GRAPH_DUMP,
    $ERR_UNK_OUTPUT_FORMAT
    )=(0..23);
my %ErrMsgs=($ERR_OK=>"",
	     $ERR_VAR=>"Unable to instantiate Alvis::Wikipedia::Variables.",
	     $ERR_PARSER=>
	         "Unable to instantiate Alvis::Wikipedia::WikitextParser.",
	     $ERR_FIRST_PASS=>"The first pass over the records failed.",
	     $ERR_SECOND_PASS=>"The main pass over the records failed.",
	     $ERR_TEMPL_ADD=>"Adding the definition of a template failed.",
	     $ERR_EXPAND=>"Variable and template expansion failed.",
	     $ERR_DUMP=>"Opening the SQL dump file failed.",
	     $ERR_TABLE_PARSE=>"Parsing a subtable failed.",
	     $ERR_REC_CB=>"Record handling callback failed.",
	     $ERR_HTML=>"Wikitext -> HTML failed.",