wiki.pl

#!/usr/bin/perl

use strict;
use warnings;
use XML::Simple;
use DBI;

my $xml= XML::Simple->new();
open(my $in, "< ./jawiki-latest-pages-articles.xml");
my $conn= DBI->connect("dbi:mysql:wikipedia;mysql_socket=/tmp/mysql.sock", "root", "", {mysql_enable_utf8 => 1});
$conn->do("SET sql_mode= ''");

my $buff;
while (my $row= <$in>)
{
  if ($row =~ qr|<page>|)
  {
    $buff= "";
  }

  $buff .= $row;

  if ($row =~ qr|</page>|)
  {
    my $ret= $xml->XMLin($buff);

    my $content= $ret->{revision}->{text}->{content};
    $content =~ s/\n/\\n/g;

    $conn->do("INSERT INTO articles SET title     = ?, " .
              "                         content   = ?, " .
              "                         timestamp = ?",
              undef, $ret->{title}, $content, $ret->{revision}->{timestamp});
  }
}

wiki.sql

CREATE DATABASE wikipedia;

CREATE TABLE wikipedia.articles (seq SERIAL, title TEXT, content LONGTEXT, timestamp DATETIME);