git.rot13.org Git - pxelator/blob - bin/couchdb2mongodb.pl

   1 #!/usr/bin/perl
   2
   3 # http://wiki.apache.org/couchdb/HTTP_Bulk_Document_API
   4
   5 use warnings;
   6 use strict;
   7
   8 use IO::Socket::INET;
   9 use Storable qw();
  10 use JSON;
  11 use Data::Dump qw(dump);
  12 use Time::HiRes qw(time);
  13 use File::Path qw(make_path remove_tree);
  14 use MongoDB;
  15
  16 my $name = 'pxelator';
  17
  18 my $conn = MongoDB::Connection->new;
  19 my $db = $conn->get_database( $name );
  20 my $audit = $db->get_collection("audit");
  21
  22 sub couchdb_socket {
  23         IO::Socket::INET->new(
  24                 PeerAddr => '10.60.0.91',
  25                 PeerPort => 5984,
  26                 Proto => 'tcp',
  27         ) || die $!;
  28 }
  29
  30 sub get_chunk {
  31         my $sock = shift;
  32         my $chunk;
  33         while(<$sock>) {
  34                 $chunk .= $_;
  35                 last if /^[\n\r]+$/;
  36         }
  37 #       warn "# $sock\n$chunk\n";
  38         return $chunk;
  39 }
  40
  41 my $sock = couchdb_socket;
  42
  43 print $sock "GET /$name/_all_docs?include_docs=true HTTP/1.0\r\n\r\n";
  44
  45 get_chunk($sock);
  46
  47 my $total = <$sock>;
  48 $total =~ s{^.*total_rows\D+(\d+).+$}{$1};
  49 warn "# total: $total\n";
  50
  51 while(<$sock>) {
  52         if ( /"id":"([^"]+)"/ ) {
  53
  54                 s/,[\r\n]+$//; # cleanup JSON
  55                 my $json = from_json( $_ );
  56                 $audit->insert( $json->{doc} );
  57                 print STDERR ".";
  58         } else {
  59                 warn "UNKNOWN: $_";
  60         }
  61 }
  62