couchdb/scripts/reblog2couchdb.pl

#!/usr/bin/perl -w

use strict;
use DBI;
use CouchDB::Client;
use Data::Dump qw/dump/;

$|++;

my $database = 'reblog';

my $dbi = "DBI:mysql:database=$database";
$dbi .= ";host=127.0.0.1;port=13306";   # XXX over ssh

my $sql = qq{
        select
                items.id as _id,
                items.*,
                feeds.url as feed_url,
                feeds.title as feed_title,
                feeds.link as feed_link,
                feeds.description as feed_description
        from items
        join items_userdata on items.id = item_id
        join feeds on items.feed_id = feeds.id
        where items.id > ?
        order by items.id asc
        limit 5000
};


my $c = CouchDB::Client->new(uri => 'http://localhost:5984/');

$c->testConnection or die "The server cannot be reached";
print "CouchDB version " . $c->serverInfo->{version} . "\n";
my $db = $c->newDB( $database );
$db->create unless $c->dbExists( $database );

my $status = $db->newDoc( '_sync' );
eval { $status->retrieve };
$status->create if $@;

print "status ",dump( $status->{data} ), "\n";

my $last_row = $status->{data}->{last_row_id} || 0;

sub commit_last_row {
        warn "commit_last_row $last_row\n";
        $status->{data}->{last_row_id} = $last_row;
        $status->update;
}

my $dbh = DBI->connect($dbi,"","") || die $DBI::errstr;

print "Fetching items from $dbi id > $last_row\n";

my $sth = $dbh->prepare($sql) || die $dbh->errstr();
$sth->execute( $last_row ) || die $sth->errstr();

warn dump( $sth->{NAME} );

print "found ",$sth->rows," items to process...";

my $pk = 'id';

my $count = 0;

while (my $row = $sth->fetchrow_hashref() ) {
        my $_id = $row->{_id} || die "row needs _id";
        my $doc = $db->newDoc( $_id );

        sub row2doc {
                my ( $row, $doc ) = @_;
                my $a = delete( $row->{xml} );
                $doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a;
                my $a = delete( $row->{content} );
                $doc->addAttachment( 'content.html', 'text/html', $a ) if $a;
                $doc->{data} = $row;
                return $doc;
        }

        row2doc( $row, $doc );

        eval { $doc->create };
        if ( $@ ) {
                $doc->retrieve;
                row2doc( $row, $doc )->update;
#               eval { $doc->update };
                warn $@ ? "ERROR $_id $@\n" : "updated $_id\n";
        } else {
                warn "created $_id\n";
        }

        $last_row = $row->{id};
        $count++;

        commit_last_row if $count % 100 == 0    # checkpoint every 100 records
}

commit_last_row;

__END__

$sql = qq{
        update items_userdata
        set value_numeric = 1
        where label = 'read' and item_id in ($ids)
};

$dbh->do( $sql );

1	dpavlin	2	#!/usr/bin/perl -w
2
3			use strict;
4			use DBI;
5			use CouchDB::Client;
6			use Data::Dump qw/dump/;
7
8			$\|++;
9
10			my $database = 'reblog';
11
12	dpavlin	6	my $dbi = "DBI:mysql:database=$database";
13			$dbi .= ";host=127.0.0.1;port=13306"; # XXX over ssh
14	dpavlin	2
15			my $sql = qq{
16			select
17	dpavlin	4	items.id as _id,
18			items.*,
19			feeds.url as feed_url,
20			feeds.title as feed_title,
21			feeds.link as feed_link,
22			feeds.description as feed_description
23	dpavlin	2	from items
24	dpavlin	4	join items_userdata on items.id = item_id
25			join feeds on items.feed_id = feeds.id
26	dpavlin	6	where items.id > ?
27	dpavlin	4	order by items.id asc
28	dpavlin	7	limit 5000
29	dpavlin	2	};
30
31
32			my $c = CouchDB::Client->new(uri => 'http://localhost:5984/');
33
34			$c->testConnection or die "The server cannot be reached";
35	dpavlin	6	print "CouchDB version " . $c->serverInfo->{version} . "\n";
36	dpavlin	2	my $db = $c->newDB( $database );
37			$db->create unless $c->dbExists( $database );
38
39	dpavlin	6	my $status = $db->newDoc( '_sync' );
40			eval { $status->retrieve };
41			$status->create if $@;
42	dpavlin	2
43	dpavlin	6	print "status ",dump( $status->{data} ), "\n";
44
45			my $last_row = $status->{data}->{last_row_id} \|\| 0;
46
47	dpavlin	7	sub commit_last_row {
48			warn "commit_last_row $last_row\n";
49			$status->{data}->{last_row_id} = $last_row;
50			$status->update;
51			}
52
53	dpavlin	6	my $dbh = DBI->connect($dbi,"","") \|\| die $DBI::errstr;
54
55			print "Fetching items from $dbi id > $last_row\n";
56
57			my $sth = $dbh->prepare($sql) \|\| die $dbh->errstr();
58			$sth->execute( $last_row ) \|\| die $sth->errstr();
59
60			warn dump( $sth->{NAME} );
61
62			print "found ",$sth->rows," items to process...";
63
64	dpavlin	2	my $pk = 'id';
65
66	dpavlin	6	my $count = 0;
67
68	dpavlin	2	while (my $row = $sth->fetchrow_hashref() ) {
69	dpavlin	4	my $_id = $row->{_id} \|\| die "row needs _id";
70	dpavlin	2	my $doc = $db->newDoc( $_id );
71	dpavlin	4
72			sub row2doc {
73			my ( $row, $doc ) = @_;
74			my $a = delete( $row->{xml} );
75			$doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a;
76			my $a = delete( $row->{content} );
77			$doc->addAttachment( 'content.html', 'text/html', $a ) if $a;
78			$doc->{data} = $row;
79			return $doc;
80			}
81
82			row2doc( $row, $doc );
83
84	dpavlin	2	eval { $doc->create };
85			if ( $@ ) {
86	dpavlin	4	$doc->retrieve;
87			row2doc( $row, $doc )->update;
88			# eval { $doc->update };
89	dpavlin	7	warn $@ ? "ERROR $_id $@\n" : "updated $_id\n";
90	dpavlin	2	} else {
91	dpavlin	7	warn "created $_id\n";
92	dpavlin	2	}
93	dpavlin	6
94	dpavlin	7	$last_row = $row->{id};
95	dpavlin	6	$count++;
96	dpavlin	7
97			commit_last_row if $count % 100 == 0 # checkpoint every 100 records
98	dpavlin	2	}
99
100	dpavlin	7	commit_last_row;
101	dpavlin	6
102	dpavlin	2	__END__
103
104			$sql = qq{
105			update items_userdata
106			set value_numeric = 1
107			where label = 'read' and item_id in ($ids)
108			};
109
110			$dbh->do( $sql );
111