couchdb/scripts/reblog2couchdb.pl

#!/usr/bin/perl -w

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

$|++;

my $debug = @ARGV ? 1 : 0;

my $database = 'reblog';

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

my $dbh = DBI->connect($dbi,"","",{ RaiseError => 1 });

$dbh->do(qq{
        create temporary table published_items as
        select
                item_id
        from
                items_userdata
        where
                label = 'published' and
                value_numeric = 1
});

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 published_items on items.id = item_id
        join feeds on items.feed_id = feeds.id
        where items.id > ?
        order by items.id asc
        limit 1000
};

my $sql_tags = qq{
select
        items_userdata.item_id,
        value_long as tags,
        timestamp
from items_userdata
join published_items p
        on items_userdata.item_id = p.item_id and label='tags'
where
        items_userdata.item_id > ?
order by items_userdata.item_id asc
};

my $c = CouchDB::Client->new(uri => 'http://192.168.1.13:5984/');

$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( 'last_sync' );
eval { $status->retrieve };
$status->create if $@;

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

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

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

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

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

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

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

my $sth_tags = $dbh->prepare($sql_tags);
$sth_tags->execute( $last_row );
print "found ",$sth_tags->rows, " tags found...\n";

my $count = 0;

my $row_tags = $sth_tags->fetchrow_hashref();

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

        while ( $row_tags && $row_tags->{item_id} < $row->{_id} ) {
                $row_tags = $sth_tags->fetchrow_hashref();
                warn "## got tags: ",dump( $row_tags ) if $debug;
        }

        sub row2doc {
                my ( $row, $doc ) = @_;
                my $a = delete( $row->{xml} );
                $doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a;
                $a = delete( $row->{content} );
                $doc->addAttachment( 'content.html', 'text/html', $a ) if $a;
                if ( $row_tags && $row_tags->{item_id} == $row->{_id} ) {
                        $row->{tags} = [ split(/\s+/, $row_tags->{tags} ) ];
                        warn "++ ",$row->{item_id}, dump( $row->{tags} );
                }
                $doc->{data} = $row;
                warn "## ",dump( $row ) if $debug;
                return $doc;
        }

        row2doc( $row, $doc );

        eval { $doc->create };
        if ( $@ ) {
                $doc->retrieve;
                row2doc( $row, $doc )->update;
#               eval { $doc->update };
                warn $@ ? "$count ERROR $_id $@\n" : "$count updated $_id\n";
        } else {
                warn "$count 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	dpavlin	15	my $debug = @ARGV ? 1 : 0;
11
12	dpavlin	2	my $database = 'reblog';
13
14	dpavlin	6	my $dbi = "DBI:mysql:database=$database";
15			$dbi .= ";host=127.0.0.1;port=13306"; # XXX over ssh
16	dpavlin	2
17	dpavlin	28	my $dbh = DBI->connect($dbi,"","",{ RaiseError => 1 });
18
19			$dbh->do(qq{
20			create temporary table published_items as
21			select
22			item_id
23			from
24			items_userdata
25			where
26			label = 'published' and
27			value_numeric = 1
28			});
29
30	dpavlin	2	my $sql = qq{
31			select
32	dpavlin	4	items.id as _id,
33			items.*,
34			feeds.url as feed_url,
35			feeds.title as feed_title,
36			feeds.link as feed_link,
37			feeds.description as feed_description
38	dpavlin	2	from items
39	dpavlin	28	join published_items on items.id = item_id
40	dpavlin	4	join feeds on items.feed_id = feeds.id
41	dpavlin	28	where items.id > ?
42	dpavlin	4	order by items.id asc
43	dpavlin	10	limit 1000
44	dpavlin	2	};
45
46	dpavlin	15	my $sql_tags = qq{
47			select
48			items_userdata.item_id,
49	dpavlin	28	value_long as tags,
50			timestamp
51	dpavlin	15	from items_userdata
52	dpavlin	28	join published_items p
53			on items_userdata.item_id = p.item_id and label='tags'
54	dpavlin	15	where
55	dpavlin	28	items_userdata.item_id > ?
56	dpavlin	15	order by items_userdata.item_id asc
57			};
58
59	dpavlin	26	my $c = CouchDB::Client->new(uri => 'http://192.168.1.13:5984/');
60	dpavlin	2
61			$c->testConnection or die "The server cannot be reached";
62	dpavlin	6	print "CouchDB version " . $c->serverInfo->{version} . "\n";
63	dpavlin	2	my $db = $c->newDB( $database );
64			$db->create unless $c->dbExists( $database );
65
66	dpavlin	27	my $status = $db->newDoc( 'last_sync' );
67	dpavlin	6	eval { $status->retrieve };
68	dpavlin	27	$status->create if $@;
69	dpavlin	2
70	dpavlin	6	print "status ",dump( $status->{data} ), "\n";
71
72			my $last_row = $status->{data}->{last_row_id} \|\| 0;
73	dpavlin	15	$last_row = 0 if $debug;
74	dpavlin	6
75	dpavlin	7	sub commit_last_row {
76			warn "commit_last_row $last_row\n";
77			$status->{data}->{last_row_id} = $last_row;
78			$status->update;
79			}
80
81	dpavlin	6	print "Fetching items from $dbi id > $last_row\n";
82
83	dpavlin	15	my $sth = $dbh->prepare($sql);
84			$sth->execute( $last_row );
85	dpavlin	6
86			warn dump( $sth->{NAME} );
87
88	dpavlin	10	print "found ",$sth->rows," items to process...\n";
89	dpavlin	6
90	dpavlin	15	my $sth_tags = $dbh->prepare($sql_tags);
91			$sth_tags->execute( $last_row );
92			print "found ",$sth_tags->rows, " tags found...\n";
93	dpavlin	2
94	dpavlin	6	my $count = 0;
95
96	dpavlin	15	my $row_tags = $sth_tags->fetchrow_hashref();
97
98	dpavlin	2	while (my $row = $sth->fetchrow_hashref() ) {
99	dpavlin	4	my $_id = $row->{_id} \|\| die "row needs _id";
100	dpavlin	2	my $doc = $db->newDoc( $_id );
101	dpavlin	4
102	dpavlin	15	while ( $row_tags && $row_tags->{item_id} < $row->{_id} ) {
103			$row_tags = $sth_tags->fetchrow_hashref();
104			warn "## got tags: ",dump( $row_tags ) if $debug;
105			}
106
107	dpavlin	4	sub row2doc {
108			my ( $row, $doc ) = @_;
109			my $a = delete( $row->{xml} );
110			$doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a;
111	dpavlin	15	$a = delete( $row->{content} );
112	dpavlin	4	$doc->addAttachment( 'content.html', 'text/html', $a ) if $a;
113	dpavlin	15	if ( $row_tags && $row_tags->{item_id} == $row->{_id} ) {
114			$row->{tags} = [ split(/\s+/, $row_tags->{tags} ) ];
115			warn "++ ",$row->{item_id}, dump( $row->{tags} );
116			}
117	dpavlin	4	$doc->{data} = $row;
118	dpavlin	15	warn "## ",dump( $row ) if $debug;
119	dpavlin	4	return $doc;
120			}
121
122			row2doc( $row, $doc );
123
124	dpavlin	2	eval { $doc->create };
125			if ( $@ ) {
126	dpavlin	4	$doc->retrieve;
127			row2doc( $row, $doc )->update;
128			# eval { $doc->update };
129	dpavlin	10	warn $@ ? "$count ERROR $_id $@\n" : "$count updated $_id\n";
130	dpavlin	2	} else {
131	dpavlin	10	warn "$count created $_id\n";
132	dpavlin	2	}
133	dpavlin	6
134	dpavlin	7	$last_row = $row->{id};
135	dpavlin	6	$count++;
136	dpavlin	7
137			commit_last_row if $count % 100 == 0 # checkpoint every 100 records
138	dpavlin	2	}
139
140	dpavlin	7	commit_last_row;
141	dpavlin	6
142	dpavlin	2	__END__
143
144			$sql = qq{
145			update items_userdata
146			set value_numeric = 1
147			where label = 'read' and item_id in ($ids)
148			};
149
150			$dbh->do( $sql );
151