7 |
|
|
8 |
$|++; |
$|++; |
9 |
|
|
10 |
|
my $debug = @ARGV ? 1 : 0; |
11 |
|
|
12 |
my $database = 'reblog'; |
my $database = 'reblog'; |
13 |
|
|
14 |
my $dbi = "DBI:mysql:database=$database"; |
my $dbi = "DBI:mysql:database=$database"; |
22 |
feeds.title as feed_title, |
feeds.title as feed_title, |
23 |
feeds.link as feed_link, |
feeds.link as feed_link, |
24 |
feeds.description as feed_description |
feeds.description as feed_description |
|
-- t.value_long as tags |
|
25 |
from items |
from items |
26 |
join items_userdata on items.id = item_id |
join items_userdata on items.id = item_id |
27 |
join feeds on items.feed_id = feeds.id |
join feeds on items.feed_id = feeds.id |
|
-- left outer join items_userdata as t on items.id = t.item_id and t.label='tags' |
|
28 |
where items.id > ? and items_userdata.label = 'published' and items_userdata.value_numeric = 1 |
where items.id > ? and items_userdata.label = 'published' and items_userdata.value_numeric = 1 |
29 |
order by items.id asc |
order by items.id asc |
30 |
limit 1000 |
limit 1000 |
31 |
}; |
}; |
32 |
|
|
33 |
|
my $sql_tags = qq{ |
34 |
|
select |
35 |
|
items_userdata.item_id, |
36 |
|
t.value_long as tags, |
37 |
|
items_userdata.timestamp |
38 |
|
from items_userdata |
39 |
|
join items_userdata as t |
40 |
|
on items_userdata.item_id = t.item_id and t.label='tags' |
41 |
|
where |
42 |
|
items_userdata.item_id > ? and |
43 |
|
items_userdata.label = 'published' and |
44 |
|
items_userdata.value_numeric = 1 |
45 |
|
order by items_userdata.item_id asc |
46 |
|
}; |
47 |
|
|
48 |
my $c = CouchDB::Client->new(uri => 'http://localhost:5984/'); |
my $c = CouchDB::Client->new(uri => 'http://localhost:5984/'); |
49 |
|
|
50 |
$c->testConnection or die "The server cannot be reached"; |
$c->testConnection or die "The server cannot be reached"; |
59 |
print "status ",dump( $status->{data} ), "\n"; |
print "status ",dump( $status->{data} ), "\n"; |
60 |
|
|
61 |
my $last_row = $status->{data}->{last_row_id} || 0; |
my $last_row = $status->{data}->{last_row_id} || 0; |
62 |
|
$last_row = 0 if $debug; |
63 |
|
|
64 |
sub commit_last_row { |
sub commit_last_row { |
65 |
warn "commit_last_row $last_row\n"; |
warn "commit_last_row $last_row\n"; |
67 |
$status->update; |
$status->update; |
68 |
} |
} |
69 |
|
|
70 |
my $dbh = DBI->connect($dbi,"","") || die $DBI::errstr; |
my $dbh = DBI->connect($dbi,"","",{ RaiseError => 1 }); |
71 |
|
|
72 |
print "Fetching items from $dbi id > $last_row\n"; |
print "Fetching items from $dbi id > $last_row\n"; |
73 |
|
|
74 |
my $sth = $dbh->prepare($sql) || die $dbh->errstr(); |
my $sth = $dbh->prepare($sql); |
75 |
$sth->execute( $last_row ) || die $sth->errstr(); |
$sth->execute( $last_row ); |
76 |
|
|
77 |
warn dump( $sth->{NAME} ); |
warn dump( $sth->{NAME} ); |
78 |
|
|
79 |
print "found ",$sth->rows," items to process...\n"; |
print "found ",$sth->rows," items to process...\n"; |
80 |
|
|
81 |
my $pk = 'id'; |
my $sth_tags = $dbh->prepare($sql_tags); |
82 |
|
$sth_tags->execute( $last_row ); |
83 |
|
print "found ",$sth_tags->rows, " tags found...\n"; |
84 |
|
|
85 |
my $count = 0; |
my $count = 0; |
86 |
|
|
87 |
|
my $row_tags = $sth_tags->fetchrow_hashref(); |
88 |
|
|
89 |
while (my $row = $sth->fetchrow_hashref() ) { |
while (my $row = $sth->fetchrow_hashref() ) { |
90 |
my $_id = $row->{_id} || die "row needs _id"; |
my $_id = $row->{_id} || die "row needs _id"; |
91 |
my $doc = $db->newDoc( $_id ); |
my $doc = $db->newDoc( $_id ); |
92 |
|
|
93 |
|
while ( $row_tags && $row_tags->{item_id} < $row->{_id} ) { |
94 |
|
$row_tags = $sth_tags->fetchrow_hashref(); |
95 |
|
warn "## got tags: ",dump( $row_tags ) if $debug; |
96 |
|
} |
97 |
|
|
98 |
sub row2doc { |
sub row2doc { |
99 |
my ( $row, $doc ) = @_; |
my ( $row, $doc ) = @_; |
100 |
my $a = delete( $row->{xml} ); |
my $a = delete( $row->{xml} ); |
101 |
$doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a; |
$doc->addAttachment( 'item.xml', 'application/xhtml+xml', $a ) if $a; |
102 |
my $a = delete( $row->{content} ); |
$a = delete( $row->{content} ); |
103 |
$doc->addAttachment( 'content.html', 'text/html', $a ) if $a; |
$doc->addAttachment( 'content.html', 'text/html', $a ) if $a; |
104 |
|
if ( $row_tags && $row_tags->{item_id} == $row->{_id} ) { |
105 |
|
$row->{tags} = [ split(/\s+/, $row_tags->{tags} ) ]; |
106 |
|
warn "++ ",$row->{item_id}, dump( $row->{tags} ); |
107 |
|
} |
108 |
$doc->{data} = $row; |
$doc->{data} = $row; |
109 |
|
warn "## ",dump( $row ) if $debug; |
110 |
return $doc; |
return $doc; |
111 |
} |
} |
112 |
|
|