socialtext-import/phpwiki/phpwiki-dump.pl

#!/usr/bin/perl -w

use DBI;

my $sql = qq{
        select
                pagename, content, lastmodified::abstime, author
        from wiki
--      where author != 'The PhpWiki programming team'
        order by lastmodified
};
my $dump_dir = './dump/';
my $st_dir = './st/';

my $connect = "DBI:Pg:dbname=phpwiki";

my $dbh = DBI->connect($connect,"","") || die $DBI::errstr;
$sth = $dbh->prepare($sql) || die $dbh->errstr();
$sth->execute() || die $sth->errstr();

while (my($page,$content,$last_change,$author) = $sth->fetchrow_array() ) {
        print "exporting: $page\n";

        my $path = $page;
        $path =~ s!/+!_!g;

        open(my $out, '>', "$dump_dir/${path}.wiki") || die "can't open $out_dir/$path: $!";
        print $out $content;
        close($out);

        # headings
        $content =~ s/^!\s*(.*?)$/^ $1/mg;
        $content =~ s/^!!\s*(.*?)$/^^ $1/mg;
        $content =~ s/^!!!\s*(.*?)$/^^^ $1/mg;
        $content =~ s/^!!!!\s*(.*?)$/^^^^ $1/mg;
        $content =~ s/^!!!!!\s*(.*?)$/^^^^^ $1/mg;

        $content =~ s/!([A-Z]\w+)/$1/mg;
        
        $content =~ s!''(.*?)''!/$1/!mg;
        $content =~ s!__(.*?)__!*$1*!mg;

        $content =~ s!\[\s*(.+?)\s*[\s\|]\s*((http://|mailto:)\S+)\s*\]!"$1"<$2>!mg;
        $content =~ s!\[\s*(http://\S+)\s*[\s\|]\s*(.+?)\s*\]!"$2"<$1>!mg;

        my $seen;
        my $out;
        foreach (split(/[\n\r]/, $content)) {
                if (
                        ( ! /^\s+/ && $seen->{space} )  # end of space preformat
                        ||
                        ( /^\s+\S+/ && ! $seen->{space} )       # begin of space preformat
                ) {

                        $out .= ".pre\n";
                        $seen->{space} = not $seen->{space};
                }
                $out .= "$_\n";
        }

        $out .= ".pre\n" if ($seen->{space});

        $content = $out;

        $content =~ s/\.pre[\n\r]+\.pre//sg;

        $content .= "\n----\nLast modified $last_change by $author\n";

        open(my $out, '>', "$st_dir/${path}.wiki") || die "can't open $out_dir/$path: $!";
        print $out $content;
        close($out);
}
1	#!/usr/bin/perl -w
2
3	use DBI;
4
5	my $sql = qq{
6	select
7	pagename, content, lastmodified::abstime, author
8	from wiki
9	-- where author != 'The PhpWiki programming team'
10	order by lastmodified
11	};
12	my $dump_dir = './dump/';
13	my $st_dir = './st/';
14
15	my $connect = "DBI:Pg:dbname=phpwiki";
16
17	my $dbh = DBI->connect($connect,"","") \|\| die $DBI::errstr;
18	$sth = $dbh->prepare($sql) \|\| die $dbh->errstr();
19	$sth->execute() \|\| die $sth->errstr();
20
21	while (my($page,$content,$last_change,$author) = $sth->fetchrow_array() ) {
22	print "exporting: $page\n";
23
24	my $path = $page;
25	$path =~ s!/+!_!g;
26
27	open(my $out, '>', "$dump_dir/${path}.wiki") \|\| die "can't open $out_dir/$path: $!";
28	print $out $content;
29	close($out);
30
31	# headings
32	$content =~ s/^!\s(.?)$/^ $1/mg;
33	$content =~ s/^!!\s(.?)$/^^ $1/mg;
34	$content =~ s/^!!!\s(.?)$/^^^ $1/mg;
35	$content =~ s/^!!!!\s(.?)$/^^^^ $1/mg;
36	$content =~ s/^!!!!!\s(.?)$/^^^^^ $1/mg;
37
38	$content =~ s/!([A-Z]\w+)/$1/mg;
39
40	$content =~ s!''(.*?)''!/$1/!mg;
41	$content =~ s!__(.?)__!$1*!mg;
42
43	$content =~ s!\[\s(.+?)\s[\s\\|]\s((http://\|mailto:)\S+)\s\]!"$1"<$2>!mg;
44	$content =~ s!\[\s(http://\S+)\s[\s\\|]\s(.+?)\s\]!"$2"<$1>!mg;
45
46	my $seen;
47	my $out;
48	foreach (split(/[\n\r]/, $content)) {
49	if (
50	( ! /^\s+/ && $seen->{space} ) # end of space preformat
51	\|\|
52	( /^\s+\S+/ && ! $seen->{space} ) # begin of space preformat
53	) {
54
55	$out .= ".pre\n";
56	$seen->{space} = not $seen->{space};
57	}
58	$out .= "$_\n";
59	}
60
61	$out .= ".pre\n" if ($seen->{space});
62
63	$content = $out;
64
65	$content =~ s/\.pre[\n\r]+\.pre//sg;
66
67	$content .= "\n----\nLast modified $last_change by $author\n";
68
69	open(my $out, '>', "$st_dir/${path}.wiki") \|\| die "can't open $out_dir/$path: $!";
70	print $out $content;
71	close($out);
72	}