1 |
#!/usr/bin/perl |
2 |
|
3 |
use warnings; |
4 |
use strict; |
5 |
|
6 |
use XML::Simple; |
7 |
use File::Find; |
8 |
use Regexp::Common qw/balanced/; |
9 |
use Socialtext::Resting; |
10 |
use Encode; |
11 |
use HTTP::Date; |
12 |
use Data::Dump qw/dump/; |
13 |
|
14 |
my $max = 999; |
15 |
|
16 |
my $page; |
17 |
my $page_date; |
18 |
|
19 |
my @page_names; |
20 |
|
21 |
find({ |
22 |
wanted => sub { |
23 |
my $path = $File::Find::name; |
24 |
return unless -f $path; |
25 |
|
26 |
warn "# $path\n"; |
27 |
my $ref = XMLin( $path ) || die "can't open $path: $!"; |
28 |
|
29 |
my $name = $ref->{name} || die "no name in $path"; |
30 |
|
31 |
return if $name =~ m/^TamSystem/; |
32 |
|
33 |
my $date = $ref->{meta}->{LastModified}; |
34 |
if ( ! $date ) { |
35 |
warn "SKIP: no LastModified in $path $name"; |
36 |
return; |
37 |
} |
38 |
my $data = |
39 |
$ref->{widgets}->{widget}->{data} || |
40 |
$ref->{widgets}->{widget}->{Body}->{data} || |
41 |
die "no data in $path ",dump( $ref ); |
42 |
|
43 |
$data .= qq{ |
44 |
---- |
45 |
|
46 |
Original: http://www.razmjenavjestina.org/$path |
47 |
}; |
48 |
|
49 |
$page->{ $name } = [ $data, $date ]; |
50 |
|
51 |
# strip path from page name |
52 |
$name =~ s,^.+/([^/]+)$,$1,; |
53 |
push @page_names, $name; |
54 |
|
55 |
# warn dump( $ref ); |
56 |
}, |
57 |
}, shift @ARGV || '.'); |
58 |
|
59 |
my @pages = ( keys %$page ); |
60 |
|
61 |
warn "found following pages: ", join(", ", @page_names),"\n"; |
62 |
|
63 |
my $page_link_re = '\b(' . join('|', @page_names) . ')\b'; |
64 |
|
65 |
my $Rester = Socialtext::Resting->new( |
66 |
username => 'tamtam', |
67 |
password => 'import', |
68 |
server => 'http://saturn.ffzg.hr/', |
69 |
); |
70 |
$Rester->workspace('razmjenavjestina'); |
71 |
$Rester->put_workspacetag('TamTam'); |
72 |
|
73 |
sub header { |
74 |
my $h = shift; |
75 |
if ( $h =~ m/^(=+)\s+(.+?)\s+\1$/ ) { |
76 |
my $level = length($1); |
77 |
return "\n" . ( '^' x $level ) . " $2\n"; |
78 |
} else { |
79 |
return $h; |
80 |
} |
81 |
} |
82 |
|
83 |
sub surround { |
84 |
my ( $with, $what ) = @_; |
85 |
return $with . $what . $with; |
86 |
} |
87 |
|
88 |
sub pre { |
89 |
my $text = shift; |
90 |
$text =~ s/^{{{//; |
91 |
$text =~ s/}}}$//; |
92 |
return '.pre' . $text . '.pre'; |
93 |
} |
94 |
|
95 |
my $count = 0; |
96 |
|
97 |
foreach my $name ( keys %$page ) { |
98 |
last if $count++ == $max; |
99 |
|
100 |
my ( $body, $date ) = @{ $page->{$name} }; |
101 |
$date = time2str( $date ); |
102 |
|
103 |
$body =~ s/\Q[[TableOfContents]]\E/{toc}/gs; |
104 |
$body =~ s/\Q[[BR]]\E/\n/gs; |
105 |
$body =~ s/$RE{balanced}{-begin => "= |== |=== |==== |===== |===== "}{-end => " =| ==| ===| ====| ====="}{-keep}/header($1)/gse; |
106 |
$body =~ s/''''(.+?)''''/surround('`',$1)/gse; |
107 |
$body =~ s/'''(.+?)'''/surround('*',$1)/gse; |
108 |
$body =~ s/''(.+?)''/surround('_',$1)/gse; |
109 |
$body =~ s/$RE{balanced}{-begin => "{{{"}{-end => "}}}"}{-keep}/pre($1)/gse; |
110 |
|
111 |
# fix bullets |
112 |
$body =~ s/^\s+([\*])/$1/gm; |
113 |
|
114 |
# fix links |
115 |
$body =~ s/\["([^"]+)"\]/[$1]/gs; |
116 |
$body =~ s,\[(http://\S+)\s+([^\]]+)\],"$2"<$1>,gs; |
117 |
$body =~ s,\[(http://[^\]]+)\],$1,gs; |
118 |
$body =~ s,$page_link_re,[$1],gs; |
119 |
|
120 |
Encode::_utf8_off( $body ); |
121 |
|
122 |
my @tags = ( 'TamTam' ); |
123 |
|
124 |
if ( $name =~ m!/! ) { |
125 |
my @page_tags = split(m!/!, $name); |
126 |
$name = pop @page_tags; # remove page name |
127 |
push @tags, @page_tags; |
128 |
} |
129 |
|
130 |
$Rester->put_page( $name, { |
131 |
content => $body, |
132 |
date => $date, |
133 |
}); |
134 |
print "+ $name\n"; |
135 |
foreach ( @tags ) { |
136 |
$Rester->put_pagetag( $name, $_ ); |
137 |
print "+ $name [$_]\n"; |
138 |
} |
139 |
|
140 |
} |
141 |
|