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 POSIX qw/strftime/; |
13 |
use Data::Dump qw/dump/; |
14 |
|
15 |
my $max = 999; |
16 |
|
17 |
my $page; |
18 |
my $page_date; |
19 |
|
20 |
my @page_names; |
21 |
|
22 |
find({ |
23 |
wanted => sub { |
24 |
my $path = $File::Find::name; |
25 |
return unless -f $path; |
26 |
|
27 |
warn "# $path\n"; |
28 |
my $ref = XMLin( $path ) || die "can't open $path: $!"; |
29 |
|
30 |
my $name = $ref->{name} || die "no name in $path"; |
31 |
|
32 |
return if $name =~ m/^TamSystem/; |
33 |
|
34 |
my $date = $ref->{meta}->{LastModified}->{value}; |
35 |
if ( ! $date ) { |
36 |
warn "SKIP: no LastModified in $path $name"; |
37 |
return; |
38 |
} |
39 |
my $data = |
40 |
$ref->{widgets}->{widget}->{data} || |
41 |
$ref->{widgets}->{widget}->{Body}->{data} || |
42 |
die "no data in $path ",dump( $ref ); |
43 |
|
44 |
$page->{ $name } = { |
45 |
content => convert_markup( $data ), |
46 |
date => convert_date( $date ), |
47 |
}; |
48 |
|
49 |
# strip path from page name |
50 |
$name =~ s,^.+/([^/]+)$,$1,; |
51 |
push @page_names, $name; |
52 |
|
53 |
# warn dump( $ref ); |
54 |
}, |
55 |
}, shift @ARGV || '.'); |
56 |
|
57 |
my @pages = ( keys %$page ); |
58 |
|
59 |
warn "found following pages: ", join(", ", @page_names),"\n"; |
60 |
|
61 |
my $page_link_re = '\b(' . join('|', @page_names) . ')\b'; |
62 |
|
63 |
my $Rester = Socialtext::Resting->new( |
64 |
username => 'tamtam', |
65 |
password => 'import', |
66 |
server => 'http://saturn.ffzg.hr/', |
67 |
); |
68 |
$Rester->workspace('razmjenavjestina'); |
69 |
$Rester->put_workspacetag('TamTam'); |
70 |
|
71 |
sub convert_date { |
72 |
my $date = shift; |
73 |
# return time2str( $date ); |
74 |
return strftime('%F %T %z', gmtime( $date )); |
75 |
} |
76 |
|
77 |
sub header { |
78 |
my $h = shift; |
79 |
if ( $h =~ m/^(=+)\s+(.+?)\s+\1$/ ) { |
80 |
my $level = length($1); |
81 |
return "\n" . ( '^' x $level ) . " $2\n"; |
82 |
} else { |
83 |
return $h; |
84 |
} |
85 |
} |
86 |
|
87 |
sub surround { |
88 |
my ( $with, $what ) = @_; |
89 |
return $with . $what . $with; |
90 |
} |
91 |
|
92 |
sub pre { |
93 |
my $text = shift; |
94 |
$text =~ s/^{{{//; |
95 |
$text =~ s/}}}$//; |
96 |
return '.pre' . $text . '.pre'; |
97 |
} |
98 |
|
99 |
sub convert_markup { |
100 |
my $body = shift; |
101 |
|
102 |
$body =~ s/\Q[[TableOfContents]]\E/{toc}/gs; |
103 |
$body =~ s/\Q[[BR]]\E/\n/gs; |
104 |
$body =~ s/$RE{balanced}{-begin => "= |== |=== |==== |===== |===== "}{-end => " =| ==| ===| ====| ====="}{-keep}/header($1)/gse; |
105 |
$body =~ s/''''(.+?)''''/surround('`',$1)/gse; |
106 |
$body =~ s/'''(.+?)'''/surround('*',$1)/gse; |
107 |
$body =~ s/''(.+?)''/surround('_',$1)/gse; |
108 |
$body =~ s/$RE{balanced}{-begin => "{{{"}{-end => "}}}"}{-keep}/pre($1)/gse; |
109 |
|
110 |
# fix bullets |
111 |
$body =~ s/^\s+([\*])/$1/gm; |
112 |
|
113 |
# fix links |
114 |
$body =~ s/\["([^"]+)"\]/[$1]/gs; |
115 |
$body =~ s,\[(http://\S+)\s+([^\]]+)\],"$2"<$1>,gs; |
116 |
$body =~ s,\[(http://[^\]]+)\],$1,gs; |
117 |
|
118 |
# fix hr |
119 |
$body =~ s,(\S+)----,$1\n----,gs; |
120 |
$body =~ s,----(\S+),----\n$1,gs; |
121 |
|
122 |
return $body; |
123 |
} |
124 |
|
125 |
my $count = 0; |
126 |
|
127 |
foreach my $name ( keys %$page ) { |
128 |
last if $count++ == $max; |
129 |
|
130 |
my $p = $page->{$name}; |
131 |
my $body = $p->{content} || die "no content?"; |
132 |
my $date = $p->{date} || die "no date?"; |
133 |
|
134 |
my @tags = ( 'TamTam' ); |
135 |
|
136 |
my $full_name = $name; |
137 |
|
138 |
if ( $name =~ m!/! ) { |
139 |
my @page_tags = split(m!/!, $name); |
140 |
$name = pop @page_tags; # remove page name |
141 |
push @tags, @page_tags; |
142 |
} |
143 |
|
144 |
# link named pages |
145 |
$body =~ s,$page_link_re,[$1],gs; |
146 |
|
147 |
$body .= qq{ |
148 |
---- |
149 |
|
150 |
Original sa http://www.razmjenavjestina.org/$full_name zadnja promjena {date: $date} |
151 |
}; |
152 |
|
153 |
Encode::_utf8_off( $body ); |
154 |
|
155 |
$Rester->put_page( $name, { |
156 |
content => $body, |
157 |
date => $date, |
158 |
}); |
159 |
print "+ $name $date\n"; |
160 |
foreach ( @tags ) { |
161 |
$Rester->put_pagetag( $name, $_ ); |
162 |
print "+ $name [$_]\n"; |
163 |
} |
164 |
|
165 |
} |
166 |
|