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 $debug = 1; |
16 |
my $max = 999; |
17 |
|
18 |
my $page; |
19 |
my $page_date; |
20 |
|
21 |
my @page_names; |
22 |
|
23 |
find({ |
24 |
wanted => sub { |
25 |
my $path = $File::Find::name; |
26 |
return unless -f $path; |
27 |
|
28 |
warn "# $path\n"; |
29 |
my $ref = XMLin( $path ) || die "can't open $path: $!"; |
30 |
|
31 |
my $name = $ref->{name} || die "no name in $path"; |
32 |
|
33 |
return if $name =~ m/^TamSystem/; |
34 |
|
35 |
my $date = $ref->{meta}->{LastModified}->{value}; |
36 |
if ( ! $date ) { |
37 |
warn "SKIP: no LastModified in $path $name"; |
38 |
return; |
39 |
} |
40 |
|
41 |
my $w = $ref->{widgets}->{widget}; |
42 |
|
43 |
my $data = |
44 |
defined( $w->{data} ) ? $w->{data} : |
45 |
defined( $w->{Body}->{date} ) ? $w->{Body}->{date} : |
46 |
die "no data in $path ",dump( $ref ); |
47 |
|
48 |
# empty data is returned like empty hash. yack. |
49 |
$data = "\n" if ref($data) eq 'HASH' and ! keys %$data; |
50 |
|
51 |
$page->{ $name } = { |
52 |
content => convert_markup( $data ), |
53 |
date => convert_date( $date ), |
54 |
}; |
55 |
|
56 |
# strip path from page name |
57 |
$name =~ s,^.+/([^/]+)$,$1,; |
58 |
push @page_names, $name; |
59 |
|
60 |
warn "## $path = ",dump( $ref ) if $debug; |
61 |
}, |
62 |
}, shift @ARGV || '.'); |
63 |
|
64 |
my @pages = ( keys %$page ); |
65 |
|
66 |
warn "found following pages: ", join(", ", @page_names),"\n"; |
67 |
|
68 |
my $page_link_re = '\b(' . join('|', @page_names) . ')\b'; |
69 |
|
70 |
my $Rester = Socialtext::Resting->new( |
71 |
username => 'tamtam', |
72 |
password => 'import', |
73 |
server => 'http://saturn.ffzg.hr/', |
74 |
); |
75 |
$Rester->workspace('razmjenavjestina'); |
76 |
$Rester->put_workspacetag('TamTam'); |
77 |
|
78 |
sub convert_date { |
79 |
my $date = shift; |
80 |
# return time2str( $date ); |
81 |
return strftime('%F %T %z', gmtime( $date )); |
82 |
} |
83 |
|
84 |
sub header { |
85 |
my $h = shift; |
86 |
if ( $h =~ m/^(=+)\s+(.+?)\s+\1$/ ) { |
87 |
my $level = length($1); |
88 |
return "\n" . ( '^' x $level ) . " $2\n"; |
89 |
} else { |
90 |
return $h; |
91 |
} |
92 |
} |
93 |
|
94 |
sub surround { |
95 |
my ( $with, $what ) = @_; |
96 |
return $with . $what . $with; |
97 |
} |
98 |
|
99 |
sub pre { |
100 |
my $text = shift; |
101 |
$text =~ s/^{{{//; |
102 |
$text =~ s/}}}$//; |
103 |
return '.pre' . $text . '.pre'; |
104 |
} |
105 |
|
106 |
sub convert_markup { |
107 |
my $body = shift; |
108 |
|
109 |
$body =~ s/\Q[[TableOfContents]]\E/{toc}/gs; |
110 |
$body =~ s/\Q[[BR]]\E/\n/gs; |
111 |
$body =~ s/$RE{balanced}{-begin => "= |== |=== |==== |===== |===== "}{-end => " =| ==| ===| ====| ====="}{-keep}/header($1)/gse; |
112 |
$body =~ s/''''(.+?)''''/surround('`',$1)/gse; |
113 |
$body =~ s/'''(.+?)'''/surround('*',$1)/gse; |
114 |
$body =~ s/''(.+?)''/surround('_',$1)/gse; |
115 |
$body =~ s/$RE{balanced}{-begin => "{{{"}{-end => "}}}"}{-keep}/pre($1)/gse; |
116 |
|
117 |
# fix bullets |
118 |
$body =~ s/^\s+([\*])/$1/gm; |
119 |
|
120 |
# fix links |
121 |
$body =~ s/\["([^"]+)"\]/[$1]/gs; |
122 |
$body =~ s,\[(http://\S+)\s+([^\]]+)\],"$2"<$1>,gs; |
123 |
$body =~ s,\[(http://[^\]]+)\],$1,gs; |
124 |
|
125 |
# fix hr |
126 |
$body =~ s,(\S+)----,$1\n----,gs; |
127 |
$body =~ s,----(\S+),----\n$1,gs; |
128 |
|
129 |
return $body; |
130 |
} |
131 |
|
132 |
my $count = 0; |
133 |
|
134 |
foreach my $name ( keys %$page ) { |
135 |
last if $count++ == $max; |
136 |
|
137 |
my $p = $page->{$name}; |
138 |
|
139 |
warn "## $name = ",dump( $p ) if $debug; |
140 |
|
141 |
my $body = $p->{content} || die "no content?"; |
142 |
my $date = $p->{date} || die "no date?"; |
143 |
|
144 |
my @tags = ( 'TamTam' ); |
145 |
|
146 |
my $full_name = $name; |
147 |
|
148 |
if ( $name =~ m!/! ) { |
149 |
my @page_tags = split(m!/!, $name); |
150 |
$name = pop @page_tags; # remove page name |
151 |
push @tags, @page_tags; |
152 |
} |
153 |
|
154 |
# link named pages |
155 |
$body =~ s,\b$page_link_re\b,[$1],gs; |
156 |
$body =~ s,``,,gs; |
157 |
|
158 |
$body .= qq{ |
159 |
---- |
160 |
|
161 |
Original sa http://www.razmjenavjestina.org/$full_name zadnja promjena {date: $date} |
162 |
}; |
163 |
|
164 |
Encode::_utf8_off( $body ); |
165 |
|
166 |
$Rester->put_page( $name, { |
167 |
content => $body, |
168 |
date => $date, |
169 |
}); |
170 |
print "+ $name $date\n"; |
171 |
foreach ( @tags ) { |
172 |
$Rester->put_pagetag( $name, $_ ); |
173 |
print "+ $name [$_]\n"; |
174 |
} |
175 |
|
176 |
} |
177 |
|