/[socialtext-import]/tamtam/tamtam2socialtext.pl
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /tamtam/tamtam2socialtext.pl

Parent Directory Parent Directory | Revision Log Revision Log


Revision 22 - (hide annotations)
Thu Dec 13 11:42:24 2007 UTC (16 years, 3 months ago) by dpavlin
File MIME type: text/plain
File size: 4630 byte(s)
fix meta handling which caused few pages to be skipped
1 dpavlin 5 #!/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 dpavlin 6 use HTTP::Date;
12 dpavlin 15 use POSIX qw/strftime/;
13 dpavlin 18 use File::Slurp;
14     use File::MMagic::XS;
15 dpavlin 21 use Getopt::Long;
16 dpavlin 5 use Data::Dump qw/dump/;
17    
18 dpavlin 18 my $debug = 0;
19 dpavlin 11 my $max = 999;
20 dpavlin 6
21 dpavlin 21 GetOptions(
22     'debug+' => \$debug,
23     'max=i' => \$max,
24     );
25    
26 dpavlin 5 my $page;
27 dpavlin 6 my $page_date;
28 dpavlin 5
29 dpavlin 9 my @page_names;
30    
31 dpavlin 18 print "Collecting pages...\n";
32    
33 dpavlin 5 find({
34     wanted => sub {
35     my $path = $File::Find::name;
36     return unless -f $path;
37 dpavlin 6
38 dpavlin 18 warn "+ $path\n";
39     my $ref = XMLin( $path,
40     KeyAttr => {
41     'attachment' => '+name',
42     'meta' => 'name',
43     },
44 dpavlin 22 ForceArray => [ 'attachment', 'meta', 'widget' ],
45 dpavlin 18 ) || die "can't open $path: $!";
46 dpavlin 6
47 dpavlin 18 warn "## $path = ",dump( $ref ) if $debug;
48    
49 dpavlin 6 my $name = $ref->{name} || die "no name in $path";
50 dpavlin 8
51     return if $name =~ m/^TamSystem/;
52    
53 dpavlin 14 my $date = $ref->{meta}->{LastModified}->{value};
54 dpavlin 6 if ( ! $date ) {
55     warn "SKIP: no LastModified in $path $name";
56     return;
57     }
58 dpavlin 17
59 dpavlin 18 my $data;
60 dpavlin 17
61 dpavlin 18 foreach my $w ( @{ $ref->{widgets}->{widget} } ) {
62 dpavlin 6
63 dpavlin 18 warn "## w = ",dump( $w ) if $debug;
64 dpavlin 17
65 dpavlin 18 $data .= "\n----\n" if $data;
66     $data .= $w->{data} || die "no data?";
67     }
68    
69     my $attachments;
70    
71     if ( my $a = $ref->{attachment} ) {
72     foreach my $name ( keys %$a ) {
73     my $full_path = $path;
74     $full_path =~ s,pages/,attachments/,;
75     $full_path .= '.' . $name;
76     die "$full_path doesn't exist" unless -e $full_path;
77     push @$attachments, {
78     full_path => $full_path,
79     name => ( $name || $a->{$name}->{desc} || 'noname' ),
80     };
81     }
82     }
83    
84 dpavlin 14 $page->{ $name } = {
85     content => convert_markup( $data ),
86 dpavlin 15 date => convert_date( $date ),
87 dpavlin 18 attachments => $attachments,
88 dpavlin 14 };
89 dpavlin 13
90 dpavlin 11 $name =~ s,^.+/([^/]+)$,$1,;
91     push @page_names, $name;
92    
93 dpavlin 5 },
94 dpavlin 18 no_chdir=>1,
95 dpavlin 5 }, shift @ARGV || '.');
96    
97 dpavlin 6 my @pages = ( keys %$page );
98 dpavlin 5
99 dpavlin 11 warn "found following pages: ", join(", ", @page_names),"\n";
100 dpavlin 5
101 dpavlin 11 my $page_link_re = '\b(' . join('|', @page_names) . ')\b';
102    
103 dpavlin 5 my $Rester = Socialtext::Resting->new(
104     username => 'tamtam',
105     password => 'import',
106     server => 'http://saturn.ffzg.hr/',
107 dpavlin 18 workspace => 'razmjenavjestina',
108 dpavlin 5 );
109     $Rester->put_workspacetag('TamTam');
110    
111 dpavlin 15 sub convert_date {
112     my $date = shift;
113     # return time2str( $date );
114     return strftime('%F %T %z', gmtime( $date ));
115     }
116    
117 dpavlin 5 sub header {
118     my $h = shift;
119     if ( $h =~ m/^(=+)\s+(.+?)\s+\1$/ ) {
120     my $level = length($1);
121     return "\n" . ( '^' x $level ) . " $2\n";
122     } else {
123 dpavlin 11 return $h;
124 dpavlin 5 }
125     }
126    
127     sub surround {
128     my ( $with, $what ) = @_;
129     return $with . $what . $with;
130     }
131    
132 dpavlin 12 sub pre {
133     my $text = shift;
134     $text =~ s/^{{{//;
135     $text =~ s/}}}$//;
136     return '.pre' . $text . '.pre';
137     }
138    
139 dpavlin 14 sub convert_markup {
140     my $body = shift;
141 dpavlin 6
142 dpavlin 5 $body =~ s/\Q[[TableOfContents]]\E/{toc}/gs;
143     $body =~ s/\Q[[BR]]\E/\n/gs;
144     $body =~ s/$RE{balanced}{-begin => "= |== |=== |==== |===== |===== "}{-end => " =| ==| ===| ====| ====="}{-keep}/header($1)/gse;
145     $body =~ s/''''(.+?)''''/surround('`',$1)/gse;
146     $body =~ s/'''(.+?)'''/surround('*',$1)/gse;
147     $body =~ s/''(.+?)''/surround('_',$1)/gse;
148 dpavlin 12 $body =~ s/$RE{balanced}{-begin => "{{{"}{-end => "}}}"}{-keep}/pre($1)/gse;
149 dpavlin 5
150 dpavlin 9 # fix bullets
151 dpavlin 7 $body =~ s/^\s+([\*])/$1/gm;
152    
153 dpavlin 9 # fix links
154     $body =~ s/\["([^"]+)"\]/[$1]/gs;
155 dpavlin 10 $body =~ s,\[(http://\S+)\s+([^\]]+)\],"$2"<$1>,gs;
156     $body =~ s,\[(http://[^\]]+)\],$1,gs;
157 dpavlin 9
158 dpavlin 14 # fix hr
159     $body =~ s,(\S+)----,$1\n----,gs;
160     $body =~ s,----(\S+),----\n$1,gs;
161 dpavlin 5
162 dpavlin 18 # attachments
163 dpavlin 19 $body =~ s,\[attachment:([^\]]+)(gif|png|jpg|jpeg)\],{image: $1$2},gis;
164 dpavlin 18 $body =~ s,\[attachment:([^\]]+)\],{file: $1},gs;
165    
166 dpavlin 14 return $body;
167     }
168    
169     my $count = 0;
170    
171 dpavlin 18 my $m = File::MMagic::XS->new;
172    
173 dpavlin 14 foreach my $name ( keys %$page ) {
174     last if $count++ == $max;
175    
176     my $p = $page->{$name};
177 dpavlin 17
178     warn "## $name = ",dump( $p ) if $debug;
179    
180 dpavlin 14 my $body = $p->{content} || die "no content?";
181     my $date = $p->{date} || die "no date?";
182    
183 dpavlin 6 my @tags = ( 'TamTam' );
184 dpavlin 5
185 dpavlin 15 my $full_name = $name;
186    
187 dpavlin 6 if ( $name =~ m!/! ) {
188     my @page_tags = split(m!/!, $name);
189 dpavlin 8 $name = pop @page_tags; # remove page name
190 dpavlin 6 push @tags, @page_tags;
191     }
192    
193 dpavlin 14 # link named pages
194 dpavlin 16 $body =~ s,\b$page_link_re\b,[$1],gs;
195     $body =~ s,``,,gs;
196 dpavlin 14
197     $body .= qq{
198 dpavlin 22
199 dpavlin 14 ----
200    
201 dpavlin 20 "original"<http://www.razmjenavjestina.org/$full_name> {date: $date}
202 dpavlin 14 };
203    
204     Encode::_utf8_off( $body );
205    
206 dpavlin 18 $Rester->put_page( $name, { content => $body, date => $date });
207     print "$name $date\n";
208 dpavlin 6 foreach ( @tags ) {
209     $Rester->put_pagetag( $name, $_ );
210 dpavlin 18 print "+ tag $_\n";
211 dpavlin 6 }
212 dpavlin 18 foreach my $a ( @{ $p->{attachments} } ) {
213     my $type = $m->get_mime( $a->{full_path} );
214     my $content = read_file( $a->{full_path} );
215     print "+ attachment ", $a->{name}," $type ", length($content), " bytes\n";
216     $Rester->post_attachment($name, $a->{name}, $content, $type );
217     }
218 dpavlin 6
219 dpavlin 5 }
220    

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26