10 |
use Encode; |
use Encode; |
11 |
use HTTP::Date; |
use HTTP::Date; |
12 |
use POSIX qw/strftime/; |
use POSIX qw/strftime/; |
13 |
|
use File::Slurp; |
14 |
|
use File::MMagic::XS; |
15 |
|
use Getopt::Long; |
16 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
17 |
|
|
18 |
|
my $debug = 0; |
19 |
my $max = 999; |
my $max = 999; |
20 |
|
my $attachments = 0; |
21 |
|
|
22 |
|
my @create_tags = (qw/ |
23 |
|
Trazi |
24 |
|
Nudi |
25 |
|
SvakodnevneDovitljivosti |
26 |
|
G33koSkop |
27 |
|
/); |
28 |
|
|
29 |
|
GetOptions( |
30 |
|
'debug+' => \$debug, |
31 |
|
'max=i' => \$max, |
32 |
|
'attachments' => \$attachments, |
33 |
|
); |
34 |
|
|
35 |
my $page; |
my $page; |
36 |
my $page_date; |
my $page_date; |
37 |
|
|
38 |
my @page_names; |
my @page_names; |
39 |
|
|
40 |
|
print "Importing $max pages", $attachments ? " with attachments" : "", "...\n"; |
41 |
|
|
42 |
find({ |
find({ |
43 |
wanted => sub { |
wanted => sub { |
44 |
my $path = $File::Find::name; |
my $path = $File::Find::name; |
45 |
return unless -f $path; |
return unless -f $path; |
46 |
|
|
47 |
warn "# $path\n"; |
warn "+ $path\n"; |
48 |
my $ref = XMLin( $path ) || die "can't open $path: $!"; |
my $ref = XMLin( $path, |
49 |
|
KeyAttr => { |
50 |
|
'attachment' => '+name', |
51 |
|
'meta' => 'name', |
52 |
|
}, |
53 |
|
ForceArray => [ 'attachment', 'meta', 'widget' ], |
54 |
|
) || die "can't open $path: $!"; |
55 |
|
|
56 |
|
warn "## $path = ",dump( $ref ) if $debug; |
57 |
|
|
58 |
my $name = $ref->{name} || die "no name in $path"; |
my $name = $ref->{name} || die "no name in $path"; |
59 |
|
|
64 |
warn "SKIP: no LastModified in $path $name"; |
warn "SKIP: no LastModified in $path $name"; |
65 |
return; |
return; |
66 |
} |
} |
67 |
my $data = |
|
68 |
$ref->{widgets}->{widget}->{data} || |
my $data; |
69 |
$ref->{widgets}->{widget}->{Body}->{data} || |
|
70 |
die "no data in $path ",dump( $ref ); |
foreach my $w ( @{ $ref->{widgets}->{widget} } ) { |
71 |
|
|
72 |
|
warn "## w = ",dump( $w ) if $debug; |
73 |
|
|
74 |
|
$data .= "\n----\n" if $data; |
75 |
|
$data .= $w->{data} || die "no data?"; |
76 |
|
} |
77 |
|
|
78 |
|
my $attachments; |
79 |
|
|
80 |
|
if ( my $a = $ref->{attachment} ) { |
81 |
|
foreach my $name ( keys %$a ) { |
82 |
|
my $full_path = $path; |
83 |
|
$full_path =~ s,pages/,attachments/,; |
84 |
|
$full_path .= '.' . $name; |
85 |
|
die "$full_path doesn't exist" unless -e $full_path; |
86 |
|
push @$attachments, { |
87 |
|
full_path => $full_path, |
88 |
|
name => ( $name || $a->{$name}->{desc} || 'noname' ), |
89 |
|
}; |
90 |
|
} |
91 |
|
} |
92 |
|
|
93 |
$page->{ $name } = { |
$page->{ $name } = { |
94 |
content => convert_markup( $data ), |
content => convert_markup( $data ), |
95 |
|
original => $data, |
96 |
date => convert_date( $date ), |
date => convert_date( $date ), |
97 |
|
attachments => $attachments, |
98 |
}; |
}; |
99 |
|
|
|
# strip path from page name |
|
100 |
$name =~ s,^.+/([^/]+)$,$1,; |
$name =~ s,^.+/([^/]+)$,$1,; |
101 |
push @page_names, $name; |
push @page_names, $name; |
102 |
|
|
|
# warn dump( $ref ); |
|
103 |
}, |
}, |
104 |
|
no_chdir=>1, |
105 |
}, shift @ARGV || '.'); |
}, shift @ARGV || '.'); |
106 |
|
|
107 |
my @pages = ( keys %$page ); |
my @pages = ( keys %$page ); |
114 |
username => 'tamtam', |
username => 'tamtam', |
115 |
password => 'import', |
password => 'import', |
116 |
server => 'http://saturn.ffzg.hr/', |
server => 'http://saturn.ffzg.hr/', |
117 |
|
workspace => 'razmjenavjestina', |
118 |
); |
); |
|
$Rester->workspace('razmjenavjestina'); |
|
119 |
$Rester->put_workspacetag('TamTam'); |
$Rester->put_workspacetag('TamTam'); |
120 |
|
|
121 |
sub convert_date { |
sub convert_date { |
141 |
|
|
142 |
sub pre { |
sub pre { |
143 |
my $text = shift; |
my $text = shift; |
144 |
$text =~ s/^{{{//; |
$text =~ s/^{{{\s*//s; |
145 |
$text =~ s/}}}$//; |
$text =~ s/\s*}}}$//s; |
146 |
return '.pre' . $text . '.pre'; |
return "\n.pre\n" . $text . "\n.pre\n"; |
147 |
} |
} |
148 |
|
|
149 |
sub convert_markup { |
sub convert_markup { |
169 |
$body =~ s,(\S+)----,$1\n----,gs; |
$body =~ s,(\S+)----,$1\n----,gs; |
170 |
$body =~ s,----(\S+),----\n$1,gs; |
$body =~ s,----(\S+),----\n$1,gs; |
171 |
|
|
172 |
|
# attachments |
173 |
|
$body =~ s,\[attachment:([^\]]+)(gif|png|jpg|jpeg)\],{image: $1$2},gis; |
174 |
|
$body =~ s,\[attachment:([^\]]+)\],{file: $1},gs; |
175 |
|
|
176 |
return $body; |
return $body; |
177 |
} |
} |
178 |
|
|
179 |
my $count = 0; |
my $count = 0; |
180 |
|
|
181 |
|
my $m = File::MMagic::XS->new; |
182 |
|
|
183 |
foreach my $name ( keys %$page ) { |
foreach my $name ( keys %$page ) { |
184 |
last if $count++ == $max; |
last if $count++ == $max; |
185 |
|
|
186 |
my $p = $page->{$name}; |
my $p = $page->{$name}; |
187 |
|
|
188 |
|
warn "## $name = ",dump( $p ) if $debug; |
189 |
|
|
190 |
my $body = $p->{content} || die "no content?"; |
my $body = $p->{content} || die "no content?"; |
191 |
my $date = $p->{date} || die "no date?"; |
my $date = $p->{date} || die "no date?"; |
192 |
|
|
205 |
$body =~ s,``,,gs; |
$body =~ s,``,,gs; |
206 |
|
|
207 |
$body .= qq{ |
$body .= qq{ |
208 |
|
|
209 |
---- |
---- |
210 |
|
|
211 |
Original sa http://www.razmjenavjestina.org/$full_name zadnja promjena {date: $date} |
"original"<http://www.razmjenavjestina.org/$full_name> {date: $date} |
212 |
}; |
}; |
213 |
|
|
214 |
Encode::_utf8_off( $body ); |
Encode::_utf8_off( $body ); |
215 |
|
|
216 |
$Rester->put_page( $name, { |
print "$name $date\n"; |
217 |
content => $body, |
|
218 |
date => $date, |
# original markup |
219 |
}); |
$Rester->put_page( $name, { content => $p->{original}, date => $date }); |
220 |
print "+ $name $date\n"; |
|
221 |
|
foreach my $t ( @create_tags ) { |
222 |
|
push @tags, $t if $full_name =~ m/$t/i; |
223 |
|
} |
224 |
|
|
225 |
foreach ( @tags ) { |
foreach ( @tags ) { |
226 |
$Rester->put_pagetag( $name, $_ ); |
$Rester->put_pagetag( $name, $_, { date => $date } ); |
227 |
print "+ $name [$_]\n"; |
print "+ tag $_\n"; |
228 |
|
} |
229 |
|
|
230 |
|
if ( $attachments ) { |
231 |
|
foreach my $a ( @{ $p->{attachments} } ) { |
232 |
|
my $type = $m->get_mime( $a->{full_path} ); |
233 |
|
my $content = read_file( $a->{full_path} ); |
234 |
|
print "+ attachment ", $a->{name}," $type ", length($content), " bytes\n"; |
235 |
|
$Rester->post_attachment($name, $a->{name}, $content, $type ); |
236 |
|
} |
237 |
} |
} |
238 |
|
|
239 |
|
# converted page |
240 |
|
$Rester->put_page( $name, { content => $body, date => $date }); |
241 |
|
|
242 |
} |
} |
243 |
|
|