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