line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package DDG::Publisher::Site::Duckduckhack::Root; |
2
|
|
|
|
|
|
|
$DDG::Publisher::Site::Duckduckhack::Root::VERSION = '1044'; |
3
|
1
|
|
|
1
|
|
1803
|
use MooX; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
9
|
|
4
|
|
|
|
|
|
|
|
5
|
1
|
|
|
1
|
|
833
|
use IO::All; |
|
1
|
|
|
|
|
3
|
|
|
1
|
|
|
|
|
10
|
|
6
|
1
|
|
|
1
|
|
662
|
use Data::Dumper; |
|
1
|
|
|
|
|
5544
|
|
|
1
|
|
|
|
|
60
|
|
7
|
1
|
|
|
1
|
|
473
|
use DDP; |
|
1
|
|
|
|
|
28542
|
|
|
1
|
|
|
|
|
11
|
|
8
|
|
|
|
|
|
|
|
9
|
1
|
|
|
1
|
|
63
|
use File::Path qw(make_path); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
56
|
|
10
|
1
|
|
|
1
|
|
6
|
use File::Find; |
|
1
|
|
|
|
|
1
|
|
|
1
|
|
|
|
|
60
|
|
11
|
1
|
|
|
1
|
|
326
|
use File::chdir; |
|
0
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
use Path::Class; |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
use Markdent::Handler::HTMLStream::Document; |
16
|
|
|
|
|
|
|
use Markdent::Parser; |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
use Text::Trim; |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
use JSON; |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
use IPC::Run qw{run timeout}; |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
with qw( |
25
|
|
|
|
|
|
|
DDG::Publisher::DirRole |
26
|
|
|
|
|
|
|
); |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
has nav_file => ( |
29
|
|
|
|
|
|
|
is => 'ro', |
30
|
|
|
|
|
|
|
lazy => 1, |
31
|
|
|
|
|
|
|
builder => sub {'ddh-index.md'}, |
32
|
|
|
|
|
|
|
); |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
has prev_next_file => ( |
35
|
|
|
|
|
|
|
is => 'ro', |
36
|
|
|
|
|
|
|
lazy => 1, |
37
|
|
|
|
|
|
|
builder => sub {'ddh-prev-next.json'}, |
38
|
|
|
|
|
|
|
); |
39
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
has source_dir => ( |
41
|
|
|
|
|
|
|
is => 'ro', |
42
|
|
|
|
|
|
|
lazy => 1, |
43
|
|
|
|
|
|
|
builder => 1, |
44
|
|
|
|
|
|
|
); |
45
|
|
|
|
|
|
|
|
46
|
|
|
|
|
|
|
sub _build_source_dir { |
47
|
|
|
|
|
|
|
my ( $self ) = @_; |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
my $cache_dir = $self->site->publisher->cache_dir; |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
my $source_dir = dir($cache_dir,'duckduckgo-documentation/duckduckhack'); |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
{ |
54
|
|
|
|
|
|
|
my ($in, $out, $err); |
55
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
# die $cache_dir; |
57
|
|
|
|
|
|
|
if (-d $source_dir) { |
58
|
|
|
|
|
|
|
local $CWD = $source_dir; |
59
|
|
|
|
|
|
|
run [ |
60
|
|
|
|
|
|
|
'git', 'pull' |
61
|
|
|
|
|
|
|
], \$in, \$out, \$err, timeout(60) or die "$err (error $?) $out"; |
62
|
|
|
|
|
|
|
} else { |
63
|
|
|
|
|
|
|
local $CWD = $cache_dir; |
64
|
|
|
|
|
|
|
run [ |
65
|
|
|
|
|
|
|
'git', 'clone', 'git@github.com:duckduckgo/duckduckgo-documentation.git' |
66
|
|
|
|
|
|
|
], \$in, \$out, \$err, timeout(60) or die "$err (error $?) $out"; |
67
|
|
|
|
|
|
|
} |
68
|
|
|
|
|
|
|
} |
69
|
|
|
|
|
|
|
|
70
|
|
|
|
|
|
|
# warn qq(DUCKDUCKHACK source_dir: $source_dir\n); |
71
|
|
|
|
|
|
|
# die; |
72
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
return $source_dir; |
74
|
|
|
|
|
|
|
} |
75
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
sub path { '/' } |
77
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
# Builds the prev/next hash to push to each template. |
79
|
|
|
|
|
|
|
sub get_prev_next { |
80
|
|
|
|
|
|
|
my $self = shift; |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
my $prev_next_path = dir($self->source_dir,$self->prev_next_file); |
83
|
|
|
|
|
|
|
my $json = io($prev_next_path)->slurp; |
84
|
|
|
|
|
|
|
$json =~ s/\.md//g; |
85
|
|
|
|
|
|
|
my $data = decode_json $json; |
86
|
|
|
|
|
|
|
|
87
|
|
|
|
|
|
|
# die p($data); |
88
|
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
return $data; |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
# Builds the nav hash to push to each template. |
93
|
|
|
|
|
|
|
sub get_nav { |
94
|
|
|
|
|
|
|
my $self = shift; |
95
|
|
|
|
|
|
|
|
96
|
|
|
|
|
|
|
my $nav_path = dir($self->source_dir,$self->nav_file); |
97
|
|
|
|
|
|
|
my $markdown = io($nav_path)->slurp; |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
# die $markdown; |
100
|
|
|
|
|
|
|
|
101
|
|
|
|
|
|
|
my %nav = (); |
102
|
|
|
|
|
|
|
my @nav = (); |
103
|
|
|
|
|
|
|
HEADING: foreach my $heading (split(/\n\-/s,$markdown)) { |
104
|
|
|
|
|
|
|
next HEADING if !$heading; |
105
|
|
|
|
|
|
|
next HEADING if $heading !~ /\*\*/; |
106
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
my $title = ''; |
108
|
|
|
|
|
|
|
my @sec = (); |
109
|
|
|
|
|
|
|
LINE: foreach my $line (split(/\n/,$heading)) { |
110
|
|
|
|
|
|
|
if ($line =~ /\*\*([^\*]+)/) { |
111
|
|
|
|
|
|
|
$title = $1; |
112
|
|
|
|
|
|
|
|
113
|
|
|
|
|
|
|
# Capturing markdown link. |
114
|
|
|
|
|
|
|
} elsif ($line =~ /\[([^\]]+)\]\(([^\)]+)/) { |
115
|
|
|
|
|
|
|
my $section = $1; |
116
|
|
|
|
|
|
|
my $link = $2; |
117
|
|
|
|
|
|
|
$link =~ s/^.*\/(.*)\.md/$1/; |
118
|
|
|
|
|
|
|
my %sec = (); |
119
|
|
|
|
|
|
|
$sec{'title'} = $section; |
120
|
|
|
|
|
|
|
$sec{'link'} = $link; |
121
|
|
|
|
|
|
|
push(@sec,\%sec); |
122
|
|
|
|
|
|
|
|
123
|
|
|
|
|
|
|
$nav{$link} = { |
124
|
|
|
|
|
|
|
'category' => $title, |
125
|
|
|
|
|
|
|
'title' => $section, |
126
|
|
|
|
|
|
|
} |
127
|
|
|
|
|
|
|
} |
128
|
|
|
|
|
|
|
} |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
if ($title) { |
131
|
|
|
|
|
|
|
my %sec = (); |
132
|
|
|
|
|
|
|
$sec{'title'} = $title; |
133
|
|
|
|
|
|
|
$sec{'sections'} = \@sec; |
134
|
|
|
|
|
|
|
push(@nav,\%sec); |
135
|
|
|
|
|
|
|
} |
136
|
|
|
|
|
|
|
} |
137
|
|
|
|
|
|
|
|
138
|
|
|
|
|
|
|
# die p(%nav); |
139
|
|
|
|
|
|
|
return \@nav, \%nav; |
140
|
|
|
|
|
|
|
} |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
# For just testing output. |
143
|
|
|
|
|
|
|
my $dir_output = '/usr/local/ddg/www-static/duckduckhack.com/tmp'; |
144
|
|
|
|
|
|
|
|
145
|
|
|
|
|
|
|
sub pages { |
146
|
|
|
|
|
|
|
my $self = shift; |
147
|
|
|
|
|
|
|
|
148
|
|
|
|
|
|
|
my %pages = (); |
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
# Old index. |
151
|
|
|
|
|
|
|
$pages{'index'} = sub {}; |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
my $prev_next_hash_ref = $self->get_prev_next; |
154
|
|
|
|
|
|
|
my %prev_next = %{$prev_next_hash_ref}; |
155
|
|
|
|
|
|
|
|
156
|
|
|
|
|
|
|
my ($nav_arr_ref,$nav_hash_ref) = $self->get_nav; |
157
|
|
|
|
|
|
|
my %nav = %{$nav_hash_ref}; |
158
|
|
|
|
|
|
|
|
159
|
|
|
|
|
|
|
find(sub { |
160
|
|
|
|
|
|
|
|
161
|
|
|
|
|
|
|
my $name = $File::Find::name; |
162
|
|
|
|
|
|
|
my $dir = $File::Find::dir; |
163
|
|
|
|
|
|
|
|
164
|
|
|
|
|
|
|
return if $name =~ /$self->{nav_file}/; |
165
|
|
|
|
|
|
|
|
166
|
|
|
|
|
|
|
return unless $name =~ /^[^.].+\.md$/; # only markdown files |
167
|
|
|
|
|
|
|
# warn qq($name\n); |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
my ($file) = $name =~ /^$self->{source_dir}\/(.*)\.md/; |
170
|
|
|
|
|
|
|
my $dir_rel = ''; |
171
|
|
|
|
|
|
|
($dir_rel,$file) = $file =~ /^(.*)\/(.*)/ if $file =~ /\//; |
172
|
|
|
|
|
|
|
# warn qq($dir_rel\t$file\n);; |
173
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
my $markdown = io($name)->slurp; |
175
|
|
|
|
|
|
|
|
176
|
|
|
|
|
|
|
# Replaces hard github links to other markdown files to our newly converted relative links |
177
|
|
|
|
|
|
|
$markdown =~ s~(\]\()https://github.com/duckduckgo/duckduckgo-documentation/blob/master/duckduckhack/(?:[^\/\.]+\/){1,4}([^\.]+?)\.md([^\)]*?\))~$1$2$3~sg; |
178
|
|
|
|
|
|
|
# die $markdown; |
179
|
|
|
|
|
|
|
|
180
|
|
|
|
|
|
|
# Replace code types the parser can't deal with, |
181
|
|
|
|
|
|
|
# that is a #### line without a space line between it and a code start line ``` |
182
|
|
|
|
|
|
|
$markdown =~ s/(\n\#+.*?)(\n`)/$1\n$2/sg; |
183
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
# die $markdown if $file eq 'spice_frontend_walkthroughs'; |
185
|
|
|
|
|
|
|
|
186
|
|
|
|
|
|
|
# Just for debugging. |
187
|
|
|
|
|
|
|
# make_path("$dir_output/$dir_rel") if $dir_rel; |
188
|
|
|
|
|
|
|
# open(IN,">$dir_output/$dir_rel/$file.html"); |
189
|
|
|
|
|
|
|
my $buffer = q{}; |
190
|
|
|
|
|
|
|
open my $fh, '>', \$buffer; |
191
|
|
|
|
|
|
|
|
192
|
|
|
|
|
|
|
my $handler = Markdent::Handler::HTMLStream::Document->new( |
193
|
|
|
|
|
|
|
title => $name, |
194
|
|
|
|
|
|
|
output => $fh, |
195
|
|
|
|
|
|
|
# output => \*IN, |
196
|
|
|
|
|
|
|
); |
197
|
|
|
|
|
|
|
|
198
|
|
|
|
|
|
|
my $parser = Markdent::Parser->new( dialect => 'GitHub', handler => $handler ); |
199
|
|
|
|
|
|
|
$parser->parse( markdown => $markdown ); |
200
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
my $html = $buffer; |
202
|
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
# Gets embedded in wrapped html. |
204
|
|
|
|
|
|
|
$html =~ s/^.*?<html>.*?<\/head>\s*<body>\s*//s; |
205
|
|
|
|
|
|
|
$html =~ s/\s*<\/body>.*?<\/html>\s*$//s; |
206
|
|
|
|
|
|
|
|
207
|
|
|
|
|
|
|
# Make anchors for headings. |
208
|
|
|
|
|
|
|
$html =~ s/(<h\d>)(.*?)(<\/h\d>)/$1 . '<a name="' . make_anchor($2) . '" class="anchor"><\/a>' . $2 . $3/ges; |
209
|
|
|
|
|
|
|
|
210
|
|
|
|
|
|
|
# die $file; |
211
|
|
|
|
|
|
|
# die $html if $file eq 'spice_frontend_walkthroughs'; |
212
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
|
214
|
|
|
|
|
|
|
my $category = $nav{$file}{'category'} || ''; |
215
|
|
|
|
|
|
|
my $title = $nav{$file}{'title'} || ''; |
216
|
|
|
|
|
|
|
|
217
|
|
|
|
|
|
|
warn qq(NO CATEGORY: $file\n) if !$category; |
218
|
|
|
|
|
|
|
# warn qq($file\t$title\t$category\n); |
219
|
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
my $prev = $prev_next{$file}{'prev'} || []; |
221
|
|
|
|
|
|
|
my $next = $prev_next{$file}{'next'} || []; |
222
|
|
|
|
|
|
|
|
223
|
|
|
|
|
|
|
warn qq(NO PREV: $file\n) if !$prev; |
224
|
|
|
|
|
|
|
warn qq(NO NEXT: $file\n) if !$next; |
225
|
|
|
|
|
|
|
|
226
|
|
|
|
|
|
|
# warn qq($file\t), p($prev); |
227
|
|
|
|
|
|
|
# warn qq($file\t), p($next); |
228
|
|
|
|
|
|
|
|
229
|
|
|
|
|
|
|
$pages{$file} = sub { |
230
|
|
|
|
|
|
|
file => $file, |
231
|
|
|
|
|
|
|
file_dir => $dir_rel, |
232
|
|
|
|
|
|
|
file_path => "https://github.com/duckduckgo/duckduckgo-documentation/tree/master/duckduckhack", |
233
|
|
|
|
|
|
|
title => $title, |
234
|
|
|
|
|
|
|
category => $category, |
235
|
|
|
|
|
|
|
html => $html, |
236
|
|
|
|
|
|
|
nav_ref => $nav_arr_ref, |
237
|
|
|
|
|
|
|
maintemplate => 'doc.tx', |
238
|
|
|
|
|
|
|
prev => $prev, |
239
|
|
|
|
|
|
|
next => $next, |
240
|
|
|
|
|
|
|
# raw_output => 1, # enable to strip out all of the wrapping for these docs and only output the main content for scraping and inclusion elsewhere |
241
|
|
|
|
|
|
|
}; |
242
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
# $ref->{$_} = $html; |
244
|
|
|
|
|
|
|
}, $self->source_dir); |
245
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
# die p($nav_ref); |
247
|
|
|
|
|
|
|
|
248
|
|
|
|
|
|
|
return \%pages; |
249
|
|
|
|
|
|
|
} |
250
|
|
|
|
|
|
|
|
251
|
|
|
|
|
|
|
sub make_anchor { |
252
|
|
|
|
|
|
|
my ($anchor) = @_; |
253
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
$anchor = lc $anchor; |
255
|
|
|
|
|
|
|
$anchor =~ s/[^a-z\s]+//g; |
256
|
|
|
|
|
|
|
$anchor = trim $anchor; |
257
|
|
|
|
|
|
|
$anchor =~ s/\s+/\-/g; |
258
|
|
|
|
|
|
|
|
259
|
|
|
|
|
|
|
return $anchor; |
260
|
|
|
|
|
|
|
} |
261
|
|
|
|
|
|
|
|
262
|
|
|
|
|
|
|
1; |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
__END__ |
265
|
|
|
|
|
|
|
|
266
|
|
|
|
|
|
|
=pod |
267
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
=head1 NAME |
269
|
|
|
|
|
|
|
|
270
|
|
|
|
|
|
|
DDG::Publisher::Site::Duckduckhack::Root |
271
|
|
|
|
|
|
|
|
272
|
|
|
|
|
|
|
=head1 VERSION |
273
|
|
|
|
|
|
|
|
274
|
|
|
|
|
|
|
version 1044 |
275
|
|
|
|
|
|
|
|
276
|
|
|
|
|
|
|
=head1 AUTHOR |
277
|
|
|
|
|
|
|
|
278
|
|
|
|
|
|
|
Torsten Raudssus <torsten@raudss.us> |
279
|
|
|
|
|
|
|
|
280
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
281
|
|
|
|
|
|
|
|
282
|
|
|
|
|
|
|
This software is Copyright (c) 2012 by DuckDuckGo, Inc. L<http://duckduckgo.com/>. |
283
|
|
|
|
|
|
|
|
284
|
|
|
|
|
|
|
This is free software, licensed under: |
285
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
The Apache License, Version 2.0, January 2004 |
287
|
|
|
|
|
|
|
|
288
|
|
|
|
|
|
|
=cut |