|
1
|
|
|
<?php |
|
2
|
|
|
/* |
|
3
|
|
|
* This file is part of the trefoil application. |
|
4
|
|
|
* |
|
5
|
|
|
* (c) Miguel Angel Gabriel <[email protected]> |
|
6
|
|
|
* |
|
7
|
|
|
* For the full copyright and license information, please view the LICENSE |
|
8
|
|
|
* file that was distributed with this source code. |
|
9
|
|
|
*/ |
|
10
|
|
|
namespace Trefoil\Plugins\Optional; |
|
11
|
|
|
|
|
12
|
|
|
use Easybook\Events\BaseEvent; |
|
13
|
|
|
use Easybook\Events\EasybookEvents; |
|
14
|
|
|
use Symfony\Component\EventDispatcher\EventSubscriberInterface; |
|
15
|
|
|
use Trefoil\Helpers\LinkChecker; |
|
16
|
|
|
use Trefoil\Util\SimpleReport; |
|
17
|
|
|
use Trefoil\Plugins\BasePlugin; |
|
18
|
|
|
/** |
|
19
|
|
|
* Plugin to check internal and external links in book. |
|
20
|
|
|
* |
|
21
|
|
|
* Internal links are checked by looking for a valid link target, i.e. an html element |
|
22
|
|
|
* whith id="the-link-target" in the whole book. |
|
23
|
|
|
* |
|
24
|
|
|
* External links are (optionally) checked for existence by performing a network lookup. |
|
25
|
|
|
* This behaviour is off by default because could be very time consuming if the book |
|
26
|
|
|
* has a large number of external links. To turn it on, set the following option in |
|
27
|
|
|
* the book's config.yml: |
|
28
|
|
|
* |
|
29
|
|
|
* editions: |
|
30
|
|
|
* <edition-name> |
|
31
|
|
|
* plugins: |
|
32
|
|
|
* ... |
|
33
|
|
|
* options: |
|
34
|
|
|
* LinkCheck: |
|
35
|
|
|
* check_external_links: true |
|
36
|
|
|
* |
|
37
|
|
|
* The plugin will generate a report in the output directory for all the links |
|
38
|
|
|
* in the book with its status from the check (OK or error). |
|
39
|
|
|
* |
|
40
|
|
|
*/ |
|
41
|
|
|
class LinkCheckPlugin extends BasePlugin implements EventSubscriberInterface |
|
42
|
|
|
{ |
|
43
|
|
|
protected $links = array( |
|
44
|
|
|
'internal' => array(), |
|
45
|
|
|
'external' => array() |
|
46
|
|
|
); |
|
47
|
|
|
|
|
48
|
|
|
protected $linkTargets = array(); |
|
49
|
|
|
|
|
50
|
1 |
|
public static function getSubscribedEvents() |
|
51
|
|
|
{ |
|
52
|
|
|
return array( |
|
53
|
1 |
|
EasybookEvents::POST_DECORATE => array('onItemPostDecorate', -1100), // the latest possible |
|
54
|
1 |
|
EasybookEvents::POST_PUBLISH => array('onPostPublish', -1100) // the latest possible |
|
55
|
1 |
|
); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
1 |
|
public function onItemPostDecorate(BaseEvent $event) |
|
59
|
|
|
{ |
|
60
|
1 |
|
$this->init($event); |
|
61
|
|
|
|
|
62
|
|
|
// retrieve all the links for this item |
|
63
|
1 |
|
$links = $this->findLinks($this->item['content'], $this->item['config']['content']); |
|
64
|
1 |
|
$this->links = array_merge_recursive($this->links, $links); |
|
65
|
|
|
|
|
66
|
|
|
// retrieve all the internal link targets for this item |
|
67
|
1 |
|
$linkTargets = $this->findLinkTargets($this->item['content']); |
|
68
|
1 |
|
$this->linkTargets = array_merge_recursive($this->linkTargets, $linkTargets); |
|
69
|
1 |
|
} |
|
70
|
|
|
|
|
71
|
1 |
|
public function onPostPublish(BaseEvent $event) |
|
72
|
|
|
{ |
|
73
|
1 |
|
$this->init($event); |
|
74
|
|
|
|
|
75
|
1 |
|
$this->checkInternalLinks(); |
|
76
|
1 |
|
$this->checkExternalLinks(); |
|
77
|
|
|
|
|
78
|
1 |
|
$this->createReport(); |
|
79
|
1 |
|
} |
|
80
|
|
|
|
|
81
|
|
|
/** |
|
82
|
|
|
* Extract all links in content |
|
83
|
|
|
* |
|
84
|
|
|
* @param string $content |
|
85
|
|
|
* @param $xref |
|
86
|
|
|
* |
|
87
|
|
|
* @return array |
|
88
|
|
|
*/ |
|
89
|
1 |
|
protected function findLinks($content, $xref) |
|
90
|
|
|
{ |
|
91
|
|
|
$links = array( |
|
92
|
|
|
'internal' => array( |
|
93
|
1 |
|
$xref => array() |
|
94
|
1 |
|
), |
|
95
|
|
|
'external' => array( |
|
96
|
1 |
|
$xref => array() |
|
97
|
1 |
|
) |
|
98
|
1 |
|
); |
|
99
|
|
|
|
|
100
|
1 |
|
preg_match_all('/<a .*href="(?<uri>.*)".*>(?<text>.*)<\/a>/Ums', $content, $matches, PREG_SET_ORDER); |
|
101
|
|
|
|
|
102
|
1 |
|
if ($matches) { |
|
103
|
1 |
|
foreach ($matches as $match) { |
|
104
|
|
|
$link = array( |
|
105
|
1 |
|
'text' => $match['text'], |
|
106
|
1 |
|
'uri' => $match['uri'] |
|
107
|
1 |
|
); |
|
108
|
|
|
|
|
109
|
1 |
|
if ('#' == substr($match['uri'], 0, 1)) { |
|
110
|
1 |
|
$links['internal'][$xref][] = $link; |
|
111
|
1 |
|
} else { |
|
112
|
1 |
|
$links['external'][$xref][] = $link; |
|
113
|
|
|
} |
|
114
|
1 |
|
} |
|
115
|
1 |
|
} |
|
116
|
|
|
|
|
117
|
1 |
|
return $links; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Extract all possible link targets in content |
|
122
|
|
|
* |
|
123
|
|
|
* @param string $content |
|
124
|
|
|
* |
|
125
|
|
|
* @return array |
|
126
|
|
|
*/ |
|
127
|
1 |
|
protected function findLinkTargets($content) |
|
128
|
|
|
{ |
|
129
|
1 |
|
$linkTargets = array(); |
|
130
|
|
|
|
|
131
|
1 |
|
preg_match_all('/<.*id="(?<id>.*)".*>/Ums', $content, $matches); |
|
132
|
|
|
|
|
133
|
1 |
|
foreach ($matches['id'] as $match) { |
|
134
|
1 |
|
$linkTargets[] = $match; |
|
135
|
1 |
|
} |
|
136
|
|
|
|
|
137
|
1 |
|
return $linkTargets; |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
/** |
|
141
|
|
|
* Check all the internal links for existence |
|
142
|
|
|
*/ |
|
143
|
1 |
|
protected function checkInternalLinks() |
|
144
|
|
|
{ |
|
145
|
1 |
|
$errors = false; |
|
146
|
|
|
|
|
147
|
1 |
|
foreach ($this->links['internal'] as $xref => $links) { |
|
148
|
1 |
|
foreach ($links as $index => $link) { |
|
149
|
1 |
|
if (!in_array(substr($link['uri'], 1), $this->linkTargets)) { |
|
150
|
1 |
|
$this->links['internal'][$xref][$index]['status'] = 'Not found'; |
|
151
|
1 |
|
$errors = true; |
|
152
|
1 |
|
} else { |
|
153
|
1 |
|
$this->links['internal'][$xref][$index]['status'] = 'OK'; |
|
154
|
|
|
} |
|
155
|
1 |
|
} |
|
156
|
1 |
|
} |
|
157
|
|
|
|
|
158
|
1 |
|
if ($errors) { |
|
159
|
1 |
|
$this->writeLn('Some internal links are not correct.', 'error'); |
|
160
|
1 |
|
} |
|
161
|
1 |
|
} |
|
162
|
|
|
|
|
163
|
|
|
/** |
|
164
|
|
|
* Check external links for existence |
|
165
|
|
|
*/ |
|
166
|
1 |
|
protected function checkExternalLinks() |
|
167
|
|
|
{ |
|
168
|
1 |
|
if (!$this->getEditionOption('plugins.options.LinkCheck.check_external_links')) { |
|
169
|
|
|
return; |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
1 |
|
$checker = new LinkChecker(); |
|
173
|
|
|
|
|
174
|
1 |
|
$this->writeLn('Checking external links....'); |
|
175
|
|
|
|
|
176
|
1 |
|
$errors = false; |
|
177
|
|
|
|
|
178
|
1 |
|
$numLinks = 0; |
|
179
|
1 |
|
foreach ($this->links['external'] as $links) { |
|
180
|
1 |
|
$numLinks += count($links); |
|
181
|
1 |
|
} |
|
182
|
|
|
|
|
183
|
1 |
|
$this->progressStart($numLinks); |
|
184
|
1 |
|
foreach ($this->links['external'] as $xref => $links) { |
|
185
|
1 |
|
foreach ($links as $index => $link) { |
|
186
|
|
|
try { |
|
187
|
1 |
|
$checker->check($link['uri']); |
|
188
|
1 |
|
$this->links['external'][$xref][$index]['status'] = 'OK'; |
|
189
|
1 |
|
} catch (\Exception $e) { |
|
190
|
1 |
|
$this->links['external'][$xref][$index]['status'] = $e->getMessage(); |
|
191
|
1 |
|
$errors = true; |
|
192
|
|
|
} |
|
193
|
1 |
|
$this->progressAdvance(); |
|
194
|
1 |
|
} |
|
195
|
1 |
|
} |
|
196
|
1 |
|
$this->progressFinish(); |
|
197
|
|
|
|
|
198
|
1 |
|
if ($errors) { |
|
199
|
1 |
|
$this->writeLn('Some external links are not correct.', 'error'); |
|
200
|
1 |
|
} else { |
|
201
|
|
|
$this->writeLn('All external links are correct.'); |
|
202
|
|
|
} |
|
203
|
1 |
|
} |
|
204
|
|
|
|
|
205
|
1 |
|
protected function createReport() |
|
206
|
|
|
{ |
|
207
|
1 |
|
$report = new SimpleReport(); |
|
208
|
1 |
|
$report->setTitle('LinkCheckPlugin'); |
|
209
|
|
|
|
|
210
|
1 |
|
$checkExternalLinks = $this->getEditionOption('plugins.options.LinkCheck.check_external_links'); |
|
211
|
1 |
|
$report->setSubtitle(sprintf('check_external_links: %s', $checkExternalLinks ? 'true' : 'false')); |
|
212
|
|
|
|
|
213
|
1 |
|
$report->setHeaders( |
|
214
|
|
|
array( |
|
215
|
1 |
|
'Type', |
|
216
|
1 |
|
'Item', |
|
217
|
1 |
|
'Status', |
|
218
|
1 |
|
'Link') |
|
219
|
1 |
|
); |
|
220
|
1 |
|
$report->setColumnsWidth( |
|
221
|
|
|
array( |
|
222
|
1 |
|
8, |
|
223
|
1 |
|
4, |
|
224
|
1 |
|
20, |
|
225
|
|
|
100 |
|
226
|
1 |
|
) |
|
227
|
1 |
|
); |
|
228
|
|
|
|
|
229
|
1 |
|
$countOk = 0; |
|
230
|
1 |
|
$countError = 0; |
|
231
|
1 |
|
$countNotChecked = 0; |
|
232
|
|
|
|
|
233
|
1 |
|
foreach ($this->links as $type => $linksByType) { |
|
234
|
|
|
|
|
235
|
1 |
|
$report->addLine(array(ucfirst($type))); |
|
236
|
1 |
|
$report->addLine(); |
|
237
|
|
|
|
|
238
|
1 |
|
foreach ($linksByType as $xref => $links) { |
|
239
|
1 |
|
$report->addLine(array('', $xref)); |
|
240
|
1 |
|
$report->addLine(); |
|
241
|
1 |
|
foreach ($links as $link) { |
|
242
|
|
|
|
|
243
|
1 |
|
$status = isset($link['status']) ? $link['status'] : 'Not checked'; |
|
244
|
|
|
|
|
245
|
1 |
|
if ('OK' == $status) { |
|
246
|
1 |
|
$countOk++; |
|
247
|
1 |
|
} elseif ('Not checked' == $status) { |
|
248
|
|
|
$countNotChecked++; |
|
249
|
|
|
} else { |
|
250
|
1 |
|
$countError++; |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
1 |
|
$report->addLine(array('', '', $status, trim($link['text']))); |
|
254
|
1 |
|
if ($link['text'] != $link['uri']) { |
|
255
|
1 |
|
$report->addLine(array('', '', '', '<' . $link['uri'] . '>')); |
|
256
|
1 |
|
} |
|
257
|
1 |
|
} |
|
258
|
1 |
|
if (!$links) { |
|
259
|
1 |
|
$report->addLine(array('', '', '== No links ==')); |
|
260
|
1 |
|
$report->addLine(); |
|
261
|
1 |
|
} |
|
262
|
1 |
|
} |
|
263
|
1 |
|
} |
|
264
|
|
|
|
|
265
|
1 |
|
$report->addSummaryLine(sprintf('Total OK.........: %s', $countOk)); |
|
266
|
1 |
|
$report->addSummaryLine(sprintf('Total Error......: %s', $countError)); |
|
267
|
1 |
|
$report->addSummaryLine(sprintf('Total Not checked: %s', $countNotChecked)); |
|
268
|
|
|
|
|
269
|
1 |
|
$text = $report->getText(); |
|
270
|
|
|
|
|
271
|
|
|
// write report |
|
272
|
1 |
|
$outputDir = $this->app['publishing.dir.output']; |
|
273
|
1 |
|
$reportFile = $outputDir . '/report-LinkCheckPlugin.txt'; |
|
274
|
|
|
|
|
275
|
1 |
|
file_put_contents($reportFile, $text); |
|
276
|
1 |
|
} |
|
277
|
|
|
} |
|
278
|
|
|
|
magabriel /
trefoil
