]> git.immae.eu Git - github/wallabag/wallabag.git/blame - src/Wallabag/CoreBundle/Helper/EntriesExport.php
EntriesExport: sanitize filename and fix tests
[github/wallabag/wallabag.git] / src / Wallabag / CoreBundle / Helper / EntriesExport.php
CommitLineData
03690d13
TC
1<?php
2
3namespace Wallabag\CoreBundle\Helper;
4
c6608783 5use Html2Text\Html2Text;
619cc453
JB
6use JMS\Serializer\SerializationContext;
7use JMS\Serializer\SerializerBuilder;
03690d13
TC
8use PHPePub\Core\EPub;
9use PHPePub\Core\Structure\OPF\DublinCore;
add597ba 10use Symfony\Component\HttpFoundation\Response;
b1428a1c 11use Symfony\Component\Translation\TranslatorInterface;
52b84c11 12use Wallabag\CoreBundle\Entity\Entry;
03690d13 13
cceca9ea
JB
14/**
15 * This class doesn't have unit test BUT it's fully covered by a functional test with ExportControllerTest.
16 */
03690d13
TC
17class EntriesExport
18{
add597ba
JB
19 private $wallabagUrl;
20 private $logoPath;
b1428a1c 21 private $translator;
add597ba 22 private $title = '';
4094ea47 23 private $entries = [];
07320a2b 24 private $author = 'wallabag';
add597ba 25 private $language = '';
03690d13 26
add597ba 27 /**
9dd67fa3
JB
28 * @param TranslatorInterface $translator Translator service
29 * @param string $wallabagUrl Wallabag instance url
30 * @param string $logoPath Path to the logo FROM THE BUNDLE SCOPE
add597ba 31 */
dc7fa8df 32 public function __construct(TranslatorInterface $translator, $wallabagUrl, $logoPath)
03690d13 33 {
dc7fa8df 34 $this->translator = $translator;
4b3c983a 35 $this->wallabagUrl = $wallabagUrl;
add597ba
JB
36 $this->logoPath = $logoPath;
37 }
38
39 /**
40 * Define entries.
41 *
42 * @param array|Entry $entries An array of entries or one entry
4094ea47
JB
43 *
44 * @return EntriesExport
add597ba
JB
45 */
46 public function setEntries($entries)
47 {
2a1ceb67 48 if (!\is_array($entries)) {
add597ba 49 $this->language = $entries->getLanguage();
4094ea47 50 $entries = [$entries];
add597ba
JB
51 }
52
03690d13
TC
53 $this->entries = $entries;
54
add597ba 55 return $this;
03690d13
TC
56 }
57
58 /**
59 * Sets the category of which we want to get articles, or just one entry.
60 *
61 * @param string $method Method to get articles
4094ea47
JB
62 *
63 * @return EntriesExport
03690d13 64 */
add597ba 65 public function updateTitle($method)
03690d13 66 {
15a6402f 67 $this->title = $method . ' articles';
add597ba
JB
68
69 if ('entry' === $method) {
70 $this->title = $this->entries[0]->getTitle();
03690d13 71 }
add597ba
JB
72
73 return $this;
03690d13
TC
74 }
75
07320a2b 76 /**
c57f69d9
ÉG
77 * Sets the author for one entry or category.
78 *
79 * The publishers are used, or the domain name if empty.
07320a2b
ÉG
80 *
81 * @param string $method Method to get articles
82 *
83 * @return EntriesExport
84 */
85 public function updateAuthor($method)
86 {
eeabca80 87 if ('entry' !== $method) {
f8108346 88 $this->author = 'Various authors';
07320a2b 89
eeabca80
ÉG
90 return $this;
91 }
92
93 $this->author = $this->entries[0]->getDomainName();
94
95 $publishedBy = $this->entries[0]->getPublishedBy();
96 if (!empty($publishedBy)) {
97 $this->author = implode(', ', $publishedBy);
07320a2b
ÉG
98 }
99
100 return $this;
101 }
102
03690d13
TC
103 /**
104 * Sets the output format.
105 *
106 * @param string $format
4094ea47
JB
107 *
108 * @return Response
03690d13
TC
109 */
110 public function exportAs($format)
111 {
15a6402f 112 $functionName = 'produce' . ucfirst($format);
8f336fda
JB
113 if (method_exists($this, $functionName)) {
114 return $this->$functionName();
03690d13 115 }
add597ba
JB
116
117 throw new \InvalidArgumentException(sprintf('The format "%s" is not yet supported.', $format));
03690d13
TC
118 }
119
8303b037
TC
120 public function exportJsonData()
121 {
122 return $this->prepareSerializingContent('json');
123 }
124
add597ba
JB
125 /**
126 * Use PHPePub to dump a .epub file.
4094ea47
JB
127 *
128 * @return Response
add597ba 129 */
03690d13
TC
130 private function produceEpub()
131 {
132 /*
133 * Start and End of the book
134 */
135 $content_start =
136 "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
15a6402f
JB
137 . "<html xmlns=\"http://www.w3.org/1999/xhtml\" xmlns:epub=\"http://www.idpf.org/2007/ops\">\n"
138 . '<head>'
139 . "<meta http-equiv=\"Default-Style\" content=\"text/html; charset=utf-8\" />\n"
140 . "<title>wallabag articles book</title>\n"
141 . "</head>\n"
142 . "<body>\n";
03690d13
TC
143
144 $bookEnd = "</body>\n</html>\n";
145
146 $book = new EPub(EPub::BOOK_VERSION_EPUB3);
147
148 /*
149 * Book metadata
150 */
151
152 $book->setTitle($this->title);
add597ba
JB
153 // Not needed, but included for the example, Language is mandatory, but EPub defaults to "en". Use RFC3066 Language codes, such as "en", "da", "fr" etc.
154 $book->setLanguage($this->language);
155 $book->setDescription('Some articles saved on my wallabag');
03690d13 156
07320a2b 157 $book->setAuthor($this->author, $this->author);
03690d13 158
add597ba
JB
159 // I hope this is a non existant address :)
160 $book->setPublisher('wallabag', 'wallabag');
161 // Strictly not needed as the book date defaults to time().
162 $book->setDate(time());
163 $book->setSourceURL($this->wallabagUrl);
03690d13
TC
164
165 $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'PHP');
166 $book->addDublinCoreMetadata(DublinCore::CONTRIBUTOR, 'wallabag');
167
168 /*
169 * Front page
170 */
add597ba
JB
171 if (file_exists($this->logoPath)) {
172 $book->setCoverImage('Cover.png', file_get_contents($this->logoPath), 'image/png');
173 }
03690d13 174
bf22266a 175 $entryIds = [];
edd1825b
KD
176 $entryCount = \count($this->entries);
177 $i = 0;
bf22266a 178
03690d13
TC
179 /*
180 * Adding actual entries
181 */
182
add597ba
JB
183 // set tags as subjects
184 foreach ($this->entries as $entry) {
edd1825b 185 ++$i;
b0458874
JB
186 foreach ($entry->getTags() as $tag) {
187 $book->setSubject($tag->getLabel());
add597ba 188 }
edd1825b 189 $filename = sha1($entry->getTitle());
45d94a98 190
4944703e
KD
191 $publishedBy = $entry->getPublishedBy();
192 if (!empty($publishedBy)) {
193 $authors = implode(',', $publishedBy);
194 } else {
195 $authors = $this->translator->trans('export.unknown');
196 }
197
198 $titlepage = $content_start .
199 '<h1>' . $entry->getTitle() . '</h1>' .
200 '<dl>' .
201 '<dt>' . $this->translator->trans('entry.view.published_by') . '</dt><dd>' . $authors . '</dd>' .
202 '<dt>' . $this->translator->trans('entry.metadata.reading_time') . '</dt><dd>' . $this->translator->trans('entry.metadata.reading_time_minutes_short', ['%readingTime%' => $entry->getReadingTime()]) . '</dd>' .
203 '<dt>' . $this->translator->trans('entry.metadata.added_on') . '</dt><dd>' . $entry->getCreatedAt()->format('Y-m-d') . '</dd>' .
204 '<dt>' . $this->translator->trans('entry.metadata.address') . '</dt><dd><a href="' . $entry->getUrl() . '">' . $entry->getUrl() . '</a></dd>' .
205 '</dl>' .
206 $bookEnd;
edd1825b 207 $book->addChapter("Entry {$i} of {$entryCount}", "{$filename}_cover.html", $titlepage, true, EPub::EXTERNAL_REF_ADD);
15a6402f 208 $chapter = $content_start . $entry->getContent() . $bookEnd;
bf22266a
KD
209
210 $entryIds[] = $entry->getId();
edd1825b 211 $book->addChapter($entry->getTitle(), "{$filename}.html", $chapter, true, EPub::EXTERNAL_REF_ADD);
03690d13 212 }
add597ba 213
30cf72bf
KD
214 $book->addChapter('Notices', 'Cover2.html', $content_start . $this->getExportInformation('PHPePub') . $bookEnd);
215
bf22266a
KD
216 // Could also be the ISBN number, prefered for published books, or a UUID.
217 $hash = sha1(sprintf('%s:%s', $this->wallabagUrl, implode(',', $entryIds)));
218 $book->setIdentifier(sprintf('urn:wallabag:%s', $hash), EPub::IDENTIFIER_URI);
219
add597ba
JB
220 return Response::create(
221 $book->getBook(),
222 200,
4094ea47 223 [
add597ba
JB
224 'Content-Description' => 'File Transfer',
225 'Content-type' => 'application/epub+zip',
dac93644 226 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.epub"',
add597ba 227 'Content-Transfer-Encoding' => 'binary',
4094ea47 228 ]
f898102c 229 );
03690d13
TC
230 }
231
add597ba
JB
232 /**
233 * Use PHPMobi to dump a .mobi file.
4094ea47
JB
234 *
235 * @return Response
add597ba 236 */
03690d13
TC
237 private function produceMobi()
238 {
239 $mobi = new \MOBI();
240 $content = new \MOBIFile();
241
242 /*
243 * Book metadata
244 */
03690d13 245 $content->set('title', $this->title);
07320a2b 246 $content->set('author', $this->author);
03690d13
TC
247 $content->set('subject', $this->title);
248
249 /*
250 * Front page
251 */
add597ba
JB
252 $content->appendParagraph($this->getExportInformation('PHPMobi'));
253 if (file_exists($this->logoPath)) {
254 $content->appendImage(imagecreatefrompng($this->logoPath));
255 }
03690d13
TC
256 $content->appendPageBreak();
257
258 /*
259 * Adding actual entries
260 */
03690d13
TC
261 foreach ($this->entries as $entry) {
262 $content->appendChapterTitle($entry->getTitle());
263 $content->appendParagraph($entry->getContent());
264 $content->appendPageBreak();
265 }
266 $mobi->setContentProvider($content);
267
add597ba
JB
268 return Response::create(
269 $mobi->toString(),
270 200,
4094ea47 271 [
add597ba
JB
272 'Accept-Ranges' => 'bytes',
273 'Content-Description' => 'File Transfer',
274 'Content-type' => 'application/x-mobipocket-ebook',
dac93644 275 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.mobi"',
add597ba 276 'Content-Transfer-Encoding' => 'binary',
4094ea47 277 ]
f898102c 278 );
03690d13
TC
279 }
280
add597ba
JB
281 /**
282 * Use TCPDF to dump a .pdf file.
4094ea47
JB
283 *
284 * @return Response
add597ba 285 */
8f336fda 286 private function producePdf()
03690d13
TC
287 {
288 $pdf = new \TCPDF(PDF_PAGE_ORIENTATION, PDF_UNIT, PDF_PAGE_FORMAT, true, 'UTF-8', false);
289
290 /*
291 * Book metadata
292 */
03690d13 293 $pdf->SetCreator(PDF_CREATOR);
07320a2b 294 $pdf->SetAuthor($this->author);
03690d13
TC
295 $pdf->SetTitle($this->title);
296 $pdf->SetSubject('Articles via wallabag');
297 $pdf->SetKeywords('wallabag');
298
03690d13
TC
299 /*
300 * Adding actual entries
301 */
03690d13 302 foreach ($this->entries as $entry) {
b0458874
JB
303 foreach ($entry->getTags() as $tag) {
304 $pdf->SetKeywords($tag->getLabel());
03690d13
TC
305 }
306
ad5ef8bc
KD
307 $publishedBy = $entry->getPublishedBy();
308 if (!empty($publishedBy)) {
309 $authors = implode(',', $publishedBy);
310 } else {
311 $authors = $this->translator->trans('export.unknown');
312 }
313
314 $pdf->addPage();
315 $html = '<h1>' . $entry->getTitle() . '</h1>' .
316 '<dl>' .
317 '<dt>' . $this->translator->trans('entry.view.published_by') . '</dt><dd>' . $authors . '</dd>' .
318 '<dt>' . $this->translator->trans('entry.metadata.reading_time') . '</dt><dd>' . $this->translator->trans('entry.metadata.reading_time_minutes_short', ['%readingTime%' => $entry->getReadingTime()]) . '</dd>' .
319 '<dt>' . $this->translator->trans('entry.metadata.added_on') . '</dt><dd>' . $entry->getCreatedAt()->format('Y-m-d') . '</dd>' .
320 '<dt>' . $this->translator->trans('entry.metadata.address') . '</dt><dd><a href="' . $entry->getUrl() . '">' . $entry->getUrl() . '</a></dd>' .
321 '</dl>';
322 $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
323
03690d13 324 $pdf->AddPage();
15a6402f 325 $html = '<h1>' . $entry->getTitle() . '</h1>';
03690d13 326 $html .= $entry->getContent();
add597ba 327
03690d13
TC
328 $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
329 }
330
ad5ef8bc
KD
331 /*
332 * Last page
333 */
334 $pdf->AddPage();
335 $html = $this->getExportInformation('tcpdf');
336
337 $pdf->writeHTMLCell(0, 0, '', '', $html, 0, 1, 0, true, '', true);
338
03690d13
TC
339 // set image scale factor
340 $pdf->setImageScale(PDF_IMAGE_SCALE_RATIO);
341
add597ba
JB
342 return Response::create(
343 $pdf->Output('', 'S'),
344 200,
4094ea47 345 [
add597ba
JB
346 'Content-Description' => 'File Transfer',
347 'Content-type' => 'application/pdf',
dac93644 348 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.pdf"',
add597ba 349 'Content-Transfer-Encoding' => 'binary',
4094ea47 350 ]
f898102c 351 );
03690d13
TC
352 }
353
add597ba
JB
354 /**
355 * Inspired from CsvFileDumper.
4094ea47
JB
356 *
357 * @return Response
add597ba 358 */
8f336fda 359 private function produceCsv()
03690d13 360 {
add597ba
JB
361 $delimiter = ';';
362 $enclosure = '"';
2a1ceb67 363 $handle = fopen('php://memory', 'b+r');
03690d13 364
9401696f 365 fputcsv($handle, ['Title', 'URL', 'Content', 'Tags', 'MIME Type', 'Language', 'Creation date'], $delimiter, $enclosure);
03690d13 366
03690d13 367 foreach ($this->entries as $entry) {
add597ba
JB
368 fputcsv(
369 $handle,
4094ea47 370 [
add597ba
JB
371 $entry->getTitle(),
372 $entry->getURL(),
cceca9ea 373 // remove new line to avoid crazy results
4094ea47 374 str_replace(["\r\n", "\r", "\n"], '', $entry->getContent()),
add597ba
JB
375 implode(', ', $entry->getTags()->toArray()),
376 $entry->getMimetype(),
377 $entry->getLanguage(),
9401696f 378 $entry->getCreatedAt()->format('d/m/Y h:i:s'),
4094ea47 379 ],
add597ba
JB
380 $delimiter,
381 $enclosure
382 );
383 }
384
385 rewind($handle);
386 $output = stream_get_contents($handle);
387 fclose($handle);
388
389 return Response::create(
390 $output,
391 200,
4094ea47 392 [
add597ba 393 'Content-type' => 'application/csv',
dac93644 394 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.csv"',
add597ba 395 'Content-Transfer-Encoding' => 'UTF-8',
4094ea47 396 ]
f898102c 397 );
add597ba
JB
398 }
399
4094ea47
JB
400 /**
401 * Dump a JSON file.
402 *
403 * @return Response
404 */
8f336fda 405 private function produceJson()
b3cc1a14 406 {
b3cc1a14 407 return Response::create(
8ac95cbf 408 $this->prepareSerializingContent('json'),
b3cc1a14 409 200,
4094ea47 410 [
b3cc1a14 411 'Content-type' => 'application/json',
dac93644 412 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.json"',
b3cc1a14 413 'Content-Transfer-Encoding' => 'UTF-8',
4094ea47 414 ]
f898102c 415 );
b3cc1a14
TC
416 }
417
4094ea47
JB
418 /**
419 * Dump a XML file.
420 *
421 * @return Response
422 */
8f336fda 423 private function produceXml()
b3cc1a14 424 {
b3cc1a14 425 return Response::create(
8ac95cbf 426 $this->prepareSerializingContent('xml'),
b3cc1a14 427 200,
4094ea47 428 [
b3cc1a14 429 'Content-type' => 'application/xml',
dac93644 430 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.xml"',
b3cc1a14 431 'Content-Transfer-Encoding' => 'UTF-8',
4094ea47 432 ]
f898102c 433 );
b3cc1a14 434 }
8ac95cbf 435
4094ea47
JB
436 /**
437 * Dump a TXT file.
438 *
439 * @return Response
440 */
8f336fda 441 private function produceTxt()
6c08fb68
TC
442 {
443 $content = '';
d3f31ec4 444 $bar = str_repeat('=', 100);
6c08fb68 445 foreach ($this->entries as $entry) {
15a6402f 446 $content .= "\n\n" . $bar . "\n\n" . $entry->getTitle() . "\n\n" . $bar . "\n\n";
c6608783
NH
447 $html = new Html2Text($entry->getContent(), ['do_links' => 'none', 'width' => 100]);
448 $content .= $html->getText();
6c08fb68 449 }
d3f31ec4 450
6c08fb68
TC
451 return Response::create(
452 $content,
453 200,
4094ea47 454 [
6c08fb68 455 'Content-type' => 'text/plain',
dac93644 456 'Content-Disposition' => 'attachment; filename="' . $this->getSanitizedFilename() . '.txt"',
6c08fb68 457 'Content-Transfer-Encoding' => 'UTF-8',
4094ea47 458 ]
f898102c 459 );
6c08fb68
TC
460 }
461
b3cc1a14
TC
462 /**
463 * Return a Serializer object for producing processes that need it (JSON & XML).
464 *
0e49487b
JB
465 * @param string $format
466 *
52b84c11 467 * @return string
b3cc1a14 468 */
8ac95cbf 469 private function prepareSerializingContent($format)
b3cc1a14 470 {
268e9e72 471 $serializer = SerializerBuilder::create()->build();
b3cc1a14 472
cceca9ea
JB
473 return $serializer->serialize(
474 $this->entries,
475 $format,
4094ea47 476 SerializationContext::create()->setGroups(['entries_for_user'])
cceca9ea 477 );
b3cc1a14
TC
478 }
479
add597ba
JB
480 /**
481 * Return a kind of footer / information for the epub.
482 *
483 * @param string $type Generator of the export, can be: tdpdf, PHPePub, PHPMobi
484 *
485 * @return string
486 */
487 private function getExportInformation($type)
488 {
b1428a1c
NL
489 $info = $this->translator->trans('export.footer_template', [
490 '%method%' => $type,
491 ]);
add597ba
JB
492
493 if ('tcpdf' === $type) {
15a6402f 494 return str_replace('%IMAGE%', '<img src="' . $this->logoPath . '" />', $info);
03690d13 495 }
add597ba
JB
496
497 return str_replace('%IMAGE%', '', $info);
03690d13 498 }
dac93644
KD
499
500 /**
501 * Return a sanitized version of the title by applying translit iconv
502 * and removing non alphanumeric characters, - and space.
503 *
504 * @return string Sanitized filename
505 */
506 private function getSanitizedFilename()
507 {
508 return preg_replace('/[^A-Za-z0-9\- \']/', '', iconv('utf-8', 'us-ascii//TRANSLIT', $this->title));
509 }
03690d13 510}