1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* This file incrementally exports a member's profile data to a downloadable file. |
5
|
|
|
* |
6
|
|
|
* Simple Machines Forum (SMF) |
7
|
|
|
* |
8
|
|
|
* @package SMF |
9
|
|
|
* @author Simple Machines https://www.simplemachines.org |
10
|
|
|
* @copyright 2020 Simple Machines and individual contributors |
11
|
|
|
* @license https://www.simplemachines.org/about/smf/license.php BSD |
12
|
|
|
* |
13
|
|
|
* @version 2.1 RC2 |
14
|
|
|
*/ |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* Class ExportProfileData_Background |
18
|
|
|
*/ |
19
|
|
|
class ExportProfileData_Background extends SMF_BackgroundTask |
20
|
|
|
{ |
21
|
|
|
/** |
22
|
|
|
* Some private variables to help the static functions in this class. |
23
|
|
|
*/ |
24
|
|
|
private static $export_details = array(); |
25
|
|
|
private static $real_modSettings = array(); |
26
|
|
|
private static $xslt_info = array('stylesheet' => '', 'doctype' => ''); |
27
|
|
|
|
28
|
|
|
/** |
29
|
|
|
* Keep track of some other things on a per-instance basis. |
30
|
|
|
*/ |
31
|
|
|
private $next_task = array(); |
32
|
|
|
private $time_limit = 30; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* This is the main dispatcher for the class. |
36
|
|
|
* It calls the correct private function based on the information stored in |
37
|
|
|
* the task details. |
38
|
|
|
* |
39
|
|
|
* @return bool Always returns true |
40
|
|
|
*/ |
41
|
|
|
public function execute() |
42
|
|
|
{ |
43
|
|
|
global $sourcedir, $smcFunc; |
44
|
|
|
|
45
|
|
|
if (!defined('EXPORTING')) |
46
|
|
|
define('EXPORTING', 1); |
47
|
|
|
|
48
|
|
|
// Avoid leaving files in an inconsistent state. |
49
|
|
|
ignore_user_abort(true); |
50
|
|
|
|
51
|
|
|
$this->time_limit = (ini_get('safe_mode') === false && @set_time_limit(MAX_CLAIM_THRESHOLD) !== false) ? MAX_CLAIM_THRESHOLD : ini_get('max_execution_time'); |
52
|
|
|
|
53
|
|
|
// This could happen if the user manually changed the URL params of the export request. |
54
|
|
|
if ($this->_details['format'] == 'HTML' && (!class_exists('DOMDocument') || !class_exists('XSLTProcessor'))) |
55
|
|
|
{ |
56
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php'); |
57
|
|
|
$export_formats = get_export_formats(); |
58
|
|
|
|
59
|
|
|
$this->_details['format'] = 'XML_XSLT'; |
60
|
|
|
$this->_details['format_settings'] = $export_formats['XML_XSLT']; |
61
|
|
|
} |
62
|
|
|
|
63
|
|
|
// Inform static functions of the export format, etc. |
64
|
|
|
self::$export_details = $this->_details; |
65
|
|
|
|
66
|
|
|
// For exports only, members can always see their own posts, even in boards that they can no longer access. |
67
|
|
|
$member_info = $this->getMinUserInfo(array($this->_details['uid'])); |
68
|
|
|
$member_info = array_merge($member_info[$this->_details['uid']], array( |
69
|
|
|
'buddies' => array(), |
70
|
|
|
'query_see_board' => '1=1', |
71
|
|
|
'query_see_message_board' => '1=1', |
72
|
|
|
'query_see_topic_board' => '1=1', |
73
|
|
|
'query_wanna_see_board' => '1=1', |
74
|
|
|
'query_wanna_see_message_board' => '1=1', |
75
|
|
|
'query_wanna_see_topic_board' => '1=1', |
76
|
|
|
)); |
77
|
|
|
|
78
|
|
|
// Use some temporary integration hooks to manipulate BBC parsing during export. |
79
|
|
|
add_integration_function('integrate_pre_parsebbc', 'ExportProfileData_Background::pre_parsebbc', false); |
80
|
|
|
add_integration_function('integrate_post_parsebbc', 'ExportProfileData_Background::post_parsebbc', false); |
81
|
|
|
add_integration_function('integrate_bbc_codes', 'ExportProfileData_Background::bbc_codes', false); |
82
|
|
|
add_integration_function('integrate_post_parseAttachBBC', 'ExportProfileData_Background::post_parseAttachBBC', false); |
83
|
|
|
add_integration_function('integrate_attach_bbc_validate', 'ExportProfileData_Background::attach_bbc_validate', false); |
84
|
|
|
|
85
|
|
|
// We currently support exporting to XML and HTML |
86
|
|
|
if ($this->_details['format'] == 'XML') |
87
|
|
|
$this->exportXml($member_info); |
88
|
|
|
elseif ($this->_details['format'] == 'HTML') |
89
|
|
|
$this->exportHtml($member_info); |
90
|
|
|
elseif ($this->_details['format'] == 'XML_XSLT') |
91
|
|
|
$this->exportXmlXslt($member_info); |
92
|
|
|
|
93
|
|
|
// If necessary, create a new background task to continue the export process. |
94
|
|
|
if (!empty($this->next_task)) |
95
|
|
|
{ |
96
|
|
|
$smcFunc['db_insert']('insert', '{db_prefix}background_tasks', |
97
|
|
|
array('task_file' => 'string-255', 'task_class' => 'string-255', 'task_data' => 'string', 'claimed_time' => 'int'), |
98
|
|
|
$this->next_task, |
99
|
|
|
array() |
100
|
|
|
); |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
ignore_user_abort(false); |
104
|
|
|
|
105
|
|
|
return true; |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
/** |
109
|
|
|
* The workhorse of this class. Compiles profile data to XML files. |
110
|
|
|
* |
111
|
|
|
* @param array $member_info Minimal $user_info about the relevant member. |
112
|
|
|
*/ |
113
|
|
|
protected function exportXml($member_info) |
114
|
|
|
{ |
115
|
|
|
global $smcFunc, $sourcedir, $context, $modSettings, $settings, $user_info, $mbname; |
116
|
|
|
global $user_profile, $txt, $scripturl, $query_this_board; |
117
|
|
|
|
118
|
|
|
// For convenience... |
119
|
|
|
$uid = $this->_details['uid']; |
120
|
|
|
$lang = $this->_details['lang']; |
121
|
|
|
$included = $this->_details['included']; |
122
|
|
|
$start = $this->_details['start']; |
123
|
|
|
$latest = $this->_details['latest']; |
124
|
|
|
$datatype = $this->_details['datatype']; |
125
|
|
|
|
126
|
|
|
if (!isset($included[$datatype]['func']) || !isset($included[$datatype]['langfile'])) |
127
|
|
|
return; |
128
|
|
|
|
129
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'News.php'); |
130
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'ScheduledTasks.php'); |
131
|
|
|
|
132
|
|
|
// Setup. |
133
|
|
|
$done = false; |
134
|
|
|
$delay = 0; |
135
|
|
|
$func = $included[$datatype]['func']; |
|
|
|
|
136
|
|
|
$context['xmlnews_uid'] = $uid; |
137
|
|
|
$context['xmlnews_limit'] = !empty($modSettings['export_rate']) ? $modSettings['export_rate'] : 250; |
138
|
|
|
$context[$datatype . '_start'] = $start[$datatype]; |
139
|
|
|
$datatypes = array_keys($included); |
140
|
|
|
|
141
|
|
|
// Fake a wee bit of $user_info so that loading the member data & language doesn't choke. |
142
|
|
|
$user_info = $member_info; |
143
|
|
|
|
144
|
|
|
loadEssentialThemeData(); |
145
|
|
|
$settings['actual_theme_dir'] = $settings['theme_dir']; |
146
|
|
|
$context['user']['id'] = $uid; |
147
|
|
|
$context['user']['language'] = $lang; |
148
|
|
|
loadMemberData($uid); |
149
|
|
|
loadLanguage(implode('+', array_unique(array('index', 'Modifications', 'Stats', 'Profile', $included[$datatype]['langfile']))), $lang); |
150
|
|
|
|
151
|
|
|
// @todo Ask lawyers whether the GDPR requires us to include posts in the recycle bin. |
152
|
|
|
$query_this_board = '{query_see_message_board}' . (!empty($modSettings['recycle_enable']) && $modSettings['recycle_board'] > 0 ? ' AND m.id_board != ' . $modSettings['recycle_board'] : ''); |
153
|
|
|
|
154
|
|
|
// We need a valid export directory. |
155
|
|
|
if (empty($modSettings['export_dir']) || !file_exists($modSettings['export_dir'])) |
156
|
|
|
{ |
157
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php'); |
158
|
|
|
if (create_export_dir() === false) |
159
|
|
|
return; |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
$export_dir_slash = $modSettings['export_dir'] . DIRECTORY_SEPARATOR; |
163
|
|
|
|
164
|
|
|
$idhash = hash_hmac('sha1', $uid, get_auth_secret()); |
165
|
|
|
$idhash_ext = $idhash . '.' . $this->_details['format_settings']['extension']; |
166
|
|
|
|
167
|
|
|
// Increment the file number until we reach one that doesn't exist. |
168
|
|
|
$filenum = 1; |
169
|
|
|
$realfile = $export_dir_slash . $filenum . '_' . $idhash_ext; |
170
|
|
|
while (file_exists($realfile)) |
171
|
|
|
$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext; |
172
|
|
|
|
173
|
|
|
$tempfile = $export_dir_slash . $idhash_ext . '.tmp'; |
174
|
|
|
$progressfile = $export_dir_slash . $idhash_ext . '.progress.json'; |
175
|
|
|
|
176
|
|
|
$feed_meta = array( |
177
|
|
|
'title' => sprintf($txt['profile_of_username'], $user_profile[$uid]['real_name']), |
178
|
|
|
'desc' => sentence_list(array_map(function ($datatype) use ($txt) { return $txt[$datatype]; }, array_keys($included))), |
179
|
|
|
'author' => $mbname, |
180
|
|
|
'source' => $scripturl . '?action=profile;u=' . $uid, |
181
|
|
|
'self' => '', // Unused, but can't be null. |
182
|
|
|
'page' => &$filenum, |
183
|
|
|
); |
184
|
|
|
|
185
|
|
|
// Some paranoid hosts disable or hamstring the disk space functions in an attempt at security via obscurity. |
186
|
|
|
$check_diskspace = !empty($modSettings['export_min_diskspace_pct']) && function_exists('disk_free_space') && function_exists('disk_total_space') && intval(@disk_total_space($modSettings['export_dir']) >= 1440); |
187
|
|
|
$minspace = $check_diskspace ? ceil(disk_total_space($modSettings['export_dir']) * $modSettings['export_min_diskspace_pct'] / 100) : 0; |
188
|
|
|
|
189
|
|
|
// If a necessary file is missing, we need to start over. |
190
|
|
|
if (!file_exists($tempfile) || !file_exists($progressfile) || filesize($progressfile) == 0) |
191
|
|
|
{ |
192
|
|
|
foreach (array_merge(array($tempfile, $progressfile), glob($export_dir_slash . '*_' . $idhash_ext)) as $fpath) |
|
|
|
|
193
|
|
|
@unlink($fpath); |
194
|
|
|
|
195
|
|
|
$filenum = 1; |
196
|
|
|
$realfile = $export_dir_slash . $filenum . '_' . $idhash_ext; |
197
|
|
|
|
198
|
|
|
buildXmlFeed('smf', array(), $feed_meta, 'profile'); |
199
|
|
|
file_put_contents($tempfile, implode('', $context['feed']), LOCK_EX); |
200
|
|
|
|
201
|
|
|
$progress = array_fill_keys($datatypes, 0); |
202
|
|
|
file_put_contents($progressfile, $smcFunc['json_encode']($progress)); |
203
|
|
|
} |
204
|
|
|
else |
205
|
|
|
$progress = $smcFunc['json_decode'](file_get_contents($progressfile), true); |
206
|
|
|
|
207
|
|
|
// Get the data, always in ascending order. |
208
|
|
|
$xml_data = call_user_func($included[$datatype]['func'], 'smf', true); |
209
|
|
|
|
210
|
|
|
// No data retrived? Just move on then. |
211
|
|
|
if (empty($xml_data)) |
212
|
|
|
$datatype_done = true; |
213
|
|
|
|
214
|
|
|
// Basic profile data is quick and easy. |
215
|
|
|
elseif ($datatype == 'profile') |
216
|
|
|
{ |
217
|
|
|
buildXmlFeed('smf', $xml_data, $feed_meta, 'profile'); |
218
|
|
|
file_put_contents($tempfile, implode('', $context['feed']), LOCK_EX); |
219
|
|
|
|
220
|
|
|
$progress[$datatype] = time(); |
221
|
|
|
$datatype_done = true; |
222
|
|
|
|
223
|
|
|
// Cache for subsequent reuse. |
224
|
|
|
$profile_basic_items = $context['feed']['items']; |
225
|
|
|
cache_put_data('export_profile_basic-' . $uid, $profile_basic_items, MAX_CLAIM_THRESHOLD); |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
// Posts and PMs... |
229
|
|
|
else |
230
|
|
|
{ |
231
|
|
|
// We need the basic profile data in every export file. |
232
|
|
|
$profile_basic_items = cache_get_data('export_profile_basic-' . $uid, MAX_CLAIM_THRESHOLD); |
233
|
|
|
if (empty($profile_basic_items)) |
234
|
|
|
{ |
235
|
|
|
$profile_data = call_user_func($included['profile']['func'], 'smf', true); |
236
|
|
|
buildXmlFeed('smf', $profile_data, $feed_meta, 'profile'); |
237
|
|
|
$profile_basic_items = $context['feed']['items']; |
238
|
|
|
cache_put_data('export_profile_basic-' . $uid, $profile_basic_items, MAX_CLAIM_THRESHOLD); |
239
|
|
|
unset($context['feed']); |
240
|
|
|
} |
241
|
|
|
|
242
|
|
|
$per_page = $this->_details['format_settings']['per_page']; |
243
|
|
|
$prev_item_count = empty($this->_details['item_count']) ? 0 : $this->_details['item_count']; |
244
|
|
|
|
245
|
|
|
// If the temp file has grown enormous, save it so we can start a new one. |
246
|
|
|
clearstatcache(); |
247
|
|
|
if (file_exists($tempfile) && filesize($tempfile) >= 1024 * 1024 * 250) |
248
|
|
|
{ |
249
|
|
|
rename($tempfile, $realfile); |
250
|
|
|
$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext; |
251
|
|
|
|
252
|
|
|
if (empty($context['feed']['header'])) |
253
|
|
|
buildXmlFeed('smf', array(), $feed_meta, 'profile'); |
254
|
|
|
|
255
|
|
|
file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])), LOCK_EX); |
256
|
|
|
|
257
|
|
|
$prev_item_count = 0; |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
// Split $xml_data into reasonably sized chunks. |
261
|
|
|
if (empty($prev_item_count)) |
262
|
|
|
{ |
263
|
|
|
$xml_data = array_chunk($xml_data, $per_page); |
264
|
|
|
} |
265
|
|
|
else |
266
|
|
|
{ |
267
|
|
|
$first_chunk = array_splice($xml_data, 0, $per_page - $prev_item_count); |
268
|
|
|
$xml_data = array_merge(array($first_chunk), array_chunk($xml_data, $per_page)); |
269
|
|
|
unset($first_chunk); |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
foreach ($xml_data as $chunk => $items) |
273
|
|
|
{ |
274
|
|
|
unset($new_item_count, $last_id); |
275
|
|
|
|
276
|
|
|
// Remember the last item so we know where to start next time. |
277
|
|
|
$last_item = end($items); |
278
|
|
|
if (isset($last_item['content'][0]['content']) && $last_item['content'][0]['tag'] === 'id') |
279
|
|
|
$last_id = $last_item['content'][0]['content']; |
280
|
|
|
|
281
|
|
|
// Build the XML string from the data. |
282
|
|
|
buildXmlFeed('smf', $items, $feed_meta, 'profile'); |
283
|
|
|
|
284
|
|
|
// If disk space is insufficient, pause for a day so the admin can fix it. |
285
|
|
|
if ($check_diskspace && disk_free_space($modSettings['export_dir']) - $minspace <= strlen(implode('', $context['feed']) . self::$xslt_info['stylesheet'])) |
286
|
|
|
{ |
287
|
|
|
loadLanguage('Errors'); |
288
|
|
|
log_error(sprintf($txt['export_low_diskspace'], $modSettings['export_min_diskspace_pct'])); |
289
|
|
|
|
290
|
|
|
$delay = 86400; |
291
|
|
|
} |
292
|
|
|
else |
293
|
|
|
{ |
294
|
|
|
// We need a file to write to, of course. |
295
|
|
|
if (!file_exists($tempfile)) |
296
|
|
|
file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])), LOCK_EX); |
297
|
|
|
|
298
|
|
|
// Insert the new data before the feed footer. |
299
|
|
|
$handle = fopen($tempfile, 'r+'); |
300
|
|
|
if (is_resource($handle)) |
301
|
|
|
{ |
302
|
|
|
flock($handle, LOCK_EX); |
303
|
|
|
|
304
|
|
|
fseek($handle, strlen($context['feed']['footer']) * -1, SEEK_END); |
305
|
|
|
|
306
|
|
|
$bytes_written = fwrite($handle, $context['feed']['items'] . $context['feed']['footer']); |
307
|
|
|
|
308
|
|
|
// If we couldn't write everything, revert the changes and consider the write to have failed. |
309
|
|
|
if ($bytes_written > 0 && $bytes_written < strlen($context['feed']['items'] . $context['feed']['footer'])) |
310
|
|
|
{ |
311
|
|
|
fseek($handle, $bytes_written * -1, SEEK_END); |
312
|
|
|
$pointer_pos = ftell($handle); |
313
|
|
|
ftruncate($handle, $pointer_pos); |
314
|
|
|
rewind($handle); |
315
|
|
|
fseek($handle, 0, SEEK_END); |
316
|
|
|
fwrite($handle, $context['feed']['footer']); |
317
|
|
|
|
318
|
|
|
$bytes_written = false; |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
flock($handle, LOCK_UN); |
322
|
|
|
fclose($handle); |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
// Write failed. We'll try again next time. |
326
|
|
|
if (empty($bytes_written)) |
327
|
|
|
{ |
328
|
|
|
$delay = MAX_CLAIM_THRESHOLD; |
329
|
|
|
break; |
330
|
|
|
} |
331
|
|
|
|
332
|
|
|
// All went well. |
333
|
|
|
else |
334
|
|
|
{ |
335
|
|
|
// Track progress by ID where appropriate, and by time otherwise. |
336
|
|
|
$progress[$datatype] = !isset($last_id) ? time() : $last_id; |
337
|
|
|
file_put_contents($progressfile, $smcFunc['json_encode']($progress)); |
338
|
|
|
|
339
|
|
|
// Are we done with this datatype yet? |
340
|
|
|
if (!isset($last_id) || (count($items) < $per_page && $last_id >= $latest[$datatype])) |
341
|
|
|
$datatype_done = true; |
342
|
|
|
|
343
|
|
|
// Finished the file for this chunk, so move on to the next one. |
344
|
|
|
if (count($items) >= $per_page - $prev_item_count) |
345
|
|
|
{ |
346
|
|
|
rename($tempfile, $realfile); |
347
|
|
|
$realfile = $export_dir_slash . ++$filenum . '_' . $idhash_ext; |
348
|
|
|
|
349
|
|
|
file_put_contents($tempfile, implode('', array($context['feed']['header'], $profile_basic_items, $context['feed']['footer'])), LOCK_EX); |
350
|
|
|
|
351
|
|
|
$prev_item_count = $new_item_count = 0; |
352
|
|
|
} |
353
|
|
|
// This was the last chunk. |
354
|
|
|
else |
355
|
|
|
{ |
356
|
|
|
// Should we append more items to this file next time? |
357
|
|
|
$new_item_count = isset($last_id) ? $prev_item_count + count($items) : 0; |
358
|
|
|
} |
359
|
|
|
} |
360
|
|
|
} |
361
|
|
|
} |
362
|
|
|
} |
363
|
|
|
|
364
|
|
|
if (!empty($datatype_done)) |
365
|
|
|
{ |
366
|
|
|
$datatype_key = array_search($datatype, $datatypes); |
367
|
|
|
$done = !isset($datatypes[$datatype_key + 1]); |
368
|
|
|
|
369
|
|
|
if (!$done) |
370
|
|
|
$datatype = $datatypes[$datatype_key + 1]; |
371
|
|
|
} |
372
|
|
|
|
373
|
|
|
// Remove the .tmp extension from the final tempfile so the system knows it's done. |
374
|
|
|
if (!empty($done)) |
375
|
|
|
{ |
376
|
|
|
rename($tempfile, $realfile); |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
// Oops. Apparently some sneaky monkey cancelled the export while we weren't looking. |
380
|
|
|
elseif (!file_exists($progressfile)) |
381
|
|
|
{ |
382
|
|
|
@unlink($tempfile); |
383
|
|
|
return; |
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
// We have more work to do again later. |
387
|
|
|
else |
388
|
|
|
{ |
389
|
|
|
$start[$datatype] = $progress[$datatype]; |
390
|
|
|
|
391
|
|
|
$new_details = array( |
392
|
|
|
'format' => $this->_details['format'], |
393
|
|
|
'uid' => $uid, |
394
|
|
|
'lang' => $lang, |
395
|
|
|
'included' => $included, |
396
|
|
|
'start' => $start, |
397
|
|
|
'latest' => $latest, |
398
|
|
|
'datatype' => $datatype, |
399
|
|
|
'format_settings' => $this->_details['format_settings'], |
400
|
|
|
'last_page' => $this->_details['last_page'], |
401
|
|
|
'dlfilename' => $this->_details['dlfilename'], |
402
|
|
|
); |
403
|
|
|
if (!empty($new_item_count)) |
404
|
|
|
$new_details['item_count'] = $new_item_count; |
405
|
|
|
|
406
|
|
|
$this->next_task = array('$sourcedir/tasks/ExportProfileData.php', 'ExportProfileData_Background', $smcFunc['json_encode']($new_details), time() - MAX_CLAIM_THRESHOLD + $delay); |
407
|
|
|
|
408
|
|
|
if (!file_exists($tempfile)) |
409
|
|
|
{ |
410
|
|
|
buildXmlFeed('smf', array(), $feed_meta, 'profile'); |
411
|
|
|
file_put_contents($tempfile, implode('', array($context['feed']['header'], !empty($profile_basic_items) ? $profile_basic_items : '', $context['feed']['footer'])), LOCK_EX); |
412
|
|
|
} |
413
|
|
|
} |
414
|
|
|
|
415
|
|
|
file_put_contents($progressfile, $smcFunc['json_encode']($progress)); |
416
|
|
|
} |
417
|
|
|
|
418
|
|
|
/** |
419
|
|
|
* Compiles profile data to HTML. |
420
|
|
|
* |
421
|
|
|
* Internally calls exportXml() and then uses an XSLT stylesheet to |
422
|
|
|
* transform the XML files into HTML. |
423
|
|
|
* |
424
|
|
|
* @param array $member_info Minimal $user_info about the relevant member. |
425
|
|
|
*/ |
426
|
|
|
protected function exportHtml($member_info) |
427
|
|
|
{ |
428
|
|
|
global $modSettings, $context, $smcFunc, $sourcedir; |
429
|
|
|
|
430
|
|
|
$context['export_last_page'] = $this->_details['last_page']; |
431
|
|
|
$context['export_dlfilename'] = $this->_details['dlfilename']; |
432
|
|
|
|
433
|
|
|
// Perform the export to XML. |
434
|
|
|
$this->exportXml($member_info); |
435
|
|
|
|
436
|
|
|
// Determine which files, if any, are ready to be transformed. |
437
|
|
|
$export_dir_slash = $modSettings['export_dir'] . DIRECTORY_SEPARATOR; |
438
|
|
|
$idhash = hash_hmac('sha1', $this->_details['uid'], get_auth_secret()); |
439
|
|
|
$idhash_ext = $idhash . '.' . $this->_details['format_settings']['extension']; |
440
|
|
|
|
441
|
|
|
$new_exportfiles = array(); |
442
|
|
|
foreach (glob($export_dir_slash . '*_' . $idhash_ext) as $completed_file) |
443
|
|
|
{ |
444
|
|
|
if (file_get_contents($completed_file, false, null, 0, 6) == '<?xml ') |
445
|
|
|
$new_exportfiles[] = $completed_file; |
446
|
|
|
} |
447
|
|
|
if (empty($new_exportfiles)) |
448
|
|
|
return; |
449
|
|
|
|
450
|
|
|
// Get the XSLT stylesheet. |
451
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php'); |
452
|
|
|
self::$xslt_info = get_xslt_stylesheet($this->_details['format'], $this->_details['uid']); |
453
|
|
|
|
454
|
|
|
// Set up the XSLT processor. |
455
|
|
|
$xslt = new DOMDocument(); |
456
|
|
|
$xslt->loadXML(self::$xslt_info['stylesheet']); |
457
|
|
|
$xsltproc = new XSLTProcessor(); |
458
|
|
|
$xsltproc->importStylesheet($xslt); |
459
|
|
|
|
460
|
|
|
$libxml_options = 0; |
461
|
|
|
if (LIBXML_VERSION >= 20621 && defined('LIBXML_COMPACT')) |
462
|
|
|
$libxml_options = $libxml_options | LIBXML_COMPACT; |
463
|
|
|
if (LIBXML_VERSION >= 20700 && defined('LIBXML_PARSEHUGE')) |
464
|
|
|
$libxml_options = $libxml_options | LIBXML_PARSEHUGE; |
465
|
|
|
if (LIBXML_VERSION >= 20900 && defined('LIBXML_BIGLINES')) |
466
|
|
|
$libxml_options = $libxml_options | LIBXML_BIGLINES; |
467
|
|
|
|
468
|
|
|
// Transform the files to HTML. |
469
|
|
|
$i = 0; |
470
|
|
|
$num_files = count($new_exportfiles); |
471
|
|
|
$max_transform_time = 0; |
472
|
|
|
$xmldoc = new DOMDocument(); |
473
|
|
|
foreach ($new_exportfiles as $exportfile) |
474
|
|
|
{ |
475
|
|
|
if (function_exists('apache_reset_timeout')) |
476
|
|
|
@apache_reset_timeout(); |
477
|
|
|
|
478
|
|
|
$started = microtime(true); |
479
|
|
|
$xmldoc->load($exportfile, $libxml_options); |
480
|
|
|
$xsltproc->transformToURI($xmldoc, $exportfile); |
481
|
|
|
$finished = microtime(true); |
482
|
|
|
|
483
|
|
|
$max_transform_time = max($max_transform_time, $finished - $started); |
484
|
|
|
|
485
|
|
|
// When deadlines loom, sometimes the best solution is procrastination. |
486
|
|
|
if (++$i < $num_files && TIME_START + $this->time_limit < $finished + $max_transform_time * 2) |
487
|
|
|
{ |
488
|
|
|
// After all, there's always next time. |
489
|
|
|
if (empty($this->next_task)) |
490
|
|
|
{ |
491
|
|
|
$progressfile = $export_dir_slash . $idhash_ext . '.progress.json'; |
492
|
|
|
|
493
|
|
|
$new_details = $this->_details; |
494
|
|
|
$new_details['start'] = $smcFunc['json_decode'](file_get_contents($progressfile), true); |
495
|
|
|
|
496
|
|
|
$this->next_task = array('$sourcedir/tasks/ExportProfileData.php', 'ExportProfileData_Background', $smcFunc['json_encode']($new_details), time() - MAX_CLAIM_THRESHOLD); |
497
|
|
|
} |
498
|
|
|
|
499
|
|
|
// So let's just relax and take a well deserved... |
500
|
|
|
break; |
501
|
|
|
} |
502
|
|
|
} |
503
|
|
|
} |
504
|
|
|
|
505
|
|
|
/** |
506
|
|
|
* Compiles profile data to XML with embedded XSLT. |
507
|
|
|
* |
508
|
|
|
* Internally calls exportXml() and then embeds an XSLT stylesheet into |
509
|
|
|
* the XML so that it can be processed by the client. |
510
|
|
|
* |
511
|
|
|
* @param array $member_info Minimal $user_info about the relevant member. |
512
|
|
|
*/ |
513
|
|
|
protected function exportXmlXslt($member_info) |
514
|
|
|
{ |
515
|
|
|
global $modSettings, $context, $smcFunc, $sourcedir; |
516
|
|
|
|
517
|
|
|
$context['export_last_page'] = $this->_details['last_page']; |
518
|
|
|
$context['export_dlfilename'] = $this->_details['dlfilename']; |
519
|
|
|
|
520
|
|
|
// Embedded XSLT requires adding a special DTD and processing instruction in the main XML document. |
521
|
|
|
add_integration_function('integrate_xml_data', 'ExportProfileData_Background::add_dtd', false); |
522
|
|
|
|
523
|
|
|
// Perform the export to XML. |
524
|
|
|
$this->exportXml($member_info); |
525
|
|
|
|
526
|
|
|
// Make sure we have everything we need. |
527
|
|
|
if (empty(self::$xslt_info['stylesheet'])) |
528
|
|
|
{ |
529
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php'); |
530
|
|
|
self::$xslt_info = get_xslt_stylesheet($this->_details['format'], $this->_details['uid']); |
531
|
|
|
} |
532
|
|
|
if (empty($context['feed']['footer'])) |
533
|
|
|
{ |
534
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'News.php'); |
535
|
|
|
buildXmlFeed('smf', array(), array_fill_keys(array('title', 'desc', 'source', 'self'), ''), 'profile'); |
536
|
|
|
} |
537
|
|
|
|
538
|
|
|
// Find any completed files that don't yet have the stylesheet embedded in them. |
539
|
|
|
$export_dir_slash = $modSettings['export_dir'] . DIRECTORY_SEPARATOR; |
540
|
|
|
$idhash = hash_hmac('sha1', $this->_details['uid'], get_auth_secret()); |
541
|
|
|
$idhash_ext = $idhash . '.' . $this->_details['format_settings']['extension']; |
542
|
|
|
|
543
|
|
|
$test_length = strlen(self::$xslt_info['stylesheet'] . $context['feed']['footer']); |
544
|
|
|
|
545
|
|
|
$new_exportfiles = array(); |
546
|
|
|
foreach (glob($export_dir_slash . '*_' . $idhash_ext) as $completed_file) |
547
|
|
|
{ |
548
|
|
|
if (filesize($completed_file) < $test_length || file_get_contents($completed_file, false, null, $test_length * -1) !== self::$xslt_info['stylesheet'] . $context['feed']['footer']) |
549
|
|
|
$new_exportfiles[] = $completed_file; |
550
|
|
|
} |
551
|
|
|
if (empty($new_exportfiles)) |
552
|
|
|
return; |
553
|
|
|
|
554
|
|
|
// Embedding the XSLT means writing to the file yet again. |
555
|
|
|
foreach ($new_exportfiles as $exportfile) |
556
|
|
|
{ |
557
|
|
|
$handle = fopen($exportfile, 'r+'); |
558
|
|
|
if (is_resource($handle)) |
559
|
|
|
{ |
560
|
|
|
flock($handle, LOCK_EX); |
561
|
|
|
|
562
|
|
|
fseek($handle, strlen($context['feed']['footer']) * -1, SEEK_END); |
563
|
|
|
|
564
|
|
|
$bytes_written = fwrite($handle, self::$xslt_info['stylesheet'] . $context['feed']['footer']); |
565
|
|
|
|
566
|
|
|
// If we couldn't write everything, revert the changes. |
567
|
|
|
if ($bytes_written > 0 && $bytes_written < strlen(self::$xslt_info['stylesheet'] . $context['feed']['footer'])) |
568
|
|
|
{ |
569
|
|
|
fseek($handle, $bytes_written * -1, SEEK_END); |
570
|
|
|
$pointer_pos = ftell($handle); |
571
|
|
|
ftruncate($handle, $pointer_pos); |
572
|
|
|
rewind($handle); |
573
|
|
|
fseek($handle, 0, SEEK_END); |
574
|
|
|
fwrite($handle, $context['feed']['footer']); |
575
|
|
|
} |
576
|
|
|
|
577
|
|
|
flock($handle, LOCK_UN); |
578
|
|
|
fclose($handle); |
579
|
|
|
} |
580
|
|
|
} |
581
|
|
|
} |
582
|
|
|
|
583
|
|
|
/** |
584
|
|
|
* Adds a custom DOCTYPE definition and an XSLT processing instruction to |
585
|
|
|
* the main XML file's header. |
586
|
|
|
*/ |
587
|
|
|
public static function add_dtd(&$xml_data, &$feed_meta, &$namespaces, &$extraFeedTags, &$forceCdataKeys, &$nsKeys, $xml_format, $subaction, &$doctype) |
588
|
|
|
{ |
589
|
|
|
global $sourcedir; |
590
|
|
|
|
591
|
|
|
require_once($sourcedir . DIRECTORY_SEPARATOR . 'Profile-Export.php'); |
592
|
|
|
self::$xslt_info = get_xslt_stylesheet(self::$export_details['format'], self::$export_details['uid']); |
593
|
|
|
|
594
|
|
|
$doctype = self::$xslt_info['doctype']; |
595
|
|
|
} |
596
|
|
|
|
597
|
|
|
/** |
598
|
|
|
* Adjusts some parse_bbc() parameters for the special case of exports. |
599
|
|
|
*/ |
600
|
|
|
public static function pre_parsebbc(&$message, &$smileys, &$cache_id, &$parse_tags) |
601
|
|
|
{ |
602
|
|
|
global $modSettings, $context, $user_info; |
603
|
|
|
|
604
|
|
|
$cache_id = ''; |
605
|
|
|
|
606
|
|
|
if (in_array(self::$export_details['format'], array('HTML', 'XML_XSLT'))) |
607
|
|
|
{ |
608
|
|
|
foreach (array('smileys_url', 'attachmentThumbnails') as $var) |
609
|
|
|
if (isset($modSettings[$var])) |
610
|
|
|
self::$real_modSettings[$var] = $modSettings[$var]; |
611
|
|
|
|
612
|
|
|
$modSettings['smileys_url'] = '.'; |
613
|
|
|
$modSettings['attachmentThumbnails'] = false; |
614
|
|
|
} |
615
|
|
|
else |
616
|
|
|
{ |
617
|
|
|
$smileys = false; |
618
|
|
|
|
619
|
|
|
if (!isset($modSettings['disabledBBC'])) |
620
|
|
|
$modSettings['disabledBBC'] = 'attach'; |
621
|
|
|
else |
622
|
|
|
{ |
623
|
|
|
self::$real_modSettings['disabledBBC'] = $modSettings['disabledBBC']; |
624
|
|
|
|
625
|
|
|
if (strpos($modSettings['disabledBBC'], 'attach') === false) |
626
|
|
|
$modSettings['disabledBBC'] = implode(',', array_merge(array_filter(explode(',', $modSettings['disabledBBC'])), array('attach'))); |
627
|
|
|
} |
628
|
|
|
} |
629
|
|
|
} |
630
|
|
|
|
631
|
|
|
/** |
632
|
|
|
* Reverses changes made by pre_parsebbc() |
633
|
|
|
*/ |
634
|
|
|
public static function post_parsebbc(&$message, &$smileys, &$cache_id, &$parse_tags) |
635
|
|
|
{ |
636
|
|
|
global $modSettings, $context; |
637
|
|
|
|
638
|
|
|
foreach (array('disabledBBC', 'smileys_url', 'attachmentThumbnails') as $var) |
639
|
|
|
if (isset(self::$real_modSettings[$var])) |
640
|
|
|
$modSettings[$var] = self::$real_modSettings[$var]; |
641
|
|
|
} |
642
|
|
|
|
643
|
|
|
/** |
644
|
|
|
* Adjusts certain BBCodes for the special case of exports. |
645
|
|
|
*/ |
646
|
|
|
public static function bbc_codes(&$codes, &$no_autolink_tags) |
647
|
|
|
{ |
648
|
|
|
foreach ($codes as &$code) |
649
|
|
|
{ |
650
|
|
|
// To make the "Select" link work we'd need to embed a bunch more JS. Not worth it. |
651
|
|
|
if ($code['tag'] === 'code') |
652
|
|
|
$code['content'] = preg_replace('~<a class="codeoperation\b.*?</a>~', '', $code['content']); |
653
|
|
|
} |
654
|
|
|
} |
655
|
|
|
|
656
|
|
|
/** |
657
|
|
|
* Adjusts the attachment download URL for the special case of exports. |
658
|
|
|
*/ |
659
|
|
|
public static function post_parseAttachBBC(&$attachContext) |
660
|
|
|
{ |
661
|
|
|
global $scripturl, $context; |
662
|
|
|
static $dltokens; |
663
|
|
|
|
664
|
|
|
if (empty($dltokens[$context['xmlnews_uid']])) |
665
|
|
|
{ |
666
|
|
|
$idhash = hash_hmac('sha1', $context['xmlnews_uid'], get_auth_secret()); |
667
|
|
|
$dltokens[$context['xmlnews_uid']] = hash_hmac('sha1', $idhash, get_auth_secret()); |
668
|
|
|
} |
669
|
|
|
|
670
|
|
|
$attachContext['orig_href'] = $scripturl . '?action=profile;area=dlattach;u=' . $context['xmlnews_uid'] . ';attach=' . $attachContext['id'] . ';t=' . $dltokens[$context['xmlnews_uid']]; |
671
|
|
|
$attachContext['href'] = rawurlencode($attachContext['id'] . ' - ' . html_entity_decode($attachContext['name'])); |
672
|
|
|
} |
673
|
|
|
|
674
|
|
|
/** |
675
|
|
|
* Adjusts the format of the HTML produced by the attach BBCode. |
676
|
|
|
*/ |
677
|
|
|
public static function attach_bbc_validate(&$returnContext, $currentAttachment, $tag, $data, $disabled, $params) |
678
|
|
|
{ |
679
|
|
|
global $smcFunc, $txt; |
680
|
|
|
|
681
|
|
|
$orig_link = '<a href="' . $currentAttachment['orig_href'] . '" class="bbc_link">' . $txt['export_download_original'] . '</a>'; |
682
|
|
|
$hidden_orig_link = ' <a href="' . $currentAttachment['orig_href'] . '" class="bbc_link dlattach_' . $currentAttachment['id'] . '" style="display:none; flex: 1 0 auto; margin: auto;">' . $txt['export_download_original'] . '</a>'; |
683
|
|
|
|
684
|
|
|
if ($params['{display}'] == 'link') |
685
|
|
|
{ |
686
|
|
|
$returnContext .= ' (' . $orig_link . ')'; |
687
|
|
|
} |
688
|
|
|
elseif (!empty($currentAttachment['is_image'])) |
689
|
|
|
{ |
690
|
|
|
$returnContext = '<span style="display: inline-flex; justify-content: center; align-items: center; position: relative;">' . preg_replace( |
691
|
|
|
array( |
692
|
|
|
'thumbnail_toggle' => '~</?a\b[^>]*>~', |
693
|
|
|
'src' => '~src="' . preg_quote($currentAttachment['href'], '~') . ';image"~', |
694
|
|
|
), |
695
|
|
|
array( |
696
|
|
|
'thumbnail_toggle' => '', |
697
|
|
|
'src' => 'src="' . $currentAttachment['href'] . '" onerror="$(\'.dlattach_' . $currentAttachment['id'] . '\').show(); $(\'.dlattach_' . $currentAttachment['id'] . '\').css({\'position\': \'absolute\'});"', |
698
|
|
|
), |
699
|
|
|
$returnContext |
700
|
|
|
) . $hidden_orig_link . '</span>' ; |
701
|
|
|
} |
702
|
|
|
elseif (strpos($currentAttachment['mime_type'], 'video/') === 0) |
703
|
|
|
{ |
704
|
|
|
$returnContext = preg_replace( |
705
|
|
|
array( |
706
|
|
|
'src' => '~src="' . preg_quote($currentAttachment['href'], '~') . '"~', |
707
|
|
|
'opening_tag' => '~^<div class="videocontainer"~', |
708
|
|
|
'closing_tag' => '~</div>$~', |
709
|
|
|
), |
710
|
|
|
array( |
711
|
|
|
'src' => '$0 onerror="$(this).fadeTo(0, 0.2); $(\'.dlattach_' . $currentAttachment['id'] . '\').show(); $(\'.dlattach_' . $currentAttachment['id'] . '\').css({\'position\': \'absolute\'});"', |
712
|
|
|
'opening_tag' => '<div class="videocontainer" style="display: flex; justify-content: center; align-items: center; position: relative;"', |
713
|
|
|
'closing_tag' => $hidden_orig_link . '</div>', |
714
|
|
|
), |
715
|
|
|
$returnContext |
716
|
|
|
); |
717
|
|
|
} |
718
|
|
|
elseif (strpos($currentAttachment['mime_type'], 'audio/') === 0) |
719
|
|
|
{ |
720
|
|
|
$returnContext = '<span style="display: inline-flex; justify-content: center; align-items: center; position: relative;">' . preg_replace( |
721
|
|
|
array( |
722
|
|
|
'opening_tag' => '~^<audio\b~', |
723
|
|
|
), |
724
|
|
|
array( |
725
|
|
|
'opening_tag' => '<audio onerror="$(this).fadeTo(0, 0); $(\'.dlattach_' . $currentAttachment['id'] . '\').show(); $(\'.dlattach_' . $currentAttachment['id'] . '\').css({\'position\': \'absolute\'});"', |
726
|
|
|
), |
727
|
|
|
$returnContext |
728
|
|
|
) . $hidden_orig_link . '</span>'; |
729
|
|
|
} |
730
|
|
|
else |
731
|
|
|
{ |
732
|
|
|
$returnContext = '<span style="display: inline-flex; justify-content: center; align-items: center; position: relative;">' . preg_replace( |
733
|
|
|
array( |
734
|
|
|
'obj_opening' => '~^<object\b~', |
735
|
|
|
'link' => '~<a href="' . preg_quote($currentAttachment['href'], '~') . '" class="bbc_link">([^<]*)</a>~', |
736
|
|
|
), |
737
|
|
|
array( |
738
|
|
|
'obj_opening' => '<object onerror="$(this).fadeTo(0, 0.2); $(\'.dlattach_' . $currentAttachment['id'] . '\').show(); $(\'.dlattach_' . $currentAttachment['id'] . '\').css({\'position\': \'absolute\'});"~', |
739
|
|
|
'link' => '$0 (' . $orig_link . ')', |
740
|
|
|
), |
741
|
|
|
$returnContext |
742
|
|
|
) . $hidden_orig_link . '</span>'; |
743
|
|
|
} |
744
|
|
|
} |
745
|
|
|
} |
746
|
|
|
|
747
|
|
|
?> |