1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Blacklight; |
4
|
|
|
|
5
|
|
|
use App\Models\Group; |
6
|
|
|
use Blacklight\db\DB; |
7
|
|
|
use App\Models\Settings; |
8
|
|
|
use Illuminate\Support\Carbon; |
9
|
|
|
use App\Models\BinaryBlacklist; |
10
|
|
|
use App\Models\MultigroupPoster; |
11
|
|
|
use Illuminate\Support\Facades\Cache; |
12
|
|
|
use Blacklight\processing\ProcessReleasesMultiGroup; |
13
|
|
|
|
14
|
|
|
/** |
15
|
|
|
* Class Binaries. |
16
|
|
|
*/ |
17
|
|
|
class Binaries |
18
|
|
|
{ |
19
|
|
|
public const OPTYPE_BLACKLIST = 1; |
20
|
|
|
public const OPTYPE_WHITELIST = 2; |
21
|
|
|
|
22
|
|
|
public const BLACKLIST_DISABLED = 0; |
23
|
|
|
public const BLACKLIST_ENABLED = 1; |
24
|
|
|
|
25
|
|
|
public const BLACKLIST_FIELD_SUBJECT = 1; |
26
|
|
|
public const BLACKLIST_FIELD_FROM = 2; |
27
|
|
|
public const BLACKLIST_FIELD_MESSAGEID = 3; |
28
|
|
|
|
29
|
|
|
/** |
30
|
|
|
* @var array |
31
|
|
|
*/ |
32
|
|
|
public $blackList = []; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @var array |
36
|
|
|
*/ |
37
|
|
|
public $whiteList = []; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @var int |
41
|
|
|
*/ |
42
|
|
|
public $messageBuffer; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* @var \Blacklight\ColorCLI |
46
|
|
|
*/ |
47
|
|
|
protected $_colorCLI; |
48
|
|
|
|
49
|
|
|
/** |
50
|
|
|
* @var \Blacklight\CollectionsCleaning |
51
|
|
|
*/ |
52
|
|
|
protected $_collectionsCleaning; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @var \Blacklight\NNTP |
56
|
|
|
*/ |
57
|
|
|
protected $_nntp; |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* Should we use header compression? |
61
|
|
|
* |
62
|
|
|
* @var bool |
63
|
|
|
*/ |
64
|
|
|
protected $_compressedHeaders; |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* Should we use part repair? |
68
|
|
|
* |
69
|
|
|
* @var bool |
70
|
|
|
*/ |
71
|
|
|
protected $_partRepair; |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* @var \Blacklight\db\DB |
75
|
|
|
*/ |
76
|
|
|
protected $_pdo; |
77
|
|
|
|
78
|
|
|
/** |
79
|
|
|
* How many days to go back on a new group? |
80
|
|
|
* |
81
|
|
|
* @var bool |
82
|
|
|
*/ |
83
|
|
|
protected $_newGroupScanByDays; |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* How many headers to download on new groups? |
87
|
|
|
* |
88
|
|
|
* @var int |
89
|
|
|
*/ |
90
|
|
|
protected $_newGroupMessagesToScan; |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* How many days to go back on new groups? |
94
|
|
|
* |
95
|
|
|
* @var int |
96
|
|
|
*/ |
97
|
|
|
protected $_newGroupDaysToScan; |
98
|
|
|
|
99
|
|
|
/** |
100
|
|
|
* How many headers to download per run of part repair? |
101
|
|
|
* |
102
|
|
|
* @var int |
103
|
|
|
*/ |
104
|
|
|
protected $_partRepairLimit; |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* Should we show dropped yEnc to CLI? |
108
|
|
|
* |
109
|
|
|
* @var bool |
110
|
|
|
*/ |
111
|
|
|
protected $_showDroppedYEncParts; |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Echo to cli? |
115
|
|
|
* |
116
|
|
|
* @var bool |
117
|
|
|
*/ |
118
|
|
|
protected $_echoCLI; |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* @var bool |
122
|
|
|
*/ |
123
|
|
|
protected $_debug = false; |
124
|
|
|
|
125
|
|
|
/** |
126
|
|
|
* Max tries to download headers. |
127
|
|
|
* @var int |
128
|
|
|
*/ |
129
|
|
|
protected $_partRepairMaxTries; |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* An array of binaryblacklist IDs that should have their activity date updated. |
133
|
|
|
* @var array(int) |
134
|
|
|
*/ |
135
|
|
|
protected $_binaryBlacklistIdsToUpdate = []; |
136
|
|
|
|
137
|
|
|
/** |
138
|
|
|
* @var float microseconds time of cleaning process start |
139
|
|
|
*/ |
140
|
|
|
protected $startCleaning; |
141
|
|
|
|
142
|
|
|
/** |
143
|
|
|
* @var float microseconds time of the start of the scan function |
144
|
|
|
*/ |
145
|
|
|
protected $startLoop; |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* @var bool Is this retrieved header a multigroup one? |
149
|
|
|
*/ |
150
|
|
|
protected $multiGroup; |
151
|
|
|
|
152
|
|
|
/** |
153
|
|
|
* @var bool |
154
|
|
|
*/ |
155
|
|
|
protected $allAsMgr; |
156
|
|
|
|
157
|
|
|
/** |
158
|
|
|
* @var string How long it took in seconds to download headers |
159
|
|
|
*/ |
160
|
|
|
protected $timeHeaders; |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* @var string How long it took in seconds to clean/parse headers |
164
|
|
|
*/ |
165
|
|
|
protected $timeCleaning; |
166
|
|
|
|
167
|
|
|
/** |
168
|
|
|
* @var float microseconds time part repair was started |
169
|
|
|
*/ |
170
|
|
|
protected $startPR; |
171
|
|
|
|
172
|
|
|
/** |
173
|
|
|
* @var array The CBP/MGR tables names |
174
|
|
|
*/ |
175
|
|
|
protected $tableNames; |
176
|
|
|
|
177
|
|
|
/** |
178
|
|
|
* @var float microseconds time header update was started |
179
|
|
|
*/ |
180
|
|
|
protected $startUpdate; |
181
|
|
|
|
182
|
|
|
/** |
183
|
|
|
* @var string The time it took to insert the headers |
184
|
|
|
*/ |
185
|
|
|
protected $timeInsert; |
186
|
|
|
|
187
|
|
|
/** |
188
|
|
|
* @var array the header currently being scanned |
189
|
|
|
*/ |
190
|
|
|
protected $header; |
191
|
|
|
|
192
|
|
|
/** |
193
|
|
|
* @var bool Should we add parts to part repair queue? |
194
|
|
|
*/ |
195
|
|
|
protected $addToPartRepair; |
196
|
|
|
|
197
|
|
|
/** |
198
|
|
|
* @var array Numbers of Headers received from the USP |
199
|
|
|
*/ |
200
|
|
|
protected $headersReceived; |
201
|
|
|
|
202
|
|
|
/** |
203
|
|
|
* @var array The current newsgroup information being updated |
204
|
|
|
*/ |
205
|
|
|
protected $groupMySQL; |
206
|
|
|
|
207
|
|
|
/** |
208
|
|
|
* @var int the last article number in the range |
209
|
|
|
*/ |
210
|
|
|
protected $last; |
211
|
|
|
|
212
|
|
|
/** |
213
|
|
|
* @var int the first article number in the range |
214
|
|
|
*/ |
215
|
|
|
protected $first; |
216
|
|
|
|
217
|
|
|
/** |
218
|
|
|
* @var int How many received headers were not yEnc encoded |
219
|
|
|
*/ |
220
|
|
|
protected $notYEnc; |
221
|
|
|
|
222
|
|
|
/** |
223
|
|
|
* @var int How many received headers were blacklist matched |
224
|
|
|
*/ |
225
|
|
|
protected $headersBlackListed; |
226
|
|
|
|
227
|
|
|
/** |
228
|
|
|
* @var array Header numbers that were not inserted |
229
|
|
|
*/ |
230
|
|
|
protected $headersNotInserted; |
231
|
|
|
|
232
|
|
|
/** |
233
|
|
|
* Constructor. |
234
|
|
|
* |
235
|
|
|
* @param array $options Class instances / echo to CLI? |
236
|
|
|
* |
237
|
|
|
* @throws \Exception |
238
|
|
|
*/ |
239
|
|
|
public function __construct(array $options = []) |
240
|
|
|
{ |
241
|
|
|
$defaults = [ |
242
|
|
|
'Echo' => true, |
243
|
|
|
'CollectionsCleaning' => null, |
244
|
|
|
'ColorCLI' => null, |
245
|
|
|
'Logger' => null, |
246
|
|
|
'Groups' => null, |
247
|
|
|
'NNTP' => null, |
248
|
|
|
'Settings' => null, |
249
|
|
|
]; |
250
|
|
|
$options += $defaults; |
251
|
|
|
|
252
|
|
|
$this->_echoCLI = ($options['Echo'] && config('nntmux.echocli')); |
253
|
|
|
|
254
|
|
|
$this->_pdo = ($options['Settings'] instanceof DB ? $options['Settings'] : new DB()); |
255
|
|
|
$this->_colorCLI = ($options['ColorCLI'] instanceof ColorCLI ? $options['ColorCLI'] : new ColorCLI()); |
256
|
|
|
$this->_nntp = ($options['NNTP'] instanceof NNTP ? $options['NNTP'] : new NNTP(['Echo' => $this->_colorCLI, 'Settings' => $this->_pdo, 'ColorCLI' => $this->_colorCLI])); |
257
|
|
|
$this->_collectionsCleaning = ($options['CollectionsCleaning'] instanceof CollectionsCleaning ? $options['CollectionsCleaning'] : new CollectionsCleaning(['Settings' => $this->_pdo])); |
258
|
|
|
|
259
|
|
|
$this->messageBuffer = Settings::settingValue('..maxmssgs') !== '' ? |
|
|
|
|
260
|
|
|
(int) Settings::settingValue('..maxmssgs') : 20000; |
261
|
|
|
$this->_compressedHeaders = (int) Settings::settingValue('..compressedheaders') === 1; |
262
|
|
|
$this->_partRepair = (int) Settings::settingValue('..partrepair') === 1; |
263
|
|
|
$this->_newGroupScanByDays = (int) Settings::settingValue('..newgroupscanmethod') === 1; |
264
|
|
|
$this->_newGroupMessagesToScan = Settings::settingValue('..newgroupmsgstoscan') !== '' ? (int) Settings::settingValue('..newgroupmsgstoscan') : 50000; |
265
|
|
|
$this->_newGroupDaysToScan = Settings::settingValue('..newgroupdaystoscan') !== '' ? (int) Settings::settingValue('..newgroupdaystoscan') : 3; |
266
|
|
|
$this->_partRepairLimit = Settings::settingValue('..maxpartrepair') !== '' ? (int) Settings::settingValue('..maxpartrepair') : 15000; |
267
|
|
|
$this->_partRepairMaxTries = (Settings::settingValue('..partrepairmaxtries') !== '' ? (int) Settings::settingValue('..partrepairmaxtries') : 3); |
268
|
|
|
$this->_showDroppedYEncParts = (int) Settings::settingValue('..showdroppedyencparts') === 1; |
269
|
|
|
$this->allAsMgr = (int) Settings::settingValue('..allasmgr') === 1; |
270
|
|
|
|
271
|
|
|
$this->blackList = $this->whiteList = []; |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
/** |
275
|
|
|
* Download new headers for all active groups. |
276
|
|
|
* |
277
|
|
|
* @param int $maxHeaders (Optional) How many headers to download max. |
278
|
|
|
* |
279
|
|
|
* @return void |
280
|
|
|
* @throws \Exception |
281
|
|
|
*/ |
282
|
|
|
public function updateAllGroups($maxHeaders = 100000): void |
283
|
|
|
{ |
284
|
|
|
$groups = Group::getActive(); |
285
|
|
|
|
286
|
|
|
$groupCount = \count($groups); |
287
|
|
|
if ($groupCount > 0) { |
288
|
|
|
$counter = 1; |
289
|
|
|
$allTime = microtime(true); |
290
|
|
|
|
291
|
|
|
$this->log( |
292
|
|
|
'Updating: '.$groupCount.' group(s) - Using compression? '.($this->_compressedHeaders ? 'Yes' : 'No'), |
293
|
|
|
__FUNCTION__, |
294
|
|
|
'header' |
295
|
|
|
); |
296
|
|
|
|
297
|
|
|
// Loop through groups. |
298
|
|
|
foreach ($groups as $group) { |
299
|
|
|
$this->log( |
300
|
|
|
'Starting group '.$counter.' of '.$groupCount, |
301
|
|
|
__FUNCTION__, |
302
|
|
|
'header' |
303
|
|
|
); |
304
|
|
|
$this->updateGroup($group, $maxHeaders); |
305
|
|
|
$counter++; |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
$this->log( |
309
|
|
|
'Updating completed in '.number_format(microtime(true) - $allTime, 2).' seconds.', |
310
|
|
|
__FUNCTION__, |
311
|
|
|
'primary' |
312
|
|
|
); |
313
|
|
|
} else { |
314
|
|
|
$this->log( |
315
|
|
|
'No groups specified. Ensure groups are added to NNTmux\'s database for updating.', |
316
|
|
|
__FUNCTION__, |
317
|
|
|
'warning' |
318
|
|
|
); |
319
|
|
|
} |
320
|
|
|
} |
321
|
|
|
|
322
|
|
|
/** |
323
|
|
|
* When the indexer is started, log the date/time. |
324
|
|
|
*/ |
325
|
|
|
public function logIndexerStart(): void |
326
|
|
|
{ |
327
|
|
|
Settings::query()->where('setting', '=', 'last_run_time')->update(['value' => Carbon::now()]); |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/** |
331
|
|
|
* Download new headers for a single group. |
332
|
|
|
* |
333
|
|
|
* @param array $groupMySQL Array of MySQL results for a single group. |
334
|
|
|
* @param int $maxHeaders (Optional) How many headers to download max. |
335
|
|
|
* |
336
|
|
|
* @return void |
337
|
|
|
* @throws \Exception |
338
|
|
|
*/ |
339
|
|
|
public function updateGroup($groupMySQL, $maxHeaders = 0): void |
340
|
|
|
{ |
341
|
|
|
$startGroup = microtime(true); |
342
|
|
|
|
343
|
|
|
$this->logIndexerStart(); |
344
|
|
|
|
345
|
|
|
// Select the group on the NNTP server, gets the latest info on it. |
346
|
|
|
$groupNNTP = $this->_nntp->selectGroup($groupMySQL['name']); |
347
|
|
|
if ($this->_nntp->isError($groupNNTP)) { |
348
|
|
|
$groupNNTP = $this->_nntp->dataError($this->_nntp, $groupMySQL['name']); |
349
|
|
|
|
350
|
|
|
if (isset($groupNNTP['code']) && (int) $groupNNTP['code'] === 411) { |
351
|
|
|
Group::disableIfNotExist($groupMySQL['id']); |
352
|
|
|
} |
353
|
|
|
if ($this->_nntp->isError($groupNNTP)) { |
354
|
|
|
return; |
355
|
|
|
} |
356
|
|
|
} |
357
|
|
|
|
358
|
|
|
if ($this->_echoCLI) { |
359
|
|
|
ColorCLI::doEcho(ColorCLI::primary('Processing '.$groupMySQL['name']), true); |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
// Attempt to repair any missing parts before grabbing new ones. |
363
|
|
|
if ((int) $groupMySQL['last_record'] !== 0) { |
364
|
|
|
if ($this->_partRepair) { |
365
|
|
|
if ($this->_echoCLI) { |
366
|
|
|
ColorCLI::doEcho(ColorCLI::primary('Part repair enabled. Checking for missing parts.'), true); |
367
|
|
|
} |
368
|
|
|
$this->partRepair($groupMySQL); |
369
|
|
|
|
370
|
|
|
$mgrPosters = $this->getMultiGroupPosters(); |
371
|
|
|
if ($this->allAsMgr === true || ! empty($mgrPosters)) { |
372
|
|
|
$tableNames = ProcessReleasesMultiGroup::tableNames(); |
373
|
|
|
$this->partRepair($groupMySQL, $tableNames); |
374
|
|
|
} |
375
|
|
|
} elseif ($this->_echoCLI) { |
376
|
|
|
ColorCLI::doEcho(ColorCLI::primary('Part repair disabled by user.'), true); |
377
|
|
|
} |
378
|
|
|
} |
379
|
|
|
|
380
|
|
|
// Generate postdate for first record, for those that upgraded. |
381
|
|
|
if ($groupMySQL['first_record_postdate'] === null && (int) $groupMySQL['first_record'] !== 0) { |
382
|
|
|
$groupMySQL['first_record_postdate'] = $this->postdate($groupMySQL['first_record'], $groupNNTP); |
383
|
|
|
Group::query()->where('id', $groupMySQL['id'])->update(['first_record_postdate' => Carbon::createFromTimestamp($groupMySQL['first_record_postdate'])]); |
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
// Get first article we want aka the oldest. |
387
|
|
|
if ((int) $groupMySQL['last_record'] === 0) { |
388
|
|
|
if ($this->_newGroupScanByDays) { |
389
|
|
|
// For new newsgroups - determine here how far we want to go back using date. |
390
|
|
|
$first = $this->daytopost($this->_newGroupDaysToScan, $groupNNTP); |
391
|
|
|
} elseif ($groupNNTP['first'] >= ($groupNNTP['last'] - ($this->_newGroupMessagesToScan + $this->messageBuffer))) { |
392
|
|
|
// If what we want is lower than the groups first article, set the wanted first to the first. |
393
|
|
|
$first = $groupNNTP['first']; |
394
|
|
|
} else { |
395
|
|
|
// Or else, use the newest article minus how much we should get for new groups. |
396
|
|
|
$first = (string) ($groupNNTP['last'] - ($this->_newGroupMessagesToScan + $this->messageBuffer)); |
397
|
|
|
} |
398
|
|
|
|
399
|
|
|
// We will use this to subtract so we leave articles for the next time (in case the server doesn't have them yet) |
400
|
|
|
$leaveOver = $this->messageBuffer; |
401
|
|
|
|
402
|
|
|
// If this is not a new group, go from our newest to the servers newest. |
403
|
|
|
} else { |
404
|
|
|
// Set our oldest wanted to our newest local article. |
405
|
|
|
$first = $groupMySQL['last_record']; |
406
|
|
|
|
407
|
|
|
// This is how many articles we will grab. (the servers newest minus our newest). |
408
|
|
|
$totalCount = (string) ($groupNNTP['last'] - $first); |
409
|
|
|
|
410
|
|
|
// Check if the server has more articles than our loop limit x 2. |
411
|
|
|
if ($totalCount > ($this->messageBuffer * 2)) { |
412
|
|
|
// Get the remainder of $totalCount / $this->message buffer |
413
|
|
|
$leaveOver = round($totalCount % $this->messageBuffer, 0, PHP_ROUND_HALF_DOWN) + $this->messageBuffer; |
414
|
|
|
} else { |
415
|
|
|
// Else get half of the available. |
416
|
|
|
$leaveOver = round($totalCount / 2, 0, PHP_ROUND_HALF_DOWN); |
417
|
|
|
} |
418
|
|
|
} |
419
|
|
|
|
420
|
|
|
// The last article we want, aka the newest. |
421
|
|
|
$last = $groupLast = (string) ($groupNNTP['last'] - $leaveOver); |
422
|
|
|
|
423
|
|
|
// If the newest we want is older than the oldest we want somehow.. set them equal. |
424
|
|
|
if ($last < $first) { |
425
|
|
|
$last = $groupLast = $first; |
426
|
|
|
} |
427
|
|
|
|
428
|
|
|
// This is how many articles we are going to get. |
429
|
|
|
$total = (string) ($groupLast - $first); |
430
|
|
|
// This is how many articles are available (without $leaveOver). |
431
|
|
|
$realTotal = (string) ($groupNNTP['last'] - $first); |
432
|
|
|
|
433
|
|
|
// Check if we should limit the amount of fetched new headers. |
434
|
|
|
if ($maxHeaders > 0) { |
435
|
|
|
if ($maxHeaders < ($groupLast - $first)) { |
436
|
|
|
$groupLast = $last = (string) ($first + $maxHeaders); |
437
|
|
|
} |
438
|
|
|
$total = (string) ($groupLast - $first); |
439
|
|
|
} |
440
|
|
|
|
441
|
|
|
// If total is bigger than 0 it means we have new parts in the newsgroup. |
442
|
|
|
if ($total > 0) { |
443
|
|
|
if ($this->_echoCLI) { |
444
|
|
|
ColorCLI::doEcho( |
445
|
|
|
ColorCLI::primary( |
446
|
|
|
( |
447
|
|
|
(int) $groupMySQL['last_record'] === 0 |
448
|
|
|
? 'New group '.$groupNNTP['group'].' starting with '. |
449
|
|
|
( |
450
|
|
|
$this->_newGroupScanByDays |
451
|
|
|
? $this->_newGroupDaysToScan.' days' |
452
|
|
|
: number_format($this->_newGroupMessagesToScan).' messages' |
453
|
|
|
).' worth.' |
454
|
|
|
: 'Group '.$groupNNTP['group'].' has '.number_format($realTotal).' new articles.' |
|
|
|
|
455
|
|
|
). |
456
|
|
|
' Leaving '.number_format($leaveOver). |
457
|
|
|
" for next pass.\nServer oldest: ".number_format($groupNNTP['first']). |
458
|
|
|
' Server newest: '.number_format($groupNNTP['last']). |
459
|
|
|
' Local newest: '.number_format($groupMySQL['last_record']) |
460
|
|
|
), |
461
|
|
|
true |
462
|
|
|
); |
463
|
|
|
} |
464
|
|
|
|
465
|
|
|
$done = false; |
466
|
|
|
// Get all the parts (in portions of $this->messageBuffer to not use too much memory). |
467
|
|
|
while ($done === false) { |
468
|
|
|
|
469
|
|
|
// Increment last until we reach $groupLast (group newest article). |
470
|
|
|
if ($total > $this->messageBuffer) { |
471
|
|
|
if ((string) ($first + $this->messageBuffer) > $groupLast) { |
472
|
|
|
$last = $groupLast; |
473
|
|
|
} else { |
474
|
|
|
$last = (string) ($first + $this->messageBuffer); |
475
|
|
|
} |
476
|
|
|
} |
477
|
|
|
// Increment first so we don't get an article we already had. |
478
|
|
|
$first++; |
479
|
|
|
|
480
|
|
|
if ($this->_echoCLI) { |
481
|
|
|
ColorCLI::doEcho( |
482
|
|
|
ColorCLI::header( |
483
|
|
|
PHP_EOL.'Getting '.number_format($last - $first + 1).' articles ('.number_format($first). |
484
|
|
|
' to '.number_format($last).') from '.$groupMySQL['name'].' - ('. |
485
|
|
|
number_format($groupLast - $last).' articles in queue).' |
486
|
|
|
) |
487
|
|
|
); |
488
|
|
|
} |
489
|
|
|
|
490
|
|
|
// Get article headers from newsgroup. |
491
|
|
|
$scanSummary = $this->scan($groupMySQL, $first, $last); |
492
|
|
|
|
493
|
|
|
// Check if we fetched headers. |
494
|
|
|
if (! empty($scanSummary)) { |
495
|
|
|
|
496
|
|
|
// If new group, update first record & postdate |
497
|
|
|
if ($groupMySQL['first_record_postdate'] === null && (int) $groupMySQL['first_record'] === 0) { |
498
|
|
|
$groupMySQL['first_record'] = $scanSummary['firstArticleNumber']; |
499
|
|
|
|
500
|
|
|
if (isset($scanSummary['firstArticleDate'])) { |
501
|
|
|
$groupMySQL['first_record_postdate'] = strtotime($scanSummary['firstArticleDate']); |
502
|
|
|
} else { |
503
|
|
|
$groupMySQL['first_record_postdate'] = $this->postdate($groupMySQL['first_record'], $groupNNTP); |
504
|
|
|
} |
505
|
|
|
|
506
|
|
|
Group::query() |
507
|
|
|
->where('id', $groupMySQL['id']) |
508
|
|
|
->update( |
509
|
|
|
[ |
510
|
|
|
'first_record' => $scanSummary['firstArticleNumber'], |
511
|
|
|
'first_record_postdate' => Carbon::createFromTimestamp( |
512
|
|
|
$groupMySQL['first_record_postdate'] |
513
|
|
|
), |
514
|
|
|
] |
515
|
|
|
); |
516
|
|
|
} |
517
|
|
|
|
518
|
|
|
$scanSummary['lastArticleDate'] = (isset($scanSummary['lastArticleDate']) ? strtotime($scanSummary['lastArticleDate']) : false); |
519
|
|
|
if (! is_numeric($scanSummary['lastArticleDate'])) { |
520
|
|
|
$scanSummary['lastArticleDate'] = $this->postdate($scanSummary['lastArticleNumber'], $groupNNTP); |
521
|
|
|
} |
522
|
|
|
|
523
|
|
|
Group::query() |
524
|
|
|
->where('id', $groupMySQL['id']) |
525
|
|
|
->update( |
526
|
|
|
[ |
527
|
|
|
'last_record' => $scanSummary['lastArticleNumber'], |
528
|
|
|
'last_record_postdate' => Carbon::createFromTimestamp($scanSummary['lastArticleDate']), |
529
|
|
|
'last_updated' => Carbon::now(), |
530
|
|
|
] |
531
|
|
|
); |
532
|
|
|
} else { |
533
|
|
|
// If we didn't fetch headers, update the record still. |
534
|
|
|
Group::query() |
535
|
|
|
->where('id', $groupMySQL['id']) |
536
|
|
|
->update( |
537
|
|
|
[ |
538
|
|
|
'last_record' => $last, |
539
|
|
|
'last_updated' => Carbon::now(), |
540
|
|
|
] |
541
|
|
|
); |
542
|
|
|
} |
543
|
|
|
|
544
|
|
|
if ((int) $last === (int) $groupLast) { |
545
|
|
|
$done = true; |
546
|
|
|
} else { |
547
|
|
|
$first = $last; |
548
|
|
|
} |
549
|
|
|
} |
550
|
|
|
|
551
|
|
|
if ($this->_echoCLI) { |
552
|
|
|
ColorCLI::doEcho( |
553
|
|
|
ColorCLI::primary( |
554
|
|
|
PHP_EOL.'Group '.$groupMySQL['name'].' processed in '. |
555
|
|
|
number_format(microtime(true) - $startGroup, 2).' seconds.' |
556
|
|
|
), |
557
|
|
|
true |
558
|
|
|
); |
559
|
|
|
} |
560
|
|
|
} elseif ($this->_echoCLI) { |
561
|
|
|
ColorCLI::doEcho( |
562
|
|
|
ColorCLI::primary( |
563
|
|
|
'No new articles for '.$groupMySQL['name'].' (first '.number_format($first). |
564
|
|
|
', last '.number_format($last).', grouplast '.number_format($groupMySQL['last_record']). |
565
|
|
|
', total '.number_format($total).")\n".'Server oldest: '.number_format($groupNNTP['first']). |
566
|
|
|
' Server newest: '.number_format($groupNNTP['last']).' Local newest: '.number_format($groupMySQL['last_record']) |
567
|
|
|
), |
568
|
|
|
true |
569
|
|
|
); |
570
|
|
|
} |
571
|
|
|
} |
572
|
|
|
|
573
|
|
|
/** |
574
|
|
|
* Loop over range of wanted headers, insert headers into DB. |
575
|
|
|
* |
576
|
|
|
* @param array $groupMySQL The group info from mysql. |
577
|
|
|
* @param int $first The oldest wanted header. |
578
|
|
|
* @param int $last The newest wanted header. |
579
|
|
|
* @param string $type Is this partrepair or update or backfill? |
580
|
|
|
* @param null|array $missingParts If we are running in partrepair, the list of missing article numbers. |
581
|
|
|
* |
582
|
|
|
* @return array Empty on failure. |
583
|
|
|
* @throws \Exception |
584
|
|
|
*/ |
585
|
|
|
public function scan($groupMySQL, $first, $last, $type = 'update', $missingParts = null): array |
586
|
|
|
{ |
587
|
|
|
// Start time of scan method and of fetching headers. |
588
|
|
|
$this->startLoop = microtime(true); |
589
|
|
|
$this->groupMySQL = $groupMySQL; |
590
|
|
|
$this->last = $last; |
591
|
|
|
$this->first = $first; |
592
|
|
|
|
593
|
|
|
$this->notYEnc = $this->headersBlackListed = 0; |
594
|
|
|
|
595
|
|
|
// Check if MySQL tables exist, create if they do not, get their names at the same time. |
596
|
|
|
$this->tableNames = (new Group())->getCBPTableNames($this->groupMySQL['id']); |
597
|
|
|
|
598
|
|
|
$mgrPosters = $this->getMultiGroupPosters(); |
599
|
|
|
|
600
|
|
|
if ($this->allAsMgr === true || ! empty($mgrPosters)) { |
601
|
|
|
$mgrActive = true; |
602
|
|
|
$mgrPosters = ! empty($mgrPosters) ? array_flip(array_column($mgrPosters, 'poster')) : ''; |
603
|
|
|
} else { |
604
|
|
|
$mgrActive = false; |
605
|
|
|
} |
606
|
|
|
|
607
|
|
|
$returnArray = $stdHeaders = $mgrHeaders = []; |
608
|
|
|
|
609
|
|
|
$partRepair = ($type === 'partrepair'); |
610
|
|
|
$this->addToPartRepair = ($type === 'update' && $this->_partRepair); |
611
|
|
|
|
612
|
|
|
// Download the headers. |
613
|
|
|
if ($partRepair === true) { |
614
|
|
|
// This is slower but possibly is better with missing headers. |
615
|
|
|
$headers = $this->_nntp->getOverview($this->first.'-'.$this->last, true, false); |
616
|
|
|
} else { |
617
|
|
|
$headers = $this->_nntp->getXOVER($this->first.'-'.$this->last); |
618
|
|
|
} |
619
|
|
|
|
620
|
|
|
// If there was an error, try to reconnect. |
621
|
|
|
if ($this->_nntp->isError($headers)) { |
622
|
|
|
|
623
|
|
|
// Increment if part repair and return false. |
624
|
|
|
if ($partRepair === true) { |
625
|
|
|
$this->_pdo->queryExec( |
626
|
|
|
sprintf( |
627
|
|
|
'UPDATE %s SET attempts = attempts + 1 WHERE groups_id = %d AND numberid %s', |
628
|
|
|
$this->tableNames['prname'], |
629
|
|
|
$this->groupMySQL['id'], |
630
|
|
|
((int) $this->first === (int) $this->last ? '= '.$this->first : 'IN ('.implode(',', range($this->first, $this->last)).')') |
631
|
|
|
) |
632
|
|
|
); |
633
|
|
|
|
634
|
|
|
return $returnArray; |
635
|
|
|
} |
636
|
|
|
|
637
|
|
|
// This is usually a compression error, so try disabling compression. |
638
|
|
|
$this->_nntp->doQuit(); |
639
|
|
|
if ($this->_nntp->doConnect(false) !== true) { |
640
|
|
|
return $returnArray; |
641
|
|
|
} |
642
|
|
|
|
643
|
|
|
// Re-select group, download headers again without compression and re-enable compression. |
644
|
|
|
$this->_nntp->selectGroup($this->groupMySQL['name']); |
645
|
|
|
$headers = $this->_nntp->getXOVER($this->first.'-'.$this->last); |
646
|
|
|
$this->_nntp->enableCompression(); |
647
|
|
|
|
648
|
|
|
// Check if the non-compression headers have an error. |
649
|
|
|
if ($this->_nntp->isError($headers)) { |
650
|
|
|
$message = ((int) $headers->code === 0 ? 'Unknown error' : $headers->message); |
651
|
|
|
$this->log( |
652
|
|
|
"Code {$headers->code}: $message\nSkipping group: {$this->groupMySQL['name']}", |
653
|
|
|
__FUNCTION__, |
654
|
|
|
'error' |
655
|
|
|
); |
656
|
|
|
|
657
|
|
|
return $returnArray; |
658
|
|
|
} |
659
|
|
|
} |
660
|
|
|
|
661
|
|
|
// Start of processing headers. |
662
|
|
|
$this->startCleaning = microtime(true); |
663
|
|
|
|
664
|
|
|
// End of the getting data from usenet. |
665
|
|
|
$this->timeHeaders = number_format($this->startCleaning - $this->startLoop, 2); |
666
|
|
|
|
667
|
|
|
// Check if we got headers. |
668
|
|
|
$msgCount = \count($headers); |
669
|
|
|
|
670
|
|
|
if ($msgCount < 1) { |
671
|
|
|
return $returnArray; |
672
|
|
|
} |
673
|
|
|
|
674
|
|
|
$this->getHighLowArticleInfo($returnArray, $headers, $msgCount); |
675
|
|
|
|
676
|
|
|
$headersRepaired = $rangeNotReceived = $this->headersReceived = $this->headersNotInserted = []; |
677
|
|
|
|
678
|
|
|
foreach ($headers as $header) { |
679
|
|
|
|
680
|
|
|
// Check if we got the article or not. |
681
|
|
|
if (isset($header['Number'])) { |
682
|
|
|
$this->headersReceived[] = $header['Number']; |
683
|
|
|
} else { |
684
|
|
|
if ($this->addToPartRepair) { |
685
|
|
|
$rangeNotReceived[] = $header['Number']; |
686
|
|
|
} |
687
|
|
|
continue; |
688
|
|
|
} |
689
|
|
|
|
690
|
|
|
// If set we are running in partRepair mode. |
691
|
|
|
if ($partRepair === true && $missingParts !== null) { |
692
|
|
|
if (! \in_array($header['Number'], $missingParts, false)) { |
693
|
|
|
// If article isn't one that is missing skip it. |
694
|
|
|
continue; |
695
|
|
|
} |
696
|
|
|
// We got the part this time. Remove article from part repair. |
697
|
|
|
$headersRepaired[] = $header['Number']; |
698
|
|
|
} |
699
|
|
|
|
700
|
|
|
/* |
701
|
|
|
* Find part / total parts. Ignore if no part count found. |
702
|
|
|
* |
703
|
|
|
* \s* Trims the leading space. |
704
|
|
|
* (?!"Usenet Index Post) ignores these types of articles, they are useless. |
705
|
|
|
* (.+) Fetches the subject. |
706
|
|
|
* \s+ Trims trailing space after the subject. |
707
|
|
|
* \((\d+)\/(\d+)\) Gets the part count. |
708
|
|
|
* No ending ($) as there are cases of subjects with extra data after the part count. |
709
|
|
|
*/ |
710
|
|
|
if (preg_match('/^\s*(?!"Usenet Index Post)(.+)\s+\((\d+)\/(\d+)\)/', $header['Subject'], $header['matches'])) { |
711
|
|
|
// Add yEnc to subjects that do not have them, but have the part number at the end of the header. |
712
|
|
|
if (stripos($header['Subject'], 'yEnc') === false) { |
713
|
|
|
$header['matches'][1] .= ' yEnc'; |
714
|
|
|
} |
715
|
|
|
} else { |
716
|
|
|
if ($this->_showDroppedYEncParts === true && strpos($header['Subject'], '"Usenet Index Post') !== 0) { |
717
|
|
|
file_put_contents( |
718
|
|
|
NN_LOGS.'not_yenc'.$this->groupMySQL['name'].'.dropped.log', |
|
|
|
|
719
|
|
|
$header['Subject'].PHP_EOL, |
720
|
|
|
FILE_APPEND |
721
|
|
|
); |
722
|
|
|
} |
723
|
|
|
$this->notYEnc++; |
724
|
|
|
continue; |
725
|
|
|
} |
726
|
|
|
|
727
|
|
|
// Filter subject based on black/white list. |
728
|
|
|
if ($this->isBlackListed($header, $this->groupMySQL['name'])) { |
729
|
|
|
$this->headersBlackListed++; |
730
|
|
|
continue; |
731
|
|
|
} |
732
|
|
|
|
733
|
|
|
if (! isset($header['Bytes'])) { |
734
|
|
|
$header['Bytes'] = (isset($this->header[':bytes']) ? $header[':bytes'] : 0); |
735
|
|
|
} |
736
|
|
|
$header['Bytes'] = (int) $header['Bytes']; |
737
|
|
|
|
738
|
|
|
if ($this->allAsMgr === true || ($mgrActive === true && array_key_exists($header['From'], $mgrPosters))) { |
739
|
|
|
$mgrHeaders[] = $header; |
740
|
|
|
} else { |
741
|
|
|
$stdHeaders[] = $header; |
742
|
|
|
} |
743
|
|
|
} |
744
|
|
|
|
745
|
|
|
unset($headers); // Reclaim memory now that headers are split. |
746
|
|
|
|
747
|
|
|
if (! empty($this->_binaryBlacklistIdsToUpdate)) { |
748
|
|
|
$this->updateBlacklistUsage(); |
749
|
|
|
} |
750
|
|
|
|
751
|
|
|
if ($this->_echoCLI && $partRepair === false) { |
752
|
|
|
$this->outputHeaderInitial(); |
753
|
|
|
} |
754
|
|
|
|
755
|
|
|
// MGR headers goes first |
756
|
|
|
if (! empty($mgrHeaders)) { |
757
|
|
|
$this->tableNames = ProcessReleasesMultiGroup::tableNames(); |
758
|
|
|
$this->storeHeaders($mgrHeaders, true); |
759
|
|
|
} |
760
|
|
|
unset($mgrHeaders); |
761
|
|
|
|
762
|
|
|
// Standard headers go second so we can switch tableNames back and do part repair to standard group tables |
763
|
|
|
if (! empty($stdHeaders)) { |
764
|
|
|
$this->tableNames = (new Group())->getCBPTableNames($this->groupMySQL['id']); |
765
|
|
|
$this->storeHeaders($stdHeaders, false); |
766
|
|
|
} |
767
|
|
|
unset($stdHeaders); |
768
|
|
|
|
769
|
|
|
// Start of part repair. |
770
|
|
|
$this->startPR = microtime(true); |
771
|
|
|
|
772
|
|
|
// End of inserting. |
773
|
|
|
$this->timeInsert = number_format($this->startPR - $this->startUpdate, 2); |
774
|
|
|
|
775
|
|
|
if ($partRepair && \count($headersRepaired) > 0) { |
776
|
|
|
$this->removeRepairedParts($headersRepaired, $this->tableNames['prname'], $this->groupMySQL['id']); |
777
|
|
|
} |
778
|
|
|
unset($headersRepaired); |
779
|
|
|
|
780
|
|
|
if ($this->addToPartRepair) { |
781
|
|
|
$notInsertedCount = \count($this->headersNotInserted); |
782
|
|
|
if ($notInsertedCount > 0) { |
783
|
|
|
$this->addMissingParts($this->headersNotInserted, $this->tableNames['prname'], $this->groupMySQL['id']); |
784
|
|
|
|
785
|
|
|
$this->log( |
786
|
|
|
$notInsertedCount.' articles failed to insert!', |
787
|
|
|
__FUNCTION__, |
788
|
|
|
'warning' |
789
|
|
|
); |
790
|
|
|
} |
791
|
|
|
unset($this->headersNotInserted); |
792
|
|
|
|
793
|
|
|
// Check if we have any missing headers. |
794
|
|
|
if (($this->last - $this->first - $this->notYEnc - $this->headersBlackListed + 1) > \count($this->headersReceived)) { |
795
|
|
|
$rangeNotReceived = array_merge($rangeNotReceived, array_diff(range($this->first, $this->last), $this->headersReceived)); |
796
|
|
|
} |
797
|
|
|
$notReceivedCount = \count($rangeNotReceived); |
798
|
|
|
if ($notReceivedCount > 0) { |
799
|
|
|
$this->addMissingParts($rangeNotReceived, $this->tableNames['prname'], $this->groupMySQL['id']); |
800
|
|
|
|
801
|
|
|
if ($this->_echoCLI) { |
802
|
|
|
ColorCLI::doEcho( |
803
|
|
|
ColorCLI::alternate( |
804
|
|
|
'Server did not return '.$notReceivedCount. |
805
|
|
|
' articles from '.$this->groupMySQL['name'].'.' |
806
|
|
|
), |
807
|
|
|
true |
808
|
|
|
); |
809
|
|
|
} |
810
|
|
|
} |
811
|
|
|
unset($rangeNotReceived); |
812
|
|
|
} |
813
|
|
|
|
814
|
|
|
$this->outputHeaderDuration(); |
815
|
|
|
|
816
|
|
|
return $returnArray; |
817
|
|
|
} |
818
|
|
|
|
819
|
|
|
/** |
820
|
|
|
* Parse headers into collections/binaries and store header data as parts. |
821
|
|
|
* |
822
|
|
|
* @param array $headers The retrieved headers |
823
|
|
|
* @param bool $multiGroup Is this task being run in MGR mode? |
824
|
|
|
* |
825
|
|
|
* @throws \Exception |
826
|
|
|
*/ |
827
|
|
|
protected function storeHeaders(array $headers, $multiGroup): void |
828
|
|
|
{ |
829
|
|
|
$this->multiGroup = $multiGroup; |
830
|
|
|
$binariesUpdate = $collectionIDs = $articles = []; |
831
|
|
|
|
832
|
|
|
$this->_pdo->beginTransaction(); |
833
|
|
|
|
834
|
|
|
$partsQuery = $partsCheck = |
835
|
|
|
"INSERT IGNORE INTO {$this->tableNames['pname']} (binaries_id, number, messageid, partnumber, size) VALUES "; |
836
|
|
|
|
837
|
|
|
// Loop articles, figure out files/parts. |
838
|
|
|
foreach ($headers as $this->header) { |
839
|
|
|
// Set up the info for inserting into parts/binaries/collections tables. |
840
|
|
|
if (! isset($articles[$this->header['matches'][1]])) { |
841
|
|
|
|
842
|
|
|
// check whether file count should be ignored (XXX packs for now only). |
843
|
|
|
$whitelistMatch = false; |
844
|
|
|
if ($this->_ignoreFileCount($this->groupMySQL['name'], $this->header['matches'][1])) { |
845
|
|
|
$whitelistMatch = true; |
846
|
|
|
$fileCount[1] = $fileCount[3] = 0; |
847
|
|
|
} |
848
|
|
|
|
849
|
|
|
// Attempt to find the file count. If it is not found, set it to 0. |
850
|
|
|
if (! $whitelistMatch && ! preg_match('/[[(\s](\d{1,5})(\/|[\s_]of[\s_]|-)(\d{1,5})[])\s$:]/i', $this->header['matches'][1], $fileCount)) { |
851
|
|
|
$fileCount[1] = $fileCount[3] = 0; |
852
|
|
|
if ($this->_showDroppedYEncParts === true) { |
853
|
|
|
file_put_contents( |
854
|
|
|
NN_LOGS.'no_files'.$this->groupMySQL['name'].'.log', |
|
|
|
|
855
|
|
|
$this->header['Subject'].PHP_EOL, |
856
|
|
|
FILE_APPEND |
857
|
|
|
); |
858
|
|
|
} |
859
|
|
|
} |
860
|
|
|
|
861
|
|
|
if ($this->multiGroup) { |
862
|
|
|
$ckName = ''; |
863
|
|
|
$ckId = ''; |
864
|
|
|
} else { |
865
|
|
|
$ckName = $this->groupMySQL['name']; |
866
|
|
|
$ckId = $this->groupMySQL['id']; |
867
|
|
|
} |
868
|
|
|
|
869
|
|
|
$collMatch = $this->_collectionsCleaning->collectionsCleaner( |
870
|
|
|
$this->header['matches'][1], |
871
|
|
|
$ckName |
872
|
|
|
); |
873
|
|
|
|
874
|
|
|
// Used to group articles together when forming the release. MGR requires this to be group irrespective |
875
|
|
|
$this->header['CollectionKey'] = $collMatch['name'].$ckId.$fileCount[3]; |
|
|
|
|
876
|
|
|
|
877
|
|
|
// If this header's collection key isn't in memory, attempt to insert the collection |
878
|
|
|
if (! isset($collectionIDs[$this->header['CollectionKey']])) { |
879
|
|
|
|
880
|
|
|
/* Date from header should be a string this format: |
881
|
|
|
* 31 Mar 2014 15:36:04 GMT or 6 Oct 1998 04:38:40 -0500 |
882
|
|
|
* Still make sure it's not unix time, convert it to unix time if it is. |
883
|
|
|
*/ |
884
|
|
|
$this->header['Date'] = (is_numeric($this->header['Date']) ? $this->header['Date'] : strtotime($this->header['Date'])); |
885
|
|
|
|
886
|
|
|
// Get the current unixtime from PHP. |
887
|
|
|
$now = Carbon::now()->timestamp; |
888
|
|
|
|
889
|
|
|
$xref = ($this->multiGroup === true ? sprintf('xref = CONCAT(xref, "\\n"%s ),', $this->_pdo->escapeString(substr($this->header['Xref'], 2, 255))) : ''); |
890
|
|
|
$date = $this->header['Date'] > $now ? $now : $this->header['Date']; |
891
|
|
|
$unixtime = is_numeric($this->header['Date']) ? $date : $now; |
892
|
|
|
|
893
|
|
|
$random = random_bytes(16); |
894
|
|
|
|
895
|
|
|
$collectionID = $this->_pdo->queryInsert( |
896
|
|
|
sprintf( |
897
|
|
|
" |
898
|
|
|
INSERT INTO %s (subject, fromname, date, xref, groups_id, |
899
|
|
|
totalfiles, collectionhash, collection_regexes_id, dateadded) |
900
|
|
|
VALUES (%s, %s, FROM_UNIXTIME(%s), %s, %d, %d, '%s', %d, NOW()) |
901
|
|
|
ON DUPLICATE KEY UPDATE %s dateadded = NOW(), noise = '%s'", |
902
|
|
|
$this->tableNames['cname'], |
903
|
|
|
$this->_pdo->escapeString(substr(utf8_encode($this->header['matches'][1]), 0, 255)), |
904
|
|
|
$this->_pdo->escapeString(utf8_encode($this->header['From'])), |
905
|
|
|
$unixtime, |
906
|
|
|
$this->_pdo->escapeString(substr($this->header['Xref'], 0, 255)), |
907
|
|
|
$this->groupMySQL['id'], |
908
|
|
|
$fileCount[3], |
909
|
|
|
sha1($this->header['CollectionKey']), |
910
|
|
|
$collMatch['id'], |
911
|
|
|
$xref, |
912
|
|
|
sodium_bin2hex($random) |
913
|
|
|
) |
914
|
|
|
); |
915
|
|
|
|
916
|
|
|
if ($collectionID === false) { |
917
|
|
|
if ($this->addToPartRepair) { |
918
|
|
|
$this->headersNotInserted[] = $this->header['Number']; |
919
|
|
|
} |
920
|
|
|
$this->_pdo->Rollback(); |
921
|
|
|
$this->_pdo->beginTransaction(); |
922
|
|
|
continue; |
923
|
|
|
} |
924
|
|
|
$collectionIDs[$this->header['CollectionKey']] = $collectionID; |
925
|
|
|
} else { |
926
|
|
|
$collectionID = $collectionIDs[$this->header['CollectionKey']]; |
927
|
|
|
} |
928
|
|
|
|
929
|
|
|
// MGR or Standard, Binary Hash should be unique to the group |
930
|
|
|
$hash = md5($this->header['matches'][1].$this->header['From'].$this->groupMySQL['id']); |
931
|
|
|
|
932
|
|
|
$binaryID = $this->_pdo->queryInsert( |
933
|
|
|
sprintf( |
934
|
|
|
" |
935
|
|
|
INSERT INTO %s (binaryhash, name, collections_id, totalparts, currentparts, filenumber, partsize) |
936
|
|
|
VALUES (UNHEX('%s'), %s, %d, %d, 1, %d, %d) |
937
|
|
|
ON DUPLICATE KEY UPDATE currentparts = currentparts + 1, partsize = partsize + %d", |
938
|
|
|
$this->tableNames['bname'], |
939
|
|
|
$hash, |
940
|
|
|
$this->_pdo->escapeString(utf8_encode($this->header['matches'][1])), |
941
|
|
|
$collectionID, |
942
|
|
|
$this->header['matches'][3], |
943
|
|
|
$fileCount[1], |
944
|
|
|
$this->header['Bytes'], |
945
|
|
|
$this->header['Bytes'] |
946
|
|
|
) |
947
|
|
|
); |
948
|
|
|
|
949
|
|
|
if ($binaryID === false) { |
950
|
|
|
if ($this->addToPartRepair) { |
951
|
|
|
$this->headersNotInserted[] = $this->header['Number']; |
952
|
|
|
} |
953
|
|
|
$this->_pdo->Rollback(); |
954
|
|
|
$this->_pdo->beginTransaction(); |
955
|
|
|
continue; |
956
|
|
|
} |
957
|
|
|
|
958
|
|
|
$binariesUpdate[$binaryID]['Size'] = 0; |
959
|
|
|
$binariesUpdate[$binaryID]['Parts'] = 0; |
960
|
|
|
|
961
|
|
|
$articles[$this->header['matches'][1]]['CollectionID'] = $collectionID; |
962
|
|
|
$articles[$this->header['matches'][1]]['BinaryID'] = $binaryID; |
963
|
|
|
} else { |
964
|
|
|
$binaryID = $articles[$this->header['matches'][1]]['BinaryID']; |
965
|
|
|
$binariesUpdate[$binaryID]['Size'] += $this->header['Bytes']; |
966
|
|
|
$binariesUpdate[$binaryID]['Parts']++; |
967
|
|
|
} |
968
|
|
|
|
969
|
|
|
// Strip the < and >, saves space in DB. |
970
|
|
|
$this->header['Message-ID'][0] = "'"; |
971
|
|
|
|
972
|
|
|
$partsQuery .= |
973
|
|
|
'('.$binaryID.','.$this->header['Number'].','.rtrim($this->header['Message-ID'], '>')."',". |
974
|
|
|
$this->header['matches'][2].','.$this->header['Bytes'].'),'; |
975
|
|
|
} |
976
|
|
|
|
977
|
|
|
unset($headers); // Reclaim memory. |
978
|
|
|
|
979
|
|
|
// Start of inserting into SQL. |
980
|
|
|
$this->startUpdate = microtime(true); |
981
|
|
|
|
982
|
|
|
// End of processing headers. |
983
|
|
|
$this->timeCleaning = number_format($this->startUpdate - $this->startCleaning, 2); |
984
|
|
|
$binariesQuery = $binariesCheck = sprintf('INSERT INTO %s (id, partsize, currentparts) VALUES ', $this->tableNames['bname']); |
985
|
|
|
foreach ($binariesUpdate as $binaryID => $binary) { |
986
|
|
|
$binariesQuery .= '('.$binaryID.','.$binary['Size'].','.$binary['Parts'].'),'; |
987
|
|
|
} |
988
|
|
|
$binariesEnd = ' ON DUPLICATE KEY UPDATE partsize = VALUES(partsize) + partsize, currentparts = VALUES(currentparts) + currentparts'; |
989
|
|
|
$binariesQuery = rtrim($binariesQuery, ',').$binariesEnd; |
990
|
|
|
|
991
|
|
|
// Check if we got any binaries. If we did, try to insert them. |
992
|
|
|
if (\strlen($binariesCheck.$binariesEnd) === \strlen($binariesQuery) ? true : $this->_pdo->queryExec($binariesQuery)) { |
993
|
|
|
if ($this->_debug) { |
994
|
|
|
ColorCLI::doEcho( |
995
|
|
|
ColorCLI::debug( |
996
|
|
|
'Sending '.round(\strlen($partsQuery) / 1024, 2). |
997
|
|
|
' KB of'.($this->multiGroup ? ' MGR' : '').' parts to MySQL' |
998
|
|
|
), true |
999
|
|
|
); |
1000
|
|
|
} |
1001
|
|
|
if (\strlen($partsQuery) === \strlen($partsCheck) ? true : $this->_pdo->queryExec(rtrim($partsQuery, ','))) { |
1002
|
|
|
$this->_pdo->Commit(); |
1003
|
|
|
} else { |
1004
|
|
|
if ($this->addToPartRepair) { |
1005
|
|
|
$this->headersNotInserted += $this->headersReceived; |
1006
|
|
|
} |
1007
|
|
|
$this->_pdo->Rollback(); |
1008
|
|
|
} |
1009
|
|
|
} else { |
1010
|
|
|
if ($this->addToPartRepair) { |
1011
|
|
|
$this->headersNotInserted += $this->headersReceived; |
1012
|
|
|
} |
1013
|
|
|
$this->_pdo->Rollback(); |
1014
|
|
|
} |
1015
|
|
|
} |
1016
|
|
|
|
1017
|
|
|
/** |
1018
|
|
|
* Gets the First and Last Article Number and Date for the received headers. |
1019
|
|
|
* |
1020
|
|
|
* @param array $returnArray |
1021
|
|
|
* @param array $headers |
1022
|
|
|
* @param int $msgCount |
1023
|
|
|
*/ |
1024
|
|
|
protected function getHighLowArticleInfo(array &$returnArray, array $headers, int $msgCount): void |
1025
|
|
|
{ |
1026
|
|
|
// Get highest and lowest article numbers/dates. |
1027
|
|
|
$iterator1 = 0; |
1028
|
|
|
$iterator2 = $msgCount - 1; |
1029
|
|
|
while (true) { |
1030
|
|
|
if (! isset($returnArray['firstArticleNumber']) && isset($headers[$iterator1]['Number'])) { |
1031
|
|
|
$returnArray['firstArticleNumber'] = $headers[$iterator1]['Number']; |
1032
|
|
|
$returnArray['firstArticleDate'] = $headers[$iterator1]['Date']; |
1033
|
|
|
} |
1034
|
|
|
|
1035
|
|
|
if (! isset($returnArray['lastArticleNumber']) && isset($headers[$iterator2]['Number'])) { |
1036
|
|
|
$returnArray['lastArticleNumber'] = $headers[$iterator2]['Number']; |
1037
|
|
|
$returnArray['lastArticleDate'] = $headers[$iterator2]['Date']; |
1038
|
|
|
} |
1039
|
|
|
|
1040
|
|
|
// Break if we found non empty articles. |
1041
|
|
|
if (isset($returnArray['firstArticleNumber, lastArticleNumber'])) { |
1042
|
|
|
break; |
1043
|
|
|
} |
1044
|
|
|
|
1045
|
|
|
// Break out if we couldn't find anything. |
1046
|
|
|
if ($iterator1++ >= $msgCount - 1 || $iterator2-- <= 0) { |
1047
|
|
|
break; |
1048
|
|
|
} |
1049
|
|
|
} |
1050
|
|
|
} |
1051
|
|
|
|
1052
|
|
|
/** |
1053
|
|
|
* Updates Blacklist Regex Timers in DB to reflect last usage. |
1054
|
|
|
*/ |
1055
|
|
|
protected function updateBlacklistUsage(): void |
1056
|
|
|
{ |
1057
|
|
|
BinaryBlacklist::query()->whereIn('id', $this->_binaryBlacklistIdsToUpdate)->update(['last_activity' => Carbon::now()]); |
1058
|
|
|
$this->_binaryBlacklistIdsToUpdate = []; |
1059
|
|
|
} |
1060
|
|
|
|
1061
|
|
|
/** |
1062
|
|
|
* Outputs the initial header scan results after yEnc check and blacklist routines. |
1063
|
|
|
*/ |
1064
|
|
|
protected function outputHeaderInitial(): void |
1065
|
|
|
{ |
1066
|
|
|
ColorCLI::doEcho( |
1067
|
|
|
ColorCLI::primary( |
1068
|
|
|
'Received '.\count($this->headersReceived). |
1069
|
|
|
' articles of '.number_format($this->last - $this->first + 1).' requested, '. |
1070
|
|
|
$this->headersBlackListed.' blacklisted, '.$this->notYEnc.' not yEnc.' |
1071
|
|
|
), true |
1072
|
|
|
); |
1073
|
|
|
} |
1074
|
|
|
|
1075
|
|
|
/** |
1076
|
|
|
* Outputs speed metrics of the scan function to CLI. |
1077
|
|
|
*/ |
1078
|
|
|
protected function outputHeaderDuration(): void |
1079
|
|
|
{ |
1080
|
|
|
$currentMicroTime = microtime(true); |
1081
|
|
|
if ($this->_echoCLI) { |
1082
|
|
|
ColorCLI::doEcho( |
1083
|
|
|
ColorCLI::alternateOver($this->timeHeaders.'s'). |
1084
|
|
|
ColorCLI::primaryOver(' to download articles, '). |
1085
|
|
|
ColorCLI::alternateOver($this->timeCleaning.'s'). |
1086
|
|
|
ColorCLI::primaryOver(' to process collections, '). |
1087
|
|
|
ColorCLI::alternateOver($this->timeInsert.'s'). |
1088
|
|
|
ColorCLI::primaryOver(' to insert binaries/parts, '). |
1089
|
|
|
ColorCLI::alternateOver(number_format($currentMicroTime - $this->startPR, 2).'s'). |
1090
|
|
|
ColorCLI::primaryOver(' for part repair, '). |
1091
|
|
|
ColorCLI::alternateOver(number_format($currentMicroTime - $this->startLoop, 2).'s'). |
1092
|
|
|
ColorCLI::primary(' total.'), true); |
1093
|
|
|
} |
1094
|
|
|
} |
1095
|
|
|
|
1096
|
|
|
/** |
1097
|
|
|
* If we failed to insert Collections/Binaries/Parts, rollback the transaction and add the parts to part repair. |
1098
|
|
|
* |
1099
|
|
|
* @param array $headers Array of headers containing sub-arrays with parts. |
1100
|
|
|
* |
1101
|
|
|
* @return array Array of article numbers to add to part repair. |
1102
|
|
|
*/ |
1103
|
|
|
protected function _rollbackAddToPartRepair(array $headers): array |
1104
|
|
|
{ |
1105
|
|
|
$headersNotInserted = []; |
1106
|
|
|
foreach ($headers as $header) { |
1107
|
|
|
foreach ($header as $file) { |
1108
|
|
|
$headersNotInserted[] = $file['Parts']['number']; |
1109
|
|
|
} |
1110
|
|
|
} |
1111
|
|
|
$this->_pdo->Rollback(); |
1112
|
|
|
|
1113
|
|
|
return $headersNotInserted; |
1114
|
|
|
} |
1115
|
|
|
|
1116
|
|
|
/** |
1117
|
|
|
* Attempt to get missing article headers. |
1118
|
|
|
* |
1119
|
|
|
* @param array|string $tables |
1120
|
|
|
* @param array $groupArr The info for this group from mysql. |
1121
|
|
|
* |
1122
|
|
|
* @return void |
1123
|
|
|
* @throws \Exception |
1124
|
|
|
*/ |
1125
|
|
|
public function partRepair($groupArr, $tables = ''): void |
1126
|
|
|
{ |
1127
|
|
|
$tableNames = $tables; |
1128
|
|
|
|
1129
|
|
|
if ($tableNames === '') { |
1130
|
|
|
$tableNames = (new Group())->getCBPTableNames($groupArr['id']); |
1131
|
|
|
} |
1132
|
|
|
// Get all parts in partrepair table. |
1133
|
|
|
$missingParts = $this->_pdo->query( |
1134
|
|
|
sprintf( |
1135
|
|
|
' |
1136
|
|
|
SELECT * FROM %s |
1137
|
|
|
WHERE groups_id = %d AND attempts < %d |
1138
|
|
|
ORDER BY numberid ASC LIMIT %d', |
1139
|
|
|
$tableNames['prname'], |
1140
|
|
|
$groupArr['id'], |
1141
|
|
|
$this->_partRepairMaxTries, |
1142
|
|
|
$this->_partRepairLimit |
1143
|
|
|
) |
1144
|
|
|
); |
1145
|
|
|
|
1146
|
|
|
$missingCount = \count($missingParts); |
1147
|
|
|
if ($missingCount > 0) { |
1148
|
|
|
if ($this->_echoCLI) { |
1149
|
|
|
ColorCLI::doEcho( |
1150
|
|
|
ColorCLI::primary( |
1151
|
|
|
'Attempting to repair '. |
1152
|
|
|
number_format($missingCount). |
1153
|
|
|
' parts.' |
1154
|
|
|
), |
1155
|
|
|
true |
1156
|
|
|
); |
1157
|
|
|
} |
1158
|
|
|
|
1159
|
|
|
// Loop through each part to group into continuous ranges with a maximum range of messagebuffer/4. |
1160
|
|
|
$ranges = $partList = []; |
1161
|
|
|
$firstPart = $lastNum = $missingParts[0]['numberid']; |
1162
|
|
|
|
1163
|
|
|
foreach ($missingParts as $part) { |
1164
|
|
|
if (($part['numberid'] - $firstPart) > ($this->messageBuffer / 4)) { |
1165
|
|
|
$ranges[] = [ |
1166
|
|
|
'partfrom' => $firstPart, |
1167
|
|
|
'partto' => $lastNum, |
1168
|
|
|
'partlist' => $partList, |
1169
|
|
|
]; |
1170
|
|
|
|
1171
|
|
|
$firstPart = $part['numberid']; |
1172
|
|
|
$partList = []; |
1173
|
|
|
} |
1174
|
|
|
$partList[] = $part['numberid']; |
1175
|
|
|
$lastNum = $part['numberid']; |
1176
|
|
|
} |
1177
|
|
|
|
1178
|
|
|
$ranges[] = [ |
1179
|
|
|
'partfrom' => $firstPart, |
1180
|
|
|
'partto' => $lastNum, |
1181
|
|
|
'partlist' => $partList, |
1182
|
|
|
]; |
1183
|
|
|
|
1184
|
|
|
// Download missing parts in ranges. |
1185
|
|
|
foreach ($ranges as $range) { |
1186
|
|
|
$partFrom = $range['partfrom']; |
1187
|
|
|
$partTo = $range['partto']; |
1188
|
|
|
$partList = $range['partlist']; |
1189
|
|
|
|
1190
|
|
|
if ($this->_echoCLI) { |
1191
|
|
|
echo \chr(random_int(45, 46)).PHP_EOL; |
1192
|
|
|
} |
1193
|
|
|
|
1194
|
|
|
// Get article headers from newsgroup. |
1195
|
|
|
$this->scan($groupArr, $partFrom, $partTo, 'missed_parts', $partList); |
1196
|
|
|
} |
1197
|
|
|
|
1198
|
|
|
// Calculate parts repaired |
1199
|
|
|
$result = $this->_pdo->queryOneRow( |
1200
|
|
|
sprintf( |
1201
|
|
|
' |
1202
|
|
|
SELECT COUNT(id) AS num |
1203
|
|
|
FROM %s |
1204
|
|
|
WHERE groups_id = %d |
1205
|
|
|
AND numberid <= %d', |
1206
|
|
|
$tableNames['prname'], |
1207
|
|
|
$groupArr['id'], |
1208
|
|
|
$missingParts[$missingCount - 1]['numberid'] |
1209
|
|
|
) |
1210
|
|
|
); |
1211
|
|
|
|
1212
|
|
|
$partsRepaired = 0; |
1213
|
|
|
if ($result !== false) { |
1214
|
|
|
$partsRepaired = ($missingCount - $result['num']); |
1215
|
|
|
} |
1216
|
|
|
|
1217
|
|
|
// Update attempts on remaining parts for active group |
1218
|
|
|
if (isset($missingParts[$missingCount - 1]['id'])) { |
1219
|
|
|
$this->_pdo->queryExec( |
1220
|
|
|
sprintf( |
1221
|
|
|
' |
1222
|
|
|
UPDATE %s |
1223
|
|
|
SET attempts = attempts + 1 |
1224
|
|
|
WHERE groups_id = %d |
1225
|
|
|
AND numberid <= %d', |
1226
|
|
|
$tableNames['prname'], |
1227
|
|
|
$groupArr['id'], |
1228
|
|
|
$missingParts[$missingCount - 1]['numberid'] |
1229
|
|
|
) |
1230
|
|
|
); |
1231
|
|
|
} |
1232
|
|
|
|
1233
|
|
|
if ($this->_echoCLI) { |
1234
|
|
|
ColorCLI::doEcho( |
1235
|
|
|
ColorCLI::primary( |
1236
|
|
|
PHP_EOL. |
1237
|
|
|
number_format($partsRepaired). |
1238
|
|
|
' parts repaired.' |
1239
|
|
|
), |
1240
|
|
|
true |
1241
|
|
|
); |
1242
|
|
|
} |
1243
|
|
|
} |
1244
|
|
|
|
1245
|
|
|
// Remove articles that we cant fetch after x attempts. |
1246
|
|
|
$this->_pdo->queryExec( |
1247
|
|
|
sprintf( |
1248
|
|
|
'DELETE FROM %s WHERE attempts >= %d AND groups_id = %d', |
1249
|
|
|
$tableNames['prname'], |
1250
|
|
|
$this->_partRepairMaxTries, |
1251
|
|
|
$groupArr['id'] |
1252
|
|
|
) |
1253
|
|
|
); |
1254
|
|
|
} |
1255
|
|
|
|
1256
|
|
|
/** |
1257
|
|
|
* Returns unix time for an article number. |
1258
|
|
|
* |
1259
|
|
|
* @param int $post The article number to get the time from. |
1260
|
|
|
* @param array $groupData Usenet group info from NNTP selectGroup method. |
1261
|
|
|
* |
1262
|
|
|
* @return int Timestamp. |
1263
|
|
|
* @throws \Exception |
1264
|
|
|
*/ |
1265
|
|
|
public function postdate($post, array $groupData): int |
1266
|
|
|
{ |
1267
|
|
|
// Set table names |
1268
|
|
|
$groupID = Group::getIDByName($groupData['group']); |
1269
|
|
|
$group = []; |
1270
|
|
|
if ($groupID !== '') { |
1271
|
|
|
$group = (new Group())->getCBPTableNames($groupID); |
|
|
|
|
1272
|
|
|
} |
1273
|
|
|
|
1274
|
|
|
$currentPost = $post; |
1275
|
|
|
|
1276
|
|
|
$attempts = $date = 0; |
1277
|
|
|
do { |
1278
|
|
|
// Try to get the article date locally first. |
1279
|
|
|
if ($groupID !== '') { |
1280
|
|
|
// Try to get locally. |
1281
|
|
|
$local = $this->_pdo->queryOneRow( |
1282
|
|
|
sprintf( |
1283
|
|
|
' |
1284
|
|
|
SELECT c.date AS date |
1285
|
|
|
FROM %s c |
1286
|
|
|
INNER JOIN %s b ON(c.id=b.collections_id) |
1287
|
|
|
INNER JOIN %s p ON(b.id=p.binaries_id) |
1288
|
|
|
WHERE p.number = %s', |
1289
|
|
|
$group['cname'], |
1290
|
|
|
$group['bname'], |
1291
|
|
|
$group['pname'], |
1292
|
|
|
$currentPost |
1293
|
|
|
) |
1294
|
|
|
); |
1295
|
|
|
if ($local !== false) { |
1296
|
|
|
$date = $local['date']; |
1297
|
|
|
break; |
1298
|
|
|
} |
1299
|
|
|
} |
1300
|
|
|
|
1301
|
|
|
// If we could not find it locally, try usenet. |
1302
|
|
|
$header = $this->_nntp->getXOVER($currentPost); |
1303
|
|
|
if (! $this->_nntp->isError($header)) { |
1304
|
|
|
// Check if the date is set. |
1305
|
|
|
if (isset($header[0]['Date']) && \strlen($header[0]['Date']) > 0) { |
1306
|
|
|
$date = $header[0]['Date']; |
1307
|
|
|
break; |
1308
|
|
|
} |
1309
|
|
|
} |
1310
|
|
|
|
1311
|
|
|
// Try to get a different article number. |
1312
|
|
|
if (abs($currentPost - $groupData['first']) > abs($groupData['last'] - $currentPost)) { |
1313
|
|
|
$tempPost = round($currentPost / (random_int(1005, 1012) / 1000), 0, PHP_ROUND_HALF_UP); |
1314
|
|
|
if ($tempPost < $groupData['first']) { |
1315
|
|
|
$tempPost = $groupData['first']; |
1316
|
|
|
} |
1317
|
|
|
} else { |
1318
|
|
|
$tempPost = round((random_int(1005, 1012) / 1000) * $currentPost, 0, PHP_ROUND_HALF_UP); |
1319
|
|
|
if ($tempPost > $groupData['last']) { |
1320
|
|
|
$tempPost = $groupData['last']; |
1321
|
|
|
} |
1322
|
|
|
} |
1323
|
|
|
// If we got the same article number as last time, give up. |
1324
|
|
|
if ($tempPost === $currentPost) { |
1325
|
|
|
break; |
1326
|
|
|
} |
1327
|
|
|
$currentPost = $tempPost; |
1328
|
|
|
|
1329
|
|
|
if ($this->_debug) { |
1330
|
|
|
ColorCLI::doEcho(ColorCLI::debug('Postdate retried '.$attempts.' time(s).'), true); |
1331
|
|
|
} |
1332
|
|
|
} while ($attempts++ <= 20); |
1333
|
|
|
|
1334
|
|
|
// If we didn't get a date, set it to now. |
1335
|
|
|
if (! $date) { |
1336
|
|
|
$date = time(); |
1337
|
|
|
} else { |
1338
|
|
|
$date = strtotime($date); |
1339
|
|
|
} |
1340
|
|
|
|
1341
|
|
|
return $date; |
1342
|
|
|
} |
1343
|
|
|
|
1344
|
|
|
/** |
1345
|
|
|
* Returns article number based on # of days. |
1346
|
|
|
* |
1347
|
|
|
* @param int $days How many days back we want to go. |
1348
|
|
|
* @param array $data Group data from usenet. |
1349
|
|
|
* |
1350
|
|
|
* @return string |
1351
|
|
|
* @throws \Exception |
1352
|
|
|
*/ |
1353
|
|
|
public function daytopost($days, $data): string |
1354
|
|
|
{ |
1355
|
|
|
$goalTime = Carbon::now()->subDays($days)->timestamp; |
1356
|
|
|
// The time we want = current unix time (ex. 1395699114) - minus 86400 (seconds in a day) |
1357
|
|
|
// times days wanted. (ie 1395699114 - 2592000 (30days)) = 1393107114 |
1358
|
|
|
|
1359
|
|
|
// The servers oldest date. |
1360
|
|
|
$firstDate = $this->postdate($data['first'], $data); |
1361
|
|
|
if ($goalTime < $firstDate) { |
1362
|
|
|
// If the date we want is older than the oldest date in the group return the groups oldest article. |
1363
|
|
|
return $data['first']; |
1364
|
|
|
} |
1365
|
|
|
|
1366
|
|
|
// The servers newest date. |
1367
|
|
|
$lastDate = $this->postdate($data['last'], $data); |
1368
|
|
|
if ($goalTime > $lastDate) { |
1369
|
|
|
// If the date we want is newer than the groups newest date, return the groups newest article. |
1370
|
|
|
return $data['last']; |
1371
|
|
|
} |
1372
|
|
|
|
1373
|
|
|
if ($this->_echoCLI) { |
1374
|
|
|
ColorCLI::doEcho( |
1375
|
|
|
ColorCLI::primary( |
1376
|
|
|
'Searching for an approximate article number for group '.$data['group'].' '.$days.' days back.' |
1377
|
|
|
), true |
1378
|
|
|
); |
1379
|
|
|
} |
1380
|
|
|
|
1381
|
|
|
// Pick the middle to start with |
1382
|
|
|
$wantedArticle = round(($data['last'] + $data['first']) / 2); |
1383
|
|
|
$aMax = $data['last']; |
1384
|
|
|
$aMin = $data['first']; |
1385
|
|
|
$reallyOldArticle = $oldArticle = $articleTime = null; |
|
|
|
|
1386
|
|
|
|
1387
|
|
|
while (true) { |
1388
|
|
|
// Article exists outside of available range, this shouldn't happen |
1389
|
|
|
if ($wantedArticle <= $data['first'] || $wantedArticle >= $data['last']) { |
1390
|
|
|
break; |
1391
|
|
|
} |
1392
|
|
|
|
1393
|
|
|
// Keep a note of the last articles we checked |
1394
|
|
|
$reallyOldArticle = $oldArticle; |
1395
|
|
|
$oldArticle = $wantedArticle; |
1396
|
|
|
|
1397
|
|
|
// Get the date of this article |
1398
|
|
|
$articleTime = $this->postdate($wantedArticle, $data); |
|
|
|
|
1399
|
|
|
|
1400
|
|
|
// Article doesn't exist, start again with something random |
1401
|
|
|
if (! $articleTime) { |
1402
|
|
|
$wantedArticle = random_int($aMin, $aMax); |
1403
|
|
|
$articleTime = $this->postdate($wantedArticle, $data); |
1404
|
|
|
} |
1405
|
|
|
|
1406
|
|
|
if ($articleTime < $goalTime) { |
1407
|
|
|
// Article is older than we want |
1408
|
|
|
$aMin = $oldArticle; |
1409
|
|
|
$wantedArticle = round(($aMax + $oldArticle) / 2); |
1410
|
|
|
if ($this->_echoCLI) { |
1411
|
|
|
echo '-'; |
1412
|
|
|
} |
1413
|
|
|
} elseif ($articleTime > $goalTime) { |
1414
|
|
|
// Article is newer than we want |
1415
|
|
|
$aMax = $oldArticle; |
1416
|
|
|
$wantedArticle = round(($aMin + $oldArticle) / 2); |
1417
|
|
|
if ($this->_echoCLI) { |
1418
|
|
|
echo '+'; |
1419
|
|
|
} |
1420
|
|
|
} elseif ($articleTime === $goalTime) { |
1421
|
|
|
// Exact match. We did it! (this will likely never happen though) |
1422
|
|
|
break; |
1423
|
|
|
} |
1424
|
|
|
|
1425
|
|
|
// We seem to be flip-flopping between 2 articles, assume we're out of articles to check. |
1426
|
|
|
// End on an article more recent than our oldest so that we don't miss any releases. |
1427
|
|
|
if ($reallyOldArticle === $wantedArticle && ($goalTime - $articleTime) <= 0) { |
1428
|
|
|
break; |
1429
|
|
|
} |
1430
|
|
|
} |
1431
|
|
|
|
1432
|
|
|
$wantedArticle = (int) $wantedArticle; |
1433
|
|
|
if ($this->_echoCLI) { |
1434
|
|
|
ColorCLI::doEcho( |
1435
|
|
|
ColorCLI::primary( |
1436
|
|
|
PHP_EOL.'Found article #'.$wantedArticle.' which has a date of '.date('r', $articleTime). |
1437
|
|
|
', vs wanted date of '.date('r', $goalTime).'. Difference from goal is '.Carbon::createFromTimestamp($goalTime)->diffInDays(Carbon::createFromTimestamp($articleTime)).'days.' |
1438
|
|
|
), true |
1439
|
|
|
); |
1440
|
|
|
} |
1441
|
|
|
|
1442
|
|
|
return $wantedArticle; |
1443
|
|
|
} |
1444
|
|
|
|
1445
|
|
|
/** |
1446
|
|
|
* Convert unix time to days ago. |
1447
|
|
|
* |
1448
|
|
|
* |
1449
|
|
|
* @param $timestamp |
1450
|
|
|
* @return int |
1451
|
|
|
*/ |
1452
|
|
|
private function daysOld($timestamp): int |
1453
|
|
|
{ |
1454
|
|
|
return Carbon::createFromTimestamp($timestamp)->diffInDays(); |
1455
|
|
|
} |
1456
|
|
|
|
1457
|
|
|
/** |
1458
|
|
|
* Add article numbers from missing headers to DB. |
1459
|
|
|
* |
1460
|
|
|
* @param array $numbers The article numbers of the missing headers. |
1461
|
|
|
* @param string $tableName Name of the partrepair table to insert into. |
1462
|
|
|
* @param int $groupID The ID of this groups. |
1463
|
|
|
* |
1464
|
|
|
* @return bool |
1465
|
|
|
*/ |
1466
|
|
|
private function addMissingParts($numbers, $tableName, $groupID): bool |
1467
|
|
|
{ |
1468
|
|
|
$insertStr = 'INSERT INTO '.$tableName.' (numberid, groups_id) VALUES '; |
1469
|
|
|
foreach ($numbers as $number) { |
1470
|
|
|
$insertStr .= '('.$number.','.$groupID.'),'; |
1471
|
|
|
} |
1472
|
|
|
|
1473
|
|
|
return $this->_pdo->queryInsert(rtrim($insertStr, ',').' ON DUPLICATE KEY UPDATE attempts=attempts+1'); |
1474
|
|
|
} |
1475
|
|
|
|
1476
|
|
|
/** |
1477
|
|
|
* Clean up part repair table. |
1478
|
|
|
* |
1479
|
|
|
* @param array $numbers The article numbers. |
1480
|
|
|
* @param string $tableName Name of the part repair table to work on. |
1481
|
|
|
* @param int $groupID The ID of the group. |
1482
|
|
|
* |
1483
|
|
|
* @return void |
1484
|
|
|
*/ |
1485
|
|
|
private function removeRepairedParts(array $numbers, $tableName, $groupID): void |
1486
|
|
|
{ |
1487
|
|
|
$sql = 'DELETE FROM '.$tableName.' WHERE numberid in ('; |
1488
|
|
|
foreach ($numbers as $number) { |
1489
|
|
|
$sql .= $number.','; |
1490
|
|
|
} |
1491
|
|
|
$this->_pdo->queryExec(rtrim($sql, ',').') AND groups_id = '.$groupID); |
1492
|
|
|
} |
1493
|
|
|
|
1494
|
|
|
/** |
1495
|
|
|
* Are white or black lists loaded for a group name? |
1496
|
|
|
* @var array |
1497
|
|
|
*/ |
1498
|
|
|
protected $_listsFound = []; |
1499
|
|
|
|
1500
|
|
|
/** |
1501
|
|
|
* Get blacklist and cache it. Return if already cached. |
1502
|
|
|
* |
1503
|
|
|
* @param string $groupName |
1504
|
|
|
* |
1505
|
|
|
* @return void |
1506
|
|
|
*/ |
1507
|
|
|
protected function _retrieveBlackList($groupName): void |
1508
|
|
|
{ |
1509
|
|
|
if (! isset($this->blackList[$groupName])) { |
1510
|
|
|
$this->blackList[$groupName] = $this->getBlacklist(true, self::OPTYPE_BLACKLIST, $groupName, true); |
1511
|
|
|
} |
1512
|
|
|
if (! isset($this->whiteList[$groupName])) { |
1513
|
|
|
$this->whiteList[$groupName] = $this->getBlacklist(true, self::OPTYPE_WHITELIST, $groupName, true); |
1514
|
|
|
} |
1515
|
|
|
$this->_listsFound[$groupName] = ($this->blackList[$groupName] || $this->whiteList[$groupName]); |
1516
|
|
|
} |
1517
|
|
|
|
1518
|
|
|
/** |
1519
|
|
|
* Check if an article is blacklisted. |
1520
|
|
|
* |
1521
|
|
|
* @param array $msg The article header (OVER format). |
1522
|
|
|
* @param string $groupName The group name. |
1523
|
|
|
* |
1524
|
|
|
* @return bool |
1525
|
|
|
*/ |
1526
|
|
|
public function isBlackListed($msg, $groupName): bool |
1527
|
|
|
{ |
1528
|
|
|
if (! isset($this->_listsFound[$groupName])) { |
1529
|
|
|
$this->_retrieveBlackList($groupName); |
1530
|
|
|
} |
1531
|
|
|
if (! $this->_listsFound[$groupName]) { |
1532
|
|
|
return false; |
1533
|
|
|
} |
1534
|
|
|
|
1535
|
|
|
$blackListed = false; |
1536
|
|
|
|
1537
|
|
|
$field = [ |
1538
|
|
|
self::BLACKLIST_FIELD_SUBJECT => $msg['Subject'], |
1539
|
|
|
self::BLACKLIST_FIELD_FROM => $msg['From'], |
1540
|
|
|
self::BLACKLIST_FIELD_MESSAGEID => $msg['Message-ID'], |
1541
|
|
|
]; |
1542
|
|
|
|
1543
|
|
|
// Try white lists first. |
1544
|
|
|
if ($this->whiteList[$groupName]) { |
1545
|
|
|
// There are white lists for this group, so anything that doesn't match a white list should be considered black listed. |
1546
|
|
|
$blackListed = true; |
1547
|
|
|
foreach ($this->whiteList[$groupName] as $whiteList) { |
1548
|
|
|
if (preg_match('/'.$whiteList['regex'].'/i', $field[$whiteList['msgcol']])) { |
1549
|
|
|
// This field matched a white list, so it might not be black listed. |
1550
|
|
|
$blackListed = false; |
1551
|
|
|
$this->_binaryBlacklistIdsToUpdate[$whiteList['id']] = $whiteList['id']; |
1552
|
|
|
break; |
1553
|
|
|
} |
1554
|
|
|
} |
1555
|
|
|
} |
1556
|
|
|
|
1557
|
|
|
// Check if the field is black listed. |
1558
|
|
|
if (! $blackListed && $this->blackList[$groupName]) { |
1559
|
|
|
foreach ($this->blackList[$groupName] as $blackList) { |
1560
|
|
|
if (preg_match('/'.$blackList['regex'].'/i', $field[$blackList['msgcol']])) { |
1561
|
|
|
$blackListed = true; |
1562
|
|
|
$this->_binaryBlacklistIdsToUpdate[$blackList['id']] = $blackList['id']; |
1563
|
|
|
break; |
1564
|
|
|
} |
1565
|
|
|
} |
1566
|
|
|
} |
1567
|
|
|
|
1568
|
|
|
return $blackListed; |
1569
|
|
|
} |
1570
|
|
|
|
1571
|
|
|
/** |
1572
|
|
|
* Return all blacklists. |
1573
|
|
|
* |
1574
|
|
|
* @param bool $activeOnly Only display active blacklists ? |
1575
|
|
|
* @param int|string $opType Optional, get white or black lists (use Binaries constants). |
1576
|
|
|
* @param string $groupName Optional, group. |
1577
|
|
|
* @param bool $groupRegex Optional Join groups / binaryblacklist using regexp for equals. |
1578
|
|
|
* |
1579
|
|
|
* @return array |
1580
|
|
|
*/ |
1581
|
|
|
public function getBlacklist($activeOnly = true, $opType = -1, $groupName = '', $groupRegex = false): array |
1582
|
|
|
{ |
1583
|
|
|
switch ($opType) { |
1584
|
|
|
case self::OPTYPE_BLACKLIST: |
1585
|
|
|
$opType = 'AND bb.optype = '.self::OPTYPE_BLACKLIST; |
1586
|
|
|
break; |
1587
|
|
|
case self::OPTYPE_WHITELIST: |
1588
|
|
|
$opType = 'AND bb.optype = '.self::OPTYPE_WHITELIST; |
1589
|
|
|
break; |
1590
|
|
|
default: |
1591
|
|
|
$opType = ''; |
1592
|
|
|
break; |
1593
|
|
|
} |
1594
|
|
|
|
1595
|
|
|
return $this->_pdo->query( |
1596
|
|
|
sprintf( |
1597
|
|
|
' |
1598
|
|
|
SELECT |
1599
|
|
|
bb.id, bb.optype, bb.status, bb.description, |
1600
|
|
|
bb.groupname AS groupname, bb.regex, g.id AS group_id, bb.msgcol, |
1601
|
|
|
bb.last_activity as last_activity |
1602
|
|
|
FROM binaryblacklist bb |
1603
|
|
|
LEFT OUTER JOIN groups g ON g.name %s bb.groupname |
1604
|
|
|
WHERE 1=1 %s %s %s |
1605
|
|
|
ORDER BY coalesce(groupname,\'zzz\')', |
1606
|
|
|
($groupRegex ? 'REGEXP' : '='), |
1607
|
|
|
($activeOnly ? 'AND bb.status = 1' : ''), |
1608
|
|
|
$opType, |
1609
|
|
|
($groupName ? ('AND g.name REGEXP '.$this->_pdo->escapeString($groupName)) : '') |
1610
|
|
|
) |
1611
|
|
|
); |
1612
|
|
|
} |
1613
|
|
|
|
1614
|
|
|
/** |
1615
|
|
|
* Return the specified blacklist. |
1616
|
|
|
* |
1617
|
|
|
* @param int $id The blacklist ID. |
1618
|
|
|
* |
1619
|
|
|
* @return \Illuminate\Database\Eloquent\Model|null|static |
1620
|
|
|
*/ |
1621
|
|
|
public function getBlacklistByID($id) |
1622
|
|
|
{ |
1623
|
|
|
return BinaryBlacklist::query()->where('id', $id)->first(); |
1624
|
|
|
} |
1625
|
|
|
|
1626
|
|
|
/** |
1627
|
|
|
* Delete a blacklist. |
1628
|
|
|
* |
1629
|
|
|
* @param int $id The ID of the blacklist. |
1630
|
|
|
*/ |
1631
|
|
|
public function deleteBlacklist($id): void |
1632
|
|
|
{ |
1633
|
|
|
BinaryBlacklist::query()->where('id', $id)->delete(); |
1634
|
|
|
} |
1635
|
|
|
|
1636
|
|
|
/** |
1637
|
|
|
* @param $blacklistArray |
1638
|
|
|
*/ |
1639
|
|
|
public function updateBlacklist($blacklistArray): void |
1640
|
|
|
{ |
1641
|
|
|
BinaryBlacklist::query()->where('id', $blacklistArray['id'])->update( |
1642
|
|
|
[ |
1643
|
|
|
'groupname' => $blacklistArray['groupname'] === '' ? 'null' : preg_replace('/a\.b\./i', 'alt.binaries.', $blacklistArray['groupname']), |
1644
|
|
|
'regex' => $blacklistArray['regex'], |
1645
|
|
|
'status' => $blacklistArray['status'], |
1646
|
|
|
'description' => $blacklistArray['description'], |
1647
|
|
|
'optype' => $blacklistArray['optype'], |
1648
|
|
|
'msgcol' => $blacklistArray['msgcol'], |
1649
|
|
|
] |
1650
|
|
|
); |
1651
|
|
|
} |
1652
|
|
|
|
1653
|
|
|
/** |
1654
|
|
|
* Adds a new blacklist from binary blacklist edit admin web page. |
1655
|
|
|
* |
1656
|
|
|
* @param array $blacklistArray |
1657
|
|
|
*/ |
1658
|
|
|
public function addBlacklist($blacklistArray): void |
1659
|
|
|
{ |
1660
|
|
|
BinaryBlacklist::query()->insert( |
1661
|
|
|
[ |
1662
|
|
|
'groupname' => $blacklistArray['groupname'] === '' ? 'null' : preg_replace('/a\.b\./i', 'alt.binaries.', $blacklistArray['groupname']), |
1663
|
|
|
'regex' => $blacklistArray['regex'], |
1664
|
|
|
'status' => $blacklistArray['status'], |
1665
|
|
|
'description' => $blacklistArray['description'], |
1666
|
|
|
'optype' => $blacklistArray['optype'], |
1667
|
|
|
'msgcol' => $blacklistArray['msgcol'], |
1668
|
|
|
] |
1669
|
|
|
); |
1670
|
|
|
} |
1671
|
|
|
|
1672
|
|
|
/** |
1673
|
|
|
* Delete Collections/Binaries/Parts for a Collection ID. |
1674
|
|
|
* |
1675
|
|
|
* @param int $collectionID Collections table ID |
1676
|
|
|
* |
1677
|
|
|
* @note A trigger automatically deletes the parts/binaries. |
1678
|
|
|
* |
1679
|
|
|
* @return void |
1680
|
|
|
*/ |
1681
|
|
|
public function delete($collectionID): void |
1682
|
|
|
{ |
1683
|
|
|
$this->_pdo->queryExec(sprintf('DELETE FROM collections WHERE id = %d', $collectionID)); |
1684
|
|
|
} |
1685
|
|
|
|
1686
|
|
|
/** |
1687
|
|
|
* Log / Echo message. |
1688
|
|
|
* |
1689
|
|
|
* @param string $message Message to log. |
1690
|
|
|
* @param string $method Method that called this. |
1691
|
|
|
* @param string $color ColorCLI method name. |
1692
|
|
|
*/ |
1693
|
|
|
private function log($message, $method, $color): void |
1694
|
|
|
{ |
1695
|
|
|
if ($this->_echoCLI) { |
1696
|
|
|
ColorCLI::doEcho( |
1697
|
|
|
ColorCLI::$color($message.' ['.__CLASS__."::$method]"), |
1698
|
|
|
true |
1699
|
|
|
); |
1700
|
|
|
} |
1701
|
|
|
} |
1702
|
|
|
|
1703
|
|
|
/** |
1704
|
|
|
* Check if we should ignore the file count and return true or false. |
1705
|
|
|
* |
1706
|
|
|
* @param string $groupName |
1707
|
|
|
* @param string $subject |
1708
|
|
|
* |
1709
|
|
|
* @return bool |
1710
|
|
|
*/ |
1711
|
|
|
protected function _ignoreFileCount($groupName, $subject): bool |
1712
|
|
|
{ |
1713
|
|
|
$ignore = false; |
1714
|
|
|
switch ($groupName) { |
1715
|
|
|
case 'alt.binaries.erotica': |
1716
|
|
|
if (preg_match('/^\[\d+\]-\[FULL\]-\[#a\.b\.erotica@EFNet\]-\[ \d{2,3}_/', $subject)) { |
1717
|
|
|
$ignore = true; |
1718
|
|
|
} |
1719
|
|
|
break; |
1720
|
|
|
} |
1721
|
|
|
|
1722
|
|
|
return $ignore; |
1723
|
|
|
} |
1724
|
|
|
|
1725
|
|
|
/** |
1726
|
|
|
* Returns all multigroup poster entries from the database. |
1727
|
|
|
* |
1728
|
|
|
* |
1729
|
|
|
* @return \Illuminate\Database\Eloquent\Collection|static[] |
1730
|
|
|
*/ |
1731
|
|
|
protected function getMultiGroupPosters() |
1732
|
|
|
{ |
1733
|
|
|
$poster = Cache::get('mgrposter'); |
1734
|
|
|
if ($poster !== null) { |
1735
|
|
|
return $poster; |
|
|
|
|
1736
|
|
|
} |
1737
|
|
|
|
1738
|
|
|
$poster = MultigroupPoster::query()->get(['poster'])->toArray(); |
1739
|
|
|
$expiresAt = Carbon::now()->addSeconds(config('nntmux.cache_expiry_short')); |
1740
|
|
|
Cache::put('mgrposter', $poster, $expiresAt); |
1741
|
|
|
|
1742
|
|
|
return $poster; |
|
|
|
|
1743
|
|
|
} |
1744
|
|
|
} |
1745
|
|
|
|