1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Blacklight; |
4
|
|
|
|
5
|
|
|
use App\Models\Group; |
6
|
|
|
use Blacklight\db\DB; |
7
|
|
|
use App\Models\Settings; |
8
|
|
|
|
9
|
|
|
class Backfill |
10
|
|
|
{ |
11
|
|
|
/** |
12
|
|
|
* @var \Blacklight\db\DB |
13
|
|
|
*/ |
14
|
|
|
public $pdo; |
15
|
|
|
|
16
|
|
|
/** |
17
|
|
|
* @var |
18
|
|
|
*/ |
19
|
|
|
protected $_binaries; |
20
|
|
|
|
21
|
|
|
/** |
22
|
|
|
* @var \Blacklight\NNTP |
23
|
|
|
*/ |
24
|
|
|
protected $_nntp; |
25
|
|
|
/** |
26
|
|
|
* Should we use compression for headers? |
27
|
|
|
* |
28
|
|
|
* @var bool |
29
|
|
|
*/ |
30
|
|
|
protected $_compressedHeaders; |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* Log and or echo debug. |
34
|
|
|
* @var bool |
35
|
|
|
*/ |
36
|
|
|
protected $_debug = false; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* Echo to cli? |
40
|
|
|
* @var bool |
41
|
|
|
*/ |
42
|
|
|
protected $_echoCLI; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* How far back should we go on safe back fill? |
46
|
|
|
* |
47
|
|
|
* @var string |
48
|
|
|
*/ |
49
|
|
|
protected $_safeBackFillDate; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var string |
53
|
|
|
*/ |
54
|
|
|
protected $_safePartRepair; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* Should we disable the group if we have backfilled far enough? |
58
|
|
|
* @var bool |
59
|
|
|
*/ |
60
|
|
|
protected $_disableBackfillGroup; |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* Constructor. |
64
|
|
|
* |
65
|
|
|
* @param array $options Class instances / Echo to cli? |
66
|
|
|
* |
67
|
|
|
* @throws \Exception |
68
|
|
|
*/ |
69
|
|
|
public function __construct(array $options = []) |
70
|
|
|
{ |
71
|
|
|
$defaults = [ |
72
|
|
|
'Echo' => true, |
73
|
|
|
'Logger' => null, |
74
|
|
|
'Groups' => null, |
75
|
|
|
'NNTP' => null, |
76
|
|
|
'Settings' => null, |
77
|
|
|
]; |
78
|
|
|
$options += $defaults; |
79
|
|
|
|
80
|
|
|
$this->_echoCLI = ($options['Echo'] && config('nntmux.echocli')); |
81
|
|
|
|
82
|
|
|
$this->pdo = ($options['Settings'] instanceof DB ? $options['Settings'] : new DB()); |
83
|
|
|
$this->_nntp = ( |
84
|
|
|
$options['NNTP'] instanceof NNTP |
85
|
|
|
? $options['NNTP'] : new NNTP(['Settings' => $this->pdo]) |
86
|
|
|
); |
87
|
|
|
|
88
|
|
|
$this->_compressedHeaders = (int) Settings::settingValue('..compressedheaders') === 1; |
|
|
|
|
89
|
|
|
$this->_safeBackFillDate = Settings::settingValue('..safebackfilldate') !== '' ? (string) Settings::settingValue('safebackfilldate') : '2008-08-14'; |
90
|
|
|
$this->_safePartRepair = (int) Settings::settingValue('..safepartrepair') === 1 ? 'update' : 'backfill'; |
91
|
|
|
$this->_disableBackfillGroup = (int) Settings::settingValue('..disablebackfillgroup') === 1; |
92
|
|
|
} |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* Backfill all the groups up to user specified time/date. |
96
|
|
|
* |
97
|
|
|
* @param string $groupName |
98
|
|
|
* @param string|int $articles |
99
|
|
|
* @param string $type |
100
|
|
|
* |
101
|
|
|
* @return void |
102
|
|
|
* @throws \Exception |
103
|
|
|
*/ |
104
|
|
|
public function backfillAllGroups($groupName = '', $articles = '', $type = ''): void |
105
|
|
|
{ |
106
|
|
|
$res = []; |
107
|
|
|
if ($groupName !== '') { |
108
|
|
|
$grp = Group::getByName($groupName); |
109
|
|
|
if ($grp) { |
110
|
|
|
$res = [$grp]; |
111
|
|
|
} |
112
|
|
|
} else { |
113
|
|
|
$res = Group::getActiveBackfill($type); |
114
|
|
|
} |
115
|
|
|
|
116
|
|
|
$groupCount = count($res); |
117
|
|
|
if ($groupCount > 0) { |
118
|
|
|
$counter = 1; |
119
|
|
|
$allTime = microtime(true); |
120
|
|
|
$dMessage = ( |
121
|
|
|
'Backfilling: '. |
122
|
|
|
$groupCount. |
123
|
|
|
' group(s) - Using compression? '. |
124
|
|
|
($this->_compressedHeaders ? 'Yes' : 'No') |
125
|
|
|
); |
126
|
|
|
|
127
|
|
|
if ($this->_echoCLI) { |
128
|
|
|
ColorCLI::doEcho(ColorCLI::header($dMessage), true); |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
$this->_binaries = new Binaries( |
132
|
|
|
['NNTP' => $this->_nntp, 'Echo' => $this->_echoCLI, 'Settings' => $this->pdo] |
133
|
|
|
); |
134
|
|
|
|
135
|
|
|
if ($articles !== '' && ! is_numeric($articles)) { |
136
|
|
|
$articles = 20000; |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
// Loop through groups. |
140
|
|
|
foreach ($res as $groupArr) { |
141
|
|
|
if ($groupName === '') { |
142
|
|
|
$dMessage = 'Starting group '.$counter.' of '.$groupCount; |
143
|
|
|
|
144
|
|
|
if ($this->_echoCLI) { |
145
|
|
|
ColorCLI::doEcho(ColorCLI::header($dMessage), true); |
146
|
|
|
} |
147
|
|
|
} |
148
|
|
|
$this->backfillGroup($groupArr, $groupCount - $counter, $articles); |
149
|
|
|
$counter++; |
150
|
|
|
} |
151
|
|
|
|
152
|
|
|
$dMessage = 'Backfilling completed in '.number_format(microtime(true) - $allTime, 2).' seconds.'; |
153
|
|
|
|
154
|
|
|
if ($this->_echoCLI) { |
155
|
|
|
ColorCLI::doEcho(ColorCLI::primary($dMessage), true); |
156
|
|
|
} |
157
|
|
|
} else { |
158
|
|
|
$dMessage = 'No groups specified. Ensure groups are added to database for updating.'; |
159
|
|
|
|
160
|
|
|
if ($this->_echoCLI) { |
161
|
|
|
ColorCLI::doEcho(ColorCLI::warning($dMessage), true); |
162
|
|
|
} |
163
|
|
|
} |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
/** |
167
|
|
|
* Backfill single group. |
168
|
|
|
* |
169
|
|
|
* @param array $groupArr |
170
|
|
|
* @param int $left |
171
|
|
|
* @param int|string $articles |
172
|
|
|
* |
173
|
|
|
* @return void |
174
|
|
|
* @throws \Exception |
175
|
|
|
*/ |
176
|
|
|
public function backfillGroup($groupArr, $left, $articles = ''): void |
177
|
|
|
{ |
178
|
|
|
// Start time for this group. |
179
|
|
|
$startGroup = microtime(true); |
180
|
|
|
|
181
|
|
|
$this->_binaries->logIndexerStart(); |
182
|
|
|
|
183
|
|
|
$groupName = str_replace('alt.binaries', 'a.b', $groupArr['name']); |
184
|
|
|
|
185
|
|
|
// If our local oldest article 0, it means we never ran update_binaries on the group. |
186
|
|
|
if ($groupArr['first_record'] <= 0) { |
187
|
|
|
$dMessage = |
188
|
|
|
'You need to run update_binaries on '. |
189
|
|
|
$groupName. |
190
|
|
|
'. Otherwise the group is dead, you must disable it.'; |
191
|
|
|
|
192
|
|
|
if ($this->_echoCLI) { |
193
|
|
|
ColorCLI::doEcho(ColorCLI::error($dMessage), true); |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
return; |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
// Select group, here, only once |
200
|
|
|
$data = $this->_nntp->selectGroup($groupArr['name']); |
201
|
|
|
if ($this->_nntp->isError($data)) { |
202
|
|
|
$data = $this->_nntp->dataError($this->_nntp, $groupArr['name']); |
203
|
|
|
if ($this->_nntp->isError($data)) { |
204
|
|
|
return; |
205
|
|
|
} |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
if ($this->_echoCLI) { |
209
|
|
|
ColorCLI::doEcho(ColorCLI::primary('Processing '.$groupName), true); |
210
|
|
|
} |
211
|
|
|
|
212
|
|
|
// Check if this is days or post backfill. |
213
|
|
|
$postCheck = $articles !== ''; |
214
|
|
|
|
215
|
|
|
// Get target post based on date or user specified number. |
216
|
|
|
$targetpost = (string) ( |
217
|
|
|
$postCheck |
218
|
|
|
? |
219
|
|
|
round($groupArr['first_record'] - $articles) |
220
|
|
|
: |
221
|
|
|
$this->_binaries->daytopost($groupArr['backfill_target'], $data) |
222
|
|
|
); |
223
|
|
|
|
224
|
|
|
// Check if target post is smaller than server's oldest, set it to oldest if so. |
225
|
|
|
if ($targetpost < $data['first']) { |
226
|
|
|
$targetpost = $data['first']; |
227
|
|
|
} |
228
|
|
|
|
229
|
|
|
// Check if our target post is newer than our oldest post or if our local oldest article is older than the servers oldest. |
230
|
|
|
if ($targetpost >= $groupArr['first_record'] || $groupArr['first_record'] <= $data['first']) { |
231
|
|
|
$dMessage = |
232
|
|
|
'We have hit the maximum we can backfill for '. |
233
|
|
|
$groupName. |
234
|
|
|
($this->_disableBackfillGroup ? ', disabling backfill on it.' : |
235
|
|
|
', skipping it, consider disabling backfill on it.'); |
236
|
|
|
|
237
|
|
|
if ($this->_disableBackfillGroup) { |
238
|
|
|
Group::updateGroupStatus($groupArr['id'], 'backfill', 0); |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
if ($this->_echoCLI) { |
242
|
|
|
ColorCLI::doEcho(ColorCLI::notice($dMessage), true); |
243
|
|
|
} |
244
|
|
|
|
245
|
|
|
return; |
246
|
|
|
} |
247
|
|
|
|
248
|
|
|
if ($this->_echoCLI) { |
249
|
|
|
ColorCLI::doEcho( |
250
|
|
|
ColorCLI::primary( |
251
|
|
|
'Group '. |
252
|
|
|
$groupName. |
253
|
|
|
"'s oldest article is ". |
254
|
|
|
number_format($data['first']). |
255
|
|
|
', newest is '. |
256
|
|
|
number_format($data['last']). |
257
|
|
|
".\nOur target article is ". |
258
|
|
|
number_format($targetpost). |
259
|
|
|
'. Our oldest article is article '. |
260
|
|
|
number_format($groupArr['first_record']). |
261
|
|
|
'.' |
262
|
|
|
), true |
263
|
|
|
); |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
// Set first and last, moving the window by max messages. |
267
|
|
|
$last = ($groupArr['first_record'] - 1); |
268
|
|
|
// Set the initial "chunk". |
269
|
|
|
$first = ($last - $this->_binaries->messageBuffer + 1); |
270
|
|
|
|
271
|
|
|
// Just in case this is the last chunk we needed. |
272
|
|
|
if ($targetpost > $first) { |
273
|
|
|
$first = $targetpost; |
274
|
|
|
} |
275
|
|
|
|
276
|
|
|
$done = false; |
277
|
|
|
while ($done === false) { |
278
|
|
|
if ($this->_echoCLI) { |
279
|
|
|
ColorCLI::doEcho( |
280
|
|
|
color('Getting '. |
281
|
|
|
number_format($last - $first + 1). |
282
|
|
|
' articles from '. |
283
|
|
|
$groupName. |
284
|
|
|
', '. |
285
|
|
|
$left. |
286
|
|
|
' group(s) left. ('. |
287
|
|
|
number_format($first - $targetpost). |
288
|
|
|
' articles in queue')->fg('yellow'), true); |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
flush(); |
292
|
|
|
$lastMsg = $this->_binaries->scan($groupArr, $first, $last, $this->_safePartRepair); |
293
|
|
|
|
294
|
|
|
// Get the oldest date. |
295
|
|
|
if (isset($lastMsg['firstArticleDate'])) { |
296
|
|
|
// Try to get it from the oldest pulled article. |
297
|
|
|
$newdate = strtotime($lastMsg['firstArticleDate']); |
298
|
|
|
} else { |
299
|
|
|
// If above failed, try to get it with postdate method. |
300
|
|
|
$newdate = $this->_binaries->postdate($first, $data); |
301
|
|
|
} |
302
|
|
|
|
303
|
|
|
$this->pdo->queryExec( |
304
|
|
|
sprintf( |
305
|
|
|
' |
306
|
|
|
UPDATE groups |
307
|
|
|
SET first_record_postdate = %s, first_record = %s, last_updated = NOW() |
308
|
|
|
WHERE id = %d', |
309
|
|
|
$this->pdo->from_unixtime($newdate), |
310
|
|
|
$this->pdo->escapeString($first), |
311
|
|
|
$groupArr['id'] |
312
|
|
|
) |
313
|
|
|
); |
314
|
|
|
if ($first === $targetpost) { |
315
|
|
|
$done = true; |
316
|
|
|
} else { |
317
|
|
|
// Keep going: set new last, new first, check for last chunk. |
318
|
|
|
$last = ($first - 1); |
319
|
|
|
$first = ($last - $this->_binaries->messageBuffer + 1); |
320
|
|
|
if ($targetpost > $first) { |
321
|
|
|
$first = $targetpost; |
322
|
|
|
} |
323
|
|
|
} |
324
|
|
|
} |
325
|
|
|
|
326
|
|
|
if ($this->_echoCLI) { |
327
|
|
|
ColorCLI::doEcho( |
328
|
|
|
ColorCLI::primary( |
329
|
|
|
PHP_EOL. |
330
|
|
|
'Group '. |
331
|
|
|
$groupName. |
332
|
|
|
' processed in '. |
333
|
|
|
number_format(microtime(true) - $startGroup, 2). |
334
|
|
|
' seconds.' |
335
|
|
|
), |
336
|
|
|
true |
337
|
|
|
); |
338
|
|
|
} |
339
|
|
|
} |
340
|
|
|
|
341
|
|
|
/** |
342
|
|
|
* Safe backfill using posts. Going back to a date specified by the user on the site settings. |
343
|
|
|
* This does 1 group for x amount of parts until it reaches the date. |
344
|
|
|
* |
345
|
|
|
* @param string $articles |
346
|
|
|
* |
347
|
|
|
* @return void |
348
|
|
|
* @throws \Exception |
349
|
|
|
*/ |
350
|
|
|
public function safeBackfill($articles = ''): void |
351
|
|
|
{ |
352
|
|
|
$groupname = $this->pdo->queryOneRow( |
353
|
|
|
sprintf( |
354
|
|
|
' |
355
|
|
|
SELECT name FROM groups |
356
|
|
|
WHERE first_record_postdate BETWEEN %s AND NOW() |
357
|
|
|
AND backfill = 1 |
358
|
|
|
ORDER BY name ASC', |
359
|
|
|
$this->pdo->escapeString($this->_safeBackFillDate) |
360
|
|
|
) |
361
|
|
|
); |
362
|
|
|
|
363
|
|
|
if (! $groupname) { |
364
|
|
|
$dMessage = |
365
|
|
|
'No groups to backfill, they are all at the target date '. |
366
|
|
|
$this->_safeBackFillDate. |
367
|
|
|
', or you have not enabled them to be backfilled in the groups page.'.PHP_EOL; |
368
|
|
|
exit($dMessage); |
|
|
|
|
369
|
|
|
} |
370
|
|
|
$this->backfillAllGroups($groupname['name'], $articles); |
371
|
|
|
} |
372
|
|
|
} |
373
|
|
|
|