|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace App\Services\Backfill; |
|
6
|
|
|
|
|
7
|
|
|
use App\Models\UsenetGroup; |
|
8
|
|
|
use App\Services\Binaries\BinariesService; |
|
9
|
|
|
use Blacklight\ColorCLI; |
|
10
|
|
|
use Blacklight\NNTP; |
|
11
|
|
|
use Illuminate\Support\Carbon; |
|
12
|
|
|
use Illuminate\Support\Facades\DB; |
|
13
|
|
|
|
|
14
|
|
|
/** |
|
15
|
|
|
* Service for backfilling Usenet groups with historical articles. |
|
16
|
|
|
* |
|
17
|
|
|
* This service handles downloading older articles from Usenet groups |
|
18
|
|
|
* to fill in historical data. It supports: |
|
19
|
|
|
* - Backfilling by article count or target date |
|
20
|
|
|
* - Safe backfill with date-based targeting |
|
21
|
|
|
* - Automatic group disable when backfill limit is reached |
|
22
|
|
|
*/ |
|
23
|
|
|
final class BackfillService |
|
24
|
|
|
{ |
|
25
|
|
|
private const DEFAULT_ARTICLE_COUNT = 20000; |
|
26
|
|
|
|
|
27
|
|
|
private BackfillConfig $config; |
|
|
|
|
|
|
28
|
|
|
|
|
29
|
|
|
private BinariesService $binaries; |
|
30
|
|
|
|
|
31
|
|
|
private NNTP $nntp; |
|
32
|
|
|
|
|
33
|
|
|
private ColorCLI $colorCli; |
|
34
|
|
|
|
|
35
|
|
|
public function __construct( |
|
36
|
|
|
?BackfillConfig $config = null, |
|
37
|
|
|
?BinariesService $binaries = null, |
|
38
|
|
|
?NNTP $nntp = null, |
|
39
|
|
|
?ColorCLI $colorCli = null, |
|
40
|
|
|
) { |
|
41
|
|
|
$this->config = $config ?? BackfillConfig::fromSettings(); |
|
42
|
|
|
$this->binaries = $binaries ?? new BinariesService; |
|
43
|
|
|
$this->nntp = $nntp ?? new NNTP; |
|
44
|
|
|
$this->colorCli = $colorCli ?? new ColorCLI; |
|
45
|
|
|
} |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Backfill all groups or a specific group. |
|
49
|
|
|
* |
|
50
|
|
|
* @param string $groupName Optional specific group to backfill |
|
51
|
|
|
* @param int|string $articles Number of articles to backfill, or empty for date-based |
|
52
|
|
|
* @param string $type Backfill type filter |
|
53
|
|
|
* |
|
54
|
|
|
* @throws \Throwable |
|
55
|
|
|
*/ |
|
56
|
|
|
public function backfillAllGroups(string $groupName = '', int|string $articles = '', string $type = ''): void |
|
57
|
|
|
{ |
|
58
|
|
|
$groups = $this->getGroupsToBackfill($groupName, $type); |
|
59
|
|
|
|
|
60
|
|
|
if ($groups === []) { |
|
61
|
|
|
$this->log('No groups specified. Ensure groups are added to database for updating.', 'warning'); |
|
62
|
|
|
|
|
63
|
|
|
return; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
$groupCount = \count($groups); |
|
67
|
|
|
$this->logBackfillStart($groupCount); |
|
68
|
|
|
|
|
69
|
|
|
$articles = $this->normalizeArticleCount($articles); |
|
70
|
|
|
$startTime = now(); |
|
71
|
|
|
|
|
72
|
|
|
foreach ($groups as $index => $group) { |
|
73
|
|
|
$this->logGroupProgress($groupName, $index + 1, $groupCount); |
|
74
|
|
|
$this->backfillGroup($group->toArray(), $groupCount - $index - 1, $articles); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
$this->logBackfillComplete($startTime); |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* Backfill a single group. |
|
82
|
|
|
* |
|
83
|
|
|
* @param array $groupArr Group data array |
|
84
|
|
|
* @param int $remainingGroups Number of groups remaining after this one |
|
85
|
|
|
* @param int|string $articles Number of articles to backfill, or empty for date-based |
|
86
|
|
|
* |
|
87
|
|
|
* @throws \Throwable |
|
88
|
|
|
*/ |
|
89
|
|
|
public function backfillGroup(array $groupArr, int $remainingGroups, int|string $articles = ''): void |
|
90
|
|
|
{ |
|
91
|
|
|
$startTime = now(); |
|
92
|
|
|
$this->binaries->logIndexerStart(); |
|
93
|
|
|
|
|
94
|
|
|
$shortGroupName = $this->getShortGroupName($groupArr['name']); |
|
95
|
|
|
|
|
96
|
|
|
if (! $this->validateGroupState($groupArr, $shortGroupName)) { |
|
97
|
|
|
return; |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
$serverData = $this->selectNntpGroup($groupArr['name']); |
|
101
|
|
|
if ($serverData === null) { |
|
102
|
|
|
return; |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
$this->log("Processing {$shortGroupName}", 'primary'); |
|
106
|
|
|
|
|
107
|
|
|
$targetPost = $this->calculateTargetPost($groupArr, $articles, $serverData); |
|
108
|
|
|
|
|
109
|
|
|
if (! $this->validateTargetPost($groupArr, $targetPost, $serverData, $shortGroupName)) { |
|
110
|
|
|
return; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
$this->logGroupInfo($groupArr, $serverData, $targetPost, $shortGroupName); |
|
114
|
|
|
|
|
115
|
|
|
$this->processBackfillChunks($groupArr, $targetPost, $remainingGroups, $shortGroupName); |
|
116
|
|
|
|
|
117
|
|
|
$this->logGroupComplete($shortGroupName, $startTime); |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Safe backfill - backfill groups that haven't reached the safe backfill date. |
|
122
|
|
|
* |
|
123
|
|
|
* @param int|string $articles Number of articles to backfill |
|
124
|
|
|
* |
|
125
|
|
|
* @throws \Throwable |
|
126
|
|
|
*/ |
|
127
|
|
|
public function safeBackfill(int|string $articles = ''): void |
|
128
|
|
|
{ |
|
129
|
|
|
$group = UsenetGroup::query() |
|
130
|
|
|
->whereBetween('first_record_postdate', [Carbon::createFromDate($this->config->safeBackFillDate), now()]) |
|
131
|
|
|
->where('backfill', '=', 1) |
|
132
|
|
|
->select(['name']) |
|
133
|
|
|
->orderBy('name') |
|
|
|
|
|
|
134
|
|
|
->first(); |
|
135
|
|
|
|
|
136
|
|
|
if ($group === null) { |
|
137
|
|
|
$message = sprintf( |
|
138
|
|
|
'No groups to backfill, they are all at the target date %s, or you have not enabled them to be backfilled in the groups page.', |
|
139
|
|
|
$this->config->safeBackFillDate |
|
140
|
|
|
); |
|
141
|
|
|
exit($message.PHP_EOL); |
|
|
|
|
|
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
$this->backfillAllGroups($group->name, $articles); |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Get groups to backfill based on criteria. |
|
149
|
|
|
*/ |
|
150
|
|
|
private function getGroupsToBackfill(string $groupName, string $type): array |
|
151
|
|
|
{ |
|
152
|
|
|
if ($groupName !== '') { |
|
153
|
|
|
$group = UsenetGroup::getByName($groupName); |
|
154
|
|
|
|
|
155
|
|
|
return $group ? [$group] : []; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
return UsenetGroup::getActiveBackfill($type)->all(); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
/** |
|
162
|
|
|
* Normalize article count parameter. |
|
163
|
|
|
*/ |
|
164
|
|
|
private function normalizeArticleCount(int|string $articles): int|string |
|
165
|
|
|
{ |
|
166
|
|
|
if ($articles !== '' && ! is_numeric($articles)) { |
|
167
|
|
|
return self::DEFAULT_ARTICLE_COUNT; |
|
168
|
|
|
} |
|
169
|
|
|
|
|
170
|
|
|
return $articles; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* Get shortened group name for display. |
|
175
|
|
|
*/ |
|
176
|
|
|
private function getShortGroupName(string $groupName): string |
|
177
|
|
|
{ |
|
178
|
|
|
return str_replace('alt.binaries', 'a.b', $groupName); |
|
179
|
|
|
} |
|
180
|
|
|
|
|
181
|
|
|
/** |
|
182
|
|
|
* Validate that group is in a valid state for backfilling. |
|
183
|
|
|
*/ |
|
184
|
|
|
private function validateGroupState(array $groupArr, string $shortGroupName): bool |
|
185
|
|
|
{ |
|
186
|
|
|
if ($groupArr['first_record'] <= 0) { |
|
187
|
|
|
$this->log( |
|
188
|
|
|
"You need to run update_binaries on {$shortGroupName}. Otherwise the group is dead, you must disable it.", |
|
189
|
|
|
'error' |
|
190
|
|
|
); |
|
191
|
|
|
|
|
192
|
|
|
return false; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
return true; |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
/** |
|
199
|
|
|
* Select NNTP group and return server data. |
|
200
|
|
|
*/ |
|
201
|
|
|
private function selectNntpGroup(string $groupName): ?array |
|
202
|
|
|
{ |
|
203
|
|
|
$data = $this->nntp->selectGroup($groupName); |
|
204
|
|
|
|
|
205
|
|
|
if ($this->nntp->isError($data)) { |
|
206
|
|
|
$data = $this->nntp->dataError($this->nntp, $groupName); |
|
207
|
|
|
if ($this->nntp->isError($data)) { |
|
208
|
|
|
return null; |
|
209
|
|
|
} |
|
210
|
|
|
} |
|
211
|
|
|
|
|
212
|
|
|
return $data; |
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
|
|
/** |
|
216
|
|
|
* Calculate target post number based on articles count or date. |
|
217
|
|
|
*/ |
|
218
|
|
|
private function calculateTargetPost(array $groupArr, int|string $articles, array $serverData): int |
|
219
|
|
|
{ |
|
220
|
|
|
$isArticleBased = $articles !== ''; |
|
221
|
|
|
|
|
222
|
|
|
$targetPost = $isArticleBased |
|
223
|
|
|
? (int) round($groupArr['first_record'] - (int) $articles) |
|
224
|
|
|
: (int) $this->binaries->daytopost($groupArr['backfill_target'], $serverData); |
|
225
|
|
|
|
|
226
|
|
|
// Ensure target is not below server's oldest article |
|
227
|
|
|
return max($targetPost, (int) $serverData['first']); |
|
228
|
|
|
} |
|
229
|
|
|
|
|
230
|
|
|
/** |
|
231
|
|
|
* Validate that target post is achievable. |
|
232
|
|
|
*/ |
|
233
|
|
|
private function validateTargetPost(array $groupArr, int $targetPost, array $serverData, string $shortGroupName): bool |
|
234
|
|
|
{ |
|
235
|
|
|
if ($targetPost >= $groupArr['first_record'] || $groupArr['first_record'] <= $serverData['first']) { |
|
236
|
|
|
$message = "We have hit the maximum we can backfill for {$shortGroupName}"; |
|
237
|
|
|
$message .= $this->config->disableBackfillGroup |
|
238
|
|
|
? ', disabling backfill on it.' |
|
239
|
|
|
: ', skipping it, consider disabling backfill on it.'; |
|
240
|
|
|
|
|
241
|
|
|
if ($this->config->disableBackfillGroup) { |
|
242
|
|
|
UsenetGroup::updateGroupStatus($groupArr['id'], 'backfill', 0); |
|
243
|
|
|
} |
|
244
|
|
|
|
|
245
|
|
|
$this->log($message, 'notice'); |
|
246
|
|
|
|
|
247
|
|
|
return false; |
|
248
|
|
|
} |
|
249
|
|
|
|
|
250
|
|
|
return true; |
|
251
|
|
|
} |
|
252
|
|
|
|
|
253
|
|
|
/** |
|
254
|
|
|
* Process backfill in chunks. |
|
255
|
|
|
*/ |
|
256
|
|
|
private function processBackfillChunks(array $groupArr, int $targetPost, int $remainingGroups, string $shortGroupName): void |
|
257
|
|
|
{ |
|
258
|
|
|
$messageBuffer = $this->binaries->getMessageBuffer(); |
|
259
|
|
|
$last = $groupArr['first_record'] - 1; |
|
260
|
|
|
$first = max($last - $messageBuffer + 1, $targetPost); |
|
261
|
|
|
|
|
262
|
|
|
while (true) { |
|
263
|
|
|
$this->logChunkProgress($first, $last, $shortGroupName, $remainingGroups, $targetPost); |
|
264
|
|
|
|
|
265
|
|
|
flush(); |
|
266
|
|
|
$scanResult = $this->binaries->scan($groupArr, $first, $last, $this->config->safePartRepair); |
|
267
|
|
|
|
|
268
|
|
|
$this->updateGroupRecord($groupArr, $first, $scanResult); |
|
269
|
|
|
|
|
270
|
|
|
if ($first === $targetPost) { |
|
271
|
|
|
break; |
|
272
|
|
|
} |
|
273
|
|
|
|
|
274
|
|
|
// Move to next chunk |
|
275
|
|
|
$last = $first - 1; |
|
276
|
|
|
$first = max($last - $messageBuffer + 1, $targetPost); |
|
277
|
|
|
} |
|
278
|
|
|
} |
|
279
|
|
|
|
|
280
|
|
|
/** |
|
281
|
|
|
* Update group record with new first_record and postdate. |
|
282
|
|
|
*/ |
|
283
|
|
|
private function updateGroupRecord(array $groupArr, int $first, ?array $scanResult): void |
|
284
|
|
|
{ |
|
285
|
|
|
$newDate = isset($scanResult['firstArticleDate']) |
|
286
|
|
|
? strtotime($scanResult['firstArticleDate']) |
|
287
|
|
|
: $this->binaries->postdate($first, $this->nntp->selectGroup($groupArr['name'])); |
|
288
|
|
|
|
|
289
|
|
|
DB::update( |
|
290
|
|
|
'UPDATE usenet_groups SET first_record_postdate = FROM_UNIXTIME(?), first_record = ?, last_updated = NOW() WHERE id = ?', |
|
291
|
|
|
[$newDate, $first, $groupArr['id']] |
|
292
|
|
|
); |
|
293
|
|
|
} |
|
294
|
|
|
|
|
295
|
|
|
/** |
|
296
|
|
|
* Log message with appropriate styling. |
|
297
|
|
|
*/ |
|
298
|
|
|
private function log(string $message, string $type = 'primary'): void |
|
299
|
|
|
{ |
|
300
|
|
|
if (! $this->config->echoCli) { |
|
301
|
|
|
return; |
|
302
|
|
|
} |
|
303
|
|
|
|
|
304
|
|
|
match ($type) { |
|
305
|
|
|
'header' => $this->colorCli->header($message), |
|
|
|
|
|
|
306
|
|
|
'warning' => $this->colorCli->warning($message), |
|
|
|
|
|
|
307
|
|
|
'error' => $this->colorCli->error($message), |
|
|
|
|
|
|
308
|
|
|
'notice' => $this->colorCli->notice($message), |
|
|
|
|
|
|
309
|
|
|
default => $this->colorCli->primary($message), |
|
|
|
|
|
|
310
|
|
|
}; |
|
311
|
|
|
} |
|
312
|
|
|
|
|
313
|
|
|
/** |
|
314
|
|
|
* Log backfill start information. |
|
315
|
|
|
*/ |
|
316
|
|
|
private function logBackfillStart(int $groupCount): void |
|
317
|
|
|
{ |
|
318
|
|
|
$compressionStatus = $this->config->compressedHeaders ? 'Yes' : 'No'; |
|
319
|
|
|
$this->log("Backfilling: {$groupCount} group(s) - Using compression? {$compressionStatus}", 'header'); |
|
320
|
|
|
} |
|
321
|
|
|
|
|
322
|
|
|
/** |
|
323
|
|
|
* Log group progress. |
|
324
|
|
|
*/ |
|
325
|
|
|
private function logGroupProgress(string $groupName, int $current, int $total): void |
|
326
|
|
|
{ |
|
327
|
|
|
if ($groupName === '') { |
|
328
|
|
|
$this->log("Starting group {$current} of {$total}", 'header'); |
|
329
|
|
|
} |
|
330
|
|
|
} |
|
331
|
|
|
|
|
332
|
|
|
/** |
|
333
|
|
|
* Log backfill completion. |
|
334
|
|
|
*/ |
|
335
|
|
|
private function logBackfillComplete(\Illuminate\Support\Carbon $startTime): void |
|
336
|
|
|
{ |
|
337
|
|
|
$duration = now()->diffInSeconds($startTime, true); |
|
338
|
|
|
$this->log("Backfilling completed in {$duration} seconds."); |
|
339
|
|
|
} |
|
340
|
|
|
|
|
341
|
|
|
/** |
|
342
|
|
|
* Log group info before processing. |
|
343
|
|
|
*/ |
|
344
|
|
|
private function logGroupInfo(array $groupArr, array $serverData, int $targetPost, string $shortGroupName): void |
|
345
|
|
|
{ |
|
346
|
|
|
$this->log(sprintf( |
|
347
|
|
|
"Group %s's oldest article is %s, newest is %s. Our target article is %s. Our oldest article is article %s.", |
|
348
|
|
|
$shortGroupName, |
|
349
|
|
|
number_format((float) $serverData['first']), |
|
350
|
|
|
number_format((float) $serverData['last']), |
|
351
|
|
|
number_format($targetPost), |
|
352
|
|
|
number_format((float) $groupArr['first_record']) |
|
353
|
|
|
)); |
|
354
|
|
|
} |
|
355
|
|
|
|
|
356
|
|
|
/** |
|
357
|
|
|
* Log chunk progress. |
|
358
|
|
|
*/ |
|
359
|
|
|
private function logChunkProgress(int $first, int $last, string $shortGroupName, int $remainingGroups, int $targetPost): void |
|
360
|
|
|
{ |
|
361
|
|
|
$this->log(sprintf( |
|
362
|
|
|
'Getting %s articles from %s, %d group(s) left. (%s articles in queue)', |
|
363
|
|
|
number_format($last - $first + 1), |
|
364
|
|
|
$shortGroupName, |
|
365
|
|
|
$remainingGroups, |
|
366
|
|
|
number_format($first - $targetPost) |
|
367
|
|
|
), 'header'); |
|
368
|
|
|
} |
|
369
|
|
|
|
|
370
|
|
|
/** |
|
371
|
|
|
* Log group completion. |
|
372
|
|
|
*/ |
|
373
|
|
|
private function logGroupComplete(string $shortGroupName, \Illuminate\Support\Carbon $startTime): void |
|
374
|
|
|
{ |
|
375
|
|
|
$duration = number_format(now()->timestamp - $startTime->timestamp, 2); |
|
376
|
|
|
$this->log(PHP_EOL."Group {$shortGroupName} processed in {$duration} seconds."); |
|
377
|
|
|
} |
|
378
|
|
|
} |
|
379
|
|
|
|
|
380
|
|
|
|
The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g.
excluded_paths: ["lib/*"], you can move it to the dependency path list as follows:For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths