1
|
|
|
<?php |
2
|
|
|
declare(strict_types = 1); |
3
|
|
|
|
4
|
|
|
namespace BrowscapPHP\Command; |
5
|
|
|
|
6
|
|
|
use BrowscapPHP\Browscap; |
7
|
|
|
use BrowscapPHP\Exception\InvalidArgumentException; |
8
|
|
|
use BrowscapPHP\Exception\ReaderException; |
9
|
|
|
use BrowscapPHP\Exception\UnknownBrowserException; |
10
|
|
|
use BrowscapPHP\Exception\UnknownBrowserTypeException; |
11
|
|
|
use BrowscapPHP\Exception\UnknownDeviceException; |
12
|
|
|
use BrowscapPHP\Exception\UnknownEngineException; |
13
|
|
|
use BrowscapPHP\Exception\UnknownPlatformException; |
14
|
|
|
use BrowscapPHP\Helper\Filesystem; |
15
|
|
|
use BrowscapPHP\Helper\LoggerHelper; |
16
|
|
|
use BrowscapPHP\Util\Logfile\ReaderCollection; |
17
|
|
|
use BrowscapPHP\Util\Logfile\ReaderFactory; |
18
|
|
|
use Doctrine\Common\Cache\FilesystemCache; |
19
|
|
|
use Psr\Log\LoggerInterface; |
20
|
|
|
use Psr\SimpleCache\CacheInterface; |
21
|
|
|
use Roave\DoctrineSimpleCache\SimpleCacheAdapter; |
22
|
|
|
use Symfony\Component\Console\Command\Command; |
23
|
|
|
use Symfony\Component\Console\Input\InputArgument; |
24
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
25
|
|
|
use Symfony\Component\Console\Input\InputOption; |
26
|
|
|
use Symfony\Component\Console\Output\OutputInterface; |
27
|
|
|
use Symfony\Component\Filesystem\Exception\IOException; |
28
|
|
|
use Symfony\Component\Finder\Finder; |
29
|
|
|
use Symfony\Component\Finder\SplFileInfo; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Commands to parse a log file and parse the useragents in it |
33
|
|
|
*/ |
34
|
|
|
class LogfileCommand extends Command |
35
|
|
|
{ |
36
|
|
|
/** |
37
|
|
|
* @var array |
38
|
|
|
*/ |
39
|
|
|
private $undefinedClients = []; |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* @var array |
43
|
|
|
*/ |
44
|
|
|
private $uas = []; |
45
|
|
|
|
46
|
|
|
/** |
47
|
|
|
* @var array |
48
|
|
|
*/ |
49
|
|
|
private $uasWithType = []; |
50
|
|
|
|
51
|
|
|
/** |
52
|
|
|
* @var int |
53
|
|
|
*/ |
54
|
|
|
private $countOk = 0; |
55
|
|
|
|
56
|
|
|
/** |
57
|
|
|
* @var int |
58
|
|
|
*/ |
59
|
|
|
private $countNok = 0; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* @var int |
63
|
|
|
*/ |
64
|
|
|
private $totalCount = 0; |
65
|
|
|
|
66
|
|
|
/** |
67
|
|
|
* @var ?CacheInterface |
68
|
|
|
*/ |
69
|
|
|
private $cache; |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @var string |
73
|
|
|
*/ |
74
|
|
|
private $defaultCacheFolder; |
75
|
|
|
|
76
|
|
|
/** |
77
|
|
|
* @var LoggerInterface |
78
|
|
|
*/ |
79
|
|
|
private $logger; |
80
|
|
|
|
81
|
1 |
|
public function __construct( |
82
|
|
|
string $defaultCacheFolder, |
83
|
|
|
?CacheInterface $cache = null, |
84
|
|
|
?LoggerInterface $logger = null |
85
|
|
|
) { |
86
|
1 |
|
$this->defaultCacheFolder = $defaultCacheFolder; |
87
|
1 |
|
$this->cache = $cache; |
88
|
1 |
|
$this->logger = $logger; |
89
|
|
|
|
90
|
1 |
|
parent::__construct(); |
91
|
1 |
|
} |
92
|
|
|
|
93
|
1 |
|
protected function configure() : void |
94
|
|
|
{ |
95
|
|
|
$this |
96
|
1 |
|
->setName('browscap:log') |
97
|
1 |
|
->setDescription('Parses the supplied webserver log file.') |
98
|
1 |
|
->addArgument( |
99
|
1 |
|
'output', |
100
|
1 |
|
InputArgument::REQUIRED, |
101
|
1 |
|
'Path to output log file', |
102
|
1 |
|
null |
103
|
|
|
) |
104
|
1 |
|
->addOption( |
105
|
1 |
|
'log-file', |
106
|
1 |
|
'f', |
107
|
1 |
|
InputOption::VALUE_REQUIRED, |
108
|
1 |
|
'Path to a webserver log file' |
109
|
|
|
) |
110
|
1 |
|
->addOption( |
111
|
1 |
|
'log-dir', |
112
|
1 |
|
'd', |
113
|
1 |
|
InputOption::VALUE_REQUIRED, |
114
|
1 |
|
'Path to webserver log directory' |
115
|
|
|
) |
116
|
1 |
|
->addOption( |
117
|
1 |
|
'include', |
118
|
1 |
|
'i', |
119
|
1 |
|
InputOption::VALUE_IS_ARRAY | InputOption::VALUE_REQUIRED, |
120
|
1 |
|
'Include glob expressions for log files in the log directory', |
121
|
1 |
|
['*.log', '*.log*.gz', '*.log*.bz2'] |
122
|
|
|
) |
123
|
1 |
|
->addOption( |
124
|
1 |
|
'exclude', |
125
|
1 |
|
'e', |
126
|
1 |
|
InputOption::VALUE_IS_ARRAY | InputOption::VALUE_REQUIRED, |
127
|
1 |
|
'Exclude glob expressions for log files in the log directory', |
128
|
1 |
|
['*error*'] |
129
|
|
|
) |
130
|
1 |
|
->addOption( |
131
|
1 |
|
'debug', |
132
|
1 |
|
null, |
133
|
1 |
|
InputOption::VALUE_NONE, |
134
|
1 |
|
'Should the debug mode entered?' |
135
|
|
|
) |
136
|
1 |
|
->addOption( |
137
|
1 |
|
'cache', |
138
|
1 |
|
'c', |
139
|
1 |
|
InputOption::VALUE_OPTIONAL, |
140
|
1 |
|
'Where the cache files are located', |
141
|
1 |
|
$this->defaultCacheFolder |
142
|
|
|
); |
143
|
1 |
|
} |
144
|
|
|
|
145
|
|
|
protected function execute(InputInterface $input, OutputInterface $output) : void |
146
|
|
|
{ |
147
|
|
|
if (! $input->getOption('log-file') && ! $input->getOption('log-dir')) { |
148
|
|
|
throw InvalidArgumentException::oneOfCommandArguments('log-file', 'log-dir'); |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
$logger = $this->getLogger($input); |
152
|
|
|
|
153
|
|
|
$browscap = new Browscap($this->getCache($input), $logger); |
154
|
|
|
$collection = ReaderFactory::factory(); |
155
|
|
|
$fs = new Filesystem(); |
156
|
|
|
|
157
|
|
|
/** @var $file \Symfony\Component\Finder\SplFileInfo */ |
158
|
|
|
foreach ($this->getFiles($input) as $file) { |
159
|
|
|
$this->uas = []; |
160
|
|
|
$path = $this->getPath($file); |
161
|
|
|
|
162
|
|
|
$this->countOk = 0; |
163
|
|
|
$this->countNok = 0; |
164
|
|
|
|
165
|
|
|
$logger->info('Analyzing file "' . $file->getPathname() . '"'); |
166
|
|
|
|
167
|
|
|
$lines = file($path); |
168
|
|
|
|
169
|
|
|
if (empty($lines)) { |
170
|
|
|
$logger->info('Skipping empty file "' . $file->getPathname() . '"'); |
171
|
|
|
continue; |
172
|
|
|
} |
173
|
|
|
|
174
|
|
|
$this->totalCount = count($lines); |
175
|
|
|
|
176
|
|
|
foreach ($lines as $line) { |
177
|
|
|
$this->handleLine( |
178
|
|
|
$output, |
179
|
|
|
$collection, |
180
|
|
|
$browscap, |
181
|
|
|
$line |
182
|
|
|
); |
183
|
|
|
} |
184
|
|
|
|
185
|
|
|
$this->outputProgress($output, '', true); |
186
|
|
|
|
187
|
|
|
arsort($this->uas, SORT_NUMERIC); |
188
|
|
|
|
189
|
|
|
try { |
190
|
|
|
$fs->dumpFile( |
191
|
|
|
$input->getArgument('output') . '/output.txt', |
192
|
|
|
implode(PHP_EOL, array_unique($this->undefinedClients)) |
193
|
|
|
); |
194
|
|
|
} catch (IOException $e) { |
195
|
|
|
// do nothing |
196
|
|
|
} |
197
|
|
|
|
198
|
|
|
try { |
199
|
|
|
$fs->dumpFile( |
200
|
|
|
$input->getArgument('output') . '/output-with-amount.txt', |
201
|
|
|
$this->createAmountContent() |
202
|
|
|
); |
203
|
|
|
} catch (IOException $e) { |
204
|
|
|
// do nothing |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
try { |
208
|
|
|
$fs->dumpFile( |
209
|
|
|
$input->getArgument('output') . '/output-with-amount-and-type.txt', |
210
|
|
|
$this->createAmountTypeContent() |
211
|
|
|
); |
212
|
|
|
} catch (IOException $e) { |
213
|
|
|
// do nothing |
214
|
|
|
} |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
$outputFile = $input->getArgument('output') . '/output.txt'; |
218
|
|
|
|
219
|
|
|
try { |
220
|
|
|
$fs->dumpFile( |
221
|
|
|
$outputFile, |
222
|
|
|
implode(PHP_EOL, array_unique($this->undefinedClients)) |
223
|
|
|
); |
224
|
|
|
} catch (IOException $e) { |
225
|
|
|
throw new \UnexpectedValueException('writing to file "' . $outputFile . '" failed', 0, $e); |
226
|
|
|
} |
227
|
|
|
|
228
|
|
|
try { |
229
|
|
|
$fs->dumpFile( |
230
|
|
|
$input->getArgument('output') . '/output-with-amount.txt', |
231
|
|
|
$this->createAmountContent() |
232
|
|
|
); |
233
|
|
|
} catch (IOException $e) { |
234
|
|
|
// do nothing |
235
|
|
|
} |
236
|
|
|
|
237
|
|
|
try { |
238
|
|
|
$fs->dumpFile( |
239
|
|
|
$input->getArgument('output') . '/output-with-amount-and-type.txt', |
240
|
|
|
$this->createAmountTypeContent() |
241
|
|
|
); |
242
|
|
|
} catch (IOException $e) { |
243
|
|
|
// do nothing |
244
|
|
|
} |
245
|
|
|
} |
246
|
|
|
|
247
|
|
|
private function createAmountContent() : string |
248
|
|
|
{ |
249
|
|
|
$counts = []; |
250
|
|
|
|
251
|
|
|
foreach ($this->uasWithType as $uas) { |
252
|
|
|
foreach ($uas as $userAgentString => $count) { |
253
|
|
|
if (isset($counts[$userAgentString])) { |
254
|
|
|
$counts[$userAgentString] += $count; |
255
|
|
|
} else { |
256
|
|
|
$counts[$userAgentString] = $count; |
257
|
|
|
} |
258
|
|
|
} |
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
$content = ''; |
262
|
|
|
|
263
|
|
|
arsort($counts, SORT_NUMERIC); |
264
|
|
|
|
265
|
|
|
foreach ($counts as $agentOfLine => $count) { |
266
|
|
|
$content .= "$count\t$agentOfLine\n"; |
267
|
|
|
} |
268
|
|
|
|
269
|
|
|
return $content; |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
private function createAmountTypeContent() : string |
273
|
|
|
{ |
274
|
|
|
$content = ''; |
275
|
|
|
$types = ['B', 'T', 'P', 'D', 'N', 'U']; |
276
|
|
|
|
277
|
|
|
foreach ($types as $type) { |
278
|
|
|
if (! isset($this->uasWithType[$type])) { |
279
|
|
|
continue; |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
arsort($this->uasWithType[$type], SORT_NUMERIC); |
283
|
|
|
|
284
|
|
|
foreach ($this->uasWithType[$type] as $agentOfLine => $count) { |
285
|
|
|
$content .= "$type\t$count\t$agentOfLine\n"; |
286
|
|
|
} |
287
|
|
|
} |
288
|
|
|
|
289
|
|
|
return $content; |
290
|
|
|
} |
291
|
|
|
|
292
|
|
|
private function handleLine( |
293
|
|
|
OutputInterface $output, |
294
|
|
|
ReaderCollection $collection, |
295
|
|
|
Browscap $browscap, |
296
|
|
|
string $line |
297
|
|
|
) : void { |
298
|
|
|
$userAgentString = ''; |
299
|
|
|
|
300
|
|
|
try { |
301
|
|
|
$userAgentString = $collection->read($line); |
302
|
|
|
|
303
|
|
|
try { |
304
|
|
|
$this->getResult($browscap->getBrowser($userAgentString)); |
305
|
|
|
} catch (\Exception $e) { |
306
|
|
|
$this->undefinedClients[] = $userAgentString; |
307
|
|
|
|
308
|
|
|
throw $e; |
309
|
|
|
} |
310
|
|
|
|
311
|
|
|
$type = '.'; |
312
|
|
|
++$this->countOk; |
313
|
|
|
} catch (ReaderException $e) { |
314
|
|
|
$type = 'E'; |
315
|
|
|
++$this->countNok; |
316
|
|
|
} catch (UnknownBrowserTypeException $e) { |
317
|
|
|
$type = 'T'; |
318
|
|
|
++$this->countNok; |
319
|
|
|
} catch (UnknownBrowserException $e) { |
320
|
|
|
$type = 'B'; |
321
|
|
|
++$this->countNok; |
322
|
|
|
} catch (UnknownPlatformException $e) { |
323
|
|
|
$type = 'P'; |
324
|
|
|
++$this->countNok; |
325
|
|
|
} catch (UnknownDeviceException $e) { |
326
|
|
|
$type = 'D'; |
327
|
|
|
++$this->countNok; |
328
|
|
|
} catch (UnknownEngineException $e) { |
329
|
|
|
$type = 'N'; |
330
|
|
|
++$this->countNok; |
331
|
|
|
} catch (\Exception $e) { |
332
|
|
|
$type = 'U'; |
333
|
|
|
++$this->countNok; |
334
|
|
|
} |
335
|
|
|
|
336
|
|
|
$this->outputProgress($output, $type); |
337
|
|
|
|
338
|
|
|
// count all useragents |
339
|
|
|
if (isset($this->uas[$userAgentString])) { |
340
|
|
|
++$this->uas[$userAgentString]; |
341
|
|
|
} else { |
342
|
|
|
$this->uas[$userAgentString] = 1; |
343
|
|
|
} |
344
|
|
|
|
345
|
|
|
if ('.' !== $type && 'E' !== $type) { |
346
|
|
|
// count all undetected useragents grouped by detection error |
347
|
|
|
if (! isset($this->uasWithType[$type])) { |
348
|
|
|
$this->uasWithType[$type] = []; |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
if (isset($this->uasWithType[$type][$userAgentString])) { |
352
|
|
|
++$this->uasWithType[$type][$userAgentString]; |
353
|
|
|
} else { |
354
|
|
|
$this->uasWithType[$type][$userAgentString] = 1; |
355
|
|
|
} |
356
|
|
|
} |
357
|
|
|
} |
358
|
|
|
|
359
|
|
|
private function outputProgress(OutputInterface $output, string $result, bool $end = false) : void |
360
|
|
|
{ |
361
|
|
|
if (($this->totalCount % 70) === 0 || $end) { |
362
|
|
|
$formatString = ' %' . strlen($this->countOk) . 'd OK, %' . strlen($this->countNok) . 'd NOK, Summary %' |
363
|
|
|
. strlen($this->totalCount) . 'd'; |
364
|
|
|
|
365
|
|
|
if ($end) { |
366
|
|
|
$result = str_pad($result, 70 - ($this->totalCount % 70), ' ', STR_PAD_RIGHT); |
367
|
|
|
} |
368
|
|
|
|
369
|
|
|
$endString = sprintf($formatString, $this->countOk, $this->countNok, $this->totalCount); |
370
|
|
|
|
371
|
|
|
$output->writeln($result . $endString); |
372
|
|
|
|
373
|
|
|
return; |
374
|
|
|
} |
375
|
|
|
|
376
|
|
|
$output->write($result); |
377
|
|
|
} |
378
|
|
|
|
379
|
|
|
private function getResult(\stdClass $result) : string |
380
|
|
|
{ |
381
|
|
|
if ('Default Browser' === $result->browser) { |
382
|
|
|
throw new UnknownBrowserException('Unknown browser found'); |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
if ('unknown' === $result->browser_type) { |
386
|
|
|
throw new UnknownBrowserTypeException('Unknown browser type found'); |
387
|
|
|
} |
388
|
|
|
|
389
|
|
|
if (in_array($result->browser_type, ['Bot/Crawler', 'Library'])) { |
390
|
|
|
return '.'; |
391
|
|
|
} |
392
|
|
|
|
393
|
|
|
if ('unknown' === $result->platform) { |
394
|
|
|
throw new UnknownPlatformException('Unknown platform found'); |
395
|
|
|
} |
396
|
|
|
|
397
|
|
|
if ('unknown' === $result->device_type) { |
398
|
|
|
throw new UnknownDeviceException('Unknown device type found'); |
399
|
|
|
} |
400
|
|
|
|
401
|
|
|
if ('unknown' === $result->renderingengine_name) { |
402
|
|
|
throw new UnknownEngineException('Unknown rendering engine found'); |
403
|
|
|
} |
404
|
|
|
|
405
|
|
|
return '.'; |
406
|
|
|
} |
407
|
|
|
|
408
|
|
|
private function getFiles(InputInterface $input) : Finder |
409
|
|
|
{ |
410
|
|
|
$finder = Finder::create(); |
411
|
|
|
|
412
|
|
|
if ($input->getOption('log-file')) { |
413
|
|
|
$file = $input->getOption('log-file'); |
414
|
|
|
$finder->append(Finder::create()->in(dirname($file))->name(basename($file))); |
415
|
|
|
} |
416
|
|
|
|
417
|
|
|
if ($input->getOption('log-dir')) { |
418
|
|
|
$dirFinder = Finder::create() |
419
|
|
|
->in($input->getOption('log-dir')); |
420
|
|
|
array_map([$dirFinder, 'name'], $input->getOption('include')); |
421
|
|
|
array_map([$dirFinder, 'notName'], $input->getOption('exclude')); |
422
|
|
|
|
423
|
|
|
$finder->append($dirFinder); |
424
|
|
|
} |
425
|
|
|
|
426
|
|
|
return $finder; |
427
|
|
|
} |
428
|
|
|
|
429
|
|
|
private function getPath(SplFileInfo $file) : string |
430
|
|
|
{ |
431
|
|
|
switch ($file->getExtension()) { |
432
|
|
|
case 'gz': |
433
|
|
|
$path = 'compress.zlib://' . $file->getPathname(); |
434
|
|
|
break; |
435
|
|
|
case 'bz2': |
436
|
|
|
$path = 'compress.bzip2://' . $file->getPathname(); |
437
|
|
|
break; |
438
|
|
|
default: |
439
|
|
|
$path = $file->getPathname(); |
440
|
|
|
break; |
441
|
|
|
} |
442
|
|
|
|
443
|
|
|
return $path; |
444
|
|
|
} |
445
|
|
|
|
446
|
|
|
private function getCache(InputInterface $input) : CacheInterface |
447
|
|
|
{ |
448
|
|
|
if (null === $this->cache) { |
449
|
|
|
$fileCache = new FilesystemCache($input->getOption('cache')); |
450
|
|
|
$this->cache = new SimpleCacheAdapter($fileCache); |
451
|
|
|
} |
452
|
|
|
|
453
|
|
|
return $this->cache; |
454
|
|
|
} |
455
|
|
|
|
456
|
|
|
private function getLogger(InputInterface $input) : LoggerInterface |
457
|
|
|
{ |
458
|
|
|
if (null === $this->logger) { |
459
|
|
|
$this->logger = LoggerHelper::createDefaultLogger($input->getOption('debug')); |
460
|
|
|
} |
461
|
|
|
|
462
|
|
|
return $this->logger; |
463
|
|
|
} |
464
|
|
|
} |
465
|
|
|
|