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