|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Vaites\ApacheTika\Clients; |
|
4
|
|
|
|
|
5
|
|
|
use Exception; |
|
6
|
|
|
|
|
7
|
|
|
use Vaites\ApacheTika\Client; |
|
8
|
|
|
use Vaites\ApacheTika\Metadata\Metadata; |
|
9
|
|
|
|
|
10
|
|
|
/** |
|
11
|
|
|
* Apache Tika command line interface client |
|
12
|
|
|
* |
|
13
|
|
|
* @author David Martínez <[email protected]> |
|
14
|
|
|
* @link http://wiki.apache.org/tika/TikaJAXRS |
|
15
|
|
|
* @link https://tika.apache.org/1.12/formats.html |
|
16
|
|
|
*/ |
|
17
|
|
|
class CLIClient extends Client |
|
18
|
|
|
{ |
|
19
|
|
|
const MODE = 'cli'; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Apache Tika app path |
|
23
|
|
|
* |
|
24
|
|
|
* @var string |
|
25
|
|
|
*/ |
|
26
|
|
|
protected $path = null; |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Java binary path |
|
30
|
|
|
* |
|
31
|
|
|
* @var string |
|
32
|
|
|
*/ |
|
33
|
|
|
protected $java = null; |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* Configure client |
|
37
|
|
|
* |
|
38
|
|
|
* @param string $path |
|
39
|
|
|
* @param string $java |
|
40
|
|
|
* |
|
41
|
|
|
* @throws Exception |
|
42
|
|
|
*/ |
|
43
|
|
|
public function __construct($path = null, $java = null) |
|
44
|
|
|
{ |
|
45
|
|
|
if($path) |
|
46
|
|
|
{ |
|
47
|
|
|
$this->setPath($path); |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
if($java) |
|
51
|
|
|
{ |
|
52
|
|
|
$this->setJava($java); |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
if(self::$check == true) |
|
|
|
|
|
|
56
|
|
|
{ |
|
57
|
|
|
$this->check(); |
|
58
|
|
|
} |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* Get the path |
|
63
|
|
|
* |
|
64
|
|
|
* @return null|string |
|
65
|
|
|
*/ |
|
66
|
|
|
public function getPath() |
|
67
|
|
|
{ |
|
68
|
|
|
return $this->path; |
|
69
|
|
|
} |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* Set the path |
|
73
|
|
|
* |
|
74
|
|
|
* @param string $path |
|
75
|
|
|
* @return $this |
|
76
|
|
|
*/ |
|
77
|
|
|
public function setPath($path) |
|
78
|
|
|
{ |
|
79
|
|
|
$this->path = $path; |
|
80
|
|
|
|
|
81
|
|
|
return $this; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
/** |
|
85
|
|
|
* Get the Java path |
|
86
|
|
|
* |
|
87
|
|
|
* @return null|int |
|
88
|
|
|
*/ |
|
89
|
|
|
public function getJava() |
|
90
|
|
|
{ |
|
91
|
|
|
return $this->java; |
|
|
|
|
|
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
/** |
|
95
|
|
|
* Set the Java path |
|
96
|
|
|
* |
|
97
|
|
|
* @param string $java |
|
98
|
|
|
* @return $this |
|
99
|
|
|
*/ |
|
100
|
|
|
public function setJava($java) |
|
101
|
|
|
{ |
|
102
|
|
|
$this->java = $java; |
|
103
|
|
|
|
|
104
|
|
|
return $this; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
/** |
|
108
|
|
|
* Check Java binary, JAR path or server connection |
|
109
|
|
|
* |
|
110
|
|
|
* @return void |
|
111
|
|
|
* @throws \Exception |
|
112
|
|
|
*/ |
|
113
|
|
|
public function check() |
|
114
|
|
|
{ |
|
115
|
|
|
if(self::$checked == false) |
|
|
|
|
|
|
116
|
|
|
{ |
|
117
|
|
|
// Java command must not return an error |
|
118
|
|
|
exec(($this->java ?: 'java') . ' -version 2> /dev/null', $output, $return); |
|
119
|
|
|
if($return != 0) |
|
120
|
|
|
{ |
|
121
|
|
|
throw new Exception('Java command not found'); |
|
122
|
|
|
} |
|
123
|
|
|
// JAR path must exists |
|
124
|
|
|
elseif(file_exists($this->path) == false) |
|
|
|
|
|
|
125
|
|
|
{ |
|
126
|
|
|
throw new Exception('Apache Tika app JAR not found'); |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
self::$checked = true; |
|
130
|
|
|
} |
|
131
|
|
|
} |
|
132
|
|
|
|
|
133
|
|
|
/** |
|
134
|
|
|
* Configure and make a request and return its results |
|
135
|
|
|
* |
|
136
|
|
|
* @param string $type |
|
137
|
|
|
* @param string $file |
|
138
|
|
|
* @return string |
|
139
|
|
|
* @throws \Exception |
|
140
|
|
|
*/ |
|
141
|
|
|
public function request($type, $file = null) |
|
142
|
|
|
{ |
|
143
|
|
|
// check if not checked |
|
144
|
|
|
$this->check(); |
|
145
|
|
|
|
|
146
|
|
|
// check if is cached |
|
147
|
|
|
if(isset($this->cache[sha1($file)][$type])) |
|
148
|
|
|
{ |
|
149
|
|
|
return $this->cache[sha1($file)][$type]; |
|
150
|
|
|
} |
|
151
|
|
|
|
|
152
|
|
|
// command arguments |
|
153
|
|
|
$arguments = $this->getArguments($type, $file); |
|
154
|
|
|
|
|
155
|
|
|
// check the request |
|
156
|
|
|
$file = parent::checkRequest($type, $file); |
|
157
|
|
|
|
|
158
|
|
|
// add last argument |
|
159
|
|
|
if($file) |
|
160
|
|
|
{ |
|
161
|
|
|
$arguments[] = escapeshellarg($file); |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
// build command |
|
165
|
|
|
$jar = escapeshellarg($this->path); |
|
166
|
|
|
$command = ($this->java ?: 'java') . " -jar $jar " . implode(' ', $arguments); |
|
167
|
|
|
|
|
168
|
|
|
// run command |
|
169
|
|
|
$response = $this->exec($command); |
|
170
|
|
|
|
|
171
|
|
|
// metadata response |
|
172
|
|
|
if($type == 'meta') |
|
173
|
|
|
{ |
|
174
|
|
|
// fix for invalid? json returned only with images |
|
175
|
|
|
$response = str_replace(basename($file) . '"}{', '", ', $response); |
|
176
|
|
|
|
|
177
|
|
|
// on Windows, response comes in another charset |
|
178
|
|
|
if(defined('PHP_WINDOWS_VERSION_MAJOR')) |
|
179
|
|
|
{ |
|
180
|
|
|
$response = utf8_encode($response); |
|
181
|
|
|
} |
|
182
|
|
|
|
|
183
|
|
|
$response = Metadata::make($response, $file); |
|
184
|
|
|
} |
|
185
|
|
|
|
|
186
|
|
|
// cache certain responses |
|
187
|
|
|
if(in_array($type, ['lang', 'meta'])) |
|
188
|
|
|
{ |
|
189
|
|
|
$this->cache[sha1($file)][$type] = $response; |
|
190
|
|
|
} |
|
191
|
|
|
|
|
192
|
|
|
return $response; |
|
|
|
|
|
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
/** |
|
196
|
|
|
* Run the command and return its results |
|
197
|
|
|
* |
|
198
|
|
|
* @param string $command |
|
199
|
|
|
* @return null|string |
|
200
|
|
|
* @throws \Exception |
|
201
|
|
|
*/ |
|
202
|
|
|
public function exec($command) |
|
203
|
|
|
{ |
|
204
|
|
|
// run command |
|
205
|
|
|
$exit = -1; |
|
206
|
|
|
$logfile = sys_get_temp_dir() . DIRECTORY_SEPARATOR . 'tika-error.log'; |
|
207
|
|
|
$descriptors = [['pipe', 'r'], ['pipe', 'w'], ['file', $logfile, 'a']]; |
|
208
|
|
|
$process = proc_open($command, $descriptors, $pipes); |
|
209
|
|
|
$callback = $this->callback; |
|
210
|
|
|
|
|
211
|
|
|
// get output if command runs ok |
|
212
|
|
|
if(is_resource($process)) |
|
213
|
|
|
{ |
|
214
|
|
|
fclose($pipes[0]); |
|
215
|
|
|
$this->response = ''; |
|
216
|
|
|
while($chunk = stream_get_line($pipes[1], $this->chunkSize)) |
|
217
|
|
|
{ |
|
218
|
|
|
if(!is_null($callback)) |
|
219
|
|
|
{ |
|
220
|
|
|
$callback($chunk); |
|
221
|
|
|
} |
|
222
|
|
|
|
|
223
|
|
|
$this->response .= $chunk; |
|
224
|
|
|
} |
|
225
|
|
|
fclose($pipes[1]); |
|
226
|
|
|
$exit = proc_close($process); |
|
227
|
|
|
} |
|
228
|
|
|
|
|
229
|
|
|
// exception if exit value is not zero |
|
230
|
|
|
if($exit > 0) |
|
231
|
|
|
{ |
|
232
|
|
|
throw new Exception("Unexpected exit value ($exit) for command $command"); |
|
233
|
|
|
} |
|
234
|
|
|
|
|
235
|
|
|
return trim($this->response); |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
|
|
/** |
|
239
|
|
|
* Get the arguments to run the command |
|
240
|
|
|
* |
|
241
|
|
|
* @param string $type |
|
242
|
|
|
* @param string $file |
|
243
|
|
|
* @return array |
|
244
|
|
|
* @throws Exception |
|
245
|
|
|
*/ |
|
246
|
|
|
protected function getArguments($type, $file = null) |
|
|
|
|
|
|
247
|
|
|
{ |
|
248
|
|
|
// parameters for command |
|
249
|
|
|
$arguments = []; |
|
250
|
|
|
switch($type) |
|
251
|
|
|
{ |
|
252
|
|
|
case 'html': |
|
253
|
|
|
$arguments[] = '--html'; |
|
254
|
|
|
break; |
|
255
|
|
|
|
|
256
|
|
|
case 'lang': |
|
257
|
|
|
$arguments[] = '--language'; |
|
258
|
|
|
break; |
|
259
|
|
|
|
|
260
|
|
|
case 'mime': |
|
261
|
|
|
$arguments[] = '--detect'; |
|
262
|
|
|
break; |
|
263
|
|
|
|
|
264
|
|
|
case 'meta': |
|
265
|
|
|
$arguments[] = '--metadata --json'; |
|
266
|
|
|
break; |
|
267
|
|
|
|
|
268
|
|
|
case 'text': |
|
269
|
|
|
$arguments[] = '--text'; |
|
270
|
|
|
break; |
|
271
|
|
|
|
|
272
|
|
|
case 'text-main': |
|
273
|
|
|
$arguments[] = '--text-main'; |
|
274
|
|
|
break; |
|
275
|
|
|
|
|
276
|
|
|
case 'mime-types': |
|
277
|
|
|
$arguments[] = '--list-supported-types'; |
|
278
|
|
|
break; |
|
279
|
|
|
|
|
280
|
|
|
case 'detectors': |
|
281
|
|
|
$arguments[] = '--list-detectors'; |
|
282
|
|
|
break; |
|
283
|
|
|
|
|
284
|
|
|
case 'parsers': |
|
285
|
|
|
$arguments[] = '--list-parsers'; |
|
286
|
|
|
break; |
|
287
|
|
|
|
|
288
|
|
|
case 'version': |
|
289
|
|
|
$arguments[] = '--version'; |
|
290
|
|
|
break; |
|
291
|
|
|
|
|
292
|
|
|
default: |
|
293
|
|
|
throw new Exception("Unknown type $type"); |
|
294
|
|
|
} |
|
295
|
|
|
|
|
296
|
|
|
return $arguments; |
|
297
|
|
|
} |
|
298
|
|
|
} |
|
299
|
|
|
|
vaites /
php-apache-tika
Loading history...
When comparing two booleans, it is generally considered safer to use the strict comparison operator.