1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Geekwright\Po; |
4
|
|
|
|
5
|
|
|
use Geekwright\Po\Exceptions\UnrecognizedInputException; |
6
|
|
|
use Geekwright\Po\Exceptions\FileNotReadableException; |
7
|
|
|
use Geekwright\Po\Exceptions\FileNotWritableException; |
8
|
|
|
|
9
|
|
|
/** |
10
|
|
|
* PoFile - represent all entries in a GNU gettext style PO or POT file as a |
11
|
|
|
* collection of PoHeader and PoEntry objects. |
12
|
|
|
* |
13
|
|
|
* @category File |
14
|
|
|
* @package Po |
15
|
|
|
* @author Richard Griffith <[email protected]> |
16
|
|
|
* @copyright 2015-2018 Richard Griffith |
17
|
|
|
* @license GNU GPL 2 or later (http://www.gnu.org/licenses/gpl-2.0.html) |
18
|
|
|
* @link https://github.com/geekwright/Po |
19
|
|
|
*/ |
20
|
|
|
class PoFile |
21
|
|
|
{ |
22
|
|
|
/** |
23
|
|
|
* @var PoHeader $header |
24
|
|
|
*/ |
25
|
|
|
protected $header = null; |
26
|
|
|
|
27
|
|
|
/** |
28
|
|
|
* @var PoEntry[] $entries |
29
|
|
|
*/ |
30
|
|
|
protected $entries = array(); |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @var PoEntry[] $unkeyedEntries |
34
|
|
|
*/ |
35
|
|
|
protected $unkeyedEntries = array(); |
36
|
|
|
|
37
|
|
|
/** |
38
|
|
|
* $var array() $unrecognizedInput |
39
|
|
|
* |
40
|
|
|
* If any lines that cannot be processed are found when reading a po file, the |
41
|
|
|
* unrecognized input will be recorded here, and an exception will be thrown. |
42
|
|
|
* No interface is supplied, but this debug data is an array in the form: |
43
|
|
|
* line number => input line |
44
|
|
|
*/ |
45
|
|
|
public $unrecognizedInput = array(); |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* Build a PoFile, empty or with provided entries |
49
|
|
|
* |
50
|
|
|
* @param PoHeader|null $header header object |
51
|
|
|
* @param PoEntry[] $entries associative array po entries |
52
|
|
|
* @param PoEntry[] $unkeyedEntries indexed array of po entries. Unkeyed entries |
53
|
|
|
* are usually comment only entries, such as for |
54
|
|
|
* obsolete entries. |
55
|
|
|
*/ |
56
|
39 |
|
public function __construct(?array $header = null, array $entries = array(), array $unkeyedEntries = array()) |
57
|
|
|
{ |
58
|
39 |
|
$this->header = $header; |
59
|
39 |
|
$this->entries = $entries; |
60
|
39 |
|
$this->unkeyedEntries = $unkeyedEntries; |
61
|
39 |
|
} |
62
|
|
|
|
63
|
|
|
/** |
64
|
|
|
* Build the internal entries array key from id, context and plural id |
65
|
|
|
* |
66
|
|
|
* @param string|null $msgid the untranslated message of the entry |
67
|
|
|
* @param string|null $msgctxt the context of the entry, if any |
68
|
|
|
* @param string|null $msgid_plural the untranslated plural message of the entry, if any |
69
|
|
|
* |
70
|
|
|
* @return string |
71
|
|
|
*/ |
72
|
18 |
|
public static function createKey(?string $msgid, ?string $msgctxt = null, ?string $msgid_plural = null): string |
73
|
|
|
{ |
74
|
18 |
|
$key = ''; |
75
|
18 |
|
if (!empty($msgctxt)) { |
76
|
4 |
|
$key .= $msgctxt . '|'; |
77
|
|
|
} |
78
|
18 |
|
$key .= (string) $msgid; |
79
|
18 |
|
if (!empty($msgid_plural)) { |
80
|
10 |
|
$key .= '|' . $msgid_plural; |
81
|
|
|
} |
82
|
18 |
|
return $key; |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Build an internal entries array key from a PoEntry |
87
|
|
|
* |
88
|
|
|
* @param PoEntry $entry the PoEntry to build key from |
89
|
|
|
* |
90
|
|
|
* @return string |
91
|
|
|
*/ |
92
|
17 |
|
public function createKeyFromEntry(PoEntry $entry): string |
93
|
|
|
{ |
94
|
17 |
|
return $this->createKey( |
95
|
17 |
|
$entry->getAsString(PoTokens::MESSAGE), |
96
|
17 |
|
$entry->getAsString(PoTokens::CONTEXT), |
97
|
17 |
|
$entry->getAsString(PoTokens::PLURAL) |
98
|
|
|
); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* Replace any existing header with the provided PoHeader |
103
|
|
|
* |
104
|
|
|
* @param PoHeader $header header object |
105
|
|
|
* |
106
|
|
|
* @return void |
107
|
|
|
*/ |
108
|
8 |
|
public function setHeaderEntry(PoHeader $header): void |
109
|
|
|
{ |
110
|
8 |
|
$this->header = $header; |
111
|
8 |
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Get the current header entry |
115
|
|
|
* |
116
|
|
|
* @return PoHeader |
117
|
|
|
*/ |
118
|
2 |
|
public function getHeaderEntry(): PoHeader |
119
|
|
|
{ |
120
|
2 |
|
return $this->header; |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
/** |
124
|
|
|
* Get an array of current entries |
125
|
|
|
* |
126
|
|
|
* @return PoEntry[] |
127
|
|
|
*/ |
128
|
11 |
|
public function getEntries(): array |
129
|
|
|
{ |
130
|
11 |
|
return $this->entries; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Replace any existing unkeyedEntries with new array of PoEntry objects |
135
|
|
|
* |
136
|
|
|
* @param PoEntry[] $entries po entries |
137
|
|
|
* |
138
|
|
|
* @return void |
139
|
|
|
*/ |
140
|
1 |
|
public function setUnkeyedEntries(array $entries): void |
141
|
|
|
{ |
142
|
1 |
|
$this->unkeyedEntries = $entries; |
143
|
1 |
|
} |
144
|
|
|
|
145
|
|
|
/** |
146
|
|
|
* Get current array of unkeyed PoEntry objects |
147
|
|
|
* |
148
|
|
|
* @return PoEntry[] |
149
|
|
|
*/ |
150
|
2 |
|
public function getUnkeyedEntries(): array |
151
|
|
|
{ |
152
|
2 |
|
return $this->unkeyedEntries; |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
/** |
156
|
|
|
* Add an entry to the PoFile using internal key |
157
|
|
|
* |
158
|
|
|
* @param PoEntry $entry the PoEntry to add |
159
|
|
|
* @param boolean $replace true to replace any existing entry matching this key, |
160
|
|
|
* false to not change the PoFile for a duplicated key |
161
|
|
|
* |
162
|
|
|
* @return boolean true if added, false if not |
163
|
|
|
*/ |
164
|
9 |
|
public function addEntry(PoEntry $entry, bool $replace = true): bool |
165
|
|
|
{ |
166
|
9 |
|
$key = $this->createKeyFromEntry($entry); |
167
|
|
|
|
168
|
|
|
// some entires, such as obsolete entries, have no key |
169
|
|
|
// for some uses, these are dead weight - need better strategy for that case |
170
|
9 |
|
if (empty($key)) { |
171
|
4 |
|
$this->unkeyedEntries[] = $entry; |
172
|
4 |
|
return true; |
173
|
|
|
} |
174
|
|
|
|
175
|
7 |
|
if (isset($this->entries[$key]) && !$replace) { |
176
|
1 |
|
return false; |
177
|
|
|
} else { |
178
|
7 |
|
$this->entries[$key] = $entry; |
179
|
7 |
|
return true; |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
/** |
184
|
|
|
* Merge an entry with any existing entry with the same key. If the key does |
185
|
|
|
* not exist, add the entry, otherwise merge comments, references, and flags. |
186
|
|
|
* |
187
|
|
|
* This is intended for use in building a POT, where the handling of translated |
188
|
|
|
* strings is not a factor. |
189
|
|
|
* |
190
|
|
|
* @param PoEntry $newEntry the PoEntry to merge |
191
|
|
|
* |
192
|
|
|
* @return boolean true if merged or added, false if not |
193
|
|
|
*/ |
194
|
7 |
|
public function mergeEntry(PoEntry $newEntry): bool |
195
|
|
|
{ |
196
|
7 |
|
$key = $this->createKeyFromEntry($newEntry); |
197
|
|
|
|
198
|
|
|
// keyed entries only |
199
|
7 |
|
if (empty($key)) { |
200
|
1 |
|
return false; |
201
|
|
|
} |
202
|
|
|
|
203
|
7 |
|
if (isset($this->entries[$key])) { |
204
|
2 |
|
$existingEntry = $this->entries[$key]; |
205
|
2 |
|
$mergeTokens = array(PoTokens::REFERENCE, PoTokens::EXTRACTED_COMMENTS); |
206
|
2 |
|
foreach ($mergeTokens as $type) { |
207
|
2 |
|
$toMerge = $newEntry->get($type); |
208
|
2 |
|
if (!empty($toMerge)) { |
209
|
2 |
|
$toMerge = is_array($toMerge) ? $toMerge : array($toMerge); |
210
|
2 |
|
foreach ($toMerge as $value) { |
211
|
2 |
|
$existingEntry->add($type, $value); |
212
|
|
|
} |
213
|
|
|
} |
214
|
|
|
} |
215
|
|
|
} else { |
216
|
7 |
|
$this->entries[$key] = $newEntry; |
217
|
|
|
} |
218
|
7 |
|
return true; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
/** |
222
|
|
|
* Get an entry based on key values - msgid, msgctxt and msgid_plural |
223
|
|
|
* |
224
|
|
|
* @param string $msgid the untranslated message of the entry |
225
|
|
|
* @param string|null $msgctxt the context of the entry, if any |
226
|
|
|
* @param string|null $msgid_plural the untranslated plural message of the entry, if any |
227
|
|
|
* |
228
|
|
|
* @return PoEntry|null matching entry, or null if not found |
229
|
|
|
*/ |
230
|
2 |
|
public function findEntry(string $msgid, ?string $msgctxt = null, ?string $msgid_plural = null): ?PoEntry |
231
|
|
|
{ |
232
|
2 |
|
$key = $this->createKey($msgid, $msgctxt, $msgid_plural); |
233
|
2 |
|
$entry = null; |
234
|
|
|
|
235
|
2 |
|
if (!empty($key) && isset($this->entries[$key])) { |
236
|
2 |
|
$entry = $this->entries[$key]; |
237
|
|
|
} |
238
|
|
|
|
239
|
2 |
|
return $entry; |
240
|
|
|
} |
241
|
|
|
|
242
|
|
|
/** |
243
|
|
|
* Remove an entry from the PoFile |
244
|
|
|
* |
245
|
|
|
* In simple cases, the entry can be found by key. There are several cases |
246
|
|
|
* where it is not that easy to locate the PoEntry to be removed: |
247
|
|
|
* - the PoEntry was altered, making the generated and stored key different |
248
|
|
|
* - the entry is not keyed and is in unkeyedEntries |
249
|
|
|
* |
250
|
|
|
* In any of these cases, we must loop thru the entry arrays looking for an |
251
|
|
|
* exact object match, so the cost of the remove goes up |
252
|
|
|
* |
253
|
|
|
* @param PoEntry $entry the PoEntry to merge |
254
|
|
|
* |
255
|
|
|
* @return boolean true if remove, false if not |
256
|
|
|
*/ |
257
|
1 |
|
public function removeEntry(PoEntry $entry): bool |
258
|
|
|
{ |
259
|
1 |
|
$key = $this->createKeyFromEntry($entry); |
260
|
|
|
|
261
|
|
|
// try by the key first. |
262
|
1 |
|
if (!empty($key) && isset($this->entries[$key])) { |
263
|
1 |
|
if ($entry === $this->entries[$key]) { |
264
|
1 |
|
unset($this->entries[$key]); |
265
|
1 |
|
return true; |
266
|
|
|
} |
267
|
|
|
} |
268
|
|
|
|
269
|
|
|
// the entry can't be matched by key, so we have to loop :( |
270
|
1 |
|
foreach ($this->entries as $key => $value) { |
271
|
1 |
|
if ($entry === $value) { |
272
|
1 |
|
unset($this->entries[$key]); |
273
|
1 |
|
return true; |
274
|
|
|
} |
275
|
|
|
} |
276
|
|
|
|
277
|
|
|
// no match found in main entries, try the unkeyedEntries |
278
|
1 |
|
foreach ($this->unkeyedEntries as $key => $value) { |
279
|
1 |
|
if ($entry === $value) { |
280
|
1 |
|
unset($this->unkeyedEntries[$key]); |
281
|
1 |
|
return true; |
282
|
|
|
} |
283
|
|
|
} |
284
|
|
|
|
285
|
1 |
|
return false; |
286
|
|
|
} |
287
|
|
|
|
288
|
|
|
/** |
289
|
|
|
* Write any current contents to a po file |
290
|
|
|
* |
291
|
|
|
* @param string $file po file to write |
292
|
|
|
* |
293
|
|
|
* @return void |
294
|
|
|
* |
295
|
|
|
* @throws FileNotWritableException |
296
|
|
|
*/ |
297
|
2 |
|
public function writePoFile(string $file): void |
298
|
|
|
{ |
299
|
2 |
|
$source = $this->dumpString(); |
300
|
2 |
|
$testName = file_exists($file) ? $file : dirname($file); |
301
|
2 |
|
$status = is_writable($testName); |
302
|
2 |
|
if ($status === true) { |
303
|
1 |
|
$status = file_put_contents($file, $source); |
304
|
|
|
} |
305
|
2 |
|
if (false === $status) { |
306
|
1 |
|
throw new FileNotWritableException($file); |
307
|
|
|
} |
308
|
1 |
|
} |
309
|
|
|
|
310
|
|
|
/** |
311
|
|
|
* Dump the current contents in PO format to a string |
312
|
|
|
* |
313
|
|
|
* @return string |
314
|
|
|
*/ |
315
|
5 |
|
public function dumpString(): string |
316
|
|
|
{ |
317
|
5 |
|
if ($this->header === null) { |
318
|
2 |
|
$this->header = new PoHeader; |
319
|
2 |
|
$this->header->buildDefaultHeader(); |
320
|
|
|
} |
321
|
5 |
|
$output = ''; |
322
|
|
|
|
323
|
5 |
|
$output .= $this->header->dumpEntry(); |
324
|
5 |
|
foreach ($this->entries as $entry) { |
325
|
3 |
|
$output .= $entry->dumpEntry(); |
326
|
|
|
} |
327
|
5 |
|
foreach ($this->unkeyedEntries as $entry) { |
328
|
2 |
|
$output .= $entry->dumpEntry(); |
329
|
|
|
} |
330
|
5 |
|
$output .= "\n"; |
331
|
|
|
|
332
|
5 |
|
return $output; |
333
|
|
|
} |
334
|
|
|
|
335
|
|
|
|
336
|
|
|
/** |
337
|
|
|
* Replace any current contents with entries from a file |
338
|
|
|
* |
339
|
|
|
* @param string $file po file/stream to read |
340
|
|
|
* @param resource|null $context context for stream if required |
341
|
|
|
* |
342
|
|
|
* @return void |
343
|
|
|
* |
344
|
|
|
* @throws FileNotReadableException |
345
|
|
|
*/ |
346
|
6 |
|
public function readPoFile(string $file, $context = null): void |
347
|
|
|
{ |
348
|
6 |
|
$oldEr = error_reporting(E_ALL ^ E_WARNING); |
349
|
6 |
|
$source = file_get_contents($file, false, $context); |
350
|
6 |
|
error_reporting($oldEr); |
351
|
6 |
|
if (false===$source) { |
352
|
1 |
|
throw new FileNotReadableException($file); |
353
|
|
|
} |
354
|
5 |
|
$this->parsePoSource($source); |
355
|
4 |
|
} |
356
|
|
|
|
357
|
|
|
/** |
358
|
|
|
* Replace any current contents with header and entries from PO souce string |
359
|
|
|
* |
360
|
|
|
* @param string $source po formatted string to parse |
361
|
|
|
* |
362
|
|
|
* @return void |
363
|
|
|
* |
364
|
|
|
* @throws UnrecognizedInputException |
365
|
|
|
*/ |
366
|
7 |
|
public function parsePoSource(string $source): void |
367
|
|
|
{ |
368
|
|
|
/** |
369
|
|
|
* This is an incredibly ugly regex pattern that breaks a line of a po file into |
370
|
|
|
* pieces that can be analyzed and acted upon. |
371
|
|
|
* |
372
|
|
|
* The matches array in preg_match will break out like this: |
373
|
|
|
* [0] full string |
374
|
|
|
* [1] mostly useless broad match of initial token, including trailing space |
375
|
|
|
* [2] bare token, or full msgstr[n] clause |
376
|
|
|
* [3] 'n' of a msgstr[n] line |
377
|
|
|
* [4] '"' if a data line |
378
|
|
|
* [5] remaining line |
379
|
|
|
* [6] a bare or malformed comment |
380
|
|
|
*/ |
381
|
7 |
|
$pattern = '/(^(#|#.|#;|#,|#\||msgid|msgid_plural|msgctxt|msgstr|msgstr\[([0-9]+)\])\s|(^"))(.+)|(^#.*)/'; |
382
|
|
|
|
383
|
7 |
|
$source_lines = explode("\n", $source); |
384
|
|
|
|
385
|
7 |
|
$wsBreak = false; |
386
|
7 |
|
$inHeader = true; |
387
|
7 |
|
$headerEntry = new PoHeader; |
388
|
7 |
|
$entry = $headerEntry; |
389
|
7 |
|
$unrecognized = array(); |
390
|
7 |
|
$lastKey = ''; |
391
|
7 |
|
$currentPlural = 0; |
392
|
7 |
|
foreach ($source_lines as $line => $s) { |
393
|
7 |
|
$result = preg_match($pattern, $s, $matches); |
394
|
7 |
|
if (!$result) { |
395
|
7 |
|
$lastKey = ''; |
396
|
7 |
|
if ($s=='' || ctype_space($s)) { |
397
|
7 |
|
if ($inHeader) { |
398
|
7 |
|
$this->setHeaderEntry($headerEntry); |
399
|
7 |
|
$entry = null; |
400
|
7 |
|
$inHeader = false; |
401
|
|
|
} |
402
|
7 |
|
if (!$wsBreak) { |
403
|
7 |
|
if (!($entry === null)) { |
404
|
6 |
|
$this->addEntry($entry); |
405
|
|
|
} |
406
|
7 |
|
$entry = null; |
407
|
7 |
|
$wsBreak=true; |
408
|
|
|
} |
409
|
|
|
} else { |
410
|
2 |
|
$wsBreak=false; |
411
|
7 |
|
$unrecognized[$line+1] = $s; |
412
|
|
|
} |
413
|
|
|
} else { |
414
|
6 |
|
if ($entry === null) { |
415
|
6 |
|
$entry = new PoEntry; |
416
|
|
|
} |
417
|
6 |
|
$wsBreak=false; |
418
|
6 |
|
$currentKey = $matches[2]; // will be used to set last key |
419
|
6 |
|
switch ($matches[2]) { |
420
|
6 |
|
case PoTokens::TRANSLATOR_COMMENTS: |
421
|
6 |
|
case PoTokens::EXTRACTED_COMMENTS: |
422
|
6 |
|
case PoTokens::REFERENCE: |
423
|
6 |
|
case PoTokens::FLAG: |
424
|
6 |
|
case PoTokens::OBSOLETE: |
425
|
6 |
|
case PoTokens::PREVIOUS: |
426
|
6 |
|
$entry->add($matches[2], $matches[5]); |
427
|
6 |
|
break; |
428
|
6 |
|
case PoTokens::CONTEXT: |
429
|
6 |
|
case PoTokens::MESSAGE: |
430
|
6 |
|
case PoTokens::PLURAL: |
431
|
6 |
|
case PoTokens::TRANSLATED: |
432
|
6 |
|
$entry->addQuoted($matches[2], $matches[5]); |
433
|
6 |
|
break; |
434
|
|
|
default: |
435
|
6 |
|
if ($matches[4]==PoTokens::CONTINUED_DATA) { |
436
|
6 |
|
$currentKey = $lastKey; // keep the previous key |
437
|
6 |
|
if ($currentKey==PoTokens::TRANSLATED_PLURAL) { |
438
|
5 |
|
$entry->addQuotedAtPosition( |
439
|
5 |
|
PoTokens::TRANSLATED, |
440
|
5 |
|
$currentPlural, |
441
|
5 |
|
'"' . $matches[5] |
442
|
|
|
); |
443
|
|
|
} else { |
444
|
6 |
|
$entry->addQuoted($currentKey, '"' . $matches[5]); |
445
|
|
|
} |
446
|
6 |
|
} elseif (substr($matches[2], 0, 7)==PoTokens::TRANSLATED_PLURAL) { |
447
|
5 |
|
$currentKey = PoTokens::TRANSLATED_PLURAL; |
448
|
5 |
|
$currentPlural = $matches[3]; |
449
|
5 |
|
$entry->addQuotedAtPosition( |
450
|
5 |
|
PoTokens::TRANSLATED, |
451
|
5 |
|
$currentPlural, |
452
|
5 |
|
$matches[5] |
453
|
|
|
); |
454
|
6 |
|
} elseif (isset($matches[6][0]) |
455
|
6 |
|
&& $matches[6][0]==PoTokens::TRANSLATOR_COMMENTS) { |
456
|
6 |
|
$value = substr($matches[6], 1); |
457
|
6 |
|
$value = empty($value) ? '' : $value; |
458
|
6 |
|
$entry->add(PoTokens::TRANSLATOR_COMMENTS, $value); |
459
|
|
|
} else { |
460
|
|
|
$unrecognized[$line+1] = $s; |
461
|
|
|
} |
462
|
6 |
|
break; |
463
|
|
|
} |
464
|
7 |
|
$lastKey = $currentKey; |
465
|
|
|
} |
466
|
|
|
} |
467
|
7 |
|
if (!($entry === null)) { |
468
|
|
|
$this->addEntry($entry); |
469
|
|
|
} |
470
|
|
|
|
471
|
|
|
// throw at the very end, anything recognized has been processed |
472
|
7 |
|
$this->unrecognizedInput = $unrecognized; |
473
|
7 |
|
if (count($unrecognized)) { |
474
|
2 |
|
throw new UnrecognizedInputException(); |
475
|
|
|
} |
476
|
5 |
|
} |
477
|
|
|
} |
478
|
|
|
|