1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* File containing the {@see Mailcode_Parser_Statement_Tokenizer} class. |
4
|
|
|
* |
5
|
|
|
* @package Mailcode |
6
|
|
|
* @subpackage Parser |
7
|
|
|
* @see Mailcode_Parser_Statement_Tokenizer |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
declare(strict_types=1); |
11
|
|
|
|
12
|
|
|
namespace Mailcode; |
13
|
|
|
|
14
|
|
|
use Mailcode\Parser\Statement\Tokenizer\EventHandler; |
15
|
|
|
use Mailcode\Parser\Statement\Tokenizer\SpecialChars; |
16
|
|
|
|
17
|
|
|
/** |
18
|
|
|
* Mailcode statement tokenizer: parses a mailcode statement |
19
|
|
|
* into its logical parts. |
20
|
|
|
* |
21
|
|
|
* @package Mailcode |
22
|
|
|
* @subpackage Parser |
23
|
|
|
* @author Sebastian Mordziol <[email protected]> |
24
|
|
|
*/ |
25
|
|
|
class Mailcode_Parser_Statement_Tokenizer |
26
|
|
|
{ |
27
|
|
|
public const ERROR_TOKENIZE_METHOD_MISSING = 49801; |
28
|
|
|
public const ERROR_INVALID_TOKEN_CREATED = 49802; |
29
|
|
|
public const ERROR_INVALID_TOKEN_CLASS = 49803; |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @var string[] |
33
|
|
|
*/ |
34
|
|
|
protected array $tokenClasses = array( |
35
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_Variables::class, |
36
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_NormalizeQuotes::class, |
37
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_EncodeSpecialChars::class, |
38
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_StringLiterals::class, |
39
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_Keywords::class, |
40
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_Numbers::class, |
41
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_Operands::class, |
42
|
|
|
Mailcode_Parser_Statement_Tokenizer_Process_ExtractTokens::class |
43
|
|
|
); |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @var Mailcode_Parser_Statement |
47
|
|
|
*/ |
48
|
|
|
protected Mailcode_Parser_Statement $statement; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @var string |
52
|
|
|
*/ |
53
|
|
|
protected string $tokenized = ''; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @var Mailcode_Parser_Statement_Tokenizer_Token[] |
57
|
|
|
*/ |
58
|
|
|
protected array $tokensOrdered = array(); |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* @var string[] |
62
|
|
|
*/ |
63
|
|
|
protected static array $ids = array(); |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* @var callable[] |
67
|
|
|
*/ |
68
|
|
|
protected array $changeHandlers = array(); |
69
|
|
|
|
70
|
|
|
private EventHandler $eventHandler; |
71
|
|
|
|
72
|
|
|
public function __construct(Mailcode_Parser_Statement $statement) |
73
|
|
|
{ |
74
|
|
|
$this->statement = $statement; |
75
|
|
|
$this->eventHandler = new EventHandler($this); |
76
|
|
|
|
77
|
|
|
$this->tokenize($statement->getStatementString()); |
78
|
|
|
} |
79
|
|
|
|
80
|
|
|
public function getSourceCommand() : ?Mailcode_Commands_Command |
81
|
|
|
{ |
82
|
|
|
return $this->statement->getSourceCommand(); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
/** |
86
|
|
|
* Retrieves all tokens detected in the statement string, in |
87
|
|
|
* the order they were found. |
88
|
|
|
* |
89
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token[] |
90
|
|
|
*/ |
91
|
|
|
public function getTokens() : array |
92
|
|
|
{ |
93
|
|
|
return $this->tokensOrdered; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
public function hasTokens() : bool |
97
|
|
|
{ |
98
|
|
|
return !empty($this->tokensOrdered); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* Whether there were any unknown tokens in the statement. |
103
|
|
|
* |
104
|
|
|
* @return bool |
105
|
|
|
*/ |
106
|
|
|
public function hasUnknown() : bool |
107
|
|
|
{ |
108
|
|
|
$unknown = $this->getUnknown(); |
109
|
|
|
|
110
|
|
|
return !empty($unknown); |
111
|
|
|
} |
112
|
|
|
|
113
|
|
|
/** |
114
|
|
|
* Retrieves all unknown content tokens, if any. |
115
|
|
|
* |
116
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token_Unknown[] |
117
|
|
|
*/ |
118
|
|
|
public function getUnknown() : array |
119
|
|
|
{ |
120
|
|
|
$result = array(); |
121
|
|
|
|
122
|
|
|
foreach($this->tokensOrdered as $token) |
123
|
|
|
{ |
124
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Unknown) |
125
|
|
|
{ |
126
|
|
|
$result[] = $token; |
127
|
|
|
} |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
return $result; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
public function getFirstUnknown() : ?Mailcode_Parser_Statement_Tokenizer_Token_Unknown |
134
|
|
|
{ |
135
|
|
|
$unknown = $this->getUnknown(); |
136
|
|
|
|
137
|
|
|
if(!empty($unknown)) |
138
|
|
|
{ |
139
|
|
|
return array_shift($unknown); |
140
|
|
|
} |
141
|
|
|
|
142
|
|
|
return null; |
143
|
|
|
} |
144
|
|
|
|
145
|
|
|
public function getNormalized() : string |
146
|
|
|
{ |
147
|
|
|
$parts = array(); |
148
|
|
|
|
149
|
|
|
foreach($this->tokensOrdered as $token) |
150
|
|
|
{ |
151
|
|
|
$string = $token->getNormalized(); |
152
|
|
|
|
153
|
|
|
if($string !== '') |
154
|
|
|
{ |
155
|
|
|
$parts[] = $string; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
return implode(' ', $parts); |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* Goes through all tokenization processors, in the order that |
164
|
|
|
* they are defined in the tokenCategories property. This filters |
165
|
|
|
* the statement string, and extracts the tokens contained within. |
166
|
|
|
* |
167
|
|
|
* @param string $statement |
168
|
|
|
* |
169
|
|
|
* @throws Mailcode_Parser_Exception |
170
|
|
|
* |
171
|
|
|
* @see Mailcode_Parser_Statement_Tokenizer_Process |
172
|
|
|
*/ |
173
|
|
|
protected function tokenize(string $statement) : void |
174
|
|
|
{ |
175
|
|
|
$statement = trim($statement); |
176
|
|
|
$tokens = array(); |
177
|
|
|
|
178
|
|
|
foreach($this->tokenClasses as $tokenClass) |
179
|
|
|
{ |
180
|
|
|
$processor = $this->createProcessor($tokenClass, $statement, $tokens); |
181
|
|
|
$processor->process(); |
182
|
|
|
|
183
|
|
|
$statement = $processor->getStatement(); |
184
|
|
|
$tokens = $processor->getTokens(); |
185
|
|
|
} |
186
|
|
|
|
187
|
|
|
$this->tokenized = $statement; |
188
|
|
|
$this->tokensOrdered = $tokens; |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
/** |
192
|
|
|
* @param string $className |
193
|
|
|
* @param string $statement |
194
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token[] $tokens |
195
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Process |
196
|
|
|
* @throws Mailcode_Parser_Exception |
197
|
|
|
*/ |
198
|
|
|
protected function createProcessor(string $className, string $statement, array $tokens) : Mailcode_Parser_Statement_Tokenizer_Process |
199
|
|
|
{ |
200
|
|
|
$instance = new $className($this, $statement, $tokens); |
201
|
|
|
|
202
|
|
|
if($instance instanceof Mailcode_Parser_Statement_Tokenizer_Process) |
203
|
|
|
{ |
204
|
|
|
return $instance; |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
throw new Mailcode_Parser_Exception( |
208
|
|
|
'Unknown statement token.', |
209
|
|
|
sprintf( |
210
|
|
|
'The tokenize class [%s] does not extend the base process class.', |
211
|
|
|
$className |
212
|
|
|
), |
213
|
|
|
self::ERROR_TOKENIZE_METHOD_MISSING |
214
|
|
|
); |
215
|
|
|
} |
216
|
|
|
|
217
|
|
|
/** |
218
|
|
|
* @param string $type |
219
|
|
|
* @param string $matchedText |
220
|
|
|
* @param mixed $subject |
221
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token |
222
|
|
|
*/ |
223
|
|
|
public function createToken(string $type, string $matchedText, $subject=null) : Mailcode_Parser_Statement_Tokenizer_Token |
224
|
|
|
{ |
225
|
|
|
$tokenID = $this->generateID(); |
226
|
|
|
|
227
|
|
|
$class = Mailcode_Parser_Statement_Tokenizer_Token::class.'_'.$type; |
228
|
|
|
|
229
|
|
|
$token = new $class($tokenID, $matchedText, $subject, $this->getSourceCommand()); |
230
|
|
|
|
231
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token) |
232
|
|
|
{ |
233
|
|
|
return $token; |
234
|
|
|
} |
235
|
|
|
|
236
|
|
|
throw new Mailcode_Parser_Exception( |
237
|
|
|
'Invalid token class', |
238
|
|
|
sprintf( |
239
|
|
|
'The class [%s] does not extend the base token class.', |
240
|
|
|
get_class($token) |
241
|
|
|
), |
242
|
|
|
self::ERROR_INVALID_TOKEN_CLASS |
243
|
|
|
); |
244
|
|
|
} |
245
|
|
|
|
246
|
|
|
private function createKeyword(string $name) : Mailcode_Parser_Statement_Tokenizer_Token_Keyword |
247
|
|
|
{ |
248
|
|
|
$name = rtrim($name, ':').':'; |
249
|
|
|
|
250
|
|
|
$token = $this->createToken('Keyword', $name); |
251
|
|
|
|
252
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Keyword) |
|
|
|
|
253
|
|
|
{ |
254
|
|
|
return $token; |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
throw new Mailcode_Parser_Exception( |
258
|
|
|
'Invalid token created', |
259
|
|
|
'', |
260
|
|
|
self::ERROR_INVALID_TOKEN_CREATED |
261
|
|
|
); |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
public function appendKeyword(string $name) : Mailcode_Parser_Statement_Tokenizer_Token_Keyword |
265
|
|
|
{ |
266
|
|
|
$token = $this->createKeyword($name); |
267
|
|
|
|
268
|
|
|
$this->appendToken($token); |
269
|
|
|
|
270
|
|
|
return $token; |
271
|
|
|
} |
272
|
|
|
|
273
|
|
|
private function createStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
274
|
|
|
{ |
275
|
|
|
$token = $this->createToken('StringLiteral', SpecialChars::encodeAll($text)); |
276
|
|
|
|
277
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral) |
|
|
|
|
278
|
|
|
{ |
279
|
|
|
return $token; |
280
|
|
|
} |
281
|
|
|
|
282
|
|
|
throw new Mailcode_Parser_Exception( |
283
|
|
|
'Invalid token created', |
284
|
|
|
'', |
285
|
|
|
self::ERROR_INVALID_TOKEN_CREATED |
286
|
|
|
); |
287
|
|
|
} |
288
|
|
|
|
289
|
|
|
public function appendStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
290
|
|
|
{ |
291
|
|
|
$token = $this->createStringLiteral($text); |
292
|
|
|
|
293
|
|
|
$this->appendToken($token); |
294
|
|
|
|
295
|
|
|
return $token; |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
public function prependStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
299
|
|
|
{ |
300
|
|
|
$token = $this->createStringLiteral($text); |
301
|
|
|
|
302
|
|
|
$this->prependToken($token); |
303
|
|
|
|
304
|
|
|
return $token; |
305
|
|
|
} |
306
|
|
|
|
307
|
|
|
public function removeToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : Mailcode_Parser_Statement_Tokenizer |
308
|
|
|
{ |
309
|
|
|
$keep = array(); |
310
|
|
|
$tokenID = $token->getID(); |
311
|
|
|
$removed = false; |
312
|
|
|
|
313
|
|
|
foreach ($this->tokensOrdered as $checkToken) |
314
|
|
|
{ |
315
|
|
|
if($checkToken->getID() === $tokenID) |
316
|
|
|
{ |
317
|
|
|
$removed = true; |
318
|
|
|
continue; |
319
|
|
|
} |
320
|
|
|
|
321
|
|
|
$keep[] = $checkToken; |
322
|
|
|
} |
323
|
|
|
|
324
|
|
|
$this->tokensOrdered = $keep; |
325
|
|
|
|
326
|
|
|
if($removed) |
327
|
|
|
{ |
328
|
|
|
$this->eventHandler->handleTokenRemoved($token); |
329
|
|
|
} |
330
|
|
|
|
331
|
|
|
return $this; |
332
|
|
|
} |
333
|
|
|
|
334
|
|
|
/** |
335
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token $token |
336
|
|
|
* @return $this |
337
|
|
|
*/ |
338
|
|
|
protected function appendToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : self |
339
|
|
|
{ |
340
|
|
|
$this->tokensOrdered[] = $token; |
341
|
|
|
|
342
|
|
|
$this->eventHandler->handleTokenAppended($token); |
343
|
|
|
|
344
|
|
|
return $this; |
345
|
|
|
} |
346
|
|
|
|
347
|
|
|
/** |
348
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token $token |
349
|
|
|
* @return $this |
350
|
|
|
*/ |
351
|
|
|
protected function prependToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : self |
352
|
|
|
{ |
353
|
|
|
array_unshift($this->tokensOrdered, $token); |
354
|
|
|
|
355
|
|
|
$this->eventHandler->handleTokenPrepended($token); |
356
|
|
|
|
357
|
|
|
return $this; |
358
|
|
|
} |
359
|
|
|
|
360
|
|
|
/** |
361
|
|
|
* Generates a unique alphabet-based ID without numbers |
362
|
|
|
* to use as token name, to avoid conflicts with the |
363
|
|
|
* numbers detection. |
364
|
|
|
* |
365
|
|
|
* @return string |
366
|
|
|
*/ |
367
|
|
|
protected function generateID() : string |
368
|
|
|
{ |
369
|
|
|
static $alphas; |
370
|
|
|
|
371
|
|
|
if(!isset($alphas)) |
372
|
|
|
{ |
373
|
|
|
$alphas = range('A', 'Z'); |
374
|
|
|
} |
375
|
|
|
|
376
|
|
|
$amount = 12; |
377
|
|
|
|
378
|
|
|
$result = ''; |
379
|
|
|
|
380
|
|
|
for($i=0; $i < $amount; $i++) |
381
|
|
|
{ |
382
|
|
|
$result .= $alphas[array_rand($alphas)]; |
383
|
|
|
} |
384
|
|
|
|
385
|
|
|
if(!in_array($result, self::$ids)) |
386
|
|
|
{ |
387
|
|
|
self::$ids[] = $result; |
388
|
|
|
return $result; |
389
|
|
|
} |
390
|
|
|
|
391
|
|
|
return $this->generateID(); |
392
|
|
|
} |
393
|
|
|
|
394
|
|
|
/** |
395
|
|
|
* @param callable $callback |
396
|
|
|
*/ |
397
|
|
|
public function onTokensChanged(callable $callback) : void |
398
|
|
|
{ |
399
|
|
|
$this->changeHandlers[] = $callback; |
400
|
|
|
} |
401
|
|
|
|
402
|
|
|
/** |
403
|
|
|
* @return EventHandler |
404
|
|
|
*/ |
405
|
|
|
public function getEventHandler() : EventHandler |
406
|
|
|
{ |
407
|
|
|
return $this->eventHandler; |
408
|
|
|
} |
409
|
|
|
} |
410
|
|
|
|