|
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\SpecialChars; |
|
15
|
|
|
use phpDocumentor\Guides\RestructuredText\Exception\InvalidTableStructure; |
|
|
|
|
|
|
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
|
|
|
public function __construct(Mailcode_Parser_Statement $statement) |
|
71
|
|
|
{ |
|
72
|
|
|
$this->statement = $statement; |
|
73
|
|
|
|
|
74
|
|
|
$this->tokenize($statement->getStatementString()); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
|
|
public function getSourceCommand() : ?Mailcode_Commands_Command |
|
78
|
|
|
{ |
|
79
|
|
|
return $this->statement->getSourceCommand(); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
/** |
|
83
|
|
|
* Retrieves all tokens detected in the statement string, in |
|
84
|
|
|
* the order they were found. |
|
85
|
|
|
* |
|
86
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token[] |
|
87
|
|
|
*/ |
|
88
|
|
|
public function getTokens() : array |
|
89
|
|
|
{ |
|
90
|
|
|
return $this->tokensOrdered; |
|
91
|
|
|
} |
|
92
|
|
|
|
|
93
|
|
|
public function hasTokens() : bool |
|
94
|
|
|
{ |
|
95
|
|
|
return !empty($this->tokensOrdered); |
|
96
|
|
|
} |
|
97
|
|
|
|
|
98
|
|
|
/** |
|
99
|
|
|
* Whether there were any unknown tokens in the statement. |
|
100
|
|
|
* |
|
101
|
|
|
* @return bool |
|
102
|
|
|
*/ |
|
103
|
|
|
public function hasUnknown() : bool |
|
104
|
|
|
{ |
|
105
|
|
|
$unknown = $this->getUnknown(); |
|
106
|
|
|
|
|
107
|
|
|
return !empty($unknown); |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Retrieves all unknown content tokens, if any. |
|
112
|
|
|
* |
|
113
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token_Unknown[] |
|
114
|
|
|
*/ |
|
115
|
|
|
public function getUnknown() : array |
|
116
|
|
|
{ |
|
117
|
|
|
$result = array(); |
|
118
|
|
|
|
|
119
|
|
|
foreach($this->tokensOrdered as $token) |
|
120
|
|
|
{ |
|
121
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Unknown) |
|
122
|
|
|
{ |
|
123
|
|
|
$result[] = $token; |
|
124
|
|
|
} |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
return $result; |
|
128
|
|
|
} |
|
129
|
|
|
|
|
130
|
|
|
public function getFirstUnknown() : ?Mailcode_Parser_Statement_Tokenizer_Token_Unknown |
|
131
|
|
|
{ |
|
132
|
|
|
$unknown = $this->getUnknown(); |
|
133
|
|
|
|
|
134
|
|
|
if(!empty($unknown)) |
|
135
|
|
|
{ |
|
136
|
|
|
return array_shift($unknown); |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
return null; |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
public function getNormalized() : string |
|
143
|
|
|
{ |
|
144
|
|
|
$parts = array(); |
|
145
|
|
|
|
|
146
|
|
|
foreach($this->tokensOrdered as $token) |
|
147
|
|
|
{ |
|
148
|
|
|
$string = $token->getNormalized(); |
|
149
|
|
|
|
|
150
|
|
|
if($string !== '') |
|
151
|
|
|
{ |
|
152
|
|
|
$parts[] = $string; |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
return implode(' ', $parts); |
|
157
|
|
|
} |
|
158
|
|
|
|
|
159
|
|
|
/** |
|
160
|
|
|
* Goes through all tokenization processors, in the order that |
|
161
|
|
|
* they are defined in the tokenCategories property. This filters |
|
162
|
|
|
* the statement string, and extracts the tokens contained within. |
|
163
|
|
|
* |
|
164
|
|
|
* @param string $statement |
|
165
|
|
|
* |
|
166
|
|
|
* @throws Mailcode_Parser_Exception |
|
167
|
|
|
* |
|
168
|
|
|
* @see Mailcode_Parser_Statement_Tokenizer_Process |
|
169
|
|
|
*/ |
|
170
|
|
|
protected function tokenize(string $statement) : void |
|
171
|
|
|
{ |
|
172
|
|
|
$statement = trim($statement); |
|
173
|
|
|
$tokens = array(); |
|
174
|
|
|
|
|
175
|
|
|
foreach($this->tokenClasses as $tokenClass) |
|
176
|
|
|
{ |
|
177
|
|
|
$processor = $this->createProcessor($tokenClass, $statement, $tokens); |
|
178
|
|
|
$processor->process(); |
|
179
|
|
|
|
|
180
|
|
|
$statement = $processor->getStatement(); |
|
181
|
|
|
$tokens = $processor->getTokens(); |
|
182
|
|
|
} |
|
183
|
|
|
|
|
184
|
|
|
$this->tokenized = $statement; |
|
185
|
|
|
$this->tokensOrdered = $tokens; |
|
186
|
|
|
} |
|
187
|
|
|
|
|
188
|
|
|
/** |
|
189
|
|
|
* @param string $className |
|
190
|
|
|
* @param string $statement |
|
191
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token[] $tokens |
|
192
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Process |
|
193
|
|
|
* @throws Mailcode_Parser_Exception |
|
194
|
|
|
*/ |
|
195
|
|
|
protected function createProcessor(string $className, string $statement, array $tokens) : Mailcode_Parser_Statement_Tokenizer_Process |
|
196
|
|
|
{ |
|
197
|
|
|
$instance = new $className($this, $statement, $tokens); |
|
198
|
|
|
|
|
199
|
|
|
if($instance instanceof Mailcode_Parser_Statement_Tokenizer_Process) |
|
200
|
|
|
{ |
|
201
|
|
|
return $instance; |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
throw new Mailcode_Parser_Exception( |
|
205
|
|
|
'Unknown statement token.', |
|
206
|
|
|
sprintf( |
|
207
|
|
|
'The tokenize class [%s] does not extend the base process class.', |
|
208
|
|
|
$className |
|
209
|
|
|
), |
|
210
|
|
|
self::ERROR_TOKENIZE_METHOD_MISSING |
|
211
|
|
|
); |
|
212
|
|
|
} |
|
213
|
|
|
|
|
214
|
|
|
/** |
|
215
|
|
|
* @param string $type |
|
216
|
|
|
* @param string $matchedText |
|
217
|
|
|
* @param mixed $subject |
|
218
|
|
|
* @return Mailcode_Parser_Statement_Tokenizer_Token |
|
219
|
|
|
*/ |
|
220
|
|
|
public function createToken(string $type, string $matchedText, $subject=null) : Mailcode_Parser_Statement_Tokenizer_Token |
|
221
|
|
|
{ |
|
222
|
|
|
$tokenID = $this->generateID(); |
|
223
|
|
|
|
|
224
|
|
|
$class = Mailcode_Parser_Statement_Tokenizer_Token::class.'_'.$type; |
|
225
|
|
|
|
|
226
|
|
|
$token = new $class($tokenID, $matchedText, $subject, $this->getSourceCommand()); |
|
227
|
|
|
|
|
228
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token) |
|
229
|
|
|
{ |
|
230
|
|
|
return $token; |
|
231
|
|
|
} |
|
232
|
|
|
|
|
233
|
|
|
throw new Mailcode_Parser_Exception( |
|
234
|
|
|
'Invalid token class', |
|
235
|
|
|
sprintf( |
|
236
|
|
|
'The class [%s] does not extend the base token class.', |
|
237
|
|
|
get_class($token) |
|
238
|
|
|
), |
|
239
|
|
|
self::ERROR_INVALID_TOKEN_CLASS |
|
240
|
|
|
); |
|
241
|
|
|
} |
|
242
|
|
|
|
|
243
|
|
|
private function createKeyword(string $name) : Mailcode_Parser_Statement_Tokenizer_Token_Keyword |
|
244
|
|
|
{ |
|
245
|
|
|
$name = rtrim($name, ':').':'; |
|
246
|
|
|
|
|
247
|
|
|
$token = $this->createToken('Keyword', $name); |
|
248
|
|
|
|
|
249
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_Keyword) |
|
|
|
|
|
|
250
|
|
|
{ |
|
251
|
|
|
return $token; |
|
252
|
|
|
} |
|
253
|
|
|
|
|
254
|
|
|
throw new Mailcode_Parser_Exception( |
|
255
|
|
|
'Invalid token created', |
|
256
|
|
|
'', |
|
257
|
|
|
self::ERROR_INVALID_TOKEN_CREATED |
|
258
|
|
|
); |
|
259
|
|
|
} |
|
260
|
|
|
|
|
261
|
|
|
public function appendKeyword(string $name) : Mailcode_Parser_Statement_Tokenizer_Token_Keyword |
|
262
|
|
|
{ |
|
263
|
|
|
$token = $this->createKeyword($name); |
|
264
|
|
|
|
|
265
|
|
|
$this->appendToken($token); |
|
266
|
|
|
|
|
267
|
|
|
return $token; |
|
268
|
|
|
} |
|
269
|
|
|
|
|
270
|
|
|
private function createStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
|
271
|
|
|
{ |
|
272
|
|
|
$token = $this->createToken('StringLiteral', SpecialChars::encodeAll($text)); |
|
273
|
|
|
|
|
274
|
|
|
if($token instanceof Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral) |
|
|
|
|
|
|
275
|
|
|
{ |
|
276
|
|
|
return $token; |
|
277
|
|
|
} |
|
278
|
|
|
|
|
279
|
|
|
throw new Mailcode_Parser_Exception( |
|
280
|
|
|
'Invalid token created', |
|
281
|
|
|
'', |
|
282
|
|
|
self::ERROR_INVALID_TOKEN_CREATED |
|
283
|
|
|
); |
|
284
|
|
|
} |
|
285
|
|
|
|
|
286
|
|
|
public function appendStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
|
287
|
|
|
{ |
|
288
|
|
|
$token = $this->createStringLiteral($text); |
|
289
|
|
|
|
|
290
|
|
|
$this->appendToken($token); |
|
291
|
|
|
|
|
292
|
|
|
return $token; |
|
293
|
|
|
} |
|
294
|
|
|
|
|
295
|
|
|
public function prependStringLiteral(string $text) : Mailcode_Parser_Statement_Tokenizer_Token_StringLiteral |
|
296
|
|
|
{ |
|
297
|
|
|
$token = $this->createStringLiteral($text); |
|
298
|
|
|
|
|
299
|
|
|
$this->prependToken($token); |
|
300
|
|
|
|
|
301
|
|
|
return $token; |
|
302
|
|
|
} |
|
303
|
|
|
|
|
304
|
|
|
public function removeToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : Mailcode_Parser_Statement_Tokenizer |
|
305
|
|
|
{ |
|
306
|
|
|
$keep = array(); |
|
307
|
|
|
$tokenID = $token->getID(); |
|
308
|
|
|
|
|
309
|
|
|
foreach ($this->tokensOrdered as $checkToken) |
|
310
|
|
|
{ |
|
311
|
|
|
if($checkToken->getID() !== $tokenID) |
|
312
|
|
|
{ |
|
313
|
|
|
$keep[] = $checkToken; |
|
314
|
|
|
} |
|
315
|
|
|
} |
|
316
|
|
|
|
|
317
|
|
|
$this->tokensOrdered = $keep; |
|
318
|
|
|
|
|
319
|
|
|
$this->triggerTokensChanged(); |
|
320
|
|
|
|
|
321
|
|
|
return $this; |
|
322
|
|
|
} |
|
323
|
|
|
|
|
324
|
|
|
/** |
|
325
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token $token |
|
326
|
|
|
* @return $this |
|
327
|
|
|
*/ |
|
328
|
|
|
protected function appendToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : self |
|
329
|
|
|
{ |
|
330
|
|
|
$this->tokensOrdered[] = $token; |
|
331
|
|
|
|
|
332
|
|
|
$this->triggerTokensChanged(); |
|
333
|
|
|
|
|
334
|
|
|
return $this; |
|
335
|
|
|
} |
|
336
|
|
|
|
|
337
|
|
|
/** |
|
338
|
|
|
* @param Mailcode_Parser_Statement_Tokenizer_Token $token |
|
339
|
|
|
* @return $this |
|
340
|
|
|
*/ |
|
341
|
|
|
protected function prependToken(Mailcode_Parser_Statement_Tokenizer_Token $token) : self |
|
342
|
|
|
{ |
|
343
|
|
|
array_unshift($this->tokensOrdered, $token); |
|
344
|
|
|
|
|
345
|
|
|
$this->triggerTokensChanged(); |
|
346
|
|
|
|
|
347
|
|
|
return $this; |
|
348
|
|
|
} |
|
349
|
|
|
|
|
350
|
|
|
/** |
|
351
|
|
|
* Generates a unique alphabet-based ID without numbers |
|
352
|
|
|
* to use as token name, to avoid conflicts with the |
|
353
|
|
|
* numbers detection. |
|
354
|
|
|
* |
|
355
|
|
|
* @return string |
|
356
|
|
|
*/ |
|
357
|
|
|
protected function generateID() : string |
|
358
|
|
|
{ |
|
359
|
|
|
static $alphas; |
|
360
|
|
|
|
|
361
|
|
|
if(!isset($alphas)) |
|
362
|
|
|
{ |
|
363
|
|
|
$alphas = range('A', 'Z'); |
|
364
|
|
|
} |
|
365
|
|
|
|
|
366
|
|
|
$amount = 12; |
|
367
|
|
|
|
|
368
|
|
|
$result = ''; |
|
369
|
|
|
|
|
370
|
|
|
for($i=0; $i < $amount; $i++) |
|
371
|
|
|
{ |
|
372
|
|
|
$result .= $alphas[array_rand($alphas)]; |
|
373
|
|
|
} |
|
374
|
|
|
|
|
375
|
|
|
if(!in_array($result, self::$ids)) |
|
376
|
|
|
{ |
|
377
|
|
|
self::$ids[] = $result; |
|
378
|
|
|
return $result; |
|
379
|
|
|
} |
|
380
|
|
|
|
|
381
|
|
|
return $this->generateID(); |
|
382
|
|
|
} |
|
383
|
|
|
|
|
384
|
|
|
/** |
|
385
|
|
|
* @param callable $callback |
|
386
|
|
|
*/ |
|
387
|
|
|
public function onTokensChanged(callable $callback) : void |
|
388
|
|
|
{ |
|
389
|
|
|
$this->changeHandlers[] = $callback; |
|
390
|
|
|
} |
|
391
|
|
|
|
|
392
|
|
|
protected function triggerTokensChanged() : void |
|
393
|
|
|
{ |
|
394
|
|
|
foreach ($this->changeHandlers as $callback) |
|
395
|
|
|
{ |
|
396
|
|
|
$callback($this); |
|
397
|
|
|
} |
|
398
|
|
|
} |
|
399
|
|
|
} |
|
400
|
|
|
|
The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g.
excluded_paths: ["lib/*"], you can move it to the dependency path list as follows:For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths