1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace Egulias\EmailValidator; |
4
|
|
|
|
5
|
|
|
use Doctrine\Common\Lexer\AbstractLexer; |
6
|
|
|
use Doctrine\Common\Lexer\Token; |
7
|
|
|
|
8
|
|
|
/** @extends AbstractLexer<int, string> */ |
9
|
|
|
class EmailLexer extends AbstractLexer |
10
|
|
|
{ |
11
|
|
|
//ASCII values |
12
|
|
|
public const S_EMPTY = -1; |
13
|
|
|
public const C_NUL = 0; |
14
|
|
|
public const S_HTAB = 9; |
15
|
|
|
public const S_LF = 10; |
16
|
|
|
public const S_CR = 13; |
17
|
|
|
public const S_SP = 32; |
18
|
|
|
public const EXCLAMATION = 33; |
19
|
|
|
public const S_DQUOTE = 34; |
20
|
|
|
public const NUMBER_SIGN = 35; |
21
|
|
|
public const DOLLAR = 36; |
22
|
|
|
public const PERCENTAGE = 37; |
23
|
|
|
public const AMPERSAND = 38; |
24
|
|
|
public const S_SQUOTE = 39; |
25
|
|
|
public const S_OPENPARENTHESIS = 40; |
26
|
|
|
public const S_CLOSEPARENTHESIS = 41; |
27
|
|
|
public const ASTERISK = 42; |
28
|
|
|
public const S_PLUS = 43; |
29
|
|
|
public const S_COMMA = 44; |
30
|
|
|
public const S_HYPHEN = 45; |
31
|
|
|
public const S_DOT = 46; |
32
|
|
|
public const S_SLASH = 47; |
33
|
|
|
public const S_COLON = 58; |
34
|
|
|
public const S_SEMICOLON = 59; |
35
|
|
|
public const S_LOWERTHAN = 60; |
36
|
|
|
public const S_EQUAL = 61; |
37
|
|
|
public const S_GREATERTHAN = 62; |
38
|
|
|
public const QUESTIONMARK = 63; |
39
|
|
|
public const S_AT = 64; |
40
|
|
|
public const S_OPENBRACKET = 91; |
41
|
|
|
public const S_BACKSLASH = 92; |
42
|
|
|
public const S_CLOSEBRACKET = 93; |
43
|
|
|
public const CARET = 94; |
44
|
|
|
public const S_UNDERSCORE = 95; |
45
|
|
|
public const S_BACKTICK = 96; |
46
|
|
|
public const S_OPENCURLYBRACES = 123; |
47
|
|
|
public const S_PIPE = 124; |
48
|
|
|
public const S_CLOSECURLYBRACES = 125; |
49
|
|
|
public const S_TILDE = 126; |
50
|
|
|
public const C_DEL = 127; |
51
|
|
|
public const INVERT_QUESTIONMARK = 168; |
52
|
|
|
public const INVERT_EXCLAMATION = 173; |
53
|
|
|
public const GENERIC = 300; |
54
|
|
|
public const S_IPV6TAG = 301; |
55
|
|
|
public const INVALID = 302; |
56
|
|
|
public const CRLF = 1310; |
57
|
|
|
public const S_DOUBLECOLON = 5858; |
58
|
|
|
public const ASCII_INVALID_FROM = 127; |
59
|
|
|
public const ASCII_INVALID_TO = 199; |
60
|
|
|
|
61
|
|
|
/** |
62
|
|
|
* US-ASCII visible characters not valid for atext (@link http://tools.ietf.org/html/rfc5322#section-3.2.3) |
63
|
|
|
* |
64
|
|
|
* @var array |
65
|
|
|
*/ |
66
|
|
|
protected $charValue = [ |
67
|
|
|
'{' => self::S_OPENCURLYBRACES, |
68
|
|
|
'}' => self::S_CLOSECURLYBRACES, |
69
|
|
|
'(' => self::S_OPENPARENTHESIS, |
70
|
|
|
')' => self::S_CLOSEPARENTHESIS, |
71
|
|
|
'<' => self::S_LOWERTHAN, |
72
|
|
|
'>' => self::S_GREATERTHAN, |
73
|
|
|
'[' => self::S_OPENBRACKET, |
74
|
|
|
']' => self::S_CLOSEBRACKET, |
75
|
|
|
':' => self::S_COLON, |
76
|
|
|
';' => self::S_SEMICOLON, |
77
|
|
|
'@' => self::S_AT, |
78
|
|
|
'\\' => self::S_BACKSLASH, |
79
|
|
|
'/' => self::S_SLASH, |
80
|
|
|
',' => self::S_COMMA, |
81
|
|
|
'.' => self::S_DOT, |
82
|
|
|
"'" => self::S_SQUOTE, |
83
|
|
|
"`" => self::S_BACKTICK, |
84
|
|
|
'"' => self::S_DQUOTE, |
85
|
|
|
'-' => self::S_HYPHEN, |
86
|
|
|
'::' => self::S_DOUBLECOLON, |
87
|
|
|
' ' => self::S_SP, |
88
|
|
|
"\t" => self::S_HTAB, |
89
|
|
|
"\r" => self::S_CR, |
90
|
|
|
"\n" => self::S_LF, |
91
|
|
|
"\r\n" => self::CRLF, |
92
|
|
|
'IPv6' => self::S_IPV6TAG, |
93
|
|
|
'' => self::S_EMPTY, |
94
|
|
|
'\0' => self::C_NUL, |
95
|
|
|
'*' => self::ASTERISK, |
96
|
|
|
'!' => self::EXCLAMATION, |
97
|
|
|
'&' => self::AMPERSAND, |
98
|
|
|
'^' => self::CARET, |
99
|
|
|
'$' => self::DOLLAR, |
100
|
|
|
'%' => self::PERCENTAGE, |
101
|
|
|
'~' => self::S_TILDE, |
102
|
|
|
'|' => self::S_PIPE, |
103
|
|
|
'_' => self::S_UNDERSCORE, |
104
|
|
|
'=' => self::S_EQUAL, |
105
|
|
|
'+' => self::S_PLUS, |
106
|
|
|
'¿' => self::INVERT_QUESTIONMARK, |
107
|
|
|
'?' => self::QUESTIONMARK, |
108
|
|
|
'#' => self::NUMBER_SIGN, |
109
|
|
|
'¡' => self::INVERT_EXCLAMATION, |
110
|
|
|
]; |
111
|
|
|
|
112
|
|
|
public const INVALID_CHARS_REGEX = "/[^\p{S}\p{C}\p{Cc}]+/iu"; |
113
|
|
|
|
114
|
|
|
public const VALID_UTF8_REGEX = '/\p{Cc}+/u'; |
115
|
|
|
|
116
|
|
|
public const CATCHABLE_PATTERNS = [ |
117
|
|
|
'[a-zA-Z]+[46]?', //ASCII and domain literal |
118
|
|
|
'[^\x00-\x7F]', //UTF-8 |
119
|
|
|
'[0-9]+', |
120
|
|
|
'\r\n', |
121
|
|
|
'::', |
122
|
|
|
'\s+?', |
123
|
|
|
'.', |
124
|
|
|
]; |
125
|
|
|
|
126
|
|
|
public const NON_CATCHABLE_PATTERNS = [ |
127
|
|
|
'[\xA0-\xff]+', |
128
|
|
|
]; |
129
|
|
|
|
130
|
|
|
public const MODIFIERS = 'iu'; |
131
|
|
|
|
132
|
|
|
/** @var bool */ |
133
|
|
|
protected $hasInvalidTokens = false; |
134
|
|
|
|
135
|
|
|
/** |
136
|
|
|
* @var Token<int, string> |
137
|
|
|
*/ |
138
|
|
|
protected Token $previous; |
139
|
|
|
|
140
|
|
|
/** |
141
|
|
|
* The last matched/seen token. |
142
|
|
|
* |
143
|
|
|
* @var Token<int, string> |
144
|
|
|
*/ |
145
|
|
|
public Token $current; |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* @var Token<int, string> |
149
|
|
|
*/ |
150
|
|
|
private Token $nullToken; |
151
|
|
|
|
152
|
|
|
/** @var string */ |
153
|
|
|
private $accumulator = ''; |
154
|
|
|
|
155
|
|
|
/** @var bool */ |
156
|
|
|
private $hasToRecord = false; |
157
|
|
|
|
158
|
|
|
public function __construct() |
159
|
|
|
{ |
160
|
|
|
/** @var Token<int, string> $nullToken */ |
161
|
|
|
$nullToken = new Token('', self::S_EMPTY, 0); |
162
|
|
|
$this->nullToken = $nullToken; |
163
|
|
|
|
164
|
|
|
$this->current = $this->previous = $this->nullToken; |
165
|
|
|
$this->lookahead = null; |
166
|
|
|
} |
167
|
|
|
|
168
|
|
|
public function reset(): void |
169
|
|
|
{ |
170
|
|
|
$this->hasInvalidTokens = false; |
171
|
|
|
parent::reset(); |
172
|
|
|
$this->current = $this->previous = $this->nullToken; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
/** |
176
|
|
|
* @param int $type |
177
|
340 |
|
* @throws \UnexpectedValueException |
178
|
|
|
* @return boolean |
179
|
340 |
|
* |
180
|
340 |
|
* @psalm-suppress InvalidScalarArgument |
181
|
|
|
*/ |
182
|
|
|
public function find($type): bool |
183
|
308 |
|
{ |
184
|
|
|
$search = clone $this; |
185
|
308 |
|
$search->skipUntil($type); |
186
|
308 |
|
|
187
|
308 |
|
if (!$search->lookahead) { |
188
|
|
|
throw new \UnexpectedValueException($type . ' not found'); |
189
|
|
|
} |
190
|
|
|
return true; |
191
|
|
|
} |
192
|
|
|
|
193
|
|
|
/** |
194
|
|
|
* moveNext |
195
|
|
|
* |
196
|
|
|
* @return boolean |
197
|
54 |
|
*/ |
198
|
|
|
public function moveNext(): bool |
199
|
54 |
|
{ |
200
|
54 |
|
if ($this->hasToRecord && $this->previous === $this->nullToken) { |
201
|
|
|
$this->accumulator .= $this->current->value; |
202
|
54 |
|
} |
203
|
6 |
|
|
204
|
|
|
$this->previous = $this->current; |
205
|
48 |
|
|
206
|
|
|
if ($this->lookahead === null) { |
207
|
|
|
$this->lookahead = $this->nullToken; |
|
|
|
|
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
$hasNext = parent::moveNext(); |
211
|
|
|
$this->current = $this->token ?? $this->nullToken; |
212
|
|
|
|
213
|
297 |
|
if ($this->hasToRecord) { |
214
|
|
|
$this->accumulator .= $this->current->value; |
215
|
297 |
|
} |
216
|
179 |
|
|
217
|
|
|
return $hasNext; |
218
|
|
|
} |
219
|
297 |
|
|
220
|
180 |
|
/** |
221
|
297 |
|
* Retrieve token type. Also processes the token value if necessary. |
222
|
|
|
* |
223
|
297 |
|
* @param string $value |
224
|
297 |
|
* @throws \InvalidArgumentException |
225
|
|
|
* @return integer |
226
|
|
|
*/ |
227
|
297 |
|
protected function getType(&$value): int |
228
|
|
|
{ |
229
|
297 |
|
$encoded = $value; |
230
|
179 |
|
|
231
|
|
|
if (mb_detect_encoding($value, 'auto', true) !== 'UTF-8') { |
232
|
|
|
$encoded = mb_convert_encoding($value, 'UTF-8', 'Windows-1252'); |
233
|
297 |
|
} |
234
|
|
|
|
235
|
|
|
if ($this->isValid($encoded)) { |
|
|
|
|
236
|
|
|
return $this->charValue[$encoded]; |
237
|
|
|
} |
238
|
|
|
|
239
|
|
|
if ($this->isNullType($encoded)) { |
|
|
|
|
240
|
|
|
return self::C_NUL; |
|
|
|
|
241
|
|
|
} |
242
|
|
|
|
243
|
300 |
|
if ($this->isInvalidChar($encoded)) { |
|
|
|
|
244
|
|
|
$this->hasInvalidTokens = true; |
245
|
300 |
|
return self::INVALID; |
|
|
|
|
246
|
|
|
} |
247
|
300 |
|
|
248
|
237 |
|
return self::GENERIC; |
|
|
|
|
249
|
|
|
} |
250
|
|
|
|
251
|
300 |
|
protected function isValid(string $value): bool |
252
|
229 |
|
{ |
253
|
|
|
return isset($this->charValue[$value]); |
254
|
|
|
} |
255
|
262 |
|
|
256
|
2 |
|
protected function isNullType(string $value): bool |
257
|
|
|
{ |
258
|
|
|
return $value === "\0"; |
259
|
261 |
|
} |
260
|
66 |
|
|
261
|
66 |
|
protected function isInvalidChar(string $value): bool |
262
|
|
|
{ |
263
|
|
|
return !preg_match(self::INVALID_CHARS_REGEX, $value); |
264
|
|
|
} |
265
|
199 |
|
|
266
|
|
|
protected function isUTF8Invalid(string $value): bool |
267
|
|
|
{ |
268
|
300 |
|
return preg_match(self::VALID_UTF8_REGEX, $value) !== false; |
269
|
|
|
} |
270
|
300 |
|
|
271
|
|
|
public function hasInvalidTokens(): bool |
272
|
|
|
{ |
273
|
262 |
|
return $this->hasInvalidTokens; |
274
|
|
|
} |
275
|
262 |
|
|
276
|
|
|
/** |
277
|
|
|
* getPrevious |
278
|
261 |
|
* |
279
|
|
|
* @return Token<int, string> |
280
|
261 |
|
*/ |
281
|
|
|
public function getPrevious(): Token |
282
|
|
|
{ |
283
|
|
|
return $this->previous; |
284
|
|
|
} |
285
|
|
|
|
286
|
|
|
/** |
287
|
|
|
* Lexical catchable patterns. |
288
|
191 |
|
* |
289
|
|
|
* @return string[] |
290
|
191 |
|
*/ |
291
|
|
|
protected function getCatchablePatterns(): array |
292
|
|
|
{ |
293
|
|
|
return self::CATCHABLE_PATTERNS; |
294
|
|
|
} |
295
|
|
|
|
296
|
|
|
/** |
297
|
|
|
* Lexical non-catchable patterns. |
298
|
178 |
|
* |
299
|
|
|
* @return string[] |
300
|
178 |
|
*/ |
301
|
|
|
protected function getNonCatchablePatterns(): array |
302
|
|
|
{ |
303
|
|
|
return self::NON_CATCHABLE_PATTERNS; |
304
|
|
|
} |
305
|
|
|
|
306
|
|
|
protected function getModifiers(): string |
307
|
|
|
{ |
308
|
301 |
|
return self::MODIFIERS; |
309
|
|
|
} |
310
|
301 |
|
|
311
|
|
|
public function getAccumulatedValues(): string |
312
|
|
|
{ |
313
|
|
|
return $this->accumulator; |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
public function startRecording(): void |
317
|
|
|
{ |
318
|
301 |
|
$this->hasToRecord = true; |
319
|
|
|
} |
320
|
301 |
|
|
321
|
|
|
public function stopRecording(): void |
322
|
|
|
{ |
323
|
301 |
|
$this->hasToRecord = false; |
324
|
|
|
} |
325
|
301 |
|
|
326
|
|
|
public function clearRecorded(): void |
327
|
|
|
{ |
328
|
151 |
|
$this->accumulator = ''; |
329
|
|
|
} |
330
|
|
|
} |
331
|
|
|
|
Our type inference engine has found an assignment to a property that is incompatible with the declared type of that property.
Either this assignment is in error or the assigned type should be added to the documentation/type hint for that property..