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