|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Egulias\EmailValidator\Parser; |
|
4
|
|
|
|
|
5
|
|
|
use Doctrine\Common\Lexer\Token; |
|
6
|
|
|
use Egulias\EmailValidator\EmailLexer; |
|
7
|
|
|
use Egulias\EmailValidator\Warning\TLD; |
|
8
|
|
|
use Egulias\EmailValidator\Result\Result; |
|
9
|
|
|
use Egulias\EmailValidator\Result\ValidEmail; |
|
10
|
|
|
use Egulias\EmailValidator\Result\InvalidEmail; |
|
11
|
|
|
use Egulias\EmailValidator\Result\Reason\DotAtEnd; |
|
12
|
|
|
use Egulias\EmailValidator\Result\Reason\DotAtStart; |
|
13
|
|
|
use Egulias\EmailValidator\Warning\DeprecatedComment; |
|
14
|
|
|
use Egulias\EmailValidator\Result\Reason\CRLFAtTheEnd; |
|
15
|
|
|
use Egulias\EmailValidator\Result\Reason\LabelTooLong; |
|
16
|
|
|
use Egulias\EmailValidator\Result\Reason\NoDomainPart; |
|
17
|
|
|
use Egulias\EmailValidator\Result\Reason\ConsecutiveAt; |
|
18
|
|
|
use Egulias\EmailValidator\Result\Reason\DomainTooLong; |
|
19
|
|
|
use Egulias\EmailValidator\Result\Reason\CharNotAllowed; |
|
20
|
|
|
use Egulias\EmailValidator\Result\Reason\DomainHyphened; |
|
21
|
|
|
use Egulias\EmailValidator\Result\Reason\ExpectingATEXT; |
|
22
|
|
|
use Egulias\EmailValidator\Parser\CommentStrategy\DomainComment; |
|
23
|
|
|
use Egulias\EmailValidator\Result\Reason\ExpectingDomainLiteralClose; |
|
24
|
|
|
use Egulias\EmailValidator\Parser\DomainLiteral as DomainLiteralParser; |
|
25
|
|
|
|
|
26
|
|
|
class DomainPart extends PartParser |
|
27
|
|
|
{ |
|
28
|
|
|
public const DOMAIN_MAX_LENGTH = 253; |
|
29
|
|
|
public const LABEL_MAX_LENGTH = 63; |
|
30
|
|
|
|
|
31
|
|
|
/** |
|
32
|
|
|
* @var string |
|
33
|
|
|
*/ |
|
34
|
|
|
protected $domainPart = ''; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* @var string |
|
38
|
|
|
*/ |
|
39
|
|
|
protected $label = ''; |
|
40
|
|
|
|
|
41
|
148 |
|
public function parse(): Result |
|
42
|
|
|
{ |
|
43
|
148 |
|
$this->lexer->clearRecorded(); |
|
44
|
148 |
|
$this->lexer->startRecording(); |
|
45
|
|
|
|
|
46
|
148 |
|
$this->lexer->moveNext(); |
|
47
|
|
|
|
|
48
|
148 |
|
$domainChecks = $this->performDomainStartChecks(); |
|
49
|
148 |
|
if ($domainChecks->isInvalid()) { |
|
50
|
12 |
|
return $domainChecks; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
136 |
|
if ($this->lexer->current->isA(EmailLexer::S_AT)) { |
|
|
|
|
|
|
54
|
1 |
|
return new InvalidEmail(new ConsecutiveAt(), $this->lexer->current->value); |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
135 |
|
$result = $this->doParseDomainPart(); |
|
58
|
135 |
|
if ($result->isInvalid()) { |
|
59
|
67 |
|
return $result; |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
68 |
|
$end = $this->checkEndOfDomain(); |
|
63
|
68 |
|
if ($end->isInvalid()) { |
|
64
|
4 |
|
return $end; |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
64 |
|
$this->lexer->stopRecording(); |
|
68
|
64 |
|
$this->domainPart = $this->lexer->getAccumulatedValues(); |
|
69
|
|
|
|
|
70
|
64 |
|
$length = strlen($this->domainPart); |
|
71
|
64 |
|
if ($length > self::DOMAIN_MAX_LENGTH) { |
|
72
|
|
|
return new InvalidEmail(new DomainTooLong(), $this->lexer->current->value); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
64 |
|
return new ValidEmail(); |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
68 |
|
private function checkEndOfDomain(): Result |
|
79
|
|
|
{ |
|
80
|
68 |
|
$prev = $this->lexer->getPrevious(); |
|
81
|
68 |
|
if ($prev->isA(EmailLexer::S_DOT)) { |
|
|
|
|
|
|
82
|
3 |
|
return new InvalidEmail(new DotAtEnd(), $this->lexer->current->value); |
|
83
|
|
|
} |
|
84
|
65 |
|
if ($prev->isA(EmailLexer::S_HYPHEN)) { |
|
85
|
1 |
|
return new InvalidEmail(new DomainHyphened('Hypen found at the end of the domain'), $prev->value); |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
64 |
|
if ($this->lexer->current->isA(EmailLexer::S_SP)) { |
|
89
|
|
|
return new InvalidEmail(new CRLFAtTheEnd(), $prev->value); |
|
90
|
|
|
} |
|
91
|
64 |
|
return new ValidEmail(); |
|
92
|
|
|
} |
|
93
|
|
|
|
|
94
|
|
|
private function performDomainStartChecks(): Result |
|
95
|
148 |
|
{ |
|
96
|
|
|
$invalidTokens = $this->checkInvalidTokensAfterAT(); |
|
97
|
148 |
|
if ($invalidTokens->isInvalid()) { |
|
98
|
148 |
|
return $invalidTokens; |
|
99
|
2 |
|
} |
|
100
|
|
|
|
|
101
|
|
|
$missingDomain = $this->checkEmptyDomain(); |
|
102
|
146 |
|
if ($missingDomain->isInvalid()) { |
|
103
|
146 |
|
return $missingDomain; |
|
104
|
10 |
|
} |
|
105
|
|
|
|
|
106
|
|
|
if ($this->lexer->current->isA(EmailLexer::S_OPENPARENTHESIS)) { |
|
|
|
|
|
|
107
|
136 |
|
$this->warnings[DeprecatedComment::CODE] = new DeprecatedComment(); |
|
108
|
3 |
|
} |
|
109
|
|
|
return new ValidEmail(); |
|
110
|
136 |
|
} |
|
111
|
|
|
|
|
112
|
|
|
private function checkEmptyDomain(): Result |
|
113
|
146 |
|
{ |
|
114
|
|
|
$thereIsNoDomain = $this->lexer->current->isA(EmailLexer::S_EMPTY) || |
|
115
|
146 |
|
($this->lexer->current->isA(EmailLexer::S_SP) && |
|
|
|
|
|
|
116
|
146 |
|
!$this->lexer->isNextToken(EmailLexer::GENERIC)); |
|
117
|
146 |
|
|
|
118
|
|
|
if ($thereIsNoDomain) { |
|
119
|
146 |
|
return new InvalidEmail(new NoDomainPart(), $this->lexer->current->value); |
|
120
|
10 |
|
} |
|
121
|
|
|
|
|
122
|
|
|
return new ValidEmail(); |
|
123
|
136 |
|
} |
|
124
|
|
|
|
|
125
|
|
|
private function checkInvalidTokensAfterAT(): Result |
|
126
|
148 |
|
{ |
|
127
|
|
|
if ($this->lexer->current->isA(EmailLexer::S_DOT)) { |
|
|
|
|
|
|
128
|
148 |
|
return new InvalidEmail(new DotAtStart(), $this->lexer->current->value); |
|
129
|
1 |
|
} |
|
130
|
|
|
if ($this->lexer->current->isA(EmailLexer::S_HYPHEN)) { |
|
131
|
147 |
|
return new InvalidEmail(new DomainHyphened('After AT'), $this->lexer->current->value); |
|
132
|
1 |
|
} |
|
133
|
|
|
return new ValidEmail(); |
|
134
|
146 |
|
} |
|
135
|
|
|
|
|
136
|
|
|
protected function parseComments(): Result |
|
137
|
7 |
|
{ |
|
138
|
|
|
$commentParser = new Comment($this->lexer, new DomainComment()); |
|
139
|
7 |
|
$result = $commentParser->parse(); |
|
140
|
7 |
|
$this->warnings = array_merge($this->warnings, $commentParser->getWarnings()); |
|
141
|
7 |
|
|
|
142
|
|
|
return $result; |
|
143
|
7 |
|
} |
|
144
|
|
|
|
|
145
|
|
|
protected function doParseDomainPart(): Result |
|
146
|
135 |
|
{ |
|
147
|
|
|
$tldMissing = true; |
|
148
|
135 |
|
$hasComments = false; |
|
149
|
135 |
|
$domain = ''; |
|
150
|
135 |
|
do { |
|
151
|
|
|
$prev = $this->lexer->getPrevious(); |
|
152
|
135 |
|
|
|
153
|
|
|
$notAllowedChars = $this->checkNotAllowedChars($this->lexer->current); |
|
154
|
135 |
|
if ($notAllowedChars->isInvalid()) { |
|
155
|
135 |
|
return $notAllowedChars; |
|
156
|
4 |
|
} |
|
157
|
|
|
|
|
158
|
|
|
if ( |
|
159
|
135 |
|
$this->lexer->current->isA(EmailLexer::S_OPENPARENTHESIS) || |
|
|
|
|
|
|
160
|
135 |
|
$this->lexer->current->isA(EmailLexer::S_CLOSEPARENTHESIS) |
|
161
|
7 |
|
) { |
|
162
|
7 |
|
$hasComments = true; |
|
163
|
|
|
$commentsResult = $this->parseComments(); |
|
164
|
|
|
|
|
165
|
7 |
|
//Invalid comment parsing |
|
166
|
5 |
|
if ($commentsResult->isInvalid()) { |
|
167
|
|
|
return $commentsResult; |
|
168
|
|
|
} |
|
169
|
|
|
} |
|
170
|
132 |
|
|
|
171
|
132 |
|
$dotsResult = $this->checkConsecutiveDots(); |
|
172
|
2 |
|
if ($dotsResult->isInvalid()) { |
|
173
|
|
|
return $dotsResult; |
|
174
|
|
|
} |
|
175
|
132 |
|
|
|
176
|
18 |
|
if ($this->lexer->current->isA(EmailLexer::S_OPENBRACKET)) { |
|
177
|
|
|
$literalResult = $this->parseDomainLiteral(); |
|
178
|
18 |
|
|
|
179
|
18 |
|
$this->addTLDWarnings($tldMissing); |
|
180
|
|
|
return $literalResult; |
|
181
|
|
|
} |
|
182
|
114 |
|
|
|
183
|
114 |
|
$labelCheck = $this->checkLabelLength(); |
|
184
|
5 |
|
if ($labelCheck->isInvalid()) { |
|
185
|
|
|
return $labelCheck; |
|
186
|
|
|
} |
|
187
|
114 |
|
|
|
188
|
114 |
|
$FwsResult = $this->parseFWS(); |
|
189
|
5 |
|
if ($FwsResult->isInvalid()) { |
|
190
|
|
|
return $FwsResult; |
|
191
|
|
|
} |
|
192
|
114 |
|
|
|
193
|
|
|
$domain .= $this->lexer->current->value; |
|
194
|
114 |
|
|
|
195
|
56 |
|
if ($this->lexer->current->isA(EmailLexer::S_DOT) && $this->lexer->isNextToken(EmailLexer::GENERIC)) { |
|
196
|
|
|
$tldMissing = false; |
|
197
|
|
|
} |
|
198
|
114 |
|
|
|
199
|
114 |
|
$exceptionsResult = $this->checkDomainPartExceptions($prev, $hasComments); |
|
200
|
41 |
|
if ($exceptionsResult->isInvalid()) { |
|
201
|
|
|
return $exceptionsResult; |
|
202
|
112 |
|
} |
|
203
|
|
|
$this->lexer->moveNext(); |
|
204
|
112 |
|
} while (!$this->lexer->current->isA(EmailLexer::S_EMPTY)); |
|
205
|
|
|
|
|
206
|
55 |
|
$labelCheck = $this->checkLabelLength(true); |
|
207
|
55 |
|
if ($labelCheck->isInvalid()) { |
|
208
|
2 |
|
return $labelCheck; |
|
209
|
|
|
} |
|
210
|
53 |
|
$this->addTLDWarnings($tldMissing); |
|
211
|
|
|
|
|
212
|
53 |
|
$this->domainPart = $domain; |
|
213
|
53 |
|
return new ValidEmail(); |
|
214
|
|
|
} |
|
215
|
|
|
|
|
216
|
|
|
/** |
|
217
|
|
|
* @param Token<int, string> $token |
|
218
|
|
|
* |
|
219
|
135 |
|
* @return Result |
|
220
|
|
|
*/ |
|
221
|
135 |
|
private function checkNotAllowedChars(Token $token): Result |
|
222
|
135 |
|
{ |
|
223
|
4 |
|
$notAllowed = [EmailLexer::S_BACKSLASH => true, EmailLexer::S_SLASH => true]; |
|
224
|
|
|
if (isset($notAllowed[$token->type])) { |
|
225
|
135 |
|
return new InvalidEmail(new CharNotAllowed(), $token->value); |
|
226
|
|
|
} |
|
227
|
|
|
return new ValidEmail(); |
|
228
|
|
|
} |
|
229
|
|
|
|
|
230
|
|
|
/** |
|
231
|
18 |
|
* @return Result |
|
232
|
|
|
*/ |
|
233
|
|
|
protected function parseDomainLiteral(): Result |
|
234
|
18 |
|
{ |
|
235
|
|
|
try { |
|
236
|
|
|
$this->lexer->find(EmailLexer::S_CLOSEBRACKET); |
|
237
|
|
|
} catch (\RuntimeException $e) { |
|
238
|
|
|
return new InvalidEmail(new ExpectingDomainLiteralClose(), $this->lexer->current->value); |
|
239
|
18 |
|
} |
|
240
|
18 |
|
|
|
241
|
18 |
|
$domainLiteralParser = new DomainLiteralParser($this->lexer); |
|
242
|
18 |
|
$result = $domainLiteralParser->parse(); |
|
243
|
|
|
$this->warnings = array_merge($this->warnings, $domainLiteralParser->getWarnings()); |
|
244
|
|
|
return $result; |
|
245
|
114 |
|
} |
|
246
|
|
|
|
|
247
|
114 |
|
/** |
|
248
|
|
|
* @param Token<int, string> $prev |
|
249
|
|
|
* @param bool $hasComments |
|
250
|
|
|
* |
|
251
|
114 |
|
* @return Result |
|
252
|
1 |
|
*/ |
|
253
|
|
|
protected function checkDomainPartExceptions(Token $prev, bool $hasComments): Result |
|
254
|
|
|
{ |
|
255
|
114 |
|
if ($this->lexer->current->isA(EmailLexer::S_OPENBRACKET) && $prev->type !== EmailLexer::S_AT) { |
|
|
|
|
|
|
256
|
114 |
|
return new InvalidEmail(new ExpectingATEXT('OPENBRACKET not after AT'), $this->lexer->current->value); |
|
257
|
|
|
} |
|
258
|
|
|
|
|
259
|
|
|
if ($this->lexer->current->isA(EmailLexer::S_HYPHEN) && $this->lexer->isNextToken(EmailLexer::S_DOT)) { |
|
260
|
114 |
|
return new InvalidEmail(new DomainHyphened('Hypen found near DOT'), $this->lexer->current->value); |
|
261
|
|
|
} |
|
262
|
|
|
|
|
263
|
108 |
|
if ( |
|
264
|
|
|
$this->lexer->current->isA(EmailLexer::S_BACKSLASH) |
|
265
|
108 |
|
&& $this->lexer->isNextToken(EmailLexer::GENERIC) |
|
266
|
108 |
|
) { |
|
267
|
108 |
|
return new InvalidEmail(new ExpectingATEXT('Escaping following "ATOM"'), $this->lexer->current->value); |
|
268
|
108 |
|
} |
|
269
|
108 |
|
|
|
270
|
|
|
return $this->validateTokens($hasComments); |
|
271
|
108 |
|
} |
|
272
|
2 |
|
|
|
273
|
2 |
|
protected function validateTokens(bool $hasComments): Result |
|
274
|
|
|
{ |
|
275
|
|
|
$validDomainTokens = array( |
|
276
|
108 |
|
EmailLexer::GENERIC => true, |
|
277
|
40 |
|
EmailLexer::S_HYPHEN => true, |
|
278
|
|
|
EmailLexer::S_DOT => true, |
|
279
|
|
|
); |
|
280
|
106 |
|
|
|
281
|
|
|
if ($hasComments) { |
|
282
|
|
|
$validDomainTokens[EmailLexer::S_OPENPARENTHESIS] = true; |
|
283
|
114 |
|
$validDomainTokens[EmailLexer::S_CLOSEPARENTHESIS] = true; |
|
284
|
|
|
} |
|
285
|
114 |
|
|
|
286
|
72 |
|
if (!isset($validDomainTokens[$this->lexer->current->type])) { |
|
287
|
7 |
|
return new InvalidEmail(new ExpectingATEXT('Invalid token in domain: ' . $this->lexer->current->value), $this->lexer->current->value); |
|
288
|
|
|
} |
|
289
|
66 |
|
|
|
290
|
|
|
return new ValidEmail(); |
|
291
|
114 |
|
} |
|
292
|
114 |
|
|
|
293
|
|
|
private function checkLabelLength(bool $isEndOfDomain = false): Result |
|
294
|
|
|
{ |
|
295
|
|
|
if ($this->lexer->current->isA(EmailLexer::S_DOT) || $isEndOfDomain) { |
|
|
|
|
|
|
296
|
72 |
|
if ($this->isLabelTooLong($this->label)) { |
|
297
|
|
|
return new InvalidEmail(new LabelTooLong(), $this->lexer->current->value); |
|
298
|
72 |
|
} |
|
299
|
10 |
|
$this->label = ''; |
|
300
|
10 |
|
} |
|
301
|
|
|
$this->label .= $this->lexer->current->value; |
|
302
|
66 |
|
return new ValidEmail(); |
|
303
|
|
|
} |
|
304
|
|
|
|
|
305
|
71 |
|
|
|
306
|
|
|
private function isLabelTooLong(string $label): bool |
|
307
|
71 |
|
{ |
|
308
|
26 |
|
if (preg_match('/[^\x00-\x7F]/', $label)) { |
|
309
|
|
|
idn_to_ascii($label, IDNA_DEFAULT, INTL_IDNA_VARIANT_UTS46, $idnaInfo); |
|
310
|
|
|
return (bool) ($idnaInfo['errors'] & IDNA_ERROR_LABEL_TOO_LONG); |
|
311
|
|
|
} |
|
312
|
148 |
|
return strlen($label) > self::LABEL_MAX_LENGTH; |
|
313
|
|
|
} |
|
314
|
148 |
|
|
|
315
|
|
|
private function addTLDWarnings(bool $isTLDMissing): void |
|
316
|
|
|
{ |
|
317
|
|
|
if ($isTLDMissing) { |
|
318
|
|
|
$this->warnings[TLD::CODE] = new TLD(); |
|
319
|
|
|
} |
|
320
|
|
|
} |
|
321
|
|
|
|
|
322
|
|
|
public function domainPart(): string |
|
323
|
|
|
{ |
|
324
|
|
|
return $this->domainPart; |
|
325
|
|
|
} |
|
326
|
|
|
} |