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