1
|
|
|
<?php |
2
|
|
|
declare(strict_types=1); |
3
|
|
|
|
4
|
|
|
namespace Genkgo\Mail; |
5
|
|
|
|
6
|
|
|
final class AlternativeText |
7
|
|
|
{ |
8
|
|
|
private const DEFAULT_CHARSET = '<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>'; |
9
|
|
|
|
10
|
|
|
/** |
11
|
|
|
* @var string |
12
|
|
|
*/ |
13
|
|
|
private $text; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* @param string $text |
17
|
|
|
*/ |
18
|
45 |
|
public function __construct(string $text) |
19
|
|
|
{ |
20
|
45 |
|
$this->text = $text; |
21
|
45 |
|
} |
22
|
|
|
|
23
|
|
|
/** |
24
|
|
|
* @return bool |
25
|
|
|
*/ |
26
|
25 |
|
public function isEmpty(): bool |
27
|
|
|
{ |
28
|
25 |
|
return $this->text === ''; |
29
|
|
|
} |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* @return string |
33
|
|
|
*/ |
34
|
15 |
|
public function getRaw(): string |
35
|
|
|
{ |
36
|
15 |
|
return $this->text; |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @return string |
41
|
|
|
*/ |
42
|
31 |
|
public function __toString(): string |
43
|
|
|
{ |
44
|
31 |
|
return $this->normalizeSpace($this->text); |
45
|
|
|
} |
46
|
|
|
|
47
|
|
|
/** |
48
|
|
|
* @param string $string |
49
|
|
|
* @return string |
50
|
|
|
*/ |
51
|
31 |
|
private function normalizeSpace(string $string): string |
52
|
|
|
{ |
53
|
31 |
|
return $this->wrap( |
54
|
31 |
|
\str_replace( |
55
|
31 |
|
[" ", "\n ", " \n", " \r\n", "\t"], |
56
|
31 |
|
[" ", "\n", "\n", "\r\n", " "], |
57
|
31 |
|
\trim($string) |
58
|
|
|
) |
59
|
|
|
); |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* @param string $text |
64
|
|
|
* @param string $charset |
65
|
|
|
* @return AlternativeText |
66
|
|
|
*/ |
67
|
19 |
|
public static function fromEncodedText(string $text, string $charset): AlternativeText |
68
|
|
|
{ |
69
|
19 |
|
if ($charset === '') { |
70
|
|
|
return new self($text); |
71
|
|
|
} |
72
|
|
|
|
73
|
19 |
|
$charset = \strtoupper($charset); |
74
|
19 |
|
if ($charset === 'UTF-8' || $charset === 'UTF8') { |
75
|
13 |
|
return new self($text); |
76
|
|
|
} |
77
|
|
|
|
78
|
6 |
|
$converted = \iconv($charset, 'UTF-8', $text); |
79
|
6 |
|
if ($converted === false) { |
80
|
|
|
throw new \InvalidArgumentException( |
81
|
|
|
'The encoded text cannot be converted to UTF-8. Is the charset ' . $charset . ' correct?' |
82
|
|
|
); |
83
|
|
|
} |
84
|
|
|
|
85
|
6 |
|
return new self($converted); |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* @param string $html |
90
|
|
|
* @return AlternativeText |
91
|
|
|
*/ |
92
|
45 |
|
public static function fromHtml(string $html): AlternativeText |
93
|
|
|
{ |
94
|
45 |
|
if ($html === '') { |
95
|
39 |
|
return new self($html); |
96
|
|
|
} |
97
|
|
|
|
98
|
25 |
|
$html = self::ensureHtmlCharset($html); |
99
|
25 |
|
$html = \preg_replace('/\h\h+/', ' ', (string)$html); |
100
|
25 |
|
$html = \preg_replace('/\v/', '', (string)$html); |
101
|
25 |
|
$text = new self((string)$html); |
102
|
|
|
|
103
|
|
|
try { |
104
|
25 |
|
$document = new \DOMDocument(); |
105
|
25 |
|
$result = @$document->loadHTML($text->text); |
106
|
25 |
|
if ($result === false) { |
107
|
|
|
throw new \DOMException('Incorrect HTML'); |
108
|
|
|
} |
109
|
|
|
|
110
|
25 |
|
$text->wrapSymbols($document); |
111
|
25 |
|
$text->updateHorizontalRule($document); |
112
|
25 |
|
$text->updateLists($document); |
113
|
25 |
|
$text->updateImages($document); |
114
|
25 |
|
$text->updateLinks($document); |
115
|
25 |
|
$text->removeHead($document); |
116
|
25 |
|
$text->updateParagraphsAndBreaksToNewLine($document); |
117
|
|
|
|
118
|
25 |
|
$text->text = $document->textContent; |
119
|
|
|
} catch (\DOMException $e) { |
120
|
|
|
$text->text = \strip_tags($text->text); |
121
|
|
|
} |
122
|
|
|
|
123
|
25 |
|
return $text; |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* @param \DOMDocument $document |
128
|
|
|
*/ |
129
|
25 |
|
private function updateParagraphsAndBreaksToNewLine(\DOMDocument $document): void |
130
|
|
|
{ |
131
|
25 |
|
$xpath = new \DOMXPath($document); |
132
|
|
|
$break = [ |
133
|
25 |
|
'br' => "\r\n", |
134
|
|
|
'ul' => "\r\n", |
135
|
|
|
'ol' => "\r\n", |
136
|
|
|
'dl' => "\r\n", |
137
|
|
|
]; |
138
|
|
|
|
139
|
25 |
|
$query = $xpath->query('//p|//br|//h1|//h2|//h3|//h4|//h5|//h6|//ul|//ol|//dl|//hr'); |
140
|
25 |
|
if ($query) { |
141
|
|
|
/** @var \DOMElement $element */ |
142
|
25 |
|
foreach ($query as $element) { |
143
|
20 |
|
if (isset($break[$element->nodeName])) { |
144
|
2 |
|
$textNode = $document->createTextNode($break[$element->nodeName]); |
145
|
|
|
} else { |
146
|
20 |
|
$textNode = $document->createTextNode("\r\n\r\n"); |
147
|
|
|
} |
148
|
|
|
|
149
|
20 |
|
$element->appendChild($textNode); |
150
|
|
|
} |
151
|
|
|
} |
152
|
25 |
|
} |
153
|
|
|
|
154
|
|
|
/** |
155
|
|
|
* @param \DOMDocument $document |
156
|
|
|
*/ |
157
|
25 |
|
private function wrapSymbols(\DOMDocument $document): void |
158
|
|
|
{ |
159
|
25 |
|
$xpath = new \DOMXPath($document); |
160
|
|
|
$wrap = [ |
161
|
25 |
|
'h1' => "*", |
162
|
|
|
'h2' => "**", |
163
|
|
|
'h3' => "***", |
164
|
|
|
'h4' => "****", |
165
|
|
|
'h5' => "*****", |
166
|
|
|
'h6' => "******", |
167
|
|
|
'strong' => "*", |
168
|
|
|
'b' => "*", |
169
|
|
|
'em' => "*", |
170
|
|
|
'i' => "*", |
171
|
|
|
]; |
172
|
|
|
|
173
|
25 |
|
$query = $xpath->query('//h1|//h2|//h3|//h4|//h5|//h6|//strong|//b|//em|//i'); |
174
|
25 |
|
if ($query) { |
175
|
|
|
/** @var \DOMElement $element */ |
176
|
25 |
|
foreach ($query as $element) { |
177
|
1 |
|
$element->appendChild( |
178
|
1 |
|
$document->createTextNode($wrap[$element->nodeName]) |
179
|
|
|
); |
180
|
|
|
|
181
|
1 |
|
if ($element->firstChild !== null) { |
182
|
1 |
|
$element->insertBefore( |
183
|
1 |
|
$document->createTextNode($wrap[$element->nodeName]), |
184
|
1 |
|
$element->firstChild |
185
|
|
|
); |
186
|
|
|
} |
187
|
|
|
} |
188
|
|
|
} |
189
|
25 |
|
} |
190
|
|
|
|
191
|
|
|
/** |
192
|
|
|
* @param \DOMDocument $document |
193
|
|
|
*/ |
194
|
25 |
|
private function updateLists(\DOMDocument $document): void |
195
|
|
|
{ |
196
|
25 |
|
$xpath = new \DOMXPath($document); |
197
|
|
|
|
198
|
25 |
|
$query = $xpath->query('//ul/li'); |
199
|
25 |
|
if ($query) { |
200
|
|
|
/** @var \DOMElement $element */ |
201
|
25 |
|
foreach ($query as $element) { |
202
|
1 |
|
if ($element->firstChild !== null) { |
203
|
1 |
|
$element->insertBefore( |
204
|
1 |
|
$document->createTextNode("\t- "), |
205
|
1 |
|
$element->firstChild |
206
|
|
|
); |
207
|
|
|
} else { |
208
|
|
|
$element->appendChild( |
209
|
|
|
$document->createTextNode("\t- ") |
210
|
|
|
); |
211
|
|
|
} |
212
|
|
|
|
213
|
1 |
|
$element->appendChild( |
214
|
1 |
|
$document->createTextNode("\r\n") |
215
|
|
|
); |
216
|
|
|
} |
217
|
|
|
} |
218
|
|
|
|
219
|
25 |
|
$query = $xpath->query('//ol/li'); |
220
|
25 |
|
if ($query) { |
221
|
|
|
/** @var \DOMElement $element */ |
222
|
25 |
|
foreach ($query as $element) { |
223
|
1 |
|
$itemPath = new \DOMXPath($document); |
224
|
1 |
|
$itemNumber = (int)$itemPath->evaluate('string(count(preceding-sibling::li))', $element) + 1; |
225
|
1 |
|
$text = \sprintf("\t%d. ", $itemNumber); |
226
|
|
|
|
227
|
1 |
|
if ($element->firstChild !== null) { |
228
|
1 |
|
$element->insertBefore( |
229
|
1 |
|
$document->createTextNode($text), |
230
|
1 |
|
$element->firstChild |
231
|
|
|
); |
232
|
|
|
} else { |
233
|
|
|
$element->appendChild( |
234
|
|
|
$document->createTextNode($text) |
235
|
|
|
); |
236
|
|
|
} |
237
|
|
|
|
238
|
1 |
|
$element->appendChild( |
239
|
1 |
|
$document->createTextNode("\r\n") |
240
|
|
|
); |
241
|
|
|
} |
242
|
|
|
} |
243
|
|
|
|
244
|
25 |
|
$query = $xpath->query('//dl/dt'); |
245
|
25 |
|
if ($query) { |
246
|
|
|
/** @var \DOMElement $element */ |
247
|
25 |
|
foreach ($query as $element) { |
248
|
1 |
|
$element->appendChild( |
249
|
1 |
|
$document->createTextNode(': ') |
250
|
|
|
); |
251
|
|
|
} |
252
|
|
|
} |
253
|
|
|
|
254
|
25 |
|
$query = $xpath->query('//dl/dd'); |
255
|
25 |
|
if ($query) { |
256
|
|
|
/** @var \DOMElement $element */ |
257
|
25 |
|
foreach ($query as $element) { |
258
|
1 |
|
$element->appendChild( |
259
|
1 |
|
$document->createTextNode("\r\n") |
260
|
|
|
); |
261
|
|
|
} |
262
|
|
|
} |
263
|
25 |
|
} |
264
|
|
|
|
265
|
|
|
/** |
266
|
|
|
* @param \DOMDocument $document |
267
|
|
|
*/ |
268
|
25 |
|
private function updateImages(\DOMDocument $document): void |
269
|
|
|
{ |
270
|
25 |
|
$xpath = new \DOMXPath($document); |
271
|
25 |
|
$query = $xpath->query('//img[@src and @alt]'); |
272
|
|
|
|
273
|
25 |
|
if ($query) { |
274
|
|
|
/** @var \DOMElement $element */ |
275
|
25 |
|
foreach ($query as $element) { |
276
|
1 |
|
$link = $document->createElement('a'); |
277
|
1 |
|
$link->setAttribute('href', $element->getAttribute('src')); |
278
|
1 |
|
$link->textContent = $element->getAttribute('alt'); |
279
|
1 |
|
$parent = $element->parentNode; |
280
|
1 |
|
if ($parent) { |
281
|
1 |
|
$parent->replaceChild($link, $element); |
282
|
|
|
} |
283
|
|
|
} |
284
|
|
|
} |
285
|
25 |
|
} |
286
|
|
|
|
287
|
|
|
/** |
288
|
|
|
* @param \DOMDocument $document |
289
|
|
|
*/ |
290
|
25 |
|
private function updateHorizontalRule(\DOMDocument $document): void |
291
|
|
|
{ |
292
|
25 |
|
$xpath = new \DOMXPath($document); |
293
|
25 |
|
$query = $xpath->query('//hr'); |
294
|
|
|
|
295
|
25 |
|
if ($query) { |
296
|
|
|
/** @var \DOMElement $element */ |
297
|
25 |
|
foreach ($query as $element) { |
298
|
1 |
|
$element->textContent = \str_repeat('=', 78); |
299
|
|
|
} |
300
|
|
|
} |
301
|
25 |
|
} |
302
|
|
|
|
303
|
|
|
/** |
304
|
|
|
* @param \DOMDocument $document |
305
|
|
|
*/ |
306
|
25 |
|
private function updateLinks(\DOMDocument $document): void |
307
|
|
|
{ |
308
|
25 |
|
$xpath = new \DOMXPath($document); |
309
|
25 |
|
$query = $xpath->query('//a[@href and @href != .]'); |
310
|
|
|
|
311
|
25 |
|
if ($query) { |
312
|
|
|
/** @var \DOMElement $element */ |
313
|
25 |
|
foreach ($query as $element) { |
314
|
1 |
|
if ($element->firstChild) { |
315
|
1 |
|
$element->insertBefore( |
316
|
1 |
|
$document->createTextNode('>> '), |
317
|
1 |
|
$element->firstChild |
318
|
|
|
); |
319
|
|
|
} |
320
|
|
|
|
321
|
1 |
|
$element->appendChild( |
322
|
1 |
|
$document->createTextNode( |
323
|
1 |
|
\sprintf( |
324
|
1 |
|
" <%s>", |
325
|
1 |
|
$element->getAttribute('href') |
326
|
|
|
) |
327
|
|
|
) |
328
|
|
|
); |
329
|
|
|
} |
330
|
|
|
} |
331
|
25 |
|
} |
332
|
|
|
|
333
|
|
|
/** |
334
|
|
|
* @param \DOMDocument $document |
335
|
|
|
*/ |
336
|
25 |
|
private function removeHead(\DOMDocument $document): void |
337
|
|
|
{ |
338
|
25 |
|
$heads = $document->getElementsByTagName('head'); |
339
|
25 |
|
while ($heads->length > 0) { |
340
|
|
|
/** @var \DOMElement $head */ |
341
|
25 |
|
$head = $heads->item(0); |
342
|
|
|
/** @var \DOMElement $parent */ |
343
|
25 |
|
$parent = $head->parentNode; |
344
|
25 |
|
$parent->removeChild($head); |
345
|
|
|
} |
346
|
25 |
|
} |
347
|
|
|
|
348
|
|
|
/** |
349
|
|
|
* @param string $unwrappedText |
350
|
|
|
* @param int $width |
351
|
|
|
* @return string |
352
|
|
|
*/ |
353
|
31 |
|
private function wrap(string $unwrappedText, int $width = 75): string |
354
|
|
|
{ |
355
|
31 |
|
$result = []; |
356
|
31 |
|
$carriageReturn = false; |
357
|
31 |
|
$lineChars = -1; |
358
|
31 |
|
$quote = false; |
359
|
31 |
|
$quoteLength = 0; |
360
|
|
|
|
361
|
31 |
|
$iterator = \IntlBreakIterator::createCharacterInstance(\Locale::getDefault()); |
362
|
31 |
|
$iterator->setText($unwrappedText); |
363
|
31 |
|
foreach ($iterator->getPartsIterator() as $char) { |
364
|
29 |
|
if ($char === "\r\n") { |
365
|
4 |
|
$lineChars = -1; |
366
|
4 |
|
$quoteLength = 0; |
367
|
4 |
|
$quote = false; |
368
|
29 |
|
} elseif ($char === "\r") { |
369
|
|
|
$carriageReturn = true; |
370
|
29 |
|
} elseif ($char === "\n") { |
371
|
15 |
|
if (!$carriageReturn) { |
372
|
15 |
|
$char = "\r\n"; |
373
|
|
|
} |
374
|
|
|
|
375
|
15 |
|
$lineChars = -1; |
376
|
15 |
|
$quoteLength = 0; |
377
|
15 |
|
$quote = false; |
378
|
|
|
} |
379
|
|
|
|
380
|
29 |
|
if ($char !== "\r") { |
381
|
29 |
|
$carriageReturn = false; |
382
|
|
|
} |
383
|
|
|
|
384
|
29 |
|
if ($lineChars >= $width && \IntlChar::isWhitespace($char)) { |
385
|
2 |
|
$char = "\r\n" . \str_pad('', $quoteLength, '>'); |
386
|
2 |
|
$lineChars = -1; |
387
|
2 |
|
$quoteLength = 0; |
388
|
2 |
|
$quote = false; |
389
|
|
|
} |
390
|
|
|
|
391
|
29 |
|
$result[] = $char; |
392
|
29 |
|
$lineChars++; |
393
|
|
|
|
394
|
29 |
|
if ($lineChars === 1 && $char === ">") { |
395
|
16 |
|
$quote = true; |
396
|
16 |
|
$quoteLength = 1; |
397
|
29 |
|
} elseif ($quote && $char === ">") { |
398
|
4 |
|
$quoteLength++; |
399
|
|
|
} else { |
400
|
29 |
|
$quote = false; |
401
|
|
|
} |
402
|
|
|
} |
403
|
|
|
|
404
|
31 |
|
return \implode('', $result); |
405
|
|
|
} |
406
|
|
|
|
407
|
|
|
/** |
408
|
|
|
* @param string $html |
409
|
|
|
* @return string |
410
|
|
|
*/ |
411
|
25 |
|
private static function ensureHtmlCharset(string $html): string |
412
|
|
|
{ |
413
|
25 |
|
if ($html === '') { |
414
|
|
|
return ''; |
415
|
|
|
} |
416
|
|
|
|
417
|
25 |
|
if (\strpos($html, 'content="text/html') !== false || \strpos($html, 'charset="') !== false) { |
418
|
24 |
|
return $html; |
419
|
|
|
} |
420
|
|
|
|
421
|
3 |
|
$headCloseStart = \strpos($html, '</head>'); |
422
|
3 |
|
if ($headCloseStart !== false) { |
423
|
1 |
|
return \substr_replace($html, self::DEFAULT_CHARSET, $headCloseStart, 0); |
424
|
|
|
} |
425
|
|
|
|
426
|
2 |
|
$bodyOpenStart = \strpos($html, '<body'); |
427
|
2 |
|
if ($bodyOpenStart !== false) { |
428
|
1 |
|
return \substr_replace($html, '<head>' . self::DEFAULT_CHARSET . '</head>', $bodyOpenStart, 0); |
429
|
|
|
} |
430
|
|
|
|
431
|
1 |
|
return '<html><head>' . self::DEFAULT_CHARSET . '</head><body>' . $html . '</body></html>'; |
432
|
|
|
} |
433
|
|
|
} |
434
|
|
|
|