|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* (c) Dennis Meckel |
|
5
|
|
|
* |
|
6
|
|
|
* For the full copyright and license information, |
|
7
|
|
|
* please view the LICENSE file that was distributed with this source code. |
|
8
|
|
|
*/ |
|
9
|
|
|
|
|
10
|
|
|
namespace Rayne\Ecoji; |
|
11
|
|
|
|
|
12
|
|
|
use InvalidArgumentException; |
|
13
|
|
|
use OutOfBoundsException; |
|
14
|
|
|
|
|
15
|
|
|
/** |
|
16
|
|
|
* TODO A stream filter implementation (`php_user_filter`) would be nice. But how would the filter be configured? |
|
17
|
|
|
*/ |
|
18
|
|
|
class EcojiStream |
|
19
|
|
|
{ |
|
20
|
|
|
/** |
|
21
|
|
|
* @var Ecoji |
|
22
|
|
|
*/ |
|
23
|
|
|
private $ecoji; |
|
24
|
|
|
|
|
25
|
|
|
/** |
|
26
|
|
|
* @var int |
|
27
|
|
|
*/ |
|
28
|
|
|
private $wrap = 0; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* @param Ecoji|null $ecoji |
|
32
|
|
|
*/ |
|
33
|
1608 |
|
public function __construct(Ecoji $ecoji = null) |
|
34
|
|
|
{ |
|
35
|
1608 |
|
$this->ecoji = $ecoji ?: new Ecoji; |
|
36
|
1608 |
|
} |
|
37
|
|
|
|
|
38
|
|
|
/** |
|
39
|
|
|
* Wrapping is disabled by default. |
|
40
|
|
|
* |
|
41
|
|
|
* @param int $wrap `0` disables wrapping. |
|
42
|
|
|
* @return $this |
|
43
|
|
|
* @throws OutOfBoundsException |
|
44
|
|
|
*/ |
|
45
|
21 |
|
public function setWrap(int $wrap) |
|
46
|
|
|
{ |
|
47
|
21 |
|
if ($wrap < 0) { |
|
48
|
3 |
|
throw new OutOfBoundsException($wrap); |
|
49
|
|
|
} |
|
50
|
|
|
|
|
51
|
18 |
|
$this->wrap = $wrap; |
|
52
|
|
|
|
|
53
|
18 |
|
return $this; |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Encodes the source stream. The destination stream will not receive a trailing newline. |
|
58
|
|
|
* |
|
59
|
|
|
* @param resource $source |
|
60
|
|
|
* @param resource $destination |
|
61
|
|
|
*/ |
|
62
|
801 |
|
public function encode($source, $destination) |
|
63
|
|
|
{ |
|
64
|
801 |
|
$printNewline = false; |
|
65
|
801 |
|
$untilWrap = $this->wrap; |
|
66
|
801 |
|
$wrap = $this->wrap; |
|
67
|
|
|
|
|
68
|
|
|
// Read multiples of five bytes at a time to prevent padding when encoding. |
|
69
|
|
|
// Padding is only allowed for the last four or less bytes. |
|
70
|
801 |
|
while ('' !== ($bytes = stream_get_contents($source, 5))) { |
|
71
|
801 |
|
if ($wrap) { |
|
72
|
18 |
|
$buffer = $this->ecoji->encode($bytes); // Up to five bytes get encoded … |
|
73
|
18 |
|
$bufferLength = mb_strlen($buffer); // … and exactly four emojis are returned. |
|
74
|
|
|
|
|
75
|
18 |
|
while ($bufferLength > 0) { |
|
76
|
18 |
|
if ($printNewline) { |
|
77
|
12 |
|
$printNewline = false; |
|
78
|
12 |
|
fwrite($destination, "\n"); |
|
79
|
|
|
} |
|
80
|
|
|
|
|
81
|
18 |
|
if ($bufferLength < $untilWrap) { |
|
82
|
12 |
|
fwrite($destination, $buffer); |
|
83
|
12 |
|
$untilWrap -= $bufferLength; |
|
84
|
12 |
|
break; |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
15 |
|
$printNewline = true; |
|
88
|
|
|
|
|
89
|
15 |
|
if ($bufferLength == $untilWrap) { |
|
90
|
12 |
|
fwrite($destination, $buffer); |
|
91
|
12 |
|
$untilWrap = $wrap; |
|
92
|
12 |
|
break; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
12 |
|
fwrite($destination, mb_substr($buffer, 0, $untilWrap)); |
|
96
|
12 |
|
$buffer = mb_substr($buffer, $untilWrap); |
|
97
|
12 |
|
$bufferLength -= $untilWrap; |
|
98
|
12 |
|
$untilWrap = $wrap; |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
18 |
|
continue; |
|
102
|
|
|
} |
|
103
|
|
|
|
|
104
|
783 |
|
fwrite($destination, $this->ecoji->encode($bytes)); |
|
105
|
|
|
} |
|
106
|
801 |
|
} |
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* @param resource $source |
|
110
|
|
|
* @param resource $destination |
|
111
|
|
|
*/ |
|
112
|
804 |
|
public function decode($source, $destination) |
|
113
|
|
|
{ |
|
114
|
804 |
|
$chars = []; |
|
115
|
|
|
|
|
116
|
804 |
|
while ('' !== ($char = $this->readUnicodeChar($source))) { |
|
117
|
|
|
// Skip newlines as Ecoji ignores them when decoding. |
|
118
|
|
|
// To be more precise: `decode()` throws newlines away |
|
119
|
|
|
// and expects groups of four remaining UTF-8 characters. |
|
120
|
789 |
|
if ($char === "\n") { |
|
121
|
3 |
|
continue; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
789 |
|
$chars[] = $char; |
|
125
|
|
|
|
|
126
|
789 |
|
if (count($chars) == 4) { |
|
127
|
789 |
|
fwrite($destination, $this->ecoji->decode(implode('', $chars))); |
|
128
|
789 |
|
$chars = []; |
|
129
|
|
|
} |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
// Un-decoded characters remaining => Invalid Ecoji encoding! |
|
133
|
789 |
|
if (count($chars) !== 0) { |
|
134
|
3 |
|
throw new InvalidArgumentException('Invalid Ecoji encoding: ' . implode('', $chars)); |
|
135
|
|
|
} |
|
136
|
786 |
|
} |
|
137
|
|
|
|
|
138
|
|
|
/** |
|
139
|
|
|
* @param resource $stream |
|
140
|
|
|
* @return string |
|
141
|
|
|
*/ |
|
142
|
804 |
|
private function readUnicodeChar($stream): string |
|
143
|
|
|
{ |
|
144
|
804 |
|
$bytes = ''; |
|
145
|
|
|
|
|
146
|
|
|
// Read up to four bytes and return when a valid UTF-8 sequence was detected. |
|
147
|
804 |
|
for ($i = 0; $i < 4; $i++) { |
|
148
|
804 |
|
$bytes .= stream_get_contents($stream, 1); |
|
149
|
|
|
|
|
150
|
|
|
// EOF reached. |
|
151
|
804 |
|
if ($bytes === '') { |
|
152
|
789 |
|
return ''; |
|
153
|
|
|
} |
|
154
|
|
|
|
|
155
|
804 |
|
if (mb_check_encoding($bytes, 'UTF-8')) { |
|
156
|
789 |
|
return $bytes; |
|
157
|
|
|
} |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
15 |
|
throw new InvalidArgumentException('Invalid Unicode: ' . $bytes); |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
|
|
// |
|
164
|
|
|
// TODO Compare the performance of the commented code below with the "simple" implementation above. |
|
165
|
|
|
// |
|
166
|
|
|
//private function readUnicodeChar($stream): string { |
|
167
|
|
|
// $byte0 = stream_get_contents($stream, 1); |
|
168
|
|
|
// |
|
169
|
|
|
// // EOF |
|
170
|
|
|
// if ($byte0 === '') { |
|
171
|
|
|
// return ''; |
|
172
|
|
|
// } |
|
173
|
|
|
// |
|
174
|
|
|
// // 0xxxxxxx |
|
175
|
|
|
// if (ord($byte0) < 0x80) { |
|
176
|
|
|
// return $byte0; |
|
177
|
|
|
// } |
|
178
|
|
|
// |
|
179
|
|
|
// // 110yyyyy 10xxxxxx |
|
180
|
|
|
// if ((ord($byte0) & 0xE0) === 0xC0) { |
|
181
|
|
|
// $byte1 = stream_get_contents($stream, 1); |
|
182
|
|
|
// |
|
183
|
|
|
// if ((ord($byte1) & 0xC0) !== 0x80) { |
|
184
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
185
|
|
|
// } |
|
186
|
|
|
// |
|
187
|
|
|
// return $byte0 . $byte1; |
|
188
|
|
|
// } |
|
189
|
|
|
// |
|
190
|
|
|
// // 1110zzzz 10yyyyyy 10xxxxxx |
|
191
|
|
|
// if ((ord($byte0) & 0xF0) === 0xE0) { |
|
192
|
|
|
// $byte1 = stream_get_contents($stream, 1); |
|
193
|
|
|
// |
|
194
|
|
|
// if ((ord($byte1) & 0xC0) !== 0x80) { |
|
195
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
196
|
|
|
// } |
|
197
|
|
|
// |
|
198
|
|
|
// $byte2 = stream_get_contents($stream, 1); |
|
199
|
|
|
// |
|
200
|
|
|
// if ((ord($byte2) & 0xC0) !== 0x80) { |
|
201
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
202
|
|
|
// } |
|
203
|
|
|
// |
|
204
|
|
|
// return $byte0 . $byte1 . $byte2; |
|
205
|
|
|
// } |
|
206
|
|
|
// |
|
207
|
|
|
// // 11110uuu 10uuzzzz 10yyyyyy 10xxxxxx |
|
208
|
|
|
// if ((ord($byte0) & 0xF8) === 0xF0) { |
|
209
|
|
|
// $byte1 = stream_get_contents($stream, 1); |
|
210
|
|
|
// |
|
211
|
|
|
// if ((ord($byte1) & 0xC0) !== 0x80) { |
|
212
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
213
|
|
|
// } |
|
214
|
|
|
// |
|
215
|
|
|
// $byte2 = stream_get_contents($stream, 1); |
|
216
|
|
|
// |
|
217
|
|
|
// if ((ord($byte2) & 0xC0) !== 0x80) { |
|
218
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
219
|
|
|
// } |
|
220
|
|
|
// |
|
221
|
|
|
// $byte3 = stream_get_contents($stream, 1); |
|
222
|
|
|
// |
|
223
|
|
|
// if ((ord($byte3) & 0xC0) !== 0x80) { |
|
224
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
225
|
|
|
// } |
|
226
|
|
|
// |
|
227
|
|
|
// return $byte0 . $byte1 . $byte2 . $byte3; |
|
228
|
|
|
// } |
|
229
|
|
|
// |
|
230
|
|
|
// throw new InvalidArgumentException('Invalid encoding.'); |
|
231
|
|
|
//} |
|
232
|
|
|
} |
|
233
|
|
|
|
Rayne /
ecoji-php
