|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* @package s9e\TextFormatter |
|
5
|
|
|
* @copyright Copyright (c) 2010-2018 The s9e Authors |
|
6
|
|
|
* @license http://www.opensource.org/licenses/mit-license.php The MIT License |
|
7
|
|
|
*/ |
|
8
|
|
|
namespace s9e\TextFormatter\Plugins\Litedown\Parser\Passes; |
|
9
|
|
|
|
|
10
|
|
|
class Emphasis extends AbstractPass |
|
11
|
|
|
{ |
|
12
|
|
|
/** |
|
13
|
|
|
* @var bool Whether current EM span is being closed by current emphasis mark |
|
14
|
|
|
*/ |
|
15
|
|
|
protected $closeEm; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* @var bool Whether current EM span is being closed by current emphasis mark |
|
19
|
|
|
*/ |
|
20
|
|
|
protected $closeStrong; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* @var integer Starting position of the current EM span in the text |
|
24
|
|
|
*/ |
|
25
|
|
|
protected $emPos; |
|
26
|
|
|
|
|
27
|
|
|
/** |
|
28
|
|
|
* @var integer Ending position of the current EM span in the text |
|
29
|
|
|
*/ |
|
30
|
|
|
protected $emEndPos; |
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* @var integer Number of emphasis characters unused in current span |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $remaining; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* @var integer Starting position of the current STRONG span in the text |
|
39
|
|
|
*/ |
|
40
|
|
|
protected $strongPos; |
|
41
|
|
|
|
|
42
|
|
|
/** |
|
43
|
|
|
* @var integer Ending position of the current STRONG span in the text |
|
44
|
|
|
*/ |
|
45
|
|
|
protected $strongEndPos; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* {@inheritdoc} |
|
49
|
|
|
*/ |
|
50
|
44 |
|
public function parse() |
|
51
|
|
|
{ |
|
52
|
44 |
|
$this->parseEmphasisByCharacter('*', '/\\*+/'); |
|
53
|
44 |
|
$this->parseEmphasisByCharacter('_', '/_+/'); |
|
54
|
44 |
|
} |
|
55
|
|
|
|
|
56
|
|
|
/** |
|
57
|
|
|
* Adjust the ending position of current EM and STRONG spans |
|
58
|
|
|
* |
|
59
|
|
|
* @return void |
|
60
|
|
|
*/ |
|
61
|
42 |
|
protected function adjustEndingPositions() |
|
62
|
|
|
{ |
|
63
|
42 |
|
if ($this->closeEm && $this->closeStrong) |
|
64
|
|
|
{ |
|
65
|
12 |
|
if ($this->emPos < $this->strongPos) |
|
66
|
|
|
{ |
|
67
|
2 |
|
$this->emEndPos += 2; |
|
68
|
|
|
} |
|
69
|
|
|
else |
|
70
|
|
|
{ |
|
71
|
11 |
|
++$this->strongEndPos; |
|
72
|
|
|
} |
|
73
|
|
|
} |
|
74
|
42 |
|
} |
|
75
|
|
|
|
|
76
|
|
|
/** |
|
77
|
|
|
* Adjust the starting position of current EM and STRONG spans |
|
78
|
|
|
* |
|
79
|
|
|
* If both EM and STRONG are set to start at the same position, we adjust their position |
|
80
|
|
|
* to match the order they are closed. If they start and end at the same position, STRONG |
|
81
|
|
|
* starts before EM to match Markdown's behaviour |
|
82
|
|
|
* |
|
83
|
|
|
* @return void |
|
84
|
|
|
*/ |
|
85
|
42 |
|
protected function adjustStartingPositions() |
|
86
|
|
|
{ |
|
87
|
42 |
|
if (isset($this->emPos) && $this->emPos === $this->strongPos) |
|
88
|
|
|
{ |
|
89
|
14 |
|
if ($this->closeEm) |
|
90
|
|
|
{ |
|
91
|
12 |
|
$this->emPos += 2; |
|
92
|
|
|
} |
|
93
|
|
|
else |
|
94
|
|
|
{ |
|
95
|
3 |
|
++$this->strongPos; |
|
96
|
|
|
} |
|
97
|
|
|
} |
|
98
|
42 |
|
} |
|
99
|
|
|
|
|
100
|
|
|
/** |
|
101
|
|
|
* End current valid EM and STRONG spans |
|
102
|
|
|
* |
|
103
|
|
|
* @return void |
|
104
|
|
|
*/ |
|
105
|
42 |
|
protected function closeSpans() |
|
106
|
|
|
{ |
|
107
|
42 |
|
if ($this->closeEm) |
|
108
|
|
|
{ |
|
109
|
30 |
|
--$this->remaining; |
|
110
|
30 |
|
$this->parser->addTagPair('EM', $this->emPos, 1, $this->emEndPos, 1); |
|
111
|
30 |
|
$this->emPos = null; |
|
112
|
|
|
} |
|
113
|
42 |
|
if ($this->closeStrong) |
|
114
|
|
|
{ |
|
115
|
23 |
|
$this->remaining -= 2; |
|
116
|
23 |
|
$this->parser->addTagPair('STRONG', $this->strongPos, 2, $this->strongEndPos, 2); |
|
117
|
23 |
|
$this->strongPos = null; |
|
118
|
|
|
} |
|
119
|
42 |
|
} |
|
120
|
|
|
|
|
121
|
|
|
/** |
|
122
|
|
|
* Parse emphasis and strong applied using given character |
|
123
|
|
|
* |
|
124
|
|
|
* @param string $character Markup character, either * or _ |
|
125
|
|
|
* @param string $regexp Regexp used to match the series of emphasis character |
|
126
|
|
|
* @return void |
|
127
|
|
|
*/ |
|
128
|
44 |
|
protected function parseEmphasisByCharacter($character, $regexp) |
|
129
|
|
|
{ |
|
130
|
44 |
|
$pos = $this->text->indexOf($character); |
|
131
|
44 |
|
if ($pos === false) |
|
132
|
|
|
{ |
|
133
|
44 |
|
return; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
43 |
|
foreach ($this->getEmphasisByBlock($regexp, $pos) as $block) |
|
137
|
|
|
{ |
|
138
|
43 |
|
$this->processEmphasisBlock($block); |
|
139
|
|
|
} |
|
140
|
43 |
|
} |
|
141
|
|
|
|
|
142
|
|
|
/** |
|
143
|
|
|
* Get emphasis markup split by block |
|
144
|
|
|
* |
|
145
|
|
|
* @param string $regexp Regexp used to match emphasis |
|
146
|
|
|
* @param integer $pos Position in the text of the first emphasis character |
|
147
|
|
|
* @return array[] Each array contains a list of [matchPos, matchLen] pairs |
|
148
|
|
|
*/ |
|
149
|
43 |
|
protected function getEmphasisByBlock($regexp, $pos) |
|
150
|
|
|
{ |
|
151
|
43 |
|
$block = []; |
|
152
|
43 |
|
$blocks = []; |
|
153
|
43 |
|
$breakPos = $this->text->indexOf("\x17", $pos); |
|
154
|
|
|
|
|
155
|
43 |
|
preg_match_all($regexp, $this->text, $matches, PREG_OFFSET_CAPTURE, $pos); |
|
156
|
43 |
|
foreach ($matches[0] as $m) |
|
157
|
|
|
{ |
|
158
|
43 |
|
$matchPos = $m[1]; |
|
159
|
43 |
|
$matchLen = strlen($m[0]); |
|
160
|
|
|
|
|
161
|
|
|
// Test whether we've just passed the limits of a block |
|
162
|
43 |
|
if ($matchPos > $breakPos) |
|
163
|
|
|
{ |
|
164
|
3 |
|
$blocks[] = $block; |
|
165
|
3 |
|
$block = []; |
|
166
|
3 |
|
$breakPos = $this->text->indexOf("\x17", $matchPos); |
|
167
|
|
|
} |
|
168
|
|
|
|
|
169
|
|
|
// Test whether we should ignore this markup |
|
170
|
43 |
|
if (!$this->ignoreEmphasis($matchPos, $matchLen)) |
|
171
|
|
|
{ |
|
172
|
43 |
|
$block[] = [$matchPos, $matchLen]; |
|
173
|
|
|
} |
|
174
|
|
|
} |
|
175
|
43 |
|
$blocks[] = $block; |
|
176
|
|
|
|
|
177
|
43 |
|
return $blocks; |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
/** |
|
181
|
|
|
* Test whether emphasis should be ignored at the given position in the text |
|
182
|
|
|
* |
|
183
|
|
|
* @param integer $matchPos Position of the emphasis in the text |
|
184
|
|
|
* @param integer $matchLen Length of the emphasis |
|
185
|
|
|
* @return bool |
|
186
|
|
|
*/ |
|
187
|
43 |
|
protected function ignoreEmphasis($matchPos, $matchLen) |
|
188
|
|
|
{ |
|
189
|
|
|
// Ignore single underscores between alphanumeric characters |
|
190
|
43 |
|
return ($this->text->charAt($matchPos) === '_' && $matchLen === 1 && $this->text->isSurroundedByAlnum($matchPos, $matchLen)); |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
/** |
|
194
|
|
|
* Open EM and STRONG spans whose content starts at given position |
|
195
|
|
|
* |
|
196
|
|
|
* @param integer $pos |
|
197
|
|
|
* @return void |
|
198
|
|
|
*/ |
|
199
|
42 |
|
protected function openSpans($pos) |
|
200
|
|
|
{ |
|
201
|
42 |
|
if ($this->remaining & 1) |
|
202
|
|
|
{ |
|
203
|
37 |
|
$this->emPos = $pos - $this->remaining; |
|
204
|
|
|
} |
|
205
|
42 |
|
if ($this->remaining & 2) |
|
206
|
|
|
{ |
|
207
|
25 |
|
$this->strongPos = $pos - $this->remaining; |
|
208
|
|
|
} |
|
209
|
42 |
|
} |
|
210
|
|
|
|
|
211
|
|
|
/** |
|
212
|
|
|
* Process a list of emphasis markup strings |
|
213
|
|
|
* |
|
214
|
|
|
* @param array[] $block List of [matchPos, matchLen] pairs |
|
215
|
|
|
* @return void |
|
216
|
|
|
*/ |
|
217
|
43 |
|
protected function processEmphasisBlock(array $block) |
|
218
|
|
|
{ |
|
219
|
43 |
|
$this->emPos = null; |
|
220
|
43 |
|
$this->strongPos = null; |
|
221
|
43 |
|
foreach ($block as list($matchPos, $matchLen)) |
|
222
|
|
|
{ |
|
223
|
42 |
|
$this->processEmphasisMatch($matchPos, $matchLen); |
|
224
|
|
|
} |
|
225
|
43 |
|
} |
|
226
|
|
|
|
|
227
|
|
|
/** |
|
228
|
|
|
* Process an emphasis mark |
|
229
|
|
|
* |
|
230
|
|
|
* @param integer $matchPos |
|
231
|
|
|
* @param integer $matchLen |
|
232
|
|
|
* @return void |
|
233
|
|
|
*/ |
|
234
|
42 |
|
protected function processEmphasisMatch($matchPos, $matchLen) |
|
235
|
|
|
{ |
|
236
|
42 |
|
$canOpen = !$this->text->isBeforeWhitespace($matchPos + $matchLen - 1); |
|
237
|
42 |
|
$canClose = !$this->text->isAfterWhitespace($matchPos); |
|
238
|
42 |
|
$closeLen = ($canClose) ? min($matchLen, 3) : 0; |
|
239
|
|
|
|
|
240
|
42 |
|
$this->closeEm = ($closeLen & 1) && isset($this->emPos); |
|
241
|
42 |
|
$this->closeStrong = ($closeLen & 2) && isset($this->strongPos); |
|
242
|
42 |
|
$this->emEndPos = $matchPos; |
|
243
|
42 |
|
$this->strongEndPos = $matchPos; |
|
244
|
42 |
|
$this->remaining = $matchLen; |
|
245
|
|
|
|
|
246
|
42 |
|
$this->adjustStartingPositions(); |
|
247
|
42 |
|
$this->adjustEndingPositions(); |
|
248
|
42 |
|
$this->closeSpans(); |
|
249
|
|
|
|
|
250
|
|
|
// Adjust the length of unused markup remaining in current match |
|
251
|
42 |
|
$this->remaining = ($canOpen) ? min($this->remaining, 3) : 0; |
|
252
|
42 |
|
$this->openSpans($matchPos + $matchLen); |
|
253
|
|
|
} |
|
254
|
|
|
} |