1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/* |
4
|
|
|
* This file is part of the HTMLUP package. |
5
|
|
|
* |
6
|
|
|
* (c) Jitendra Adhikari <[email protected]> |
7
|
|
|
* <https://github.com/adhocore> |
8
|
|
|
* |
9
|
|
|
* Licensed under MIT license. |
10
|
|
|
*/ |
11
|
|
|
|
12
|
|
|
namespace Ahc; |
13
|
|
|
|
14
|
|
|
abstract class BlockElementParser |
15
|
|
|
{ |
16
|
|
|
use HtmlHelper; |
17
|
|
|
|
18
|
|
|
const RE_MD_QUOTE = '~^\s*(>+)\s+~'; |
19
|
|
|
const RE_RAW = '/^<\/?\w.*?\/?>/'; |
20
|
|
|
const RE_MD_SETEXT = '~^\s*(={3,}|-{3,})\s*$~'; |
21
|
|
|
const RE_MD_CODE = '/^```\s*([\w-]+)?/'; |
22
|
|
|
const RE_MD_RULE = '~^(_{3,}|\*{3,}|\-{3,})$~'; |
23
|
|
|
const RE_MD_TCOL = '~(\|\s*\:)?\s*\-{3,}\s*(\:\s*\|)?~'; |
24
|
|
|
const RE_MD_OL = '/^\d+\. /'; |
25
|
|
|
|
26
|
|
|
protected $lines = []; |
27
|
|
|
protected $stackList = []; |
28
|
|
|
protected $stackBlock = []; |
29
|
|
|
protected $stackTable = []; |
30
|
|
|
|
31
|
|
|
protected $pointer = -1; |
32
|
|
|
protected $listLevel = 0; |
33
|
|
|
protected $quoteLevel = 0; |
34
|
|
|
protected $indent = 0; |
35
|
|
|
protected $nextIndent = 0; |
36
|
|
|
protected $indentLen = 4; |
37
|
|
|
|
38
|
|
|
protected $indentStr = ' '; |
39
|
|
|
protected $line = ''; |
40
|
|
|
protected $trimmedLine = ''; |
41
|
|
|
protected $prevLine = ''; |
42
|
|
|
protected $trimmedPrevLine = ''; |
43
|
|
|
protected $nextLine = ''; |
44
|
|
|
protected $trimmedNextLine = ''; |
45
|
|
|
protected $markup = ''; |
46
|
|
|
|
47
|
|
|
protected $inList = \false; |
48
|
|
|
protected $inQuote = \false; |
49
|
|
|
protected $inPara = \false; |
50
|
|
|
protected $inHtml = \false; |
51
|
|
|
protected $inTable = \false; |
52
|
|
|
|
53
|
|
|
protected function parseBlockElements() |
54
|
|
|
{ |
55
|
|
|
while (isset($this->lines[++$this->pointer])) { |
56
|
|
|
$this->init(); |
|
|
|
|
57
|
|
|
|
58
|
|
|
if ($this->flush() || $this->raw()) { |
|
|
|
|
59
|
|
|
continue; |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
$this->quote(); |
|
|
|
|
63
|
|
|
|
64
|
|
|
if (($block = $this->isBlock()) || $this->inList) { |
65
|
|
|
$this->markup .= $block ? '' : $this->trimmedLine; |
66
|
|
|
|
67
|
|
|
continue; |
68
|
|
|
} |
69
|
|
|
|
70
|
|
|
$this->table() || $this->paragraph(); |
|
|
|
|
71
|
|
|
} |
72
|
|
|
} |
73
|
|
|
|
74
|
|
|
protected function isBlock() |
75
|
|
|
{ |
76
|
|
|
return $this->atx() || $this->setext() || $this->code() || $this->rule() || $this->listt(); |
77
|
|
|
} |
78
|
|
|
|
79
|
|
|
protected function atx() |
80
|
|
|
{ |
81
|
|
|
if (isset($this->trimmedLine[0]) && $this->trimmedLine[0] === '#') { |
82
|
|
|
$level = \strlen($this->trimmedLine) - \strlen(\ltrim($this->trimmedLine, '#')); |
83
|
|
|
$head = $this->h($level, $this->trimmedLine); |
84
|
|
|
|
85
|
|
|
$this->markup .= $head; |
86
|
|
|
|
87
|
|
|
return (bool) $head; |
88
|
|
|
} |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
protected function setext() |
92
|
|
|
{ |
93
|
|
|
if (\preg_match(static::RE_MD_SETEXT, $this->nextLine)) { |
94
|
|
|
$this->markup .= $this->h($this->nextLine, $this->trimmedLine); |
95
|
|
|
|
96
|
|
|
$this->pointer++; |
97
|
|
|
|
98
|
|
|
return \true; |
99
|
|
|
} |
100
|
|
|
} |
101
|
|
|
|
102
|
|
|
protected function code() |
103
|
|
|
{ |
104
|
|
|
$isShifted = ($this->indent - $this->nextIndent) >= $this->indentLen; |
105
|
|
|
$codeBlock = \preg_match(static::RE_MD_CODE, $this->line, $codeMatch); |
106
|
|
|
|
107
|
|
|
if ($codeBlock || (!$this->inList && !$this->inQuote && $isShifted)) { |
108
|
|
|
$lang = isset($codeMatch[1]) |
109
|
|
|
? ' class="language-' . $codeMatch[1] . '"' |
110
|
|
|
: ''; |
111
|
|
|
|
112
|
|
|
$this->markup .= "\n<pre><code{$lang}>"; |
113
|
|
|
|
114
|
|
|
if (!$codeBlock) { |
115
|
|
|
$this->markup .= $this->escape(\substr($this->line, $this->indentLen)); |
116
|
|
|
} |
117
|
|
|
|
118
|
|
|
$this->codeInternal($codeBlock); |
119
|
|
|
|
120
|
|
|
$this->pointer++; |
121
|
|
|
|
122
|
|
|
$this->markup .= '</code></pre>'; |
123
|
|
|
|
124
|
|
|
return \true; |
125
|
|
|
} |
126
|
|
|
} |
127
|
|
|
|
128
|
|
|
private function codeInternal($codeBlock) |
129
|
|
|
{ |
130
|
|
|
while (isset($this->lines[$this->pointer + 1])) { |
131
|
|
|
$this->line = $this->escape($this->lines[$this->pointer + 1]); |
132
|
|
|
|
133
|
|
|
if (($codeBlock && \substr(\ltrim($this->line), 0, 3) !== '```') |
134
|
|
|
|| \strpos($this->line, $this->indentStr) === 0 |
135
|
|
|
) { |
136
|
|
|
$this->markup .= "\n"; // @todo: donot use \n for first line |
137
|
|
|
$this->markup .= $codeBlock ? $this->line : \substr($this->line, $this->indentLen); |
138
|
|
|
|
139
|
|
|
$this->pointer++; |
140
|
|
|
|
141
|
|
|
continue; |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
break; |
145
|
|
|
} |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
protected function rule() |
149
|
|
|
{ |
150
|
|
|
if ($this->trimmedPrevLine === '' |
151
|
|
|
&& \preg_match(static::RE_MD_RULE, $this->trimmedLine) |
152
|
|
|
) { |
153
|
|
|
$this->markup .= "\n<hr />"; |
154
|
|
|
|
155
|
|
|
return \true; |
156
|
|
|
} |
157
|
|
|
} |
158
|
|
|
|
159
|
|
|
protected function listt() |
160
|
|
|
{ |
161
|
|
|
$isUl = \in_array(\substr($this->trimmedLine, 0, 2), ['- ', '* ', '+ ']); |
162
|
|
|
|
163
|
|
|
if ($isUl || \preg_match(static::RE_MD_OL, $this->trimmedLine)) { |
164
|
|
|
$wrapper = $isUl ? 'ul' : 'ol'; |
165
|
|
|
|
166
|
|
|
if (!$this->inList) { |
167
|
|
|
$this->stackList[] = "</$wrapper>"; |
168
|
|
|
|
169
|
|
|
$this->markup .= "\n<$wrapper>\n"; |
170
|
|
|
$this->inList = \true; |
171
|
|
|
|
172
|
|
|
$this->listLevel++; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
$this->markup .= '<li>' . \ltrim($this->trimmedLine, '+-*0123456789. '); |
176
|
|
|
|
177
|
|
|
$this->listInternal(); |
178
|
|
|
|
179
|
|
|
return \true; |
180
|
|
|
} |
181
|
|
|
} |
182
|
|
|
|
183
|
|
|
private function listInternal() |
184
|
|
|
{ |
185
|
|
|
$isUl = \in_array(\substr($this->trimmedNextLine, 0, 2), ['- ', '* ', '+ ']); |
186
|
|
|
|
187
|
|
|
if ($isUl || \preg_match(static::RE_MD_OL, $this->trimmedNextLine)) { |
188
|
|
|
$wrapper = $isUl ? 'ul' : 'ol'; |
189
|
|
|
if ($this->nextIndent > $this->indent) { |
190
|
|
|
$this->stackList[] = "</li>\n"; |
191
|
|
|
$this->stackList[] = "</$wrapper>"; |
192
|
|
|
$this->markup .= "\n<$wrapper>\n"; |
193
|
|
|
|
194
|
|
|
$this->listLevel++; |
195
|
|
|
} else { |
196
|
|
|
$this->markup .= "</li>\n"; |
197
|
|
|
} |
198
|
|
|
|
199
|
|
|
if ($this->nextIndent < $this->indent) { |
200
|
|
|
$shift = \intval(($this->indent - $this->nextIndent) / $this->indentLen); |
201
|
|
|
|
202
|
|
|
while ($shift--) { |
203
|
|
|
$this->markup .= \array_pop($this->stackList); |
204
|
|
|
|
205
|
|
|
if ($this->listLevel > 2) { |
206
|
|
|
$this->markup .= \array_pop($this->stackList); |
207
|
|
|
} |
208
|
|
|
} |
209
|
|
|
} |
210
|
|
|
} else { |
211
|
|
|
$this->markup .= "</li>\n"; |
212
|
|
|
} |
213
|
|
|
} |
214
|
|
|
|
215
|
|
|
protected function table() |
216
|
|
|
{ |
217
|
|
|
static $headerCount = 0; |
218
|
|
|
|
219
|
|
|
if (!$this->inTable) { |
220
|
|
|
$headerCount = \substr_count(\trim($this->trimmedLine, '|'), '|'); |
221
|
|
|
|
222
|
|
|
return $this->tableInternal($headerCount); |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
$this->markup .= $this->tableRow($this->trimmedLine, $headerCount); |
226
|
|
|
|
227
|
|
|
if (empty($this->trimmedNextLine) |
228
|
|
|
|| !\substr_count(\trim($this->trimmedNextLine, '|'), '|') |
229
|
|
|
) { |
230
|
|
|
$headerCount = 0; |
231
|
|
|
$this->inTable = \false; |
232
|
|
|
$this->stackTable[] = "</tbody>\n</table>"; |
233
|
|
|
} |
234
|
|
|
|
235
|
|
|
return \true; |
236
|
|
|
} |
237
|
|
|
|
238
|
|
|
private function tableInternal($headerCount) |
239
|
|
|
{ |
240
|
|
|
$columnCount = \preg_match_all(static::RE_MD_TCOL, \trim($this->trimmedNextLine, '|')); |
241
|
|
|
|
242
|
|
|
if ($headerCount > 0 && $headerCount <= $columnCount) { |
243
|
|
|
$this->pointer++; |
244
|
|
|
|
245
|
|
|
$this->inTable = \true; |
246
|
|
|
$this->markup .= $this->tableStart($this->trimmedLine); |
247
|
|
|
|
248
|
|
|
return \true; |
249
|
|
|
} |
250
|
|
|
} |
251
|
|
|
} |
252
|
|
|
|