|
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($this->trimmedLine, '#')); |
|
83
|
|
|
|
|
84
|
|
|
if ($level < 7) { |
|
85
|
|
|
$this->markup .= "\n<h{$level}>" . \ltrim(\ltrim($this->trimmedLine, '# ')) . "</h{$level}>"; |
|
86
|
|
|
|
|
87
|
|
|
return \true; |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
protected function setext() |
|
93
|
|
|
{ |
|
94
|
|
|
if (\preg_match(static::RE_MD_SETEXT, $this->nextLine)) { |
|
95
|
|
|
$level = \trim($this->nextLine, '- ') === '' ? 2 : 1; |
|
96
|
|
|
|
|
97
|
|
|
$this->markup .= "\n<h{$level}>{$this->trimmedLine}</h{$level}>"; |
|
98
|
|
|
|
|
99
|
|
|
$this->pointer++; |
|
100
|
|
|
|
|
101
|
|
|
return \true; |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
protected function code() |
|
106
|
|
|
{ |
|
107
|
|
|
$isShifted = ($this->indent - $this->nextIndent) >= $this->indentLen; |
|
108
|
|
|
$codeBlock = \preg_match(static::RE_MD_CODE, $this->line, $codeMatch); |
|
109
|
|
|
|
|
110
|
|
|
if ($codeBlock || (!$this->inList && !$this->inQuote && $isShifted)) { |
|
111
|
|
|
$lang = isset($codeMatch[1]) |
|
112
|
|
|
? ' class="language-' . $codeMatch[1] . '"' |
|
113
|
|
|
: ''; |
|
114
|
|
|
|
|
115
|
|
|
$this->markup .= "\n<pre><code{$lang}>"; |
|
116
|
|
|
|
|
117
|
|
|
if (!$codeBlock) { |
|
118
|
|
|
$this->markup .= $this->escape(\substr($this->line, $this->indentLen)); |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
$this->codeInternal($codeBlock); |
|
122
|
|
|
|
|
123
|
|
|
$this->pointer++; |
|
124
|
|
|
|
|
125
|
|
|
$this->markup .= '</code></pre>'; |
|
126
|
|
|
|
|
127
|
|
|
return \true; |
|
128
|
|
|
} |
|
129
|
|
|
} |
|
130
|
|
|
|
|
131
|
|
|
private function codeInternal($codeBlock) |
|
132
|
|
|
{ |
|
133
|
|
|
while (isset($this->lines[$this->pointer + 1])) { |
|
134
|
|
|
$this->line = $this->escape($this->lines[$this->pointer + 1]); |
|
135
|
|
|
|
|
136
|
|
|
if (($codeBlock && \substr(\ltrim($this->line), 0, 3) !== '```') |
|
137
|
|
|
|| \strpos($this->line, $this->indentStr) === 0 |
|
138
|
|
|
) { |
|
139
|
|
|
$this->markup .= "\n"; // @todo: donot use \n for first line |
|
140
|
|
|
$this->markup .= $codeBlock ? $this->line : \substr($this->line, $this->indentLen); |
|
141
|
|
|
|
|
142
|
|
|
$this->pointer++; |
|
143
|
|
|
} else { |
|
144
|
|
|
break; |
|
145
|
|
|
} |
|
146
|
|
|
} |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
protected function rule() |
|
150
|
|
|
{ |
|
151
|
|
|
if ($this->trimmedPrevLine === '' |
|
152
|
|
|
&& \preg_match(static::RE_MD_RULE, $this->trimmedLine) |
|
153
|
|
|
) { |
|
154
|
|
|
$this->markup .= "\n<hr />"; |
|
155
|
|
|
|
|
156
|
|
|
return \true; |
|
157
|
|
|
} |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
|
|
protected function listt() |
|
161
|
|
|
{ |
|
162
|
|
|
$isUl = \in_array(\substr($this->trimmedLine, 0, 2), ['- ', '* ', '+ ']); |
|
163
|
|
|
|
|
164
|
|
|
if ($isUl || \preg_match(static::RE_MD_OL, $this->trimmedLine)) { |
|
165
|
|
|
$wrapper = $isUl ? 'ul' : 'ol'; |
|
166
|
|
|
|
|
167
|
|
|
if (!$this->inList) { |
|
168
|
|
|
$this->stackList[] = "</$wrapper>"; |
|
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
|
|
|
$this->markup .= "<tr>\n"; |
|
226
|
|
|
|
|
227
|
|
|
foreach (\explode('|', \trim($this->trimmedLine, '|')) as $i => $col) { |
|
228
|
|
|
if ($i > $headerCount) { |
|
229
|
|
|
break; |
|
230
|
|
|
} |
|
231
|
|
|
|
|
232
|
|
|
$col = \trim($col); |
|
233
|
|
|
$this->markup .= "<td>{$col}</td>\n"; |
|
234
|
|
|
} |
|
235
|
|
|
|
|
236
|
|
|
$this->markup .= "</tr>\n"; |
|
237
|
|
|
|
|
238
|
|
|
if (empty($this->trimmedNextLine) |
|
239
|
|
|
|| !\substr_count(\trim($this->trimmedNextLine, '|'), '|') |
|
240
|
|
|
) { |
|
241
|
|
|
$headerCount = 0; |
|
242
|
|
|
$this->inTable = \false; |
|
243
|
|
|
$this->stackTable[] = "</tbody>\n</table>"; |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
|
|
return \true; |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
private function tableInternal($headerCount) |
|
250
|
|
|
{ |
|
251
|
|
|
$columnCount = \preg_match_all(static::RE_MD_TCOL, \trim($this->trimmedNextLine, '|')); |
|
252
|
|
|
|
|
253
|
|
|
if ($headerCount > 0 && $headerCount <= $columnCount) { |
|
254
|
|
|
$this->pointer++; |
|
255
|
|
|
|
|
256
|
|
|
$this->inTable = \true; |
|
257
|
|
|
$this->markup .= "<table>\n<thead>\n<tr>\n"; |
|
258
|
|
|
$this->trimmedLine = \trim($this->trimmedLine, '|'); |
|
259
|
|
|
|
|
260
|
|
|
foreach (\explode('|', $this->trimmedLine) as $hdr) { |
|
261
|
|
|
$this->markup .= '<th>' . \trim($hdr) . "</th>\n"; |
|
262
|
|
|
} |
|
263
|
|
|
|
|
264
|
|
|
$this->markup .= "</tr>\n</thead>\n<tbody>\n"; |
|
265
|
|
|
|
|
266
|
|
|
return \true; |
|
267
|
|
|
} |
|
268
|
|
|
} |
|
269
|
|
|
} |
|
270
|
|
|
|
adhocore /
htmlup
Loading history...