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
|
|
|
/** |
15
|
|
|
* HtmlUp - A **lightweight** && **fast** `markdown` to HTML Parser. |
16
|
|
|
* |
17
|
|
|
* Supports most of the markdown specs except deep nested elements. |
18
|
|
|
* Check readme.md for the details of its features && limitations. |
19
|
|
|
* |
20
|
|
|
* @author adhocore | Jitendra Adhikari <[email protected]> |
21
|
|
|
* @copyright (c) 2014 Jitendra Adhikari |
22
|
|
|
*/ |
23
|
|
|
class HtmlUp extends BlockElementParser |
24
|
|
|
{ |
25
|
|
|
/** |
26
|
|
|
* Constructor. |
27
|
|
|
* |
28
|
|
|
* @param string $markdown |
29
|
|
|
* @param int $indentWidth |
30
|
|
|
*/ |
31
|
|
|
public function __construct($markdown = \null, $indentWidth = 4) |
32
|
|
|
{ |
33
|
|
|
$this->scan($markdown, $indentWidth); |
34
|
|
|
} |
35
|
|
|
|
36
|
|
|
protected function scan($markdown, $indentWidth = 4) |
37
|
|
|
{ |
38
|
|
|
if ('' === \trim($markdown)) { |
39
|
|
|
return; |
40
|
|
|
} |
41
|
|
|
|
42
|
|
|
$this->indentLen = $indentWidth == 2 ? 2 : 4; |
43
|
|
|
$this->indentStr = $indentWidth == 2 ? ' ' : ' '; |
44
|
|
|
|
45
|
|
|
// Normalize whitespaces |
46
|
|
|
$markdown = \str_replace("\t", $this->indentStr, $markdown); |
47
|
|
|
$markdown = \str_replace(["\r\n", "\r"], "\n", $markdown); |
48
|
|
|
|
49
|
|
|
$this->lines = \array_merge([''], \explode("\n", $markdown), ['']); |
50
|
|
|
} |
51
|
|
|
|
52
|
|
|
public function __toString() |
53
|
|
|
{ |
54
|
|
|
return $this->parse(); |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
/** |
58
|
|
|
* Parse markdown. |
59
|
|
|
* |
60
|
|
|
* @param string $markdown |
61
|
|
|
* @param int $indentWidth |
62
|
|
|
* |
63
|
|
|
* @return string |
64
|
|
|
*/ |
65
|
|
|
public function parse($markdown = \null, $indentWidth = 4) |
66
|
|
|
{ |
67
|
|
|
if (\null !== $markdown) { |
68
|
|
|
$this->reset(\true); |
69
|
|
|
|
70
|
|
|
$this->scan($markdown, $indentWidth); |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
if (empty($this->lines)) { |
74
|
|
|
return ''; |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
$this->parseBlockElements(); |
78
|
|
|
|
79
|
|
|
return (new SpanElementParser)->parse($this->markup); |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
protected function parseBlockElements() |
83
|
|
|
{ |
84
|
|
|
while (isset($this->lines[++$this->pointer])) { |
85
|
|
|
$this->init(); |
86
|
|
|
|
87
|
|
|
if ($this->flush() || $this->raw()) { |
88
|
|
|
continue; |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
$this->quote(); |
92
|
|
|
|
93
|
|
|
if (($block = $this->isBlock()) || $this->inList) { |
94
|
|
|
$this->markup .= $block ? '' : $this->trimmedLine; |
95
|
|
|
|
96
|
|
|
continue; |
97
|
|
|
} |
98
|
|
|
|
99
|
|
|
$this->table() || $this->paragraph(); |
100
|
|
|
} |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
protected function isBlock() |
104
|
|
|
{ |
105
|
|
|
return $this->atx() || $this->setext() || $this->code() || $this->rule() || $this->listt(); |
106
|
|
|
} |
107
|
|
|
|
108
|
|
|
protected function init() |
109
|
|
|
{ |
110
|
|
|
list($this->prevLine, $this->trimmedPrevLine) = [$this->line, $this->trimmedLine]; |
111
|
|
|
|
112
|
|
|
$this->line = $this->lines[$this->pointer]; |
113
|
|
|
$this->trimmedLine = \trim($this->line); |
114
|
|
|
|
115
|
|
|
$this->indent = \strlen($this->line) - \strlen(\ltrim($this->line)); |
116
|
|
|
$this->nextLine = isset($this->lines[$this->pointer + 1]) |
117
|
|
|
? $this->lines[$this->pointer + 1] |
118
|
|
|
: ''; |
119
|
|
|
$this->trimmedNextLine = \trim($this->nextLine); |
120
|
|
|
$this->nextIndent = \strlen($this->nextLine) - \strlen(\ltrim($this->nextLine)); |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
protected function reset($all = \false) |
124
|
|
|
{ |
125
|
|
|
$except = $all ? [] : \array_flip(['lines', 'pointer', 'markup', 'indentStr', 'indentLen']); |
126
|
|
|
|
127
|
|
|
// Reset all current values. |
128
|
|
|
foreach (\get_class_vars(__CLASS__) as $prop => $value) { |
129
|
|
|
isset($except[$prop]) || $this->{$prop} = $value; |
130
|
|
|
} |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
protected function flush() |
134
|
|
|
{ |
135
|
|
|
if ('' !== $this->trimmedLine) { |
136
|
|
|
return \false; |
137
|
|
|
} |
138
|
|
|
|
139
|
|
|
while (!empty($this->stackList)) { |
140
|
|
|
$this->markup .= \array_pop($this->stackList); |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
while (!empty($this->stackBlock)) { |
144
|
|
|
$this->markup .= \array_pop($this->stackBlock); |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
while (!empty($this->stackTable)) { |
148
|
|
|
$this->markup .= \array_pop($this->stackTable); |
149
|
|
|
} |
150
|
|
|
|
151
|
|
|
$this->markup .= "\n"; |
152
|
|
|
|
153
|
|
|
$this->reset(\false); |
154
|
|
|
|
155
|
|
|
return \true; |
156
|
|
|
} |
157
|
|
|
|
158
|
|
|
protected function raw() |
159
|
|
|
{ |
160
|
|
|
if ($this->inHtml || \preg_match(static::RE_RAW, $this->trimmedLine)) { |
161
|
|
|
$this->markup .= "\n$this->line"; |
162
|
|
|
if (!$this->inHtml && empty($this->lines[$this->pointer - 1])) { |
163
|
|
|
$this->inHtml = \true; |
164
|
|
|
} |
165
|
|
|
|
166
|
|
|
return \true; |
167
|
|
|
} |
168
|
|
|
} |
169
|
|
|
|
170
|
|
|
protected function quote() |
171
|
|
|
{ |
172
|
|
|
if (\preg_match(static::RE_MD_QUOTE, $this->line, $quoteMatch)) { |
173
|
|
|
$this->line = \substr($this->line, \strlen($quoteMatch[0])); |
174
|
|
|
$this->trimmedLine = \trim($this->line); |
175
|
|
|
|
176
|
|
|
if (!$this->inQuote || $this->quoteLevel < \strlen($quoteMatch[1])) { |
177
|
|
|
$this->markup .= "\n<blockquote>"; |
178
|
|
|
|
179
|
|
|
$this->stackBlock[] = "\n</blockquote>"; |
180
|
|
|
|
181
|
|
|
$this->quoteLevel++; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
return $this->inQuote = \true; |
185
|
|
|
} |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
protected function atx() |
189
|
|
|
{ |
190
|
|
|
if (isset($this->trimmedLine[0]) && $this->trimmedLine[0] === '#') { |
191
|
|
|
$level = \strlen($this->trimmedLine) - \strlen(\ltrim($this->trimmedLine, '#')); |
192
|
|
|
|
193
|
|
|
if ($level < 7) { |
194
|
|
|
$this->markup .= "\n<h{$level}>" . \ltrim(\ltrim($this->trimmedLine, '# ')) . "</h{$level}>"; |
195
|
|
|
|
196
|
|
|
return \true; |
197
|
|
|
} |
198
|
|
|
} |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
protected function setext() |
202
|
|
|
{ |
203
|
|
|
if (\preg_match(static::RE_MD_SETEXT, $this->nextLine)) { |
204
|
|
|
$level = \trim($this->nextLine, '- ') === '' ? 2 : 1; |
205
|
|
|
|
206
|
|
|
$this->markup .= "\n<h{$level}>{$this->trimmedLine}</h{$level}>"; |
207
|
|
|
|
208
|
|
|
$this->pointer++; |
209
|
|
|
|
210
|
|
|
return \true; |
211
|
|
|
} |
212
|
|
|
} |
213
|
|
|
|
214
|
|
|
protected function code() |
215
|
|
|
{ |
216
|
|
|
$isShifted = ($this->indent - $this->nextIndent) >= $this->indentLen; |
217
|
|
|
$codeBlock = \preg_match(static::RE_MD_CODE, $this->line, $codeMatch); |
218
|
|
|
|
219
|
|
|
if ($codeBlock || (!$this->inList && !$this->inQuote && $isShifted)) { |
220
|
|
|
$lang = isset($codeMatch[1]) |
221
|
|
|
? ' class="language-' . $codeMatch[1] . '"' |
222
|
|
|
: ''; |
223
|
|
|
|
224
|
|
|
$this->markup .= "\n<pre><code{$lang}>"; |
225
|
|
|
|
226
|
|
|
if (!$codeBlock) { |
227
|
|
|
$this->markup .= $this->escape(\substr($this->line, $this->indentLen)); |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
$this->codeInternal($codeBlock); |
231
|
|
|
|
232
|
|
|
$this->pointer++; |
233
|
|
|
|
234
|
|
|
$this->markup .= '</code></pre>'; |
235
|
|
|
|
236
|
|
|
return \true; |
237
|
|
|
} |
238
|
|
|
} |
239
|
|
|
|
240
|
|
|
protected function rule() |
241
|
|
|
{ |
242
|
|
|
if ($this->trimmedPrevLine === '' |
243
|
|
|
&& \preg_match(static::RE_MD_RULE, $this->trimmedLine) |
244
|
|
|
) { |
245
|
|
|
$this->markup .= "\n<hr />"; |
246
|
|
|
|
247
|
|
|
return \true; |
248
|
|
|
} |
249
|
|
|
} |
250
|
|
|
|
251
|
|
|
protected function listt() |
252
|
|
|
{ |
253
|
|
|
$isUl = \in_array(\substr($this->trimmedLine, 0, 2), ['- ', '* ', '+ ']); |
254
|
|
|
|
255
|
|
|
if ($isUl || \preg_match(static::RE_MD_OL, $this->trimmedLine)) { |
256
|
|
|
$wrapper = $isUl ? 'ul' : 'ol'; |
257
|
|
|
|
258
|
|
|
if (!$this->inList) { |
259
|
|
|
$this->stackList[] = "</$wrapper>"; |
260
|
|
|
$this->markup .= "\n<$wrapper>\n"; |
261
|
|
|
$this->inList = \true; |
262
|
|
|
|
263
|
|
|
$this->listLevel++; |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
$this->markup .= '<li>' . \ltrim($this->trimmedLine, '+-*0123456789. '); |
267
|
|
|
|
268
|
|
|
$this->listInternal(); |
269
|
|
|
|
270
|
|
|
return \true; |
271
|
|
|
} |
272
|
|
|
} |
273
|
|
|
|
274
|
|
|
protected function table() |
275
|
|
|
{ |
276
|
|
|
static $headerCount = 0; |
277
|
|
|
|
278
|
|
|
if (!$this->inTable) { |
279
|
|
|
$headerCount = \substr_count(\trim($this->trimmedLine, '|'), '|'); |
280
|
|
|
|
281
|
|
|
return $this->tableInternal($headerCount); |
282
|
|
|
} |
283
|
|
|
|
284
|
|
|
$this->markup .= "<tr>\n"; |
285
|
|
|
|
286
|
|
|
foreach (\explode('|', \trim($this->trimmedLine, '|')) as $i => $col) { |
287
|
|
|
if ($i > $headerCount) { |
288
|
|
|
break; |
289
|
|
|
} |
290
|
|
|
|
291
|
|
|
$col = \trim($col); |
292
|
|
|
$this->markup .= "<td>{$col}</td>\n"; |
293
|
|
|
} |
294
|
|
|
|
295
|
|
|
$this->markup .= "</tr>\n"; |
296
|
|
|
|
297
|
|
|
if (empty($this->trimmedNextLine) |
298
|
|
|
|| !\substr_count(\trim($this->trimmedNextLine, '|'), '|') |
299
|
|
|
) { |
300
|
|
|
$headerCount = 0; |
301
|
|
|
$this->inTable = \false; |
302
|
|
|
$this->stackTable[] = "</tbody>\n</table>"; |
303
|
|
|
} |
304
|
|
|
|
305
|
|
|
return \true; |
306
|
|
|
} |
307
|
|
|
|
308
|
|
|
protected function paragraph() |
309
|
|
|
{ |
310
|
|
|
$this->markup .= $this->inPara ? "\n<br />" : "\n<p>"; |
311
|
|
|
$this->markup .= $this->trimmedLine; |
312
|
|
|
|
313
|
|
|
if (empty($this->trimmedNextLine)) { |
314
|
|
|
$this->markup .= '</p>'; |
315
|
|
|
$this->inPara = \false; |
316
|
|
|
} else { |
317
|
|
|
$this->inPara = \true; |
318
|
|
|
} |
319
|
|
|
} |
320
|
|
|
} |
321
|
|
|
|