|
1
|
|
|
<?php |
|
2
|
|
|
/** |
|
3
|
|
|
* @copyright Copyright (c) 2014 Carsten Brandt |
|
4
|
|
|
* @license https://github.com/cebe/markdown/blob/master/LICENSE |
|
5
|
|
|
* @link https://github.com/cebe/markdown#readme |
|
6
|
|
|
*/ |
|
7
|
|
|
|
|
8
|
|
|
namespace cebe\markdown\latex; |
|
9
|
|
|
|
|
10
|
|
|
use cebe\markdown\block\CodeTrait; |
|
11
|
|
|
use cebe\markdown\block\HeadlineTrait; |
|
12
|
|
|
use cebe\markdown\block\ListTrait; |
|
13
|
|
|
use cebe\markdown\block\QuoteTrait; |
|
14
|
|
|
use cebe\markdown\block\RuleTrait; |
|
15
|
|
|
|
|
16
|
|
|
use cebe\markdown\inline\CodeTrait as InlineCodeTrait; |
|
17
|
|
|
use cebe\markdown\inline\EmphStrongTrait; |
|
18
|
|
|
use cebe\markdown\inline\LinkTrait; |
|
19
|
|
|
|
|
20
|
|
|
use MikeVanRiel\TextToLatex; |
|
21
|
|
|
|
|
22
|
|
|
/** |
|
23
|
|
|
* Markdown parser for the [initial markdown spec](http://daringfireball.net/projects/markdown/syntax). |
|
24
|
|
|
* |
|
25
|
|
|
* @author Carsten Brandt <[email protected]> |
|
26
|
|
|
*/ |
|
27
|
|
|
class Markdown extends \cebe\markdown\Parser |
|
28
|
|
|
{ |
|
29
|
|
|
// include block element parsing using traits |
|
30
|
|
|
use CodeTrait; |
|
31
|
|
|
use HeadlineTrait; |
|
32
|
|
|
use ListTrait { |
|
33
|
|
|
// Check Ul List before headline |
|
34
|
|
|
identifyUl as protected identifyBUl; |
|
35
|
|
|
consumeUl as protected consumeBUl; |
|
36
|
|
|
} |
|
37
|
|
|
use QuoteTrait; |
|
38
|
|
|
use RuleTrait { |
|
39
|
|
|
// Check Hr before checking lists |
|
40
|
|
|
identifyHr as protected identifyAHr; |
|
41
|
|
|
consumeHr as protected consumeAHr; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
// include inline element parsing using traits |
|
45
|
|
|
use InlineCodeTrait; |
|
46
|
|
|
use EmphStrongTrait; |
|
47
|
|
|
use LinkTrait; |
|
48
|
|
|
|
|
49
|
|
|
/** |
|
50
|
|
|
* @var string this string will be prefixed to all auto generated labels. |
|
51
|
|
|
* This can be used to disambiguate labels when combining multiple markdown files into one document. |
|
52
|
|
|
*/ |
|
53
|
|
|
public $labelPrefix = ''; |
|
54
|
|
|
|
|
55
|
|
|
const LINK_STYLE_FOOTNOTE = 'footnote'; |
|
56
|
|
|
const LINK_STYLE_HREF = 'href'; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* @var string link style defines how links are rendered in LaTeX, there are two different options: |
|
60
|
|
|
* |
|
61
|
|
|
* - `footnote` (default) - render all links with a footnote, which contains the full URL of the link. This is good for printing the PDF. |
|
62
|
|
|
* - `href` - render all links with a hyperref, similar to HTML, the link target is not visible in this case. |
|
63
|
|
|
*/ |
|
64
|
|
|
public $linkStyle = self::LINK_STYLE_FOOTNOTE; |
|
65
|
|
|
|
|
66
|
|
|
/** |
|
67
|
|
|
* @var array these are "escapeable" characters. When using one of these prefixed with a |
|
68
|
|
|
* backslash, the character will be outputted without the backslash and is not interpreted |
|
69
|
|
|
* as markdown. |
|
70
|
|
|
*/ |
|
71
|
|
|
protected $escapeCharacters = [ |
|
72
|
|
|
'\\', // backslash |
|
73
|
|
|
'`', // backtick |
|
74
|
|
|
'*', // asterisk |
|
75
|
|
|
'_', // underscore |
|
76
|
|
|
'{', '}', // curly braces |
|
77
|
|
|
'[', ']', // square brackets |
|
78
|
|
|
'(', ')', // parentheses |
|
79
|
|
|
'#', // hash mark |
|
80
|
|
|
'+', // plus sign |
|
81
|
|
|
'-', // minus sign (hyphen) |
|
82
|
|
|
'.', // dot |
|
83
|
|
|
'!', // exclamation mark |
|
84
|
|
|
'<', '>', |
|
85
|
|
|
]; |
|
86
|
|
|
|
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* @inheritDoc |
|
90
|
|
|
*/ |
|
91
|
23 |
|
protected function prepare() |
|
92
|
|
|
{ |
|
93
|
|
|
// reset references |
|
94
|
23 |
|
$this->references = []; |
|
95
|
23 |
|
} |
|
96
|
|
|
|
|
97
|
|
|
/** |
|
98
|
|
|
* Consume lines for a paragraph |
|
99
|
|
|
* |
|
100
|
|
|
* Allow headlines and code to break paragraphs |
|
101
|
|
|
*/ |
|
102
|
8 |
|
protected function consumeParagraph($lines, $current) |
|
103
|
|
|
{ |
|
104
|
|
|
// consume until newline |
|
105
|
8 |
|
$content = []; |
|
106
|
8 |
|
for ($i = $current, $count = count($lines); $i < $count; $i++) { |
|
107
|
8 |
|
$line = $lines[$i]; |
|
108
|
8 |
|
if (!empty($line) && ltrim($line) !== '' && |
|
109
|
8 |
|
!($line[0] === "\t" || $line[0] === " " && strncmp($line, ' ', 4) === 0) && |
|
110
|
8 |
|
!$this->identifyHeadline($line, $lines, $i)) |
|
111
|
8 |
|
{ |
|
112
|
8 |
|
$content[] = $line; |
|
113
|
8 |
|
} else { |
|
114
|
7 |
|
break; |
|
115
|
|
|
} |
|
116
|
8 |
|
} |
|
117
|
|
|
$block = [ |
|
118
|
8 |
|
'paragraph', |
|
119
|
8 |
|
'content' => $this->parseInline(implode("\n", $content)), |
|
120
|
8 |
|
]; |
|
121
|
8 |
|
return [$block, --$i]; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
|
|
125
|
|
|
// rendering adjusted for LaTeX output |
|
126
|
|
|
|
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* @inheritdoc |
|
130
|
|
|
*/ |
|
131
|
21 |
|
protected function renderParagraph($block) |
|
132
|
|
|
{ |
|
133
|
21 |
|
return $this->renderAbsy($block['content']) . "\n\n"; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
/** |
|
137
|
|
|
* @inheritdoc |
|
138
|
|
|
*/ |
|
139
|
2 |
|
protected function renderQuote($block) |
|
140
|
|
|
{ |
|
141
|
2 |
|
return '\begin{quote}' . $this->renderAbsy($block['content']) . "\\end{quote}\n"; |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
/** |
|
145
|
|
|
* @inheritdoc |
|
146
|
|
|
*/ |
|
147
|
3 |
|
protected function renderCode($block) |
|
148
|
|
|
{ |
|
149
|
3 |
|
$language = isset($block['language']) ? "\\lstset{language={$block['language']}}" : '\lstset{language={}}'; |
|
150
|
|
|
|
|
151
|
3 |
|
$content = $block['content']; |
|
152
|
|
|
// replace No-Break Space characters in code block, which do not render in LaTeX |
|
153
|
3 |
|
$content = preg_replace("/[\x{00a0}\x{202f}]/u", ' ', $content); |
|
154
|
|
|
|
|
155
|
3 |
|
return "$language\\begin{lstlisting}\n{$content}\n\\end{lstlisting}\n"; |
|
156
|
|
|
} |
|
157
|
|
|
|
|
158
|
|
|
/** |
|
159
|
|
|
* @inheritdoc |
|
160
|
|
|
*/ |
|
161
|
2 |
|
protected function renderList($block) |
|
162
|
|
|
{ |
|
163
|
2 |
|
$type = ($block['list'] === 'ol') ? 'enumerate' : 'itemize'; |
|
164
|
2 |
|
$output = "\\begin{{$type}}\n"; |
|
165
|
|
|
|
|
166
|
2 |
|
foreach ($block['items'] as $item => $itemLines) { |
|
167
|
2 |
|
$output .= '\item ' . $this->renderAbsy($itemLines). "\n"; |
|
168
|
2 |
|
} |
|
169
|
|
|
|
|
170
|
2 |
|
return "$output\\end{{$type}}\n"; |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
/** |
|
174
|
|
|
* @inheritdoc |
|
175
|
|
|
*/ |
|
176
|
3 |
|
protected function renderHeadline($block) |
|
177
|
|
|
{ |
|
178
|
3 |
|
$content = $this->renderAbsy($block['content']); |
|
179
|
3 |
|
switch($block['level']) { |
|
180
|
3 |
|
case 1: return "\\section{{$content}}\n"; |
|
181
|
3 |
|
case 2: return "\\subsection{{$content}}\n"; |
|
182
|
2 |
|
case 3: return "\\subsubsection{{$content}}\n"; |
|
183
|
2 |
|
default: return "\\paragraph{{$content}}\n"; |
|
184
|
2 |
|
} |
|
185
|
|
|
} |
|
186
|
|
|
|
|
187
|
|
|
/** |
|
188
|
|
|
* @inheritdoc |
|
189
|
|
|
*/ |
|
190
|
2 |
|
protected function renderHr($block) |
|
|
|
|
|
|
191
|
|
|
{ |
|
192
|
2 |
|
return "\n\\noindent\\rule{\\textwidth}{0.4pt}\n"; |
|
193
|
|
|
} |
|
194
|
|
|
|
|
195
|
|
|
/** |
|
196
|
|
|
* @inheritdoc |
|
197
|
|
|
*/ |
|
198
|
2 |
|
protected function renderLink($block) |
|
199
|
|
|
{ |
|
200
|
2 |
View Code Duplication |
if (isset($block['refkey'])) { |
|
|
|
|
|
|
201
|
|
|
if (($ref = $this->lookupReference($block['refkey'])) !== false) { |
|
202
|
|
|
$block = array_merge($block, $ref); |
|
203
|
|
|
} else { |
|
204
|
|
|
return $block['orig']; |
|
205
|
|
|
} |
|
206
|
|
|
} |
|
207
|
|
|
|
|
208
|
2 |
|
$url = $block['url']; |
|
209
|
2 |
|
$text = $this->renderAbsy($block['text']); |
|
210
|
2 |
|
if (strpos($url, '://') === false) { |
|
211
|
|
|
// consider all non absolute links as relative in the document |
|
212
|
|
|
// $title is ignored in this case. |
|
213
|
|
|
if (isset($url[0]) && $url[0] === '#') { |
|
214
|
|
|
$url = $this->labelPrefix . $url; |
|
215
|
|
|
} |
|
216
|
|
|
return '\hyperref['.str_replace('#', '::', $url).']{' . $text . '}'; |
|
217
|
|
|
} else { |
|
218
|
2 |
|
if ($this->linkStyle === self::LINK_STYLE_HREF) { |
|
219
|
|
|
return '\href{' . $this->escapeUrl($url) . '}{' . $text . '}'; |
|
220
|
|
|
} |
|
221
|
2 |
|
return $text . '\\footnote{' . (empty($block['title']) ? '' : $this->escapeLatex($block['title']) . ': ') . '\url{' . $this->escapeUrl($url) . '}}'; |
|
222
|
|
|
} |
|
223
|
|
|
} |
|
224
|
|
|
|
|
225
|
|
|
/** |
|
226
|
|
|
* @inheritdoc |
|
227
|
|
|
*/ |
|
228
|
2 |
|
protected function renderImage($block) |
|
229
|
|
|
{ |
|
230
|
2 |
View Code Duplication |
if (isset($block['refkey'])) { |
|
|
|
|
|
|
231
|
|
|
if (($ref = $this->lookupReference($block['refkey'])) !== false) { |
|
232
|
|
|
$block = array_merge($block, $ref); |
|
233
|
|
|
} else { |
|
234
|
|
|
return $block['orig']; |
|
235
|
|
|
} |
|
236
|
|
|
} |
|
237
|
|
|
|
|
238
|
|
|
// TODO create figure with caption with title |
|
239
|
|
|
$replaces = [ |
|
240
|
2 |
|
'%' => '\\%', |
|
241
|
2 |
|
'{' => '\\%7B', |
|
242
|
2 |
|
'}' => '\\%7D', |
|
243
|
2 |
|
'\\' => '\\\\', |
|
244
|
2 |
|
'#' => '\\#', |
|
245
|
2 |
|
'$' => '\\%24', |
|
246
|
2 |
|
]; |
|
247
|
2 |
|
$url = str_replace(array_keys($replaces), array_values($replaces), $block['url']); |
|
248
|
2 |
|
return "\\noindent\\includegraphics[width=\\textwidth]{{$url}}"; |
|
249
|
|
|
} |
|
250
|
|
|
|
|
251
|
|
|
/** |
|
252
|
|
|
* Parses <a name="..."></a> tags as reference labels |
|
253
|
|
|
*/ |
|
254
|
2 |
|
private function parseInlineHtml($text) |
|
255
|
|
|
{ |
|
256
|
2 |
|
if (strpos($text, '>') !== false) { |
|
257
|
|
|
// convert a name markers to \labels |
|
258
|
2 |
|
if (preg_match('~^<((a|span)) (name|id)="(.*?)">.*?</\1>~i', $text, $matches)) { |
|
259
|
|
|
return [ |
|
260
|
2 |
|
['label', 'name' => str_replace('#', '::', $this->labelPrefix . $matches[4])], |
|
261
|
2 |
|
strlen($matches[0]) |
|
262
|
2 |
|
]; |
|
263
|
|
|
} |
|
264
|
|
|
} |
|
265
|
|
|
return [['text', '<'], 1]; |
|
266
|
|
|
} |
|
267
|
|
|
|
|
268
|
|
|
/** |
|
269
|
|
|
* renders a reference label |
|
270
|
|
|
*/ |
|
271
|
2 |
|
protected function renderLabel($block) |
|
272
|
|
|
{ |
|
273
|
2 |
|
return "\\label{{$block['name']}}"; |
|
274
|
|
|
} |
|
275
|
|
|
|
|
276
|
|
|
/** |
|
277
|
|
|
* @inheritdoc |
|
278
|
|
|
*/ |
|
279
|
2 |
|
protected function renderEmail($block) |
|
280
|
|
|
{ |
|
281
|
2 |
|
$email = $this->escapeUrl($block[1]); |
|
282
|
2 |
|
return "\\href{mailto:{$email}}{{$email}}"; |
|
283
|
|
|
} |
|
284
|
|
|
|
|
285
|
|
|
/** |
|
286
|
|
|
* @inheritdoc |
|
287
|
|
|
*/ |
|
288
|
2 |
|
protected function renderUrl($block) |
|
289
|
|
|
{ |
|
290
|
2 |
|
return '\url{' . $this->escapeUrl($block[1]) . '}'; |
|
291
|
|
|
} |
|
292
|
|
|
|
|
293
|
|
|
/** |
|
294
|
|
|
* @inheritdoc |
|
295
|
|
|
*/ |
|
296
|
1 |
|
protected function renderInlineCode($block) |
|
297
|
|
|
{ |
|
298
|
|
|
// replace No-Break Space characters in code block, which do not render in LaTeX |
|
299
|
1 |
|
$content = preg_replace("/[\x{00a0}\x{202f}]/u", ' ', $block[1]); |
|
300
|
|
|
|
|
301
|
1 |
|
if (strpos($content, '|') !== false) { |
|
302
|
1 |
|
return '\\lstinline`' . str_replace("\n", ' ', $content) . '`'; // TODO make this more robust against code containing backticks |
|
303
|
|
|
} else { |
|
304
|
|
|
return '\\lstinline|' . str_replace("\n", ' ', $content) . '|'; |
|
305
|
|
|
} |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
/** |
|
309
|
|
|
* @inheritdoc |
|
310
|
|
|
*/ |
|
311
|
1 |
|
protected function renderStrong($block) |
|
312
|
|
|
{ |
|
313
|
1 |
|
return '\textbf{' . $this->renderAbsy($block[1]) . '}'; |
|
314
|
|
|
} |
|
315
|
|
|
|
|
316
|
|
|
/** |
|
317
|
|
|
* @inheritdoc |
|
318
|
|
|
*/ |
|
319
|
1 |
|
protected function renderEmph($block) |
|
320
|
|
|
{ |
|
321
|
1 |
|
return '\textit{' . $this->renderAbsy($block[1]) . '}'; |
|
322
|
|
|
} |
|
323
|
|
|
|
|
324
|
|
|
/** |
|
325
|
|
|
* Parses escaped special characters. |
|
326
|
|
|
* This allow a backslash to be interpreted as LaTeX |
|
327
|
|
|
* @marker \ |
|
328
|
|
|
*/ |
|
329
|
1 |
|
protected function parseEscape($text) |
|
330
|
|
|
{ |
|
331
|
1 |
|
if (isset($text[1]) && in_array($text[1], $this->escapeCharacters)) { |
|
332
|
1 |
|
if ($text[1] === '\\') { |
|
333
|
|
|
return [['backslash'], 2]; |
|
|
|
|
|
|
334
|
|
|
} |
|
335
|
1 |
|
return [['text', $text[1]], 2]; |
|
336
|
|
|
} |
|
337
|
|
|
return [['text', $text[0]], 1]; |
|
338
|
|
|
} |
|
339
|
|
|
|
|
340
|
|
|
protected function renderBackslash() |
|
341
|
|
|
{ |
|
342
|
|
|
return '\\'; |
|
343
|
|
|
} |
|
344
|
|
|
|
|
345
|
|
|
private $_escaper; |
|
346
|
|
|
|
|
347
|
|
|
/** |
|
348
|
|
|
* Escape special characters in URLs |
|
349
|
|
|
*/ |
|
350
|
4 |
|
protected function escapeUrl($string) |
|
351
|
|
|
{ |
|
352
|
4 |
|
return str_replace('%', '\\%', $this->escapeLatex($string)); |
|
353
|
|
|
} |
|
354
|
|
|
|
|
355
|
|
|
/** |
|
356
|
|
|
* Escape special LaTeX characters |
|
357
|
|
|
*/ |
|
358
|
21 |
|
protected function escapeLatex($string) |
|
359
|
|
|
{ |
|
360
|
21 |
|
if ($this->_escaper === null) { |
|
361
|
21 |
|
$this->_escaper = new TextToLatex(); |
|
362
|
21 |
|
} |
|
363
|
21 |
|
return $this->_escaper->convert($string); |
|
364
|
|
|
} |
|
365
|
|
|
|
|
366
|
|
|
/** |
|
367
|
|
|
* @inheritdocs |
|
368
|
|
|
* |
|
369
|
|
|
* Parses a newline indicated by two spaces on the end of a markdown line. |
|
370
|
|
|
*/ |
|
371
|
21 |
|
protected function renderText($text) |
|
372
|
|
|
{ |
|
373
|
21 |
|
$output = str_replace(" \n", "\\\\\n", $this->escapeLatex($text[1])); |
|
374
|
|
|
// support No-Break Space in LaTeX |
|
375
|
21 |
|
$output = preg_replace("/\x{00a0}/u", '~', $output); |
|
376
|
|
|
// support Narrow No-Break Space spaces in LaTeX |
|
377
|
|
|
// http://unicode-table.com/en/202F/ |
|
378
|
|
|
// http://tex.stackexchange.com/questions/76132/how-to-typeset-a-small-non-breaking-space |
|
379
|
21 |
|
$output = preg_replace("/\x{202f}/u", '\nobreak\hspace{.16667em plus .08333em}', $output); |
|
380
|
21 |
|
return $output; |
|
381
|
|
|
} |
|
382
|
|
|
} |
|
383
|
|
|
|
This check looks from parameters that have been defined for a function or method, but which are not used in the method body.