| 1 |
efrain |
1 |
<?php
|
|
|
2 |
|
|
|
3 |
/*
|
|
|
4 |
* This file is part of Mustache.php.
|
|
|
5 |
*
|
|
|
6 |
* (c) 2010-2017 Justin Hileman
|
|
|
7 |
*
|
|
|
8 |
* For the full copyright and license information, please view the LICENSE
|
|
|
9 |
* file that was distributed with this source code.
|
|
|
10 |
*/
|
|
|
11 |
|
|
|
12 |
/**
|
|
|
13 |
* Mustache Tokenizer class.
|
|
|
14 |
*
|
|
|
15 |
* This class is responsible for turning raw template source into a set of Mustache tokens.
|
|
|
16 |
*/
|
|
|
17 |
class Mustache_Tokenizer
|
|
|
18 |
{
|
|
|
19 |
// Finite state machine states
|
|
|
20 |
const IN_TEXT = 0;
|
|
|
21 |
const IN_TAG_TYPE = 1;
|
|
|
22 |
const IN_TAG = 2;
|
|
|
23 |
|
|
|
24 |
// Token types
|
|
|
25 |
const T_SECTION = '#';
|
|
|
26 |
const T_INVERTED = '^';
|
|
|
27 |
const T_END_SECTION = '/';
|
|
|
28 |
const T_COMMENT = '!';
|
|
|
29 |
const T_PARTIAL = '>';
|
|
|
30 |
const T_PARENT = '<';
|
|
|
31 |
const T_DELIM_CHANGE = '=';
|
|
|
32 |
const T_ESCAPED = '_v';
|
|
|
33 |
const T_UNESCAPED = '{';
|
|
|
34 |
const T_UNESCAPED_2 = '&';
|
|
|
35 |
const T_TEXT = '_t';
|
|
|
36 |
const T_PRAGMA = '%';
|
|
|
37 |
const T_BLOCK_VAR = '$';
|
|
|
38 |
const T_BLOCK_ARG = '$arg';
|
|
|
39 |
|
|
|
40 |
// Valid token types
|
|
|
41 |
private static $tagTypes = array(
|
|
|
42 |
self::T_SECTION => true,
|
|
|
43 |
self::T_INVERTED => true,
|
|
|
44 |
self::T_END_SECTION => true,
|
|
|
45 |
self::T_COMMENT => true,
|
|
|
46 |
self::T_PARTIAL => true,
|
|
|
47 |
self::T_PARENT => true,
|
|
|
48 |
self::T_DELIM_CHANGE => true,
|
|
|
49 |
self::T_ESCAPED => true,
|
|
|
50 |
self::T_UNESCAPED => true,
|
|
|
51 |
self::T_UNESCAPED_2 => true,
|
|
|
52 |
self::T_PRAGMA => true,
|
|
|
53 |
self::T_BLOCK_VAR => true,
|
|
|
54 |
);
|
|
|
55 |
|
|
|
56 |
// Token properties
|
|
|
57 |
const TYPE = 'type';
|
|
|
58 |
const NAME = 'name';
|
|
|
59 |
const OTAG = 'otag';
|
|
|
60 |
const CTAG = 'ctag';
|
|
|
61 |
const LINE = 'line';
|
|
|
62 |
const INDEX = 'index';
|
|
|
63 |
const END = 'end';
|
|
|
64 |
const INDENT = 'indent';
|
|
|
65 |
const NODES = 'nodes';
|
|
|
66 |
const VALUE = 'value';
|
|
|
67 |
const FILTERS = 'filters';
|
|
|
68 |
|
|
|
69 |
private $state;
|
|
|
70 |
private $tagType;
|
|
|
71 |
private $buffer;
|
|
|
72 |
private $tokens;
|
|
|
73 |
private $seenTag;
|
|
|
74 |
private $line;
|
|
|
75 |
|
|
|
76 |
private $otag;
|
|
|
77 |
private $otagChar;
|
|
|
78 |
private $otagLen;
|
|
|
79 |
|
|
|
80 |
private $ctag;
|
|
|
81 |
private $ctagChar;
|
|
|
82 |
private $ctagLen;
|
|
|
83 |
|
|
|
84 |
/**
|
|
|
85 |
* Scan and tokenize template source.
|
|
|
86 |
*
|
|
|
87 |
* @throws Mustache_Exception_SyntaxException when mismatched section tags are encountered
|
|
|
88 |
* @throws Mustache_Exception_InvalidArgumentException when $delimiters string is invalid
|
|
|
89 |
*
|
|
|
90 |
* @param string $text Mustache template source to tokenize
|
|
|
91 |
* @param string $delimiters Optionally, pass initial opening and closing delimiters (default: empty string)
|
|
|
92 |
*
|
|
|
93 |
* @return array Set of Mustache tokens
|
|
|
94 |
*/
|
|
|
95 |
public function scan($text, $delimiters = '')
|
|
|
96 |
{
|
|
|
97 |
// Setting mbstring.func_overload makes things *really* slow.
|
|
|
98 |
// Let's do everyone a favor and scan this string as ASCII instead.
|
|
|
99 |
//
|
|
|
100 |
// The INI directive was removed in PHP 8.0 so we don't need to check there (and can drop it
|
|
|
101 |
// when we remove support for older versions of PHP).
|
|
|
102 |
//
|
|
|
103 |
// @codeCoverageIgnoreStart
|
|
|
104 |
$encoding = null;
|
|
|
105 |
if (version_compare(PHP_VERSION, '8.0.0', '<')) {
|
|
|
106 |
if (function_exists('mb_internal_encoding') && ini_get('mbstring.func_overload') & 2) {
|
|
|
107 |
$encoding = mb_internal_encoding();
|
|
|
108 |
mb_internal_encoding('ASCII');
|
|
|
109 |
}
|
|
|
110 |
}
|
|
|
111 |
// @codeCoverageIgnoreEnd
|
|
|
112 |
|
|
|
113 |
$this->reset();
|
|
|
114 |
|
|
|
115 |
if (is_string($delimiters) && $delimiters = trim($delimiters)) {
|
|
|
116 |
$this->setDelimiters($delimiters);
|
|
|
117 |
}
|
|
|
118 |
|
|
|
119 |
$len = strlen($text);
|
|
|
120 |
for ($i = 0; $i < $len; $i++) {
|
|
|
121 |
switch ($this->state) {
|
|
|
122 |
case self::IN_TEXT:
|
|
|
123 |
$char = $text[$i];
|
|
|
124 |
// Test whether it's time to change tags.
|
|
|
125 |
if ($char === $this->otagChar && substr($text, $i, $this->otagLen) === $this->otag) {
|
|
|
126 |
$i--;
|
|
|
127 |
$this->flushBuffer();
|
|
|
128 |
$this->state = self::IN_TAG_TYPE;
|
|
|
129 |
} else {
|
|
|
130 |
$this->buffer .= $char;
|
|
|
131 |
if ($char === "\n") {
|
|
|
132 |
$this->flushBuffer();
|
|
|
133 |
$this->line++;
|
|
|
134 |
}
|
|
|
135 |
}
|
|
|
136 |
break;
|
|
|
137 |
|
|
|
138 |
case self::IN_TAG_TYPE:
|
|
|
139 |
$i += $this->otagLen - 1;
|
|
|
140 |
$char = $text[$i + 1];
|
|
|
141 |
if (isset(self::$tagTypes[$char])) {
|
|
|
142 |
$tag = $char;
|
|
|
143 |
$this->tagType = $tag;
|
|
|
144 |
} else {
|
|
|
145 |
$tag = null;
|
|
|
146 |
$this->tagType = self::T_ESCAPED;
|
|
|
147 |
}
|
|
|
148 |
|
|
|
149 |
if ($this->tagType === self::T_DELIM_CHANGE) {
|
|
|
150 |
$i = $this->changeDelimiters($text, $i);
|
|
|
151 |
$this->state = self::IN_TEXT;
|
|
|
152 |
} elseif ($this->tagType === self::T_PRAGMA) {
|
|
|
153 |
$i = $this->addPragma($text, $i);
|
|
|
154 |
$this->state = self::IN_TEXT;
|
|
|
155 |
} else {
|
|
|
156 |
if ($tag !== null) {
|
|
|
157 |
$i++;
|
|
|
158 |
}
|
|
|
159 |
$this->state = self::IN_TAG;
|
|
|
160 |
}
|
|
|
161 |
$this->seenTag = $i;
|
|
|
162 |
break;
|
|
|
163 |
|
|
|
164 |
default:
|
|
|
165 |
$char = $text[$i];
|
|
|
166 |
// Test whether it's time to change tags.
|
|
|
167 |
if ($char === $this->ctagChar && substr($text, $i, $this->ctagLen) === $this->ctag) {
|
|
|
168 |
$token = array(
|
|
|
169 |
self::TYPE => $this->tagType,
|
|
|
170 |
self::NAME => trim($this->buffer),
|
|
|
171 |
self::OTAG => $this->otag,
|
|
|
172 |
self::CTAG => $this->ctag,
|
|
|
173 |
self::LINE => $this->line,
|
|
|
174 |
self::INDEX => ($this->tagType === self::T_END_SECTION) ? $this->seenTag - $this->otagLen : $i + $this->ctagLen,
|
|
|
175 |
);
|
|
|
176 |
|
|
|
177 |
if ($this->tagType === self::T_UNESCAPED) {
|
|
|
178 |
// Clean up `{{{ tripleStache }}}` style tokens.
|
|
|
179 |
if ($this->ctag === '}}') {
|
|
|
180 |
if (($i + 2 < $len) && $text[$i + 2] === '}') {
|
|
|
181 |
$i++;
|
|
|
182 |
} else {
|
|
|
183 |
$msg = sprintf(
|
|
|
184 |
'Mismatched tag delimiters: %s on line %d',
|
|
|
185 |
$token[self::NAME],
|
|
|
186 |
$token[self::LINE]
|
|
|
187 |
);
|
|
|
188 |
|
|
|
189 |
throw new Mustache_Exception_SyntaxException($msg, $token);
|
|
|
190 |
}
|
|
|
191 |
} else {
|
|
|
192 |
$lastName = $token[self::NAME];
|
|
|
193 |
if (substr($lastName, -1) === '}') {
|
|
|
194 |
$token[self::NAME] = trim(substr($lastName, 0, -1));
|
|
|
195 |
} else {
|
|
|
196 |
$msg = sprintf(
|
|
|
197 |
'Mismatched tag delimiters: %s on line %d',
|
|
|
198 |
$token[self::NAME],
|
|
|
199 |
$token[self::LINE]
|
|
|
200 |
);
|
|
|
201 |
|
|
|
202 |
throw new Mustache_Exception_SyntaxException($msg, $token);
|
|
|
203 |
}
|
|
|
204 |
}
|
|
|
205 |
}
|
|
|
206 |
|
|
|
207 |
$this->buffer = '';
|
|
|
208 |
$i += $this->ctagLen - 1;
|
|
|
209 |
$this->state = self::IN_TEXT;
|
|
|
210 |
$this->tokens[] = $token;
|
|
|
211 |
} else {
|
|
|
212 |
$this->buffer .= $char;
|
|
|
213 |
}
|
|
|
214 |
break;
|
|
|
215 |
}
|
|
|
216 |
}
|
|
|
217 |
|
|
|
218 |
if ($this->state !== self::IN_TEXT) {
|
|
|
219 |
$this->throwUnclosedTagException();
|
|
|
220 |
}
|
|
|
221 |
|
|
|
222 |
$this->flushBuffer();
|
|
|
223 |
|
|
|
224 |
// Restore the user's encoding...
|
|
|
225 |
// @codeCoverageIgnoreStart
|
|
|
226 |
if ($encoding) {
|
|
|
227 |
mb_internal_encoding($encoding);
|
|
|
228 |
}
|
|
|
229 |
// @codeCoverageIgnoreEnd
|
|
|
230 |
|
|
|
231 |
return $this->tokens;
|
|
|
232 |
}
|
|
|
233 |
|
|
|
234 |
/**
|
|
|
235 |
* Helper function to reset tokenizer internal state.
|
|
|
236 |
*/
|
|
|
237 |
private function reset()
|
|
|
238 |
{
|
|
|
239 |
$this->state = self::IN_TEXT;
|
|
|
240 |
$this->tagType = null;
|
|
|
241 |
$this->buffer = '';
|
|
|
242 |
$this->tokens = array();
|
|
|
243 |
$this->seenTag = false;
|
|
|
244 |
$this->line = 0;
|
|
|
245 |
|
|
|
246 |
$this->otag = '{{';
|
|
|
247 |
$this->otagChar = '{';
|
|
|
248 |
$this->otagLen = 2;
|
|
|
249 |
|
|
|
250 |
$this->ctag = '}}';
|
|
|
251 |
$this->ctagChar = '}';
|
|
|
252 |
$this->ctagLen = 2;
|
|
|
253 |
}
|
|
|
254 |
|
|
|
255 |
/**
|
|
|
256 |
* Flush the current buffer to a token.
|
|
|
257 |
*/
|
|
|
258 |
private function flushBuffer()
|
|
|
259 |
{
|
|
|
260 |
if (strlen($this->buffer) > 0) {
|
|
|
261 |
$this->tokens[] = array(
|
|
|
262 |
self::TYPE => self::T_TEXT,
|
|
|
263 |
self::LINE => $this->line,
|
|
|
264 |
self::VALUE => $this->buffer,
|
|
|
265 |
);
|
|
|
266 |
$this->buffer = '';
|
|
|
267 |
}
|
|
|
268 |
}
|
|
|
269 |
|
|
|
270 |
/**
|
|
|
271 |
* Change the current Mustache delimiters. Set new `otag` and `ctag` values.
|
|
|
272 |
*
|
|
|
273 |
* @throws Mustache_Exception_SyntaxException when delimiter string is invalid
|
|
|
274 |
*
|
|
|
275 |
* @param string $text Mustache template source
|
|
|
276 |
* @param int $index Current tokenizer index
|
|
|
277 |
*
|
|
|
278 |
* @return int New index value
|
|
|
279 |
*/
|
|
|
280 |
private function changeDelimiters($text, $index)
|
|
|
281 |
{
|
|
|
282 |
$startIndex = strpos($text, '=', $index) + 1;
|
|
|
283 |
$close = '=' . $this->ctag;
|
|
|
284 |
$closeIndex = strpos($text, $close, $index);
|
|
|
285 |
|
|
|
286 |
if ($closeIndex === false) {
|
|
|
287 |
$this->throwUnclosedTagException();
|
|
|
288 |
}
|
|
|
289 |
|
|
|
290 |
$token = array(
|
|
|
291 |
self::TYPE => self::T_DELIM_CHANGE,
|
|
|
292 |
self::LINE => $this->line,
|
|
|
293 |
);
|
|
|
294 |
|
|
|
295 |
try {
|
|
|
296 |
$this->setDelimiters(trim(substr($text, $startIndex, $closeIndex - $startIndex)));
|
|
|
297 |
} catch (Mustache_Exception_InvalidArgumentException $e) {
|
|
|
298 |
throw new Mustache_Exception_SyntaxException($e->getMessage(), $token);
|
|
|
299 |
}
|
|
|
300 |
|
|
|
301 |
$this->tokens[] = $token;
|
|
|
302 |
|
|
|
303 |
return $closeIndex + strlen($close) - 1;
|
|
|
304 |
}
|
|
|
305 |
|
|
|
306 |
/**
|
|
|
307 |
* Set the current Mustache `otag` and `ctag` delimiters.
|
|
|
308 |
*
|
|
|
309 |
* @throws Mustache_Exception_InvalidArgumentException when delimiter string is invalid
|
|
|
310 |
*
|
|
|
311 |
* @param string $delimiters
|
|
|
312 |
*/
|
|
|
313 |
private function setDelimiters($delimiters)
|
|
|
314 |
{
|
|
|
315 |
if (!preg_match('/^\s*(\S+)\s+(\S+)\s*$/', $delimiters, $matches)) {
|
|
|
316 |
throw new Mustache_Exception_InvalidArgumentException(sprintf('Invalid delimiters: %s', $delimiters));
|
|
|
317 |
}
|
|
|
318 |
|
|
|
319 |
list($_, $otag, $ctag) = $matches;
|
|
|
320 |
|
|
|
321 |
$this->otag = $otag;
|
|
|
322 |
$this->otagChar = $otag[0];
|
|
|
323 |
$this->otagLen = strlen($otag);
|
|
|
324 |
|
|
|
325 |
$this->ctag = $ctag;
|
|
|
326 |
$this->ctagChar = $ctag[0];
|
|
|
327 |
$this->ctagLen = strlen($ctag);
|
|
|
328 |
}
|
|
|
329 |
|
|
|
330 |
/**
|
|
|
331 |
* Add pragma token.
|
|
|
332 |
*
|
|
|
333 |
* Pragmas are hoisted to the front of the template, so all pragma tokens
|
|
|
334 |
* will appear at the front of the token list.
|
|
|
335 |
*
|
|
|
336 |
* @param string $text
|
|
|
337 |
* @param int $index
|
|
|
338 |
*
|
|
|
339 |
* @return int New index value
|
|
|
340 |
*/
|
|
|
341 |
private function addPragma($text, $index)
|
|
|
342 |
{
|
|
|
343 |
$end = strpos($text, $this->ctag, $index);
|
|
|
344 |
if ($end === false) {
|
|
|
345 |
$this->throwUnclosedTagException();
|
|
|
346 |
}
|
|
|
347 |
|
|
|
348 |
$pragma = trim(substr($text, $index + 2, $end - $index - 2));
|
|
|
349 |
|
|
|
350 |
// Pragmas are hoisted to the front of the template.
|
|
|
351 |
array_unshift($this->tokens, array(
|
|
|
352 |
self::TYPE => self::T_PRAGMA,
|
|
|
353 |
self::NAME => $pragma,
|
|
|
354 |
self::LINE => 0,
|
|
|
355 |
));
|
|
|
356 |
|
|
|
357 |
return $end + $this->ctagLen - 1;
|
|
|
358 |
}
|
|
|
359 |
|
|
|
360 |
private function throwUnclosedTagException()
|
|
|
361 |
{
|
|
|
362 |
$name = trim($this->buffer);
|
|
|
363 |
if ($name !== '') {
|
|
|
364 |
$msg = sprintf('Unclosed tag: %s on line %d', $name, $this->line);
|
|
|
365 |
} else {
|
|
|
366 |
$msg = sprintf('Unclosed tag on line %d', $this->line);
|
|
|
367 |
}
|
|
|
368 |
|
|
|
369 |
throw new Mustache_Exception_SyntaxException($msg, array(
|
|
|
370 |
self::TYPE => $this->tagType,
|
|
|
371 |
self::NAME => $name,
|
|
|
372 |
self::OTAG => $this->otag,
|
|
|
373 |
self::CTAG => $this->ctag,
|
|
|
374 |
self::LINE => $this->line,
|
|
|
375 |
self::INDEX => $this->seenTag - $this->otagLen,
|
|
|
376 |
));
|
|
|
377 |
}
|
|
|
378 |
}
|