profile-builder / assets / lib / Mustache / Tokenizer.php

Tokenizer.php in User Profile Builder – Beautiful User Registration Forms, User Profiles & User Role Editor 2.0.4, at assets/lib/Mustache/Tokenizer.php

279 lines 8.6 KB

Raw Download Zip

1	<?php
2
3	/*
4	* This file is part of Mustache.php.
5	*
6	* (c) 2012 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
20	// Finite state machine states
21	const IN_TEXT = 0;
22	const IN_TAG_TYPE = 1;
23	const IN_TAG = 2;
24
25	// Token types
26	const T_SECTION = '#';
27	const T_INVERTED = '^';
28	const T_END_SECTION = '/';
29	const T_COMMENT = '!';
30	const T_PARTIAL = '>';
31	const T_PARTIAL_2 = '<';
32	const T_DELIM_CHANGE = '=';
33	const T_ESCAPED = '_v';
34	const T_UNESCAPED = '{';
35	const T_UNESCAPED_2 = '&';
36	const T_TEXT = '_t';
37	const T_PRAGMA = '%';
38
39	// Valid token types
40	private static $tagTypes = array(
41	self::T_SECTION => true,
42	self::T_INVERTED => true,
43	self::T_END_SECTION => true,
44	self::T_COMMENT => true,
45	self::T_PARTIAL => true,
46	self::T_PARTIAL_2 => true,
47	self::T_DELIM_CHANGE => true,
48	self::T_ESCAPED => true,
49	self::T_UNESCAPED => true,
50	self::T_UNESCAPED_2 => true,
51	self::T_PRAGMA => true,
52	);
53
54	// Interpolated tags
55	private static $interpolatedTags = array(
56	self::T_ESCAPED => true,
57	self::T_UNESCAPED => true,
58	self::T_UNESCAPED_2 => true,
59	);
60
61	// Token properties
62	const TYPE = 'type';
63	const NAME = 'name';
64	const OTAG = 'otag';
65	const CTAG = 'ctag';
66	const LINE = 'line';
67	const INDEX = 'index';
68	const END = 'end';
69	const INDENT = 'indent';
70	const NODES = 'nodes';
71	const VALUE = 'value';
72
73	private $state;
74	private $tagType;
75	private $tag;
76	private $buffer;
77	private $tokens;
78	private $seenTag;
79	private $line;
80	private $otag;
81	private $ctag;
82
83	/**
84	* Scan and tokenize template source.
85	*
86	* @param string $text Mustache template source to tokenize
87	* @param string $delimiters Optionally, pass initial opening and closing delimiters (default: null)
88	*
89	* @return array Set of Mustache tokens
90	*/
91	public function scan($text, $delimiters = null)
92	{
93	$this->reset();
94
95	if ($delimiters = trim($delimiters)) {
96	list($otag, $ctag) = explode(' ', $delimiters);
97	$this->otag = $otag;
98	$this->ctag = $ctag;
99	}
100
101	$len = strlen($text);
102	for ($i = 0; $i < $len; $i++) {
103	switch ($this->state) {
104	case self::IN_TEXT:
105	if ($this->tagChange($this->otag, $text, $i)) {
106	$i--;
107	$this->flushBuffer();
108	$this->state = self::IN_TAG_TYPE;
109	} else {
110	$char = substr($text, $i, 1);
111	$this->buffer .= $char;
112	if ($char == "\n") {
113	$this->flushBuffer();
114	$this->line++;
115	}
116	}
117	break;
118
119	case self::IN_TAG_TYPE:
120	$i += strlen($this->otag) - 1;
121	$char = substr($text, $i + 1, 1);
122	if (isset(self::$tagTypes[$char])) {
123	$tag = $char;
124	$this->tagType = $tag;
125	} else {
126	$tag = null;
127	$this->tagType = self::T_ESCAPED;
128	}
129
130	if ($this->tagType === self::T_DELIM_CHANGE) {
131	$i = $this->changeDelimiters($text, $i);
132	$this->state = self::IN_TEXT;
133	} elseif ($this->tagType === self::T_PRAGMA) {
134	$i = $this->addPragma($text, $i);
135	$this->state = self::IN_TEXT;
136	} else {
137	if ($tag !== null) {
138	$i++;
139	}
140	$this->state = self::IN_TAG;
141	}
142	$this->seenTag = $i;
143	break;
144
145	default:
146	if ($this->tagChange($this->ctag, $text, $i)) {
147	$this->tokens[] = array(
148	self::TYPE => $this->tagType,
149	self::NAME => trim($this->buffer),
150	self::OTAG => $this->otag,
151	self::CTAG => $this->ctag,
152	self::LINE => $this->line,
153	self::INDEX => ($this->tagType == self::T_END_SECTION) ? $this->seenTag - strlen($this->otag) : $i + strlen($this->ctag)
154	);
155
156	$this->buffer = '';
157	$i += strlen($this->ctag) - 1;
158	$this->state = self::IN_TEXT;
159	if ($this->tagType == self::T_UNESCAPED) {
160	if ($this->ctag == '}}') {
161	$i++;
162	} else {
163	// Clean up `{{{ tripleStache }}}` style tokens.
164	$lastName = $this->tokens[count($this->tokens) - 1][self::NAME];
165	if (substr($lastName, -1) === '}') {
166	$this->tokens[count($this->tokens) - 1][self::NAME] = trim(substr($lastName, 0, -1));
167	}
168	}
169	}
170	} else {
171	$this->buffer .= substr($text, $i, 1);
172	}
173	break;
174	}
175	}
176
177	$this->flushBuffer();
178
179	return $this->tokens;
180	}
181
182	/**
183	* Helper function to reset tokenizer internal state.
184	*/
185	private function reset()
186	{
187	$this->state = self::IN_TEXT;
188	$this->tagType = null;
189	$this->tag = null;
190	$this->buffer = '';
191	$this->tokens = array();
192	$this->seenTag = false;
193	$this->line = 0;
194	$this->otag = '{{';
195	$this->ctag = '}}';
196	}
197
198	/**
199	* Flush the current buffer to a token.
200	*/
201	private function flushBuffer()
202	{
203	if (!empty($this->buffer)) {
204	$this->tokens[] = array(
205	self::TYPE => self::T_TEXT,
206	self::LINE => $this->line,
207	self::VALUE => $this->buffer
208	);
209	$this->buffer = '';
210	}
211	}
212
213	/**
214	* Change the current Mustache delimiters. Set new `otag` and `ctag` values.
215	*
216	* @param string $text Mustache template source
217	* @param int $index Current tokenizer index
218	*
219	* @return int New index value
220	*/
221	private function changeDelimiters($text, $index)
222	{
223	$startIndex = strpos($text, '=', $index) + 1;
224	$close = '='.$this->ctag;
225	$closeIndex = strpos($text, $close, $index);
226
227	list($otag, $ctag) = explode(' ', trim(substr($text, $startIndex, $closeIndex - $startIndex)));
228	$this->otag = $otag;
229	$this->ctag = $ctag;
230
231	$this->tokens[] = array(
232	self::TYPE => self::T_DELIM_CHANGE,
233	self::LINE => $this->line,
234	);
235
236	return $closeIndex + strlen($close) - 1;
237	}
238
239	/**
240	* Add pragma token.
241	*
242	* Pragmas are hoisted to the front of the template, so all pragma tokens
243	* will appear at the front of the token list.
244	*
245	* @param string $text
246	* @param int $index
247	*
248	* @return int New index value
249	*/
250	private function addPragma($text, $index)
251	{
252	$end = strpos($text, $this->ctag, $index);
253	$pragma = trim(substr($text, $index + 2, $end - $index - 2));
254
255	// Pragmas are hoisted to the front of the template.
256	array_unshift($this->tokens, array(
257	self::TYPE => self::T_PRAGMA,
258	self::NAME => $pragma,
259	self::LINE => 0,
260	));
261
262	return $end + strlen($this->ctag) - 1;
263	}
264
265	/**
266	* Test whether it's time to change tags.
267	*
268	* @param string $tag Current tag name
269	* @param string $text Mustache template source
270	* @param int $index Current tokenizer index
271	*
272	* @return boolean True if this is a closing section tag
273	*/
274	private function tagChange($tag, $text, $index)
275	{
276	return substr($text, $index, strlen($tag)) === $tag;
277	}
278	}
279