easy-invoice / vendor / sabberworm / php-css-parser / src / Parsing / ParserState.php

ParserState.php in Easy Invoice – Invoice Generator, PDF Quotes & Payments 2.4.4, at vendor/sabberworm/php-css-parser/src/Parsing/ParserState.php

553 lines 14.1 KB

Raw Download Zip

1	<?php
2
3	namespace Sabberworm\CSS\Parsing;
4
5	use Sabberworm\CSS\Comment\Comment;
6	use Sabberworm\CSS\Settings;
7
8	/**
9	* @internal since 8.7.0
10	*/
11	class ParserState
12	{
13	/**
14	* @var null
15	*
16	* @internal since 8.5.2
17	*/
18	const EOF = null;
19
20	/**
21	* @var Settings
22	*/
23	private $oParserSettings;
24
25	/**
26	* @var string
27	*/
28	private $sText;
29
30	/**
31	* @var array<int, string>
32	*/
33	private $aText;
34
35	/**
36	* @var int
37	*/
38	private $iCurrentPosition;
39
40	/**
41	* will only be used if the CSS does not contain an `@charset` declaration
42	*
43	* @var string
44	*/
45	private $sCharset;
46
47	/**
48	* @var int
49	*/
50	private $iLength;
51
52	/**
53	* @var int
54	*/
55	private $iLineNo;
56
57	/**
58	* @param string $sText the complete CSS as text (i.e., usually the contents of a CSS file)
59	* @param int $iLineNo
60	*/
61	public function __construct($sText, Settings $oParserSettings, $iLineNo = 1)
62	{
63	$this->oParserSettings = $oParserSettings;
64	$this->sText = $sText;
65	$this->iCurrentPosition = 0;
66	$this->iLineNo = $iLineNo;
67	$this->setCharset($this->oParserSettings->sDefaultCharset);
68	}
69
70	/**
71	* Sets the charset to be used if the CSS does not contain an `@charset` declaration.
72	*
73	* @param string $sCharset
74	*
75	* @return void
76	*/
77	public function setCharset($sCharset)
78	{
79	$this->sCharset = $sCharset;
80	$this->aText = $this->strsplit($this->sText);
81	if (is_array($this->aText)) {
82	$this->iLength = count($this->aText);
83	}
84	}
85
86	/**
87	* Returns the charset that is used if the CSS does not contain an `@charset` declaration.
88	*
89	* @return string
90	*/
91	public function getCharset()
92	{
93	return $this->sCharset;
94	}
95
96	/**
97	* @return int
98	*/
99	public function currentLine()
100	{
101	return $this->iLineNo;
102	}
103
104	/**
105	* @return int
106	*/
107	public function currentColumn()
108	{
109	return $this->iCurrentPosition;
110	}
111
112	/**
113	* @return Settings
114	*/
115	public function getSettings()
116	{
117	return $this->oParserSettings;
118	}
119
120	/**
121	* @return \Sabberworm\CSS\Parsing\Anchor
122	*/
123	public function anchor()
124	{
125	return new Anchor($this->iCurrentPosition, $this);
126	}
127
128	/**
129	* @param int $iPosition
130	*
131	* @return void
132	*/
133	public function setPosition($iPosition)
134	{
135	$this->iCurrentPosition = $iPosition;
136	}
137
138	/**
139	* @param bool $bIgnoreCase
140	*
141	* @return string
142	*
143	* @throws UnexpectedTokenException
144	*
145	* @internal since V8.8.0
146	*/
147	public function parseIdentifier($bIgnoreCase = true)
148	{
149	if ($this->isEnd()) {
150	throw new UnexpectedEOFException('', '', 'identifier', $this->iLineNo);
151	}
152	$sResult = $this->parseCharacter(true);
153	if ($sResult === null) {
154	throw new UnexpectedTokenException($sResult, $this->peek(5), 'identifier', $this->iLineNo);
155	}
156	$sCharacter = null;
157	while (!$this->isEnd() && ($sCharacter = $this->parseCharacter(true)) !== null) {
158	if (preg_match('/[a-zA-Z0-9\x{00A0}-\x{FFFF}_-]/Sux', $sCharacter)) {
159	$sResult .= $sCharacter;
160	} else {
161	$sResult .= '\\' . $sCharacter;
162	}
163	}
164	if ($bIgnoreCase) {
165	$sResult = $this->strtolower($sResult);
166	}
167	return $sResult;
168	}
169
170	/**
171	* @param bool $bIsForIdentifier
172	*
173	* @return string\|null
174	*
175	* @throws UnexpectedEOFException
176	* @throws UnexpectedTokenException
177	*
178	* @internal since V8.8.0
179	*/
180	public function parseCharacter($bIsForIdentifier)
181	{
182	if ($this->peek() === '\\') {
183	if (
184	$bIsForIdentifier && $this->oParserSettings->bLenientParsing
185	&& ($this->comes('\0') \|\| $this->comes('\9'))
186	) {
187	// Non-strings can contain \0 or \9 which is an IE hack supported in lenient parsing.
188	return null;
189	}
190	$this->consume('\\');
191	if ($this->comes('\n') \|\| $this->comes('\r')) {
192	return '';
193	}
194	if (preg_match('/[0-9a-fA-F]/Su', $this->peek()) === 0) {
195	return $this->consume(1);
196	}
197	$sUnicode = $this->consumeExpression('/^[0-9a-fA-F]{1,6}/u', 6);
198	if ($this->strlen($sUnicode) < 6) {
199	// Consume whitespace after incomplete unicode escape
200	if (preg_match('/\\s/isSu', $this->peek())) {
201	if ($this->comes('\r\n')) {
202	$this->consume(2);
203	} else {
204	$this->consume(1);
205	}
206	}
207	}
208	$iUnicode = intval($sUnicode, 16);
209	$sUtf32 = "";
210	for ($i = 0; $i < 4; ++$i) {
211	$sUtf32 .= chr($iUnicode & 0xff);
212	$iUnicode = $iUnicode >> 8;
213	}
214	return iconv('utf-32le', $this->sCharset, $sUtf32);
215	}
216	if ($bIsForIdentifier) {
217	$peek = ord($this->peek());
218	// Ranges: a-z A-Z 0-9 - _
219	if (
220	($peek >= 97 && $peek <= 122)
221	\|\| ($peek >= 65 && $peek <= 90)
222	\|\| ($peek >= 48 && $peek <= 57)
223	\|\| ($peek === 45)
224	\|\| ($peek === 95)
225	\|\| ($peek > 0xa1)
226	) {
227	return $this->consume(1);
228	}
229	} else {
230	return $this->consume(1);
231	}
232	return null;
233	}
234
235	/**
236	* @return array<int, Comment>\|void
237	*
238	* @throws UnexpectedEOFException
239	* @throws UnexpectedTokenException
240	*/
241	public function consumeWhiteSpace()
242	{
243	$aComments = [];
244	do {
245	while (preg_match('/\\s/isSu', $this->peek()) === 1) {
246	$this->consume(1);
247	}
248	if ($this->oParserSettings->bLenientParsing) {
249	try {
250	$oComment = $this->consumeComment();
251	} catch (UnexpectedEOFException $e) {
252	$this->iCurrentPosition = $this->iLength;
253	return $aComments;
254	}
255	} else {
256	$oComment = $this->consumeComment();
257	}
258	if ($oComment !== false) {
259	$aComments[] = $oComment;
260	}
261	} while ($oComment !== false);
262	return $aComments;
263	}
264
265	/**
266	* @param string $sString
267	* @param bool $bCaseInsensitive
268	*
269	* @return bool
270	*/
271	public function comes($sString, $bCaseInsensitive = false)
272	{
273	$sPeek = $this->peek(strlen($sString));
274	return ($sPeek == '')
275	? false
276	: $this->streql($sPeek, $sString, $bCaseInsensitive);
277	}
278
279	/**
280	* @param int $iLength
281	* @param int $iOffset
282	*
283	* @return string
284	*/
285	public function peek($iLength = 1, $iOffset = 0)
286	{
287	$iOffset += $this->iCurrentPosition;
288	if ($iOffset >= $this->iLength) {
289	return '';
290	}
291	return $this->substr($iOffset, $iLength);
292	}
293
294	/**
295	* @param int $mValue
296	*
297	* @return string
298	*
299	* @throws UnexpectedEOFException
300	* @throws UnexpectedTokenException
301	*/
302	public function consume($mValue = 1)
303	{
304	if (is_string($mValue)) {
305	$iLineCount = substr_count($mValue, "\n");
306	$iLength = $this->strlen($mValue);
307	if (!$this->streql($this->substr($this->iCurrentPosition, $iLength), $mValue)) {
308	throw new UnexpectedTokenException($mValue, $this->peek(max($iLength, 5)), $this->iLineNo);
309	}
310	$this->iLineNo += $iLineCount;
311	$this->iCurrentPosition += $this->strlen($mValue);
312	return $mValue;
313	} else {
314	if ($this->iCurrentPosition + $mValue > $this->iLength) {
315	throw new UnexpectedEOFException($mValue, $this->peek(5), 'count', $this->iLineNo);
316	}
317	$sResult = $this->substr($this->iCurrentPosition, $mValue);
318	$iLineCount = substr_count($sResult, "\n");
319	$this->iLineNo += $iLineCount;
320	$this->iCurrentPosition += $mValue;
321	return $sResult;
322	}
323	}
324
325	/**
326	* @param string $mExpression
327	* @param int\|null $iMaxLength
328	*
329	* @return string
330	*
331	* @throws UnexpectedEOFException
332	* @throws UnexpectedTokenException
333	*/
334	public function consumeExpression($mExpression, $iMaxLength = null)
335	{
336	$aMatches = null;
337	$sInput = $iMaxLength !== null ? $this->peek($iMaxLength) : $this->inputLeft();
338	if (preg_match($mExpression, $sInput, $aMatches, PREG_OFFSET_CAPTURE) === 1) {
339	return $this->consume($aMatches[0][0]);
340	}
341	throw new UnexpectedTokenException($mExpression, $this->peek(5), 'expression', $this->iLineNo);
342	}
343
344	/**
345	* @return Comment\|false
346	*/
347	public function consumeComment()
348	{
349	$mComment = false;
350	if ($this->comes('/*')) {
351	$iLineNo = $this->iLineNo;
352	$this->consume(1);
353	$mComment = '';
354	while (($char = $this->consume(1)) !== '') {
355	$mComment .= $char;
356	if ($this->comes('*/')) {
357	$this->consume(2);
358	break;
359	}
360	}
361	}
362
363	if ($mComment !== false) {
364	// We skip the * which was included in the comment.
365	return new Comment(substr($mComment, 1), $iLineNo);
366	}
367
368	return $mComment;
369	}
370
371	/**
372	* @return bool
373	*/
374	public function isEnd()
375	{
376	return $this->iCurrentPosition >= $this->iLength;
377	}
378
379	/**
380	* @param array<array-key, string>\|string $aEnd
381	* @param string $bIncludeEnd
382	* @param string $consumeEnd
383	* @param array<int, Comment> $comments
384	*
385	* @return string
386	*
387	* @throws UnexpectedEOFException
388	* @throws UnexpectedTokenException
389	*/
390	public function consumeUntil($aEnd, $bIncludeEnd = false, $consumeEnd = false, array &$comments = [])
391	{
392	$aEnd = is_array($aEnd) ? $aEnd : [$aEnd];
393	$out = '';
394	$start = $this->iCurrentPosition;
395
396	while (!$this->isEnd()) {
397	$char = $this->consume(1);
398	if (in_array($char, $aEnd)) {
399	if ($bIncludeEnd) {
400	$out .= $char;
401	} elseif (!$consumeEnd) {
402	$this->iCurrentPosition -= $this->strlen($char);
403	}
404	return $out;
405	}
406	$out .= $char;
407	if ($comment = $this->consumeComment()) {
408	$comments[] = $comment;
409	}
410	}
411
412	if (in_array(self::EOF, $aEnd)) {
413	return $out;
414	}
415
416	$this->iCurrentPosition = $start;
417	throw new UnexpectedEOFException(
418	'One of ("' . implode('","', $aEnd) . '")',
419	$this->peek(5),
420	'search',
421	$this->iLineNo
422	);
423	}
424
425	/**
426	* @return string
427	*/
428	private function inputLeft()
429	{
430	return $this->substr($this->iCurrentPosition, -1);
431	}
432
433	/**
434	* @param string $sString1
435	* @param string $sString2
436	* @param bool $bCaseInsensitive
437	*
438	* @return bool
439	*/
440	public function streql($sString1, $sString2, $bCaseInsensitive = true)
441	{
442	if ($bCaseInsensitive) {
443	return $this->strtolower($sString1) === $this->strtolower($sString2);
444	} else {
445	return $sString1 === $sString2;
446	}
447	}
448
449	/**
450	* @param int $iAmount
451	*
452	* @return void
453	*/
454	public function backtrack($iAmount)
455	{
456	$this->iCurrentPosition -= $iAmount;
457	}
458
459	/**
460	* @param string $sString
461	*
462	* @return int
463	*/
464	public function strlen($sString)
465	{
466	if ($this->oParserSettings->bMultibyteSupport) {
467	return mb_strlen($sString, $this->sCharset);
468	} else {
469	return strlen($sString);
470	}
471	}
472
473	/**
474	* @param int $iStart
475	* @param int $iLength
476	*
477	* @return string
478	*/
479	private function substr($iStart, $iLength)
480	{
481	if ($iLength < 0) {
482	$iLength = $this->iLength - $iStart + $iLength;
483	}
484	if ($iStart + $iLength > $this->iLength) {
485	$iLength = $this->iLength - $iStart;
486	}
487	$sResult = '';
488	while ($iLength > 0) {
489	$sResult .= $this->aText[$iStart];
490	$iStart++;
491	$iLength--;
492	}
493	return $sResult;
494	}
495
496	/**
497	* @param string $sString
498	*
499	* @return string
500	*/
501	private function strtolower($sString)
502	{
503	if ($this->oParserSettings->bMultibyteSupport) {
504	return mb_strtolower($sString, $this->sCharset);
505	} else {
506	return strtolower($sString);
507	}
508	}
509
510	/**
511	* @param string $sString
512	*
513	* @return array<int, string>
514	*/
515	private function strsplit($sString)
516	{
517	if ($this->oParserSettings->bMultibyteSupport) {
518	if ($this->streql($this->sCharset, 'utf-8')) {
519	return preg_split('//u', $sString, -1, PREG_SPLIT_NO_EMPTY);
520	} else {
521	$iLength = mb_strlen($sString, $this->sCharset);
522	$aResult = [];
523	for ($i = 0; $i < $iLength; ++$i) {
524	$aResult[] = mb_substr($sString, $i, 1, $this->sCharset);
525	}
526	return $aResult;
527	}
528	} else {
529	if ($sString === '') {
530	return [];
531	} else {
532	return str_split($sString);
533	}
534	}
535	}
536
537	/**
538	* @param string $sString
539	* @param string $sNeedle
540	* @param int $iOffset
541	*
542	* @return int\|false
543	*/
544	private function strpos($sString, $sNeedle, $iOffset)
545	{
546	if ($this->oParserSettings->bMultibyteSupport) {
547	return mb_strpos($sString, $sNeedle, $iOffset, $this->sCharset);
548	} else {
549	return strpos($sString, $sNeedle, $iOffset);
550	}
551	}
552	}
553