|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PHPHtmlParser; |
|
6
|
|
|
|
|
7
|
|
|
use PHPHtmlParser\Enum\StringToken; |
|
8
|
|
|
use PHPHtmlParser\Exceptions\ContentLengthException; |
|
9
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
|
10
|
|
|
|
|
11
|
|
|
/** |
|
12
|
|
|
* Class Content. |
|
13
|
|
|
*/ |
|
14
|
|
|
class Content |
|
15
|
|
|
{ |
|
16
|
|
|
/** |
|
17
|
|
|
* The content string. |
|
18
|
|
|
* |
|
19
|
|
|
* @var string |
|
20
|
|
|
*/ |
|
21
|
|
|
protected $content; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* The size of the content. |
|
25
|
|
|
* |
|
26
|
|
|
* @var int |
|
27
|
|
|
*/ |
|
28
|
|
|
protected $size; |
|
29
|
|
|
|
|
30
|
|
|
/** |
|
31
|
|
|
* The current position we are in the content. |
|
32
|
|
|
* |
|
33
|
|
|
* @var int |
|
34
|
|
|
*/ |
|
35
|
|
|
protected $pos; |
|
36
|
|
|
|
|
37
|
|
|
/** |
|
38
|
|
|
* The following 4 strings are tags that are important to us. |
|
39
|
|
|
* |
|
40
|
|
|
* @var string |
|
41
|
|
|
*/ |
|
42
|
|
|
protected $blank = " \t\r\n"; |
|
43
|
|
|
protected $equal = ' =/>'; |
|
44
|
|
|
protected $slash = " />\r\n\t"; |
|
45
|
|
|
protected $attr = ' >'; |
|
46
|
|
|
|
|
47
|
|
|
/** |
|
48
|
|
|
* Content constructor. |
|
49
|
|
|
*/ |
|
50
|
333 |
|
public function __construct(string $content = '') |
|
51
|
|
|
{ |
|
52
|
333 |
|
$this->content = $content; |
|
53
|
333 |
|
$this->size = \strlen($content); |
|
54
|
333 |
|
$this->pos = 0; |
|
55
|
333 |
|
} |
|
56
|
|
|
|
|
57
|
|
|
/** |
|
58
|
|
|
* Returns the current position of the content. |
|
59
|
|
|
*/ |
|
60
|
75 |
|
public function getPosition(): int |
|
61
|
|
|
{ |
|
62
|
75 |
|
return $this->pos; |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
/** |
|
66
|
|
|
* Gets the current character we are at. |
|
67
|
|
|
* |
|
68
|
|
|
* @param ?int $char |
|
69
|
|
|
*/ |
|
70
|
318 |
|
public function char(?int $char = null): string |
|
71
|
|
|
{ |
|
72
|
318 |
|
return $this->content[$char ?? $this->pos] ?? ''; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* Gets a string from the current character position. |
|
77
|
|
|
* |
|
78
|
|
|
* @param int $length |
|
79
|
|
|
* @return string |
|
80
|
|
|
*/ |
|
81
|
288 |
|
public function string(int $length = 1): string |
|
82
|
|
|
{ |
|
83
|
288 |
|
$string = ''; |
|
84
|
288 |
|
$position = $this->pos; |
|
85
|
|
|
do { |
|
86
|
288 |
|
$string .= $this->char($position++); |
|
87
|
288 |
|
} while ($position < $this->pos + $length); |
|
88
|
288 |
|
return $string; |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
/** |
|
92
|
|
|
* Moves the current position forward. |
|
93
|
|
|
* |
|
94
|
|
|
* @throws ContentLengthException |
|
95
|
|
|
*/ |
|
96
|
303 |
|
public function fastForward(int $count): Content |
|
97
|
|
|
{ |
|
98
|
303 |
|
if (!$this->canFastForward($count)) { |
|
99
|
|
|
// trying to go over the content length, throw exception |
|
100
|
3 |
|
throw new ContentLengthException('Attempt to fastForward pass the length of the content.'); |
|
101
|
|
|
} |
|
102
|
303 |
|
$this->pos += $count; |
|
103
|
|
|
|
|
104
|
303 |
|
return $this; |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
|
|
/** |
|
108
|
|
|
* Checks if we can move the position forward. |
|
109
|
|
|
*/ |
|
110
|
303 |
|
public function canFastForward(int $count): bool |
|
111
|
|
|
{ |
|
112
|
303 |
|
return \strlen($this->content) >= $this->pos + $count; |
|
113
|
|
|
} |
|
114
|
|
|
|
|
115
|
|
|
/** |
|
116
|
|
|
* Moves the current position backward. |
|
117
|
|
|
*/ |
|
118
|
27 |
|
public function rewind(int $count): Content |
|
119
|
|
|
{ |
|
120
|
27 |
|
$this->pos -= $count; |
|
121
|
27 |
|
if ($this->pos < 0) { |
|
122
|
3 |
|
$this->pos = 0; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
27 |
|
return $this; |
|
126
|
|
|
} |
|
127
|
|
|
|
|
128
|
|
|
/** |
|
129
|
|
|
* Copy the content until we find the given string. |
|
130
|
|
|
*/ |
|
131
|
309 |
|
public function copyUntil(string $string, bool $char = false, bool $escape = false): string |
|
132
|
|
|
{ |
|
133
|
309 |
|
if ($this->pos >= $this->size) { |
|
134
|
|
|
// nothing left |
|
135
|
288 |
|
return ''; |
|
136
|
|
|
} |
|
137
|
|
|
|
|
138
|
309 |
|
if ($escape) { |
|
139
|
261 |
|
$position = $this->pos; |
|
140
|
261 |
|
$found = false; |
|
141
|
261 |
|
while (!$found) { |
|
142
|
261 |
|
$position = \strpos($this->content, $string, $position); |
|
143
|
261 |
|
if ($position === false) { |
|
144
|
|
|
// reached the end |
|
145
|
240 |
|
break; |
|
146
|
|
|
} |
|
147
|
|
|
|
|
148
|
237 |
|
if ($this->char($position - 1) == '\\') { |
|
149
|
|
|
// this character is escaped |
|
150
|
3 |
|
++$position; |
|
151
|
3 |
|
continue; |
|
152
|
|
|
} |
|
153
|
|
|
|
|
154
|
237 |
|
$found = true; |
|
155
|
|
|
} |
|
156
|
306 |
|
} elseif ($char) { |
|
157
|
294 |
|
$position = \strcspn($this->content, $string, $this->pos); |
|
158
|
294 |
|
$position += $this->pos; |
|
159
|
|
|
} else { |
|
160
|
300 |
|
$position = \strpos($this->content, $string, $this->pos); |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
309 |
|
if ($position === false) { |
|
|
|
|
|
|
164
|
|
|
// could not find character, just return the remaining of the content |
|
165
|
249 |
|
$return = \substr($this->content, $this->pos, $this->size - $this->pos); |
|
166
|
249 |
|
if ($return === false) { |
|
167
|
|
|
throw new LogicalException('Substr returned false with position ' . $this->pos . '.'); |
|
168
|
|
|
} |
|
169
|
249 |
|
$this->pos = $this->size; |
|
170
|
|
|
|
|
171
|
249 |
|
return $return; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
300 |
|
if ($position == $this->pos) { |
|
175
|
|
|
// we are at the right place |
|
176
|
288 |
|
return ''; |
|
177
|
|
|
} |
|
178
|
|
|
|
|
179
|
300 |
|
$return = \substr($this->content, $this->pos, $position - $this->pos); |
|
180
|
300 |
|
if ($return === false) { |
|
181
|
|
|
throw new LogicalException('Substr returned false with position ' . $this->pos . '.'); |
|
182
|
|
|
} |
|
183
|
|
|
// set the new position |
|
184
|
300 |
|
$this->pos = $position; |
|
185
|
|
|
|
|
186
|
300 |
|
return $return; |
|
187
|
|
|
} |
|
188
|
|
|
|
|
189
|
|
|
/** |
|
190
|
|
|
* Copies the content until the string is found and return it |
|
191
|
|
|
* unless the 'unless' is found in the substring. |
|
192
|
|
|
*/ |
|
193
|
258 |
|
public function copyUntilUnless(string $string, string $unless): string |
|
194
|
|
|
{ |
|
195
|
258 |
|
$lastPos = $this->pos; |
|
196
|
258 |
|
$this->fastForward(1); |
|
197
|
258 |
|
$foundString = $this->copyUntil($string, true, true); |
|
198
|
|
|
|
|
199
|
258 |
|
$position = \strcspn($foundString, $unless); |
|
200
|
258 |
|
if ($position == \strlen($foundString)) { |
|
201
|
69 |
|
return $string . $foundString; |
|
202
|
|
|
} |
|
203
|
|
|
// rewind changes and return nothing |
|
204
|
258 |
|
$this->pos = $lastPos; |
|
205
|
|
|
|
|
206
|
258 |
|
return ''; |
|
207
|
|
|
} |
|
208
|
|
|
|
|
209
|
|
|
/** |
|
210
|
|
|
* Copies the content until it reaches the token string.,. |
|
211
|
|
|
* |
|
212
|
|
|
* @uses $this->copyUntil() |
|
213
|
|
|
*/ |
|
214
|
291 |
|
public function copyByToken(StringToken $stringToken, bool $char = false, bool $escape = false): string |
|
215
|
|
|
{ |
|
216
|
291 |
|
$string = $stringToken->getValue(); |
|
217
|
|
|
|
|
218
|
291 |
|
return $this->copyUntil($string, $char, $escape); |
|
219
|
|
|
} |
|
220
|
|
|
|
|
221
|
|
|
/** |
|
222
|
|
|
* Skip a given set of characters. |
|
223
|
|
|
* |
|
224
|
|
|
* @throws LogicalException |
|
225
|
|
|
*/ |
|
226
|
297 |
|
public function skip(string $string, bool $copy = false): string |
|
227
|
|
|
{ |
|
228
|
297 |
|
$len = \strspn($this->content, $string, $this->pos); |
|
229
|
297 |
|
if ($len === false) { |
|
230
|
|
|
throw new LogicalException('Strspn returned false with position ' . $this->pos . '.'); |
|
231
|
|
|
} |
|
232
|
297 |
|
$return = ''; |
|
233
|
297 |
|
if ($copy) { |
|
234
|
273 |
|
$return = \substr($this->content, $this->pos, $len); |
|
235
|
273 |
|
if ($return === false) { |
|
236
|
|
|
throw new LogicalException('Substr returned false with position ' . $this->pos . '.'); |
|
237
|
|
|
} |
|
238
|
|
|
} |
|
239
|
|
|
|
|
240
|
|
|
// update the position |
|
241
|
297 |
|
$this->pos += $len; |
|
242
|
|
|
|
|
243
|
297 |
|
return $return; |
|
244
|
|
|
} |
|
245
|
|
|
|
|
246
|
|
|
/** |
|
247
|
|
|
* Skip a given token of pre-defined characters. |
|
248
|
|
|
* |
|
249
|
|
|
* @uses $this->skip() |
|
250
|
|
|
*/ |
|
251
|
291 |
|
public function skipByToken(StringToken $skipToken, bool $copy = false): string |
|
252
|
|
|
{ |
|
253
|
291 |
|
$string = $skipToken->getValue(); |
|
254
|
|
|
|
|
255
|
291 |
|
return $this->skip($string, $copy); |
|
256
|
|
|
} |
|
257
|
|
|
} |
|
258
|
|
|
|