|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Inspirum\XML\Services; |
|
4
|
|
|
|
|
5
|
|
|
use Exception; |
|
6
|
|
|
use XMLReader as BaseXMLReader; |
|
7
|
|
|
|
|
8
|
|
|
class XMLReader |
|
9
|
|
|
{ |
|
10
|
|
|
/** |
|
11
|
|
|
* XML Reader |
|
12
|
|
|
* |
|
13
|
|
|
* @var \XMLReader |
|
14
|
|
|
*/ |
|
15
|
|
|
private $reader; |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* XML document |
|
19
|
|
|
* |
|
20
|
|
|
* @var \Inspirum\XML\Services\XML |
|
21
|
|
|
*/ |
|
22
|
|
|
private $xml; |
|
23
|
|
|
|
|
24
|
|
|
/** |
|
25
|
|
|
* XMLReader constructor |
|
26
|
|
|
* |
|
27
|
|
|
* @param string $filepath |
|
28
|
|
|
*/ |
|
29
|
17 |
|
public function __construct(string $filepath) |
|
30
|
|
|
{ |
|
31
|
17 |
|
$this->reader = $this->open($filepath); |
|
32
|
15 |
|
$this->xml = new XML(); |
|
33
|
15 |
|
} |
|
34
|
|
|
|
|
35
|
|
|
/** |
|
36
|
|
|
* XMLReader destructor |
|
37
|
|
|
*/ |
|
38
|
13 |
|
public function __destruct() |
|
39
|
|
|
{ |
|
40
|
13 |
|
$this->reader->close(); |
|
41
|
13 |
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* Open file |
|
45
|
|
|
* |
|
46
|
|
|
* @param string $filepath |
|
47
|
|
|
* |
|
48
|
|
|
* @return \XMLReader |
|
49
|
|
|
* |
|
50
|
|
|
* @throws \Exception |
|
51
|
|
|
*/ |
|
52
|
17 |
|
private function open(string $filepath): BaseXMLReader |
|
53
|
|
|
{ |
|
54
|
17 |
|
$xmlReader = new BaseXMLReader(); |
|
55
|
|
|
|
|
56
|
17 |
|
$opened = $this->withErrorHandler(function () use ($xmlReader, $filepath) { |
|
57
|
17 |
|
return $xmlReader->open($filepath); |
|
58
|
17 |
|
}); |
|
59
|
|
|
|
|
60
|
15 |
|
if ($opened == false) { |
|
61
|
|
|
// @codeCoverageIgnoreStart |
|
62
|
|
|
throw new Exception('\XMLReader::open() method failed'); |
|
63
|
|
|
// @codeCoverageIgnoreEnd |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
15 |
|
return $xmlReader; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Parse file and yield next node |
|
71
|
|
|
* |
|
72
|
|
|
* @param string $nodeName |
|
73
|
|
|
* |
|
74
|
|
|
* @return \Generator|\Inspirum\XML\Services\XMLNode[] |
|
75
|
|
|
*/ |
|
76
|
7 |
|
public function iterateNode(string $nodeName): iterable |
|
77
|
|
|
{ |
|
78
|
7 |
|
$found = $this->moveToNode($nodeName); |
|
79
|
|
|
|
|
80
|
6 |
|
if ($found === false) { |
|
81
|
1 |
|
return yield from []; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
do { |
|
85
|
5 |
|
$item = $this->readNode(); |
|
86
|
|
|
|
|
87
|
5 |
|
if ($item !== null) { |
|
88
|
5 |
|
yield $item; |
|
89
|
|
|
} |
|
90
|
5 |
|
} while ($this->moveToNextNode($nodeName)); |
|
91
|
5 |
|
} |
|
92
|
|
|
|
|
93
|
|
|
/** |
|
94
|
|
|
* Get next node |
|
95
|
|
|
* |
|
96
|
|
|
* @param string $nodeName |
|
97
|
|
|
* |
|
98
|
|
|
* @return \Inspirum\XML\Services\XMLNode|null |
|
99
|
|
|
*/ |
|
100
|
8 |
|
public function nextNode(string $nodeName): ?XMLNode |
|
101
|
|
|
{ |
|
102
|
8 |
|
$found = $this->moveToNode($nodeName); |
|
103
|
|
|
|
|
104
|
7 |
|
if ($found === false) { |
|
105
|
2 |
|
return null; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
6 |
|
return $this->readNode(); |
|
109
|
|
|
} |
|
110
|
|
|
|
|
111
|
|
|
/** |
|
112
|
|
|
* Move to first element by tag name |
|
113
|
|
|
* |
|
114
|
|
|
* @param string $nodeName |
|
115
|
|
|
* |
|
116
|
|
|
* @return bool |
|
117
|
|
|
*/ |
|
118
|
14 |
|
private function moveToNode(string $nodeName): bool |
|
119
|
|
|
{ |
|
120
|
14 |
|
while ($this->read()) { |
|
121
|
12 |
|
if ($this->isNodeElementType() && $this->getNodeName() === $nodeName) { |
|
122
|
10 |
|
return true; |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
|
|
126
|
3 |
|
return false; |
|
127
|
|
|
} |
|
128
|
|
|
|
|
129
|
|
|
/** |
|
130
|
|
|
* Move to next sibling element by tag name |
|
131
|
|
|
* |
|
132
|
|
|
* @param string $nodeName |
|
133
|
|
|
* |
|
134
|
|
|
* @return bool |
|
135
|
|
|
*/ |
|
136
|
5 |
|
private function moveToNextNode(string $nodeName): bool |
|
137
|
|
|
{ |
|
138
|
5 |
|
while ($this->reader->next(Formatter::getLocalName($nodeName))) { |
|
139
|
5 |
|
if ($this->getNodeName() === $nodeName) { |
|
140
|
4 |
|
return true; |
|
141
|
|
|
} |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
5 |
|
return false; |
|
145
|
|
|
} |
|
146
|
|
|
|
|
147
|
|
|
/** |
|
148
|
|
|
* Return associative array of element by name |
|
149
|
|
|
* |
|
150
|
|
|
* @return \Inspirum\XML\Services\XMLNode|null |
|
151
|
|
|
*/ |
|
152
|
10 |
|
private function readNode(): ?XMLNode |
|
153
|
|
|
{ |
|
154
|
10 |
|
$nodeName = $this->getNodeName(); |
|
155
|
10 |
|
$attributes = $this->getNodeAttributes(); |
|
156
|
|
|
|
|
157
|
10 |
|
if ($this->isNodeEmptyElementType()) { |
|
158
|
1 |
|
return $this->xml->createElement($nodeName, $attributes); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
9 |
|
$node = null; |
|
162
|
9 |
|
$text = null; |
|
163
|
9 |
|
$elements = []; |
|
164
|
|
|
|
|
165
|
9 |
|
while ($this->read()) { |
|
166
|
9 |
|
if ($this->isNodeElementEndType() && $this->getNodeName() == $nodeName) { |
|
167
|
9 |
|
$node = $this->xml->createTextElement($nodeName, $text, $attributes); |
|
168
|
|
|
|
|
169
|
9 |
|
foreach ($elements as $element) { |
|
170
|
6 |
|
$node->append($element); |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
9 |
|
break; |
|
174
|
8 |
|
} elseif ($this->isNodeTextType()) { |
|
175
|
8 |
|
$text = $this->getNodeValue(); |
|
176
|
6 |
|
} elseif ($this->isNodeElementType()) { |
|
177
|
6 |
|
if ($this->isNodeEmptyElementType()) { |
|
178
|
1 |
|
$elements[] = $this->xml->createElement($this->getNodeName()); |
|
179
|
1 |
|
continue; |
|
180
|
|
|
} |
|
181
|
|
|
|
|
182
|
6 |
|
$element = $this->readNode(); |
|
183
|
|
|
|
|
184
|
6 |
|
if ($element instanceof XMLNode) { |
|
185
|
6 |
|
$elements[] = $element; |
|
186
|
|
|
} |
|
187
|
|
|
} |
|
188
|
|
|
} |
|
189
|
|
|
|
|
190
|
9 |
|
return $node; |
|
191
|
|
|
} |
|
192
|
|
|
|
|
193
|
|
|
/** |
|
194
|
|
|
* Move to next node in document |
|
195
|
|
|
* |
|
196
|
|
|
* @return bool |
|
197
|
|
|
* |
|
198
|
|
|
* @throws \DOMException |
|
199
|
|
|
*/ |
|
200
|
14 |
|
private function read(): bool |
|
201
|
|
|
{ |
|
202
|
14 |
|
return $this->withErrorHandler(function () { |
|
203
|
14 |
|
return $this->reader->read(); |
|
204
|
14 |
|
}); |
|
205
|
|
|
} |
|
206
|
|
|
|
|
207
|
|
|
/** |
|
208
|
|
|
* Get current node name |
|
209
|
|
|
* |
|
210
|
|
|
* @return string |
|
211
|
|
|
*/ |
|
212
|
12 |
|
private function getNodeName(): string |
|
213
|
|
|
{ |
|
214
|
12 |
|
return $this->reader->name; |
|
215
|
|
|
} |
|
216
|
|
|
|
|
217
|
|
|
/** |
|
218
|
|
|
* Get current node type |
|
219
|
|
|
* |
|
220
|
|
|
* @return int |
|
221
|
|
|
*/ |
|
222
|
12 |
|
private function getNodeType(): int |
|
223
|
|
|
{ |
|
224
|
12 |
|
return $this->reader->nodeType; |
|
225
|
|
|
} |
|
226
|
|
|
|
|
227
|
|
|
/** |
|
228
|
|
|
* Get current node value |
|
229
|
|
|
* |
|
230
|
|
|
* @return string |
|
231
|
|
|
*/ |
|
232
|
10 |
|
private function getNodeValue(): string |
|
233
|
|
|
{ |
|
234
|
10 |
|
return $this->reader->value; |
|
235
|
|
|
} |
|
236
|
|
|
|
|
237
|
|
|
/** |
|
238
|
|
|
* If current node is element open tag |
|
239
|
|
|
* |
|
240
|
|
|
* @return bool |
|
241
|
|
|
*/ |
|
242
|
12 |
|
private function isNodeElementType(): bool |
|
243
|
|
|
{ |
|
244
|
12 |
|
return $this->isNodeType(BaseXMLReader::ELEMENT); |
|
245
|
|
|
} |
|
246
|
|
|
|
|
247
|
|
|
/** |
|
248
|
|
|
* If current node is element open tag |
|
249
|
|
|
* |
|
250
|
|
|
* @return bool |
|
251
|
|
|
*/ |
|
252
|
10 |
|
private function isNodeEmptyElementType(): bool |
|
253
|
|
|
{ |
|
254
|
10 |
|
return $this->reader->isEmptyElement; |
|
255
|
|
|
} |
|
256
|
|
|
|
|
257
|
|
|
/** |
|
258
|
|
|
* If current node is element close tag |
|
259
|
|
|
* |
|
260
|
|
|
* @return bool |
|
261
|
|
|
*/ |
|
262
|
9 |
|
private function isNodeElementEndType(): bool |
|
263
|
|
|
{ |
|
264
|
9 |
|
return $this->isNodeType(BaseXMLReader::END_ELEMENT); |
|
265
|
|
|
} |
|
266
|
|
|
|
|
267
|
|
|
/** |
|
268
|
|
|
* If current node is text content |
|
269
|
|
|
* |
|
270
|
|
|
* @return bool |
|
271
|
|
|
*/ |
|
272
|
8 |
|
private function isNodeTextType(): bool |
|
273
|
|
|
{ |
|
274
|
8 |
|
return $this->isNodeType(BaseXMLReader::TEXT) || $this->isNodeType(BaseXMLReader::CDATA); |
|
275
|
|
|
} |
|
276
|
|
|
|
|
277
|
|
|
/** |
|
278
|
|
|
* If current node is given node type |
|
279
|
|
|
* |
|
280
|
|
|
* @param int $type |
|
281
|
|
|
* |
|
282
|
|
|
* @return bool |
|
283
|
|
|
*/ |
|
284
|
12 |
|
private function isNodeType(int $type): bool |
|
285
|
|
|
{ |
|
286
|
12 |
|
return $this->getNodeType() === $type; |
|
287
|
|
|
} |
|
288
|
|
|
|
|
289
|
|
|
/** |
|
290
|
|
|
* Get current node attributes |
|
291
|
|
|
* |
|
292
|
|
|
* @return array<string,string> |
|
293
|
|
|
*/ |
|
294
|
10 |
|
private function getNodeAttributes(): array |
|
295
|
|
|
{ |
|
296
|
10 |
|
$attributes = []; |
|
297
|
|
|
|
|
298
|
10 |
|
if ($this->reader->hasAttributes) { |
|
299
|
6 |
|
while ($this->reader->moveToNextAttribute()) { |
|
300
|
6 |
|
$attributes[$this->getNodeName()] = $this->getNodeValue(); |
|
301
|
|
|
} |
|
302
|
6 |
|
$this->reader->moveToElement(); |
|
303
|
|
|
} |
|
304
|
|
|
|
|
305
|
10 |
|
return $attributes; |
|
306
|
|
|
} |
|
307
|
|
|
|
|
308
|
|
|
/** |
|
309
|
|
|
* Register custom error handler to throw Exception on warning message |
|
310
|
|
|
* |
|
311
|
|
|
* @param callable $callback |
|
312
|
|
|
* |
|
313
|
|
|
* @return mixed |
|
314
|
|
|
* |
|
315
|
|
|
* @throws \DOMException |
|
316
|
|
|
*/ |
|
317
|
26 |
|
protected function withErrorHandler(callable $callback) |
|
318
|
|
|
{ |
|
319
|
17 |
|
set_error_handler(function (int $code, string $message) { |
|
320
|
13 |
|
if (strpos($message, 'XMLReader::') !== false) { |
|
321
|
4 |
|
throw new Exception($message, $code); |
|
322
|
|
|
} |
|
323
|
26 |
|
}); |
|
324
|
|
|
|
|
325
|
17 |
|
$response = $callback(); |
|
326
|
|
|
|
|
327
|
15 |
|
restore_error_handler(); |
|
328
|
|
|
|
|
329
|
15 |
|
return $response; |
|
330
|
|
|
} |
|
331
|
|
|
} |
|
332
|
|
|
|