1
|
|
|
<?php |
2
|
|
|
declare(strict_types=1); |
3
|
|
|
|
4
|
|
|
namespace JsonDecodeStream; |
5
|
|
|
|
6
|
|
|
use Generator; |
7
|
|
|
use JsonDecodeStream\Collector\Collector; |
8
|
|
|
use JsonDecodeStream\Collector\CollectorInterface; |
9
|
|
|
use JsonDecodeStream\Exception\ParserException; |
10
|
|
|
use JsonDecodeStream\Internal\SourceBuffer; |
11
|
|
|
use JsonDecodeStream\Internal\Stack; |
12
|
|
|
use JsonDecodeStream\Internal\StackFrame; |
13
|
|
|
use JsonDecodeStream\Source\FileSource; |
14
|
|
|
use JsonDecodeStream\Source\Psr7Source; |
15
|
|
|
use JsonDecodeStream\Source\SourceInterface; |
16
|
|
|
use JsonDecodeStream\Source\StreamSource; |
17
|
|
|
use JsonDecodeStream\Source\StringSource; |
18
|
|
|
use Psr\Http\Message\StreamInterface; |
19
|
|
|
|
20
|
|
|
class Parser |
21
|
|
|
{ |
22
|
|
|
/** @var SourceBuffer */ |
23
|
|
|
protected $buffer; |
24
|
|
|
|
25
|
|
|
/** @var Stack */ |
26
|
|
|
protected $stack; |
27
|
|
|
|
28
|
44 |
|
public function __construct(SourceInterface $source) |
29
|
|
|
{ |
30
|
44 |
|
$this->buffer = new SourceBuffer($source); |
31
|
44 |
|
} |
32
|
|
|
|
33
|
17 |
|
public static function fromString(string $string) |
34
|
|
|
{ |
35
|
17 |
|
return new static(new StringSource($string)); |
36
|
|
|
} |
37
|
|
|
|
38
|
27 |
|
public static function fromFile(string $path) |
39
|
|
|
{ |
40
|
27 |
|
return new static(new FileSource($path)); |
41
|
|
|
} |
42
|
|
|
|
43
|
|
|
public static function fromStream($stream) |
44
|
|
|
{ |
45
|
|
|
return new static(new StreamSource($stream)); |
46
|
|
|
} |
47
|
|
|
|
48
|
|
|
public static function fromPsr7(StreamInterface $stream) |
49
|
|
|
{ |
50
|
|
|
return new static(new Psr7Source($stream)); |
51
|
|
|
} |
52
|
|
|
|
53
|
|
|
/** |
54
|
|
|
* @param string|string[]|Collector|Collector[] $selectors |
55
|
|
|
* single or coma-separated selector string |
56
|
|
|
* or custom CollectorInterface implementation |
57
|
|
|
* or array of any of both |
58
|
|
|
* @param bool $objectsAsAssoc |
59
|
|
|
* @return iterable|Generator |
60
|
|
|
* @throws Exception\CollectorException |
61
|
|
|
* @throws Exception\SelectorException |
62
|
|
|
* @throws Exception\TokenizerException |
63
|
|
|
* @throws ParserException |
64
|
|
|
*/ |
65
|
7 |
|
public function items($selectors, bool $objectsAsAssoc = false) |
66
|
|
|
{ |
67
|
7 |
|
if (is_string($selectors)) { |
68
|
5 |
|
$selectorsArray = explode(',', $selectors); |
69
|
2 |
|
} else if (is_array($selectors)) { |
70
|
|
|
$selectorsArray = $selectors; |
71
|
2 |
|
} else if ($selectors instanceof CollectorInterface) { |
|
|
|
|
72
|
2 |
|
$selectorsArray = [ $selectors ]; |
73
|
|
|
} else { |
74
|
|
|
throw new ParserException('Unexpected selectors are provided', ParserException::CODE_INVALID_ARGUMENT); |
75
|
|
|
} |
76
|
7 |
|
if (empty($selectorsArray)) { |
77
|
|
|
throw new ParserException('No selectors are provided', ParserException::CODE_INVALID_ARGUMENT); |
78
|
|
|
} |
79
|
7 |
|
$collectors = []; |
80
|
7 |
|
foreach ($selectorsArray as $selector) { |
81
|
7 |
|
if (is_string($selector)) { |
82
|
5 |
|
$collectors[] = new Collector($selector, $objectsAsAssoc); |
83
|
2 |
|
} elseif ($selector instanceof CollectorInterface) { |
84
|
2 |
|
$collectors[] = $selector; |
85
|
|
|
} else { |
86
|
|
|
throw new ParserException( |
87
|
|
|
'Invalid collector: ' |
88
|
|
|
. is_object($selector) ? get_class($selector) : gettype($selector), |
89
|
|
|
ParserException::CODE_INVALID_ARGUMENT |
90
|
|
|
); |
91
|
|
|
} |
92
|
|
|
} |
93
|
|
|
|
94
|
7 |
|
foreach ($this->events() as $event) { |
95
|
7 |
|
foreach ($collectors as $collector) { |
96
|
7 |
|
$yielded = $collector->processEvent($event); |
97
|
7 |
|
if (is_array($yielded)) { |
98
|
6 |
|
if (count($yielded) != 2) { |
99
|
|
|
throw ParserException::unexpectedCollectorReturn($yielded, $event); |
|
|
|
|
100
|
|
|
} |
101
|
6 |
|
[ $key, $value ] = $yielded; |
102
|
6 |
|
yield $key => $value; |
103
|
7 |
|
} else if ($yielded instanceof Generator) { |
104
|
1 |
|
foreach ($yielded as $yieldedSingle) { |
105
|
1 |
|
if (!is_array($yieldedSingle) || count($yieldedSingle) != 2) { |
106
|
|
|
throw ParserException::unexpectedCollectorReturn($yielded, $event); |
|
|
|
|
107
|
|
|
} |
108
|
1 |
|
[ $key, $value ] = $yieldedSingle; |
109
|
1 |
|
yield $key => $value; |
110
|
|
|
} |
111
|
6 |
|
} else if ($yielded === null) { |
112
|
6 |
|
continue; |
113
|
|
|
} else { |
114
|
|
|
throw ParserException::unexpectedCollectorReturn($yielded, $event); |
|
|
|
|
115
|
|
|
} |
116
|
|
|
} |
117
|
|
|
} |
118
|
7 |
|
} |
119
|
|
|
|
120
|
|
|
/** |
121
|
|
|
* @return Generator|Event[] |
122
|
|
|
* @psalm-return \Generator<Event> |
123
|
|
|
* @throws ParserException |
124
|
|
|
* @noinspection PhpStatementHasEmptyBodyInspection |
125
|
|
|
*/ |
126
|
44 |
|
public function events(): Generator |
127
|
|
|
{ |
128
|
44 |
|
$stack = new Stack(); |
129
|
44 |
|
$tokens = $this->tokens(); |
130
|
|
|
|
131
|
|
|
// shortcut to event factory |
132
|
44 |
|
$createEvent = function (string $eventId, $value = null) use ($stack, &$token): Event { |
133
|
44 |
|
return $this->createEvent($eventId, $value, $stack, $token->getLineNumber(), $token->getCharNumber()); |
134
|
44 |
|
}; |
135
|
|
|
|
136
|
44 |
|
foreach ($tokens as $token) { |
137
|
44 |
|
if ($token->getId() == Token::WHITESPACE) { |
138
|
|
|
// ignore whitespaces |
139
|
42 |
|
continue; |
140
|
|
|
} |
141
|
|
|
|
142
|
44 |
|
if ($stack->isEmpty()) { |
143
|
44 |
|
switch ($token->getId()) { |
144
|
|
|
case Token::OBJECT_START: |
145
|
29 |
|
yield $createEvent(Event::DOCUMENT_START); |
146
|
29 |
|
yield $createEvent(Event::OBJECT_START); |
147
|
|
|
|
148
|
29 |
|
$stack->push(StackFrame::object()); |
149
|
29 |
|
break; |
150
|
|
|
|
151
|
|
|
case Token::ARRAY_START: |
152
|
15 |
|
yield $createEvent(Event::DOCUMENT_START); |
153
|
15 |
|
yield $createEvent(Event::ARRAY_START); |
154
|
|
|
|
155
|
15 |
|
$stack->push(StackFrame::array()); |
156
|
15 |
|
break; |
157
|
|
|
|
158
|
|
|
default: |
159
|
|
|
throw ParserException::unexpectedToken($token); |
160
|
|
|
} |
161
|
44 |
|
continue; |
162
|
|
|
} |
163
|
44 |
|
if ($stack->current()->isAwaitsComa()) { |
164
|
42 |
|
if ($token->getId() == Token::COMA) { |
165
|
42 |
|
$stack->current()->setAwaitsComa(false); |
166
|
42 |
|
if ($stack->current()->isObject()) { |
167
|
29 |
|
$stack->current()->setLastKey(null); |
168
|
29 |
|
$stack->current()->setAwaitsKey(true); |
169
|
|
|
} |
170
|
42 |
|
continue; |
171
|
42 |
|
} elseif ($stack->current()->isObject() && $token->getId() == Token::OBJECT_END) { |
172
|
|
|
// pass |
173
|
36 |
|
} elseif ($stack->current()->isArray() && $token->getId() == Token::ARRAY_END) { |
174
|
|
|
// pass |
175
|
|
|
} else { |
176
|
|
|
throw ParserException::expectedButGot('","', $token); |
177
|
|
|
} |
178
|
|
|
} |
179
|
44 |
|
if ($stack->current()->isAwaitsKeyDelimiter()) { |
180
|
34 |
|
if ($token->getId() == Token::KEY_DELIMITER) { |
181
|
34 |
|
$stack->current()->setAwaitsKeyDelimiter(false); |
182
|
34 |
|
continue; |
183
|
|
|
} else { |
184
|
|
|
throw ParserException::expectedButGot('":"', $token); |
185
|
|
|
} |
186
|
|
|
} |
187
|
44 |
|
if ($stack->current()->isAwaitsKey()) { |
188
|
35 |
|
if ($token->getId() != Token::STRING && $token->getId() != Token::OBJECT_END) { |
189
|
|
|
throw ParserException::expectedButGot('object key', $token); |
190
|
|
|
} |
191
|
|
|
} |
192
|
|
|
|
193
|
44 |
|
if ($stack->current()->isArray()) { |
194
|
37 |
|
switch ($token->getId()) { |
195
|
|
|
case Token::STRING: |
196
|
|
|
case Token::NUMBER: |
197
|
|
|
case Token::NULL: |
198
|
|
|
case Token::TRUE: |
199
|
|
|
case Token::FALSE: |
200
|
35 |
|
$stack->current()->setAwaitsComa(true); |
201
|
35 |
|
$stack->current()->setLastKey( |
202
|
35 |
|
$stack->current()->getElementCount() |
203
|
|
|
); |
204
|
35 |
|
$stack->current()->incrementElementCount(); |
205
|
35 |
|
yield $createEvent(Event::VALUE, $token->getValue()); |
206
|
35 |
|
break; |
207
|
|
|
|
208
|
|
|
case Token::ARRAY_START: |
209
|
7 |
|
yield $createEvent(Event::ARRAY_START); |
210
|
7 |
|
$stack->current()->setAwaitsComa(true); |
211
|
7 |
|
$stack->current()->setLastKey( |
212
|
7 |
|
$stack->current()->getElementCount() |
213
|
|
|
); |
214
|
7 |
|
$stack->current()->incrementElementCount(); |
215
|
7 |
|
$stack->push(StackFrame::array()); |
216
|
7 |
|
break; |
217
|
|
|
|
218
|
|
|
case Token::ARRAY_END: |
219
|
37 |
|
$stack->pop(); |
220
|
37 |
|
yield $createEvent(Event::ARRAY_END); |
221
|
37 |
|
if ($stack->isEmpty()) { |
222
|
15 |
|
yield $createEvent(Event::DOCUMENT_END); |
223
|
|
|
} |
224
|
37 |
|
break; |
225
|
|
|
|
226
|
|
|
case Token::OBJECT_START: |
227
|
27 |
|
$stack->current()->setLastKey( |
228
|
27 |
|
$stack->current()->getElementCount() |
229
|
|
|
); |
230
|
27 |
|
yield $createEvent(Event::OBJECT_START); |
231
|
27 |
|
$stack->current()->setAwaitsComa(true); |
232
|
27 |
|
$stack->current()->incrementElementCount(); |
233
|
27 |
|
$stack->push(StackFrame::object()); |
234
|
27 |
|
break; |
235
|
|
|
|
236
|
|
|
case Token::OBJECT_END: |
237
|
|
|
$stack->pop(); |
238
|
|
|
yield $createEvent(Event::OBJECT_END); |
239
|
|
|
break; |
240
|
|
|
|
241
|
|
|
default: |
242
|
|
|
throw ParserException::unexpectedToken($token); |
243
|
|
|
} |
244
|
|
|
|
245
|
37 |
|
continue; |
246
|
|
|
} |
247
|
|
|
|
248
|
35 |
|
if ($stack->current()->isObject()) { |
249
|
35 |
|
switch ($token->getId()) { |
250
|
|
|
case Token::STRING: |
251
|
34 |
|
if ($stack->current()->isAwaitsKey()) { |
252
|
34 |
|
yield $createEvent(Event::KEY, $token->getValue()); |
253
|
34 |
|
$stack->current()->setLastKey($token->getValue()); |
254
|
34 |
|
$stack->current()->setAwaitsKeyDelimiter(true); |
255
|
34 |
|
$stack->current()->setAwaitsKey(false); |
256
|
|
|
} else { |
257
|
25 |
|
yield $createEvent(Event::VALUE, $token->getValue()); |
258
|
25 |
|
$stack->current()->setAwaitsComa(true); |
259
|
25 |
|
$stack->current()->incrementElementCount(); |
260
|
|
|
} |
261
|
34 |
|
break; |
262
|
|
|
|
263
|
|
|
case Token::NUMBER: |
264
|
|
|
case Token::NULL: |
265
|
|
|
case Token::TRUE: |
266
|
|
|
case Token::FALSE: |
267
|
28 |
|
yield $createEvent(Event::VALUE, $token->getValue()); |
268
|
28 |
|
$stack->current()->setAwaitsComa(true); |
269
|
28 |
|
$stack->current()->incrementElementCount(); |
270
|
28 |
|
break; |
271
|
|
|
|
272
|
|
|
case Token::ARRAY_START: |
273
|
28 |
|
yield $createEvent(Event::ARRAY_START); |
274
|
28 |
|
$stack->current()->setAwaitsComa(true); |
275
|
28 |
|
$stack->current()->incrementElementCount(); |
276
|
28 |
|
$stack->push(StackFrame::array()); |
277
|
28 |
|
break; |
278
|
|
|
|
279
|
|
|
case Token::ARRAY_END: |
280
|
|
|
yield $createEvent(Event::ARRAY_END); |
281
|
|
|
$stack->pop(); |
282
|
|
|
break; |
283
|
|
|
|
284
|
|
|
case Token::OBJECT_START: |
285
|
27 |
|
yield $createEvent(Event::OBJECT_START); |
286
|
27 |
|
$stack->current()->setAwaitsComa(true); |
287
|
27 |
|
$stack->current()->incrementElementCount(); |
288
|
27 |
|
$stack->push(StackFrame::object()); |
289
|
27 |
|
break; |
290
|
|
|
|
291
|
|
|
case Token::OBJECT_END: |
292
|
35 |
|
$stack->pop(); |
293
|
35 |
|
yield $createEvent(Event::OBJECT_END); |
294
|
35 |
|
if ($stack->isEmpty()) { |
295
|
29 |
|
yield $createEvent(Event::DOCUMENT_END); |
296
|
|
|
} |
297
|
35 |
|
break; |
298
|
|
|
|
299
|
|
|
default: |
300
|
|
|
throw ParserException::unexpectedToken($token); |
301
|
|
|
} |
302
|
|
|
|
303
|
35 |
|
continue; |
304
|
|
|
} |
305
|
|
|
} |
306
|
44 |
|
} |
307
|
|
|
|
308
|
|
|
/** |
309
|
|
|
* @return iterable|Tokenizer|Token[] |
310
|
|
|
*/ |
311
|
44 |
|
public function tokens(): iterable |
312
|
|
|
{ |
313
|
44 |
|
return new Tokenizer($this->buffer); |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
/** |
317
|
|
|
* @param string $eventId |
318
|
|
|
* @param $value |
319
|
|
|
* @param Stack $stack |
320
|
|
|
* @param int $lineNumber |
321
|
|
|
* @param int $charNumber |
322
|
|
|
* @return Event |
323
|
|
|
*/ |
324
|
44 |
|
protected function createEvent(string $eventId, $value, Stack $stack, int $lineNumber, int $charNumber): Event |
325
|
|
|
{ |
326
|
44 |
|
return new Event($eventId, $value, $stack, $lineNumber, $charNumber); |
327
|
|
|
} |
328
|
|
|
} |
329
|
|
|
|