1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
|
4
|
|
|
namespace TgScraper\Common; |
5
|
|
|
|
6
|
|
|
|
7
|
|
|
use Composer\InstalledVersions; |
8
|
|
|
use InvalidArgumentException; |
9
|
|
|
use JetBrains\PhpStorm\ArrayShape; |
10
|
|
|
use OutOfBoundsException; |
11
|
|
|
use PHPHtmlParser\Dom; |
12
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
13
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
14
|
|
|
use PHPHtmlParser\Exceptions\ContentLengthException; |
15
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
16
|
|
|
use PHPHtmlParser\Exceptions\NotLoadedException; |
17
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
18
|
|
|
use PHPHtmlParser\Exceptions\StrictException; |
19
|
|
|
use Psr\Http\Client\ClientExceptionInterface; |
20
|
|
|
use Psr\Log\LoggerInterface; |
21
|
|
|
use TgScraper\Constants\Versions; |
22
|
|
|
use Throwable; |
23
|
|
|
|
24
|
|
|
/** |
25
|
|
|
* Class SchemaExtractor |
26
|
|
|
* @package TgScraper\Common |
27
|
|
|
*/ |
28
|
|
|
class SchemaExtractor |
29
|
|
|
{ |
30
|
|
|
|
31
|
|
|
/** |
32
|
|
|
* Additional methods with boolean return value. |
33
|
|
|
*/ |
34
|
|
|
private const BOOL_RETURNS = [ |
35
|
|
|
'answerShippingQuery', |
36
|
|
|
'answerPreCheckoutQuery' |
37
|
|
|
]; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @var string |
41
|
|
|
*/ |
42
|
|
|
private string $version; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* SchemaExtractor constructor. |
46
|
|
|
* @param LoggerInterface $logger |
47
|
|
|
* @param Dom $dom |
48
|
|
|
* @throws ChildNotFoundException |
49
|
|
|
* @throws NotLoadedException |
50
|
|
|
*/ |
51
|
|
|
public function __construct(private LoggerInterface $logger, private Dom $dom) |
52
|
|
|
{ |
53
|
|
|
$this->version = $this->parseVersion(); |
54
|
|
|
$this->logger->info('Bot API version: ' . $this->version); |
55
|
|
|
} |
56
|
|
|
|
57
|
|
|
|
58
|
|
|
/** |
59
|
|
|
* @param LoggerInterface $logger |
60
|
|
|
* @param string $version |
61
|
|
|
* @return SchemaExtractor |
62
|
|
|
* @throws OutOfBoundsException |
63
|
|
|
* @throws Throwable |
64
|
|
|
*/ |
65
|
|
|
public static function fromVersion(LoggerInterface $logger, string $version = Versions::LATEST): SchemaExtractor |
66
|
|
|
{ |
67
|
|
|
if (InstalledVersions::isInstalled('sysbot/tgscraper-cache') and class_exists('\TgScraper\Cache\CacheLoader')) { |
68
|
|
|
$logger->info('Cache package detected, searching for a cached version.'); |
69
|
|
|
try { |
70
|
|
|
$path = \TgScraper\Cache\CacheLoader::getCachedVersion($version); |
|
|
|
|
71
|
|
|
$logger->info('Cached version found.'); |
72
|
|
|
return self::fromFile($logger, $path); |
73
|
|
|
} catch (OutOfBoundsException) { |
74
|
|
|
$logger->info('Cached version not found, continuing with URL.'); |
75
|
|
|
} |
76
|
|
|
} |
77
|
|
|
$url = Versions::getUrlFromText($version); |
78
|
|
|
$logger->info(sprintf('Using URL: %s', $url)); |
79
|
|
|
return self::fromUrl($logger, $url); |
80
|
|
|
} |
81
|
|
|
|
82
|
|
|
/** |
83
|
|
|
* @param LoggerInterface $logger |
84
|
|
|
* @param string $path |
85
|
|
|
* @return SchemaExtractor |
86
|
|
|
* @throws Throwable |
87
|
|
|
*/ |
88
|
|
|
public static function fromFile(LoggerInterface $logger, string $path): SchemaExtractor |
89
|
|
|
{ |
90
|
|
|
$dom = new Dom; |
91
|
|
|
if (!file_exists($path)) { |
92
|
|
|
throw new InvalidArgumentException('File not found'); |
93
|
|
|
} |
94
|
|
|
$path = realpath($path); |
95
|
|
|
try { |
96
|
|
|
$logger->info(sprintf('Loading data from file "%s".', $path)); |
97
|
|
|
$dom->loadFromFile($path); |
98
|
|
|
$logger->info('Data loaded.'); |
99
|
|
|
} catch (Throwable $e) { |
100
|
|
|
$logger->critical(sprintf('Unable to load data from "%s": %s', $path, $e->getMessage())); |
101
|
|
|
throw $e; |
102
|
|
|
} |
103
|
|
|
return new self($logger, $dom); |
104
|
|
|
} |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* @param LoggerInterface $logger |
108
|
|
|
* @param string $url |
109
|
|
|
* @return SchemaExtractor |
110
|
|
|
* @throws ChildNotFoundException |
111
|
|
|
* @throws CircularException |
112
|
|
|
* @throws ClientExceptionInterface |
113
|
|
|
* @throws ContentLengthException |
114
|
|
|
* @throws LogicalException |
115
|
|
|
* @throws StrictException |
116
|
|
|
* @throws NotLoadedException |
117
|
|
|
*/ |
118
|
|
|
public static function fromUrl(LoggerInterface $logger, string $url): SchemaExtractor |
119
|
|
|
{ |
120
|
|
|
$dom = new Dom; |
121
|
|
|
try { |
122
|
|
|
$dom->loadFromURL($url); |
123
|
|
|
} catch (Throwable $e) { |
124
|
|
|
$logger->critical(sprintf('Unable to load data from URL "%s": %s', $url, $e->getMessage())); |
125
|
|
|
throw $e; |
126
|
|
|
} |
127
|
|
|
$logger->info(sprintf('Data loaded from "%s".', $url)); |
128
|
|
|
return new self($logger, $dom); |
129
|
|
|
} |
130
|
|
|
|
131
|
|
|
/** |
132
|
|
|
* @throws ParentNotFoundException |
133
|
|
|
* @throws ChildNotFoundException |
134
|
|
|
*/ |
135
|
|
|
#[ArrayShape(['description' => "string", 'table' => "mixed", 'extended_by' => "array"])] |
136
|
|
|
private static function parseNode(Dom\Node\AbstractNode $node): ?array |
137
|
|
|
{ |
138
|
|
|
$description = ''; |
139
|
|
|
$table = null; |
140
|
|
|
$extendedBy = []; |
141
|
|
|
$tag = ''; |
142
|
|
|
$sibling = $node; |
143
|
|
|
while (!str_starts_with($tag, 'h')) { |
144
|
|
|
$sibling = $sibling->nextSibling(); |
145
|
|
|
$tag = $sibling?->tag?->name(); |
146
|
|
|
if (empty($node->text()) or empty($tag) or $tag == 'text') { |
147
|
|
|
continue; |
148
|
|
|
} elseif ($tag == 'p') { |
149
|
|
|
$description .= PHP_EOL . $sibling->innerHtml(); |
150
|
|
|
} elseif ($tag == 'ul') { |
151
|
|
|
$items = $sibling->find('li'); |
152
|
|
|
/* @var Dom\Node\AbstractNode $item */ |
153
|
|
|
foreach ($items as $item) { |
154
|
|
|
$extendedBy[] = $item->innerText; |
155
|
|
|
} |
156
|
|
|
break; |
157
|
|
|
} elseif ($tag == 'table') { |
158
|
|
|
$table = $sibling->find('tbody')->find('tr'); |
159
|
|
|
break; |
160
|
|
|
} |
161
|
|
|
} |
162
|
|
|
return ['description' => $description, 'table' => $table, 'extended_by' => $extendedBy]; |
163
|
|
|
} |
164
|
|
|
|
165
|
|
|
/** |
166
|
|
|
* @throws ChildNotFoundException |
167
|
|
|
* @throws NotLoadedException |
168
|
|
|
*/ |
169
|
|
|
private function parseVersion(): string |
170
|
|
|
{ |
171
|
|
|
/** @var Dom\Node\AbstractNode $element */ |
172
|
|
|
$element = $this->dom->find('h3')[0]; |
173
|
|
|
$tag = ''; |
174
|
|
|
while ($tag != 'p') { |
175
|
|
|
try { |
176
|
|
|
$element = $element->nextSibling(); |
177
|
|
|
} catch (ChildNotFoundException | ParentNotFoundException) { |
178
|
|
|
continue; |
179
|
|
|
} |
180
|
|
|
$tag = $element->tag->name(); |
181
|
|
|
} |
182
|
|
|
$versionNumbers = explode('.', str_replace('Bot API ', '', $element->innerText)); |
183
|
|
|
return sprintf( |
184
|
|
|
'%s.%s.%s', |
185
|
|
|
$versionNumbers[0] ?? '1', |
186
|
|
|
$versionNumbers[1] ?? '0', |
187
|
|
|
$versionNumbers[2] ?? '0' |
188
|
|
|
); |
189
|
|
|
} |
190
|
|
|
|
191
|
|
|
/** |
192
|
|
|
* @return array |
193
|
|
|
* @throws Throwable |
194
|
|
|
*/ |
195
|
|
|
#[ArrayShape(['version' => "string", 'methods' => "array", 'types' => "array"])] |
196
|
|
|
public function extract(): array |
197
|
|
|
{ |
198
|
|
|
try { |
199
|
|
|
$elements = $this->dom->find('h4'); |
200
|
|
|
} catch (Throwable $e) { |
201
|
|
|
$this->logger->critical(sprintf('Unable to parse data: %s', $e->getMessage())); |
202
|
|
|
throw $e; |
203
|
|
|
} |
204
|
|
|
$data = ['version' => $this->version]; |
205
|
|
|
/* @var Dom\Node\AbstractNode $element */ |
206
|
|
|
foreach ($elements as $element) { |
207
|
|
|
if (!str_contains($name = $element->text, ' ')) { |
208
|
|
|
$isMethod = lcfirst($name) == $name; |
209
|
|
|
$path = $isMethod ? 'methods' : 'types'; |
210
|
|
|
['description' => $description, 'table' => $table, 'extended_by' => $extendedBy] = self::parseNode( |
211
|
|
|
$element |
212
|
|
|
); |
213
|
|
|
$data[$path][] = self::generateElement( |
214
|
|
|
$name, |
215
|
|
|
trim($description), |
216
|
|
|
$table, |
217
|
|
|
$extendedBy, |
218
|
|
|
$isMethod |
219
|
|
|
); |
220
|
|
|
} |
221
|
|
|
} |
222
|
|
|
return $data; |
223
|
|
|
} |
224
|
|
|
|
225
|
|
|
/** |
226
|
|
|
* @param string $name |
227
|
|
|
* @param string $description |
228
|
|
|
* @param Dom\Node\Collection|null $unparsedFields |
229
|
|
|
* @param array $extendedBy |
230
|
|
|
* @param bool $isMethod |
231
|
|
|
* @return array |
232
|
|
|
* @throws ChildNotFoundException |
233
|
|
|
* @throws CircularException |
234
|
|
|
* @throws ContentLengthException |
235
|
|
|
* @throws LogicalException |
236
|
|
|
* @throws NotLoadedException |
237
|
|
|
* @throws StrictException |
238
|
|
|
*/ |
239
|
|
|
private static function generateElement( |
240
|
|
|
string $name, |
241
|
|
|
string $description, |
242
|
|
|
?Dom\Node\Collection $unparsedFields, |
243
|
|
|
array $extendedBy, |
244
|
|
|
bool $isMethod |
245
|
|
|
): array { |
246
|
|
|
$fields = self::parseFields($unparsedFields, $isMethod); |
247
|
|
|
$result = [ |
248
|
|
|
'name' => $name, |
249
|
|
|
'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES), |
250
|
|
|
'fields' => $fields |
251
|
|
|
]; |
252
|
|
|
if ($isMethod) { |
253
|
|
|
$returnTypes = self::parseReturnTypes($description); |
254
|
|
|
if (empty($returnTypes) and in_array($name, self::BOOL_RETURNS)) { |
255
|
|
|
$returnTypes[] = 'bool'; |
256
|
|
|
} |
257
|
|
|
$result['return_types'] = $returnTypes; |
258
|
|
|
return $result; |
259
|
|
|
} |
260
|
|
|
$result['extended_by'] = $extendedBy; |
261
|
|
|
return $result; |
262
|
|
|
} |
263
|
|
|
|
264
|
|
|
/** |
265
|
|
|
* @param Dom\Node\Collection|null $fields |
266
|
|
|
* @param bool $isMethod |
267
|
|
|
* @return array |
268
|
|
|
*/ |
269
|
|
|
private static function parseFields(?Dom\Node\Collection $fields, bool $isMethod): array |
270
|
|
|
{ |
271
|
|
|
$parsedFields = []; |
272
|
|
|
$fields = $fields ?? []; |
273
|
|
|
foreach ($fields as $field) { |
274
|
|
|
/* @var Dom\Node\AbstractNode $fieldData */ |
275
|
|
|
$fieldData = $field->find('td'); |
276
|
|
|
$name = $fieldData[0]->text; |
277
|
|
|
if (empty($name)) { |
278
|
|
|
continue; |
279
|
|
|
} |
280
|
|
|
$parsedData = [ |
281
|
|
|
'name' => $name, |
282
|
|
|
'type' => strip_tags($fieldData[1]->innerHtml) |
283
|
|
|
]; |
284
|
|
|
$parsedData['types'] = self::parseFieldTypes($parsedData['type']); |
285
|
|
|
unset($parsedData['type']); |
286
|
|
|
if ($isMethod) { |
287
|
|
|
$parsedData['optional'] = $fieldData[2]->text != 'Yes'; |
288
|
|
|
$parsedData['description'] = htmlspecialchars_decode( |
289
|
|
|
strip_tags($fieldData[3]->innerHtml ?? $fieldData[3]->text ?? ''), |
290
|
|
|
ENT_QUOTES |
291
|
|
|
); |
292
|
|
|
} else { |
293
|
|
|
$description = htmlspecialchars_decode(strip_tags($fieldData[2]->innerHtml), ENT_QUOTES); |
294
|
|
|
$parsedData['optional'] = str_starts_with($description, 'Optional.'); |
295
|
|
|
$parsedData['description'] = $description; |
296
|
|
|
} |
297
|
|
|
$parsedFields[] = $parsedData; |
298
|
|
|
} |
299
|
|
|
return $parsedFields; |
300
|
|
|
} |
301
|
|
|
|
302
|
|
|
/** |
303
|
|
|
* @param string $rawType |
304
|
|
|
* @return array |
305
|
|
|
*/ |
306
|
|
|
private static function parseFieldTypes(string $rawType): array |
307
|
|
|
{ |
308
|
|
|
$types = []; |
309
|
|
|
foreach (explode(' or ', $rawType) as $rawOrType) { |
310
|
|
|
if (stripos($rawOrType, 'array') === 0) { |
311
|
|
|
$types[] = str_replace(' and', ',', $rawOrType); |
312
|
|
|
continue; |
313
|
|
|
} |
314
|
|
|
foreach (explode(' and ', $rawOrType) as $unparsedType) { |
315
|
|
|
$types[] = $unparsedType; |
316
|
|
|
} |
317
|
|
|
} |
318
|
|
|
$parsedTypes = []; |
319
|
|
|
foreach ($types as $type) { |
320
|
|
|
$type = trim(str_replace(['number', 'of'], '', $type)); |
321
|
|
|
$multiplesCount = substr_count(strtolower($type), 'array'); |
322
|
|
|
$parsedType = trim( |
323
|
|
|
str_replace( |
324
|
|
|
['Array', 'Integer', 'String', 'Boolean', 'Float', 'True'], |
325
|
|
|
['', 'int', 'string', 'bool', 'float', 'bool'], |
326
|
|
|
$type |
327
|
|
|
) |
328
|
|
|
); |
329
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
330
|
|
|
$parsedType = sprintf('Array<%s>', $parsedType); |
331
|
|
|
} |
332
|
|
|
$parsedTypes[] = $parsedType; |
333
|
|
|
} |
334
|
|
|
return $parsedTypes; |
335
|
|
|
} |
336
|
|
|
|
337
|
|
|
/** |
338
|
|
|
* @param string $description |
339
|
|
|
* @return array |
340
|
|
|
* @throws ChildNotFoundException |
341
|
|
|
* @throws CircularException |
342
|
|
|
* @throws NotLoadedException |
343
|
|
|
* @throws StrictException |
344
|
|
|
* @throws ContentLengthException |
345
|
|
|
* @throws LogicalException |
346
|
|
|
* @noinspection PhpUndefinedFieldInspection |
347
|
|
|
*/ |
348
|
|
|
private static function parseReturnTypes(string $description): array |
349
|
|
|
{ |
350
|
|
|
$returnTypes = []; |
351
|
|
|
$phrases = explode('.', $description); |
352
|
|
|
$phrases = array_filter( |
353
|
|
|
$phrases, |
354
|
|
|
function ($phrase) { |
355
|
|
|
return (false !== stripos($phrase, 'returns') or false !== stripos($phrase, 'is returned')); |
356
|
|
|
} |
357
|
|
|
); |
358
|
|
|
foreach ($phrases as $phrase) { |
359
|
|
|
$dom = new Dom; |
360
|
|
|
$dom->loadStr($phrase); |
361
|
|
|
$a = $dom->find('a'); |
362
|
|
|
$em = $dom->find('em'); |
363
|
|
|
foreach ($a as $element) { |
364
|
|
|
if ($element->text == 'Messages') { |
365
|
|
|
$returnTypes[] = 'Array<Message>'; |
366
|
|
|
continue; |
367
|
|
|
} |
368
|
|
|
|
369
|
|
|
$multiplesCount = substr_count(strtolower($phrase), 'array'); |
370
|
|
|
$returnType = $element->text; |
371
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
372
|
|
|
$returnType = sprintf('Array<%s>', $returnType); |
373
|
|
|
} |
374
|
|
|
$returnTypes[] = $returnType; |
375
|
|
|
} |
376
|
|
|
foreach ($em as $element) { |
377
|
|
|
if (in_array($element->text, ['False', 'force', 'Array'])) { |
378
|
|
|
continue; |
379
|
|
|
} |
380
|
|
|
$type = str_replace(['True', 'Int', 'String'], ['bool', 'int', 'string'], $element->text); |
381
|
|
|
$returnTypes[] = $type; |
382
|
|
|
} |
383
|
|
|
} |
384
|
|
|
return $returnTypes; |
385
|
|
|
} |
386
|
|
|
|
387
|
|
|
} |
The issue could also be caused by a filter entry in the build configuration. If the path has been excluded in your configuration, e.g.
excluded_paths: ["lib/*"]
, you can move it to the dependency path list as follows:For further information see https://scrutinizer-ci.com/docs/tools/php/php-scrutinizer/#list-dependency-paths