|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
|
|
4
|
|
|
namespace TgScraper\Common; |
|
5
|
|
|
|
|
6
|
|
|
|
|
7
|
|
|
use JetBrains\PhpStorm\ArrayShape; |
|
8
|
|
|
use PHPHtmlParser\Dom; |
|
9
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
|
10
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
|
11
|
|
|
use PHPHtmlParser\Exceptions\ContentLengthException; |
|
12
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
|
13
|
|
|
use PHPHtmlParser\Exceptions\NotLoadedException; |
|
14
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
|
15
|
|
|
use PHPHtmlParser\Exceptions\StrictException; |
|
16
|
|
|
use Psr\Http\Client\ClientExceptionInterface; |
|
17
|
|
|
use Psr\Log\LoggerInterface; |
|
18
|
|
|
use TgScraper\Constants\Versions; |
|
19
|
|
|
use Throwable; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* Class SchemaExtractor |
|
23
|
|
|
* @package TgScraper\Common |
|
24
|
|
|
*/ |
|
25
|
|
|
class SchemaExtractor |
|
26
|
|
|
{ |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Additional methods with boolean return value. |
|
30
|
|
|
*/ |
|
31
|
|
|
private const BOOL_RETURNS = [ |
|
32
|
|
|
'answerShippingQuery', |
|
33
|
|
|
'answerPreCheckoutQuery' |
|
34
|
|
|
]; |
|
35
|
|
|
|
|
36
|
|
|
/** |
|
37
|
|
|
* SchemaExtractor constructor. |
|
38
|
|
|
* @param LoggerInterface $logger |
|
39
|
|
|
* @param string $url |
|
40
|
|
|
*/ |
|
41
|
|
|
public function __construct(private LoggerInterface $logger, private string $url = Versions::LATEST) |
|
42
|
|
|
{ |
|
43
|
|
|
} |
|
44
|
|
|
|
|
45
|
|
|
/** |
|
46
|
|
|
* @throws ParentNotFoundException |
|
47
|
|
|
* @throws ChildNotFoundException |
|
48
|
|
|
*/ |
|
49
|
|
|
#[ArrayShape(['description' => "string", 'table' => "mixed", 'extended_by' => "array"])] |
|
50
|
|
|
private function parseNode(Dom\Node\AbstractNode $node): ?array |
|
51
|
|
|
{ |
|
52
|
|
|
$description = ''; |
|
53
|
|
|
$table = null; |
|
54
|
|
|
$extendedBy = []; |
|
55
|
|
|
$tag = ''; |
|
56
|
|
|
$sibling = $node; |
|
57
|
|
|
while (!str_starts_with($tag, 'h')) { |
|
58
|
|
|
$sibling = $sibling->nextSibling(); |
|
59
|
|
|
$tag = $sibling?->tag?->name(); |
|
60
|
|
|
if (empty($node->text()) or empty($tag) or $tag == 'text') { |
|
61
|
|
|
continue; |
|
62
|
|
|
} elseif ($tag == 'p') { |
|
63
|
|
|
$description .= PHP_EOL . $sibling->innerHtml(); |
|
64
|
|
|
} elseif ($tag == 'ul') { |
|
65
|
|
|
$items = $sibling->find('li'); |
|
66
|
|
|
/* @var Dom\Node\AbstractNode $item */ |
|
67
|
|
|
foreach ($items as $item) { |
|
68
|
|
|
$extendedBy[] = $item->innerText; |
|
69
|
|
|
} |
|
70
|
|
|
break; |
|
71
|
|
|
} elseif ($tag == 'table') { |
|
72
|
|
|
$table = $sibling->find('tbody')->find('tr'); |
|
73
|
|
|
break; |
|
74
|
|
|
} |
|
75
|
|
|
} |
|
76
|
|
|
return ['description' => $description, 'table' => $table, 'extended_by' => $extendedBy]; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @return array |
|
81
|
|
|
* @throws ChildNotFoundException |
|
82
|
|
|
* @throws CircularException |
|
83
|
|
|
* @throws ContentLengthException |
|
84
|
|
|
* @throws LogicalException |
|
85
|
|
|
* @throws NotLoadedException |
|
86
|
|
|
* @throws ParentNotFoundException |
|
87
|
|
|
* @throws StrictException |
|
88
|
|
|
* @throws ClientExceptionInterface |
|
89
|
|
|
* @throws Throwable |
|
90
|
|
|
*/ |
|
91
|
|
|
public function extract(): array |
|
92
|
|
|
{ |
|
93
|
|
|
$dom = new Dom; |
|
94
|
|
|
try { |
|
95
|
|
|
$dom->loadFromURL($this->url); |
|
96
|
|
|
} catch (Throwable $e) { |
|
97
|
|
|
$this->logger->critical(sprintf('Unable to load data from URL "%s": %s', $this->url, $e->getMessage())); |
|
98
|
|
|
throw $e; |
|
99
|
|
|
} |
|
100
|
|
|
try { |
|
101
|
|
|
$elements = $dom->find('h4'); |
|
102
|
|
|
} catch (Throwable $e) { |
|
103
|
|
|
$this->logger->critical(sprintf('Unable to load data from URL "%s": %s', $this->url, $e->getMessage())); |
|
104
|
|
|
throw $e; |
|
105
|
|
|
} |
|
106
|
|
|
$data = []; |
|
107
|
|
|
/* @var Dom\Node\AbstractNode $element */ |
|
108
|
|
|
foreach ($elements as $element) { |
|
109
|
|
|
if (!str_contains($name = $element->text, ' ')) { |
|
110
|
|
|
$isMethod = lcfirst($name) == $name; |
|
111
|
|
|
$path = $isMethod ? 'methods' : 'types'; |
|
112
|
|
|
['description' => $description, 'table' => $table, 'extended_by' => $extendedBy] = self::parseNode($element); |
|
|
|
|
|
|
113
|
|
|
$data[$path][] = self::generateElement( |
|
114
|
|
|
$name, |
|
115
|
|
|
trim($description), |
|
116
|
|
|
$table, |
|
117
|
|
|
$extendedBy, |
|
118
|
|
|
$isMethod |
|
119
|
|
|
); |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
return $data; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* @param string $name |
|
127
|
|
|
* @param string $description |
|
128
|
|
|
* @param Dom\Node\Collection|null $unparsedFields |
|
129
|
|
|
* @param array $extendedBy |
|
130
|
|
|
* @param bool $isMethod |
|
131
|
|
|
* @return array |
|
132
|
|
|
* @throws ChildNotFoundException |
|
133
|
|
|
* @throws CircularException |
|
134
|
|
|
* @throws ContentLengthException |
|
135
|
|
|
* @throws LogicalException |
|
136
|
|
|
* @throws NotLoadedException |
|
137
|
|
|
* @throws StrictException |
|
138
|
|
|
*/ |
|
139
|
|
|
private static function generateElement( |
|
140
|
|
|
string $name, |
|
141
|
|
|
string $description, |
|
142
|
|
|
?Dom\Node\Collection $unparsedFields, |
|
143
|
|
|
array $extendedBy, |
|
144
|
|
|
bool $isMethod |
|
145
|
|
|
): array { |
|
146
|
|
|
$fields = self::parseFields($unparsedFields, $isMethod); |
|
147
|
|
|
$result = [ |
|
148
|
|
|
'name' => $name, |
|
149
|
|
|
'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES), |
|
150
|
|
|
'fields' => $fields, |
|
151
|
|
|
'extended_by' => $extendedBy |
|
152
|
|
|
]; |
|
153
|
|
|
if ($isMethod) { |
|
154
|
|
|
$returnTypes = self::parseReturnTypes($description); |
|
155
|
|
|
if (empty($returnTypes) and in_array($name, self::BOOL_RETURNS)) { |
|
156
|
|
|
$returnTypes[] = 'bool'; |
|
157
|
|
|
} |
|
158
|
|
|
$result['return_types'] = $returnTypes; |
|
159
|
|
|
return $result; |
|
160
|
|
|
} |
|
161
|
|
|
return $result; |
|
162
|
|
|
} |
|
163
|
|
|
|
|
164
|
|
|
/** |
|
165
|
|
|
* @param Dom\Node\Collection|null $fields |
|
166
|
|
|
* @param bool $isMethod |
|
167
|
|
|
* @return array |
|
168
|
|
|
* @throws ChildNotFoundException |
|
169
|
|
|
* @throws NotLoadedException |
|
170
|
|
|
*/ |
|
171
|
|
|
private static function parseFields(?Dom\Node\Collection $fields, bool $isMethod): array |
|
172
|
|
|
{ |
|
173
|
|
|
$parsedFields = []; |
|
174
|
|
|
$fields = $fields ?? []; |
|
175
|
|
|
foreach ($fields as $field) { |
|
176
|
|
|
/* @var Dom $field */ |
|
177
|
|
|
$fieldData = $field->find('td'); |
|
178
|
|
|
$name = $fieldData[0]->text; |
|
179
|
|
|
if (empty($name)) { |
|
180
|
|
|
continue; |
|
181
|
|
|
} |
|
182
|
|
|
$parsedData = [ |
|
183
|
|
|
'name' => $name, |
|
184
|
|
|
'type' => strip_tags($fieldData[1]->innerHtml) |
|
185
|
|
|
]; |
|
186
|
|
|
$parsedData['types'] = self::parseFieldTypes($parsedData['type']); |
|
187
|
|
|
unset($parsedData['type']); |
|
188
|
|
|
if ($isMethod) { |
|
189
|
|
|
$parsedData['required'] = $fieldData[2]->text == 'Yes'; |
|
190
|
|
|
$parsedData['description'] = htmlspecialchars_decode( |
|
191
|
|
|
strip_tags($fieldData[3]->innerHtml ?? $fieldData[3]->text ?? ''), |
|
192
|
|
|
ENT_QUOTES |
|
193
|
|
|
); |
|
194
|
|
|
} else { |
|
195
|
|
|
$description = htmlspecialchars_decode(strip_tags($fieldData[2]->innerHtml), ENT_QUOTES); |
|
196
|
|
|
$parsedData['optional'] = str_starts_with($description, 'Optional.'); |
|
197
|
|
|
$parsedData['description'] = $description; |
|
198
|
|
|
} |
|
199
|
|
|
$parsedFields[] = $parsedData; |
|
200
|
|
|
} |
|
201
|
|
|
return $parsedFields; |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
/** |
|
205
|
|
|
* @param string $rawType |
|
206
|
|
|
* @return array |
|
207
|
|
|
*/ |
|
208
|
|
|
private static function parseFieldTypes(string $rawType): array |
|
209
|
|
|
{ |
|
210
|
|
|
$types = []; |
|
211
|
|
|
foreach (explode(' or ', $rawType) as $rawOrType) { |
|
212
|
|
|
if (stripos($rawOrType, 'array') === 0) { |
|
213
|
|
|
$types[] = str_replace(' and', ',', $rawOrType); |
|
214
|
|
|
continue; |
|
215
|
|
|
} |
|
216
|
|
|
foreach (explode(' and ', $rawOrType) as $unparsedType) { |
|
217
|
|
|
$types[] = $unparsedType; |
|
218
|
|
|
} |
|
219
|
|
|
} |
|
220
|
|
|
$parsedTypes = []; |
|
221
|
|
|
foreach ($types as $type) { |
|
222
|
|
|
$type = trim(str_replace(['number', 'of'], '', $type)); |
|
223
|
|
|
$multiplesCount = substr_count(strtolower($type), 'array'); |
|
224
|
|
|
$parsedType = trim( |
|
225
|
|
|
str_replace( |
|
226
|
|
|
['Array', 'Integer', 'String', 'Boolean', 'Float', 'True'], |
|
227
|
|
|
['', 'int', 'string', 'bool', 'float', 'bool'], |
|
228
|
|
|
$type |
|
229
|
|
|
) |
|
230
|
|
|
); |
|
231
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
|
232
|
|
|
$parsedType = sprintf('Array<%s>', $parsedType); |
|
233
|
|
|
} |
|
234
|
|
|
$parsedTypes[] = $parsedType; |
|
235
|
|
|
} |
|
236
|
|
|
return $parsedTypes; |
|
237
|
|
|
} |
|
238
|
|
|
|
|
239
|
|
|
/** |
|
240
|
|
|
* @param string $description |
|
241
|
|
|
* @return array |
|
242
|
|
|
* @throws ChildNotFoundException |
|
243
|
|
|
* @throws CircularException |
|
244
|
|
|
* @throws NotLoadedException |
|
245
|
|
|
* @throws StrictException |
|
246
|
|
|
* @throws ContentLengthException |
|
247
|
|
|
* @throws LogicalException |
|
248
|
|
|
* @noinspection PhpUndefinedFieldInspection |
|
249
|
|
|
*/ |
|
250
|
|
|
private static function parseReturnTypes(string $description): array |
|
251
|
|
|
{ |
|
252
|
|
|
$returnTypes = []; |
|
253
|
|
|
$phrases = explode('.', $description); |
|
254
|
|
|
$phrases = array_filter( |
|
255
|
|
|
$phrases, |
|
256
|
|
|
function ($phrase) { |
|
257
|
|
|
return (false !== stripos($phrase, 'returns') or false !== stripos($phrase, 'is returned')); |
|
258
|
|
|
} |
|
259
|
|
|
); |
|
260
|
|
|
foreach ($phrases as $phrase) { |
|
261
|
|
|
$dom = new Dom; |
|
262
|
|
|
$dom->loadStr($phrase); |
|
263
|
|
|
$a = $dom->find('a'); |
|
264
|
|
|
$em = $dom->find('em'); |
|
265
|
|
|
foreach ($a as $element) { |
|
266
|
|
|
if ($element->text == 'Messages') { |
|
267
|
|
|
$returnTypes[] = 'Array<Message>'; |
|
268
|
|
|
continue; |
|
269
|
|
|
} |
|
270
|
|
|
|
|
271
|
|
|
$multiplesCount = substr_count(strtolower($phrase), 'array'); |
|
272
|
|
|
$returnType = $element->text; |
|
273
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
|
274
|
|
|
$returnType = sprintf('Array<%s>', $returnType); |
|
275
|
|
|
} |
|
276
|
|
|
$returnTypes[] = $returnType; |
|
277
|
|
|
} |
|
278
|
|
|
foreach ($em as $element) { |
|
279
|
|
|
if (in_array($element->text, ['False', 'force', 'Array'])) { |
|
280
|
|
|
continue; |
|
281
|
|
|
} |
|
282
|
|
|
$type = str_replace(['True', 'Int', 'String'], ['bool', 'int', 'string'], $element->text); |
|
283
|
|
|
$returnTypes[] = $type; |
|
284
|
|
|
} |
|
285
|
|
|
} |
|
286
|
|
|
return $returnTypes; |
|
287
|
|
|
} |
|
288
|
|
|
|
|
289
|
|
|
} |
Sysbot-org /
tgscraper
Loading history...