1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace TgScraper; |
4
|
|
|
|
5
|
|
|
use Exception; |
6
|
|
|
use JsonException; |
7
|
|
|
use PHPHtmlParser\Dom; |
8
|
|
|
use PHPHtmlParser\Exceptions\ChildNotFoundException; |
9
|
|
|
use PHPHtmlParser\Exceptions\CircularException; |
10
|
|
|
use PHPHtmlParser\Exceptions\ContentLengthException; |
11
|
|
|
use PHPHtmlParser\Exceptions\LogicalException; |
12
|
|
|
use PHPHtmlParser\Exceptions\NotLoadedException; |
13
|
|
|
use PHPHtmlParser\Exceptions\ParentNotFoundException; |
14
|
|
|
use PHPHtmlParser\Exceptions\StrictException; |
15
|
|
|
use Psr\Http\Client\ClientExceptionInterface; |
16
|
|
|
use Symfony\Component\Yaml\Yaml; |
17
|
|
|
|
18
|
|
|
class Generator |
19
|
|
|
{ |
20
|
|
|
|
21
|
|
|
private const BOOL_RETURNS = [ |
22
|
|
|
'answerShippingQuery', |
23
|
|
|
'answerPreCheckoutQuery' |
24
|
|
|
]; |
25
|
|
|
|
26
|
|
|
public const BOT_API_URL = 'https://core.telegram.org/bots/api'; |
27
|
|
|
|
28
|
|
|
public function __construct(private string $url = self::BOT_API_URL) |
29
|
|
|
{ |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
/** |
33
|
|
|
* @param string $directory |
34
|
|
|
* @param string $namespace |
35
|
|
|
* @param string|null $scheme |
36
|
|
|
* @return bool |
37
|
|
|
* @throws ClientExceptionInterface |
38
|
|
|
*/ |
39
|
|
|
public function toStubs(string $directory = '', string $namespace = '', string $scheme = null): bool |
40
|
|
|
{ |
41
|
|
|
try { |
42
|
|
|
$directory = self::getTargetDirectory($directory); |
43
|
|
|
} catch (Exception $e) { |
44
|
|
|
echo 'Unable to use target directory:' . $e->getMessage() . PHP_EOL; |
45
|
|
|
return false; |
46
|
|
|
} |
47
|
|
|
mkdir($directory . '/Types', 0755); |
48
|
|
|
try { |
49
|
|
|
if (!empty($scheme)) { |
50
|
|
|
try { |
51
|
|
|
$data = json_decode($scheme, true, flags: JSON_THROW_ON_ERROR); |
52
|
|
|
} /** @noinspection PhpRedundantCatchClauseInspection */ catch (JsonException) { |
53
|
|
|
$data = null; |
54
|
|
|
} |
55
|
|
|
} |
56
|
|
|
$data = $data ?? $this->extractScheme(); |
57
|
|
|
$creator = new StubCreator($data, $namespace); |
58
|
|
|
$code = $creator->generateCode(); |
59
|
|
|
foreach ($code['types'] as $className => $type) { |
60
|
|
|
$filename = sprintf('%s/Types/%s.php', $directory, $className); |
61
|
|
|
file_put_contents($filename, $type); |
62
|
|
|
} |
63
|
|
|
file_put_contents($directory . '/API.php', $code['api']); |
64
|
|
|
} catch (Exception $e) { |
65
|
|
|
echo $e->getMessage() . PHP_EOL; |
66
|
|
|
return false; |
67
|
|
|
} |
68
|
|
|
return true; |
69
|
|
|
} |
70
|
|
|
|
71
|
|
|
/** |
72
|
|
|
* @param string $path |
73
|
|
|
* @return string |
74
|
|
|
* @throws Exception |
75
|
|
|
*/ |
76
|
|
|
private static function getTargetDirectory(string $path): string |
77
|
|
|
{ |
78
|
|
|
$path = realpath($path); |
79
|
|
|
if (false === $path) { |
80
|
|
|
if (!mkdir($path)) { |
81
|
|
|
$path = __DIR__ . '/../generated'; |
82
|
|
|
if (!file_exists($path)) { |
83
|
|
|
mkdir($path, 0755); |
84
|
|
|
} |
85
|
|
|
} |
86
|
|
|
} |
87
|
|
|
if (realpath($path) === false) { |
88
|
|
|
throw new Exception('Could not create target directory'); |
89
|
|
|
} |
90
|
|
|
return $path; |
91
|
|
|
} |
92
|
|
|
|
93
|
|
|
/** |
94
|
|
|
* @return array |
95
|
|
|
* @throws ChildNotFoundException |
96
|
|
|
* @throws CircularException |
97
|
|
|
* @throws ContentLengthException |
98
|
|
|
* @throws LogicalException |
99
|
|
|
* @throws NotLoadedException |
100
|
|
|
* @throws ParentNotFoundException |
101
|
|
|
* @throws StrictException |
102
|
|
|
* @throws ClientExceptionInterface |
103
|
|
|
*/ |
104
|
|
|
private function extractScheme(): array |
105
|
|
|
{ |
106
|
|
|
$dom = new Dom; |
107
|
|
|
$dom->loadFromURL($this->url); |
108
|
|
|
$elements = $dom->find('h4'); |
109
|
|
|
$data = []; |
110
|
|
|
/* @var Dom\Node\AbstractNode $element */ |
111
|
|
|
foreach ($elements as $element) { |
112
|
|
|
if (!str_contains($name = $element->text, ' ')) { |
113
|
|
|
$isMethod = self::isMethod($name); |
114
|
|
|
$path = $isMethod ? 'methods' : 'types'; |
115
|
|
|
$temp = $element; |
116
|
|
|
$description = ''; |
117
|
|
|
$table = null; |
118
|
|
|
while (true) { |
119
|
|
|
try { |
120
|
|
|
$element = $element->nextSibling(); |
121
|
|
|
} catch (ChildNotFoundException) { |
122
|
|
|
break; |
123
|
|
|
} |
124
|
|
|
$tag = $element->tag->name() ?? null; |
125
|
|
|
if (empty($temp->text()) or empty($tag) or $tag == 'text') { |
126
|
|
|
continue; |
127
|
|
|
} elseif (str_starts_with($tag, 'h')) { |
128
|
|
|
break; |
129
|
|
|
} elseif ($tag == 'p') { |
130
|
|
|
$description .= PHP_EOL . $element->innerHtml(); |
131
|
|
|
} elseif ($tag == 'table') { |
132
|
|
|
$table = $element->find('tbody')->find('tr'); |
133
|
|
|
break; |
134
|
|
|
} |
135
|
|
|
} |
136
|
|
|
/* @var Dom\Node\AbstractNode $element */ |
137
|
|
|
$data[$path][] = self::generateElement( |
138
|
|
|
$name, |
139
|
|
|
trim($description), |
140
|
|
|
$table, |
141
|
|
|
$isMethod |
142
|
|
|
); |
143
|
|
|
} |
144
|
|
|
} |
145
|
|
|
return $data; |
146
|
|
|
} |
147
|
|
|
|
148
|
|
|
private static function isMethod(string $name): bool |
149
|
|
|
{ |
150
|
|
|
return lcfirst($name) == $name; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* @param string $name |
155
|
|
|
* @param string $description |
156
|
|
|
* @param Dom\Node\Collection|null $unparsedFields |
157
|
|
|
* @param bool $isMethod |
158
|
|
|
* @return array |
159
|
|
|
* @throws ChildNotFoundException |
160
|
|
|
* @throws CircularException |
161
|
|
|
* @throws ContentLengthException |
162
|
|
|
* @throws LogicalException |
163
|
|
|
* @throws NotLoadedException |
164
|
|
|
* @throws StrictException |
165
|
|
|
*/ |
166
|
|
|
private static function generateElement( |
167
|
|
|
string $name, |
168
|
|
|
string $description, |
169
|
|
|
?Dom\Node\Collection $unparsedFields, |
170
|
|
|
bool $isMethod |
171
|
|
|
): array { |
172
|
|
|
$fields = self::parseFields($unparsedFields, $isMethod); |
173
|
|
|
if (!$isMethod) { |
174
|
|
|
return [ |
175
|
|
|
'name' => $name, |
176
|
|
|
'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES), |
177
|
|
|
'fields' => $fields |
178
|
|
|
]; |
179
|
|
|
} |
180
|
|
|
$returnTypes = self::parseReturnTypes($description); |
181
|
|
|
if (empty($returnTypes) and in_array($name, self::BOOL_RETURNS)) { |
182
|
|
|
$returnTypes[] = 'bool'; |
183
|
|
|
} |
184
|
|
|
return [ |
185
|
|
|
'name' => $name, |
186
|
|
|
'description' => htmlspecialchars_decode(strip_tags($description), ENT_QUOTES), |
187
|
|
|
'fields' => $fields, |
188
|
|
|
'return_types' => $returnTypes |
189
|
|
|
]; |
190
|
|
|
} |
191
|
|
|
|
192
|
|
|
/** |
193
|
|
|
* @param Dom\Node\Collection|null $fields |
194
|
|
|
* @param bool $isMethod |
195
|
|
|
* @return array |
196
|
|
|
* @throws ChildNotFoundException |
197
|
|
|
* @throws NotLoadedException |
198
|
|
|
*/ |
199
|
|
|
private static function parseFields(?Dom\Node\Collection $fields, bool $isMethod): array |
200
|
|
|
{ |
201
|
|
|
$parsedFields = []; |
202
|
|
|
$fields = $fields ?? []; |
203
|
|
|
foreach ($fields as $field) { |
204
|
|
|
/* @var Dom $field */ |
205
|
|
|
$fieldData = $field->find('td'); |
206
|
|
|
$name = $fieldData[0]->text; |
207
|
|
|
if (empty($name)) { |
208
|
|
|
continue; |
209
|
|
|
} |
210
|
|
|
$parsedData = [ |
211
|
|
|
'name' => $name, |
212
|
|
|
'type' => strip_tags($fieldData[1]->innerHtml) |
213
|
|
|
]; |
214
|
|
|
$parsedData['types'] = self::parseFieldTypes($parsedData['type']); |
215
|
|
|
unset($parsedData['type']); |
216
|
|
|
if ($isMethod) { |
217
|
|
|
$parsedData['required'] = $fieldData[2]->text == 'Yes'; |
218
|
|
|
$parsedData['description'] = htmlspecialchars_decode( |
219
|
|
|
strip_tags($fieldData[3]->innerHtml ?? $fieldData[3]->text ?? ''), |
220
|
|
|
ENT_QUOTES |
221
|
|
|
); |
222
|
|
|
} else { |
223
|
|
|
$description = htmlspecialchars_decode(strip_tags($fieldData[2]->innerHtml), ENT_QUOTES); |
224
|
|
|
$parsedData['optional'] = str_starts_with($description, 'Optional.'); |
225
|
|
|
$parsedData['description'] = $description; |
226
|
|
|
} |
227
|
|
|
$parsedFields[] = $parsedData; |
228
|
|
|
} |
229
|
|
|
return $parsedFields; |
230
|
|
|
} |
231
|
|
|
|
232
|
|
|
/** |
233
|
|
|
* @param string $rawType |
234
|
|
|
* @return array |
235
|
|
|
*/ |
236
|
|
|
private static function parseFieldTypes(string $rawType): array |
237
|
|
|
{ |
238
|
|
|
$types = []; |
239
|
|
|
foreach (explode(' or ', $rawType) as $rawOrType) { |
240
|
|
|
if (stripos($rawOrType, 'array') === 0) { |
241
|
|
|
$types[] = str_replace(' and', ',', $rawOrType); |
242
|
|
|
continue; |
243
|
|
|
} |
244
|
|
|
foreach (explode(' and ', $rawOrType) as $unparsedType) { |
245
|
|
|
$types[] = $unparsedType; |
246
|
|
|
} |
247
|
|
|
} |
248
|
|
|
$parsedTypes = []; |
249
|
|
|
foreach ($types as $type) { |
250
|
|
|
$type = trim(str_replace(['number', 'of'], '', $type)); |
251
|
|
|
$multiplesCount = substr_count(strtolower($type), 'array'); |
252
|
|
|
$parsedType = trim( |
253
|
|
|
str_replace( |
254
|
|
|
['Array', 'Integer', 'String', 'Boolean', 'Float', 'True'], |
255
|
|
|
['', 'int', 'string', 'bool', 'float', 'bool'], |
256
|
|
|
$type |
257
|
|
|
) |
258
|
|
|
); |
259
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
260
|
|
|
$parsedType = sprintf('Array<%s>', $parsedType); |
261
|
|
|
} |
262
|
|
|
$parsedTypes[] = $parsedType; |
263
|
|
|
} |
264
|
|
|
return $parsedTypes; |
265
|
|
|
} |
266
|
|
|
|
267
|
|
|
/** |
268
|
|
|
* @param string $description |
269
|
|
|
* @return array |
270
|
|
|
* @throws ChildNotFoundException |
271
|
|
|
* @throws CircularException |
272
|
|
|
* @throws NotLoadedException |
273
|
|
|
* @throws StrictException |
274
|
|
|
* @throws ContentLengthException |
275
|
|
|
* @throws LogicalException |
276
|
|
|
* @noinspection PhpUndefinedFieldInspection |
277
|
|
|
*/ |
278
|
|
|
private static function parseReturnTypes(string $description): array |
279
|
|
|
{ |
280
|
|
|
$returnTypes = []; |
281
|
|
|
$phrases = explode('.', $description); |
282
|
|
|
$phrases = array_filter( |
283
|
|
|
$phrases, |
284
|
|
|
function ($phrase) { |
285
|
|
|
return (false !== stripos($phrase, 'returns') or false !== stripos($phrase, 'is returned')); |
286
|
|
|
} |
287
|
|
|
); |
288
|
|
|
foreach ($phrases as $phrase) { |
289
|
|
|
$dom = new Dom; |
290
|
|
|
$dom->loadStr($phrase); |
291
|
|
|
$a = $dom->find('a'); |
292
|
|
|
$em = $dom->find('em'); |
293
|
|
|
foreach ($a as $element) { |
294
|
|
|
if ($element->text == 'Messages') { |
295
|
|
|
$returnTypes[] = 'Array<Message>'; |
296
|
|
|
continue; |
297
|
|
|
} |
298
|
|
|
|
299
|
|
|
$multiplesCount = substr_count(strtolower($phrase), 'array'); |
300
|
|
|
$returnType = $element->text; |
301
|
|
|
for ($i = 0; $i < $multiplesCount; $i++) { |
302
|
|
|
$returnType = sprintf('Array<%s>', $returnType); |
303
|
|
|
} |
304
|
|
|
$returnTypes[] = $returnType; |
305
|
|
|
} |
306
|
|
|
foreach ($em as $element) { |
307
|
|
|
if (in_array($element->text, ['False', 'force', 'Array'])) { |
308
|
|
|
continue; |
309
|
|
|
} |
310
|
|
|
$type = str_replace(['True', 'Int', 'String'], ['bool', 'int', 'string'], $element->text); |
311
|
|
|
$returnTypes[] = $type; |
312
|
|
|
} |
313
|
|
|
} |
314
|
|
|
return $returnTypes; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
/** |
318
|
|
|
* @param int $options |
319
|
|
|
* @return string |
320
|
|
|
* @throws ChildNotFoundException |
321
|
|
|
* @throws CircularException |
322
|
|
|
* @throws ClientExceptionInterface |
323
|
|
|
* @throws ContentLengthException |
324
|
|
|
* @throws LogicalException |
325
|
|
|
* @throws NotLoadedException |
326
|
|
|
* @throws ParentNotFoundException |
327
|
|
|
* @throws StrictException |
328
|
|
|
*/ |
329
|
|
|
public function toJson(int $options = 0): string |
330
|
|
|
{ |
331
|
|
|
$scheme = $this->extractScheme(); |
332
|
|
|
return json_encode($scheme, $options); |
333
|
|
|
} |
334
|
|
|
|
335
|
|
|
/** |
336
|
|
|
* @throws ChildNotFoundException |
337
|
|
|
* @throws CircularException |
338
|
|
|
* @throws ParentNotFoundException |
339
|
|
|
* @throws StrictException |
340
|
|
|
* @throws ClientExceptionInterface |
341
|
|
|
* @throws NotLoadedException |
342
|
|
|
* @throws ContentLengthException |
343
|
|
|
* @throws LogicalException |
344
|
|
|
*/ |
345
|
|
|
public function toYaml(int $inline = 6, int $indent = 4, int $flags = 0): string |
346
|
|
|
{ |
347
|
|
|
$scheme = $this->extractScheme(); |
348
|
|
|
return Yaml::dump($scheme, $inline, $indent, $flags); |
349
|
|
|
} |
350
|
|
|
|
351
|
|
|
} |
352
|
|
|
|