|
1
|
|
|
<?php declare(strict_types=1); |
|
2
|
|
|
|
|
3
|
|
|
namespace Bigwhoop\Trumpet\Commands; |
|
4
|
|
|
|
|
5
|
|
|
use Bigwhoop\SentenceBreaker\SentenceBreaker; |
|
6
|
|
|
|
|
7
|
|
|
final class WikiCommand implements Command |
|
8
|
|
|
{ |
|
9
|
|
|
const ENDPOINT = 'https://en.wikipedia.org/w/api.php'; |
|
10
|
|
|
|
|
11
|
|
|
/** @var CommandExecutionContext */ |
|
12
|
|
|
private $executionContext; |
|
13
|
|
|
|
|
14
|
|
|
/** @var SentenceBreaker */ |
|
15
|
|
|
private $sentenceBreaker; |
|
16
|
|
|
|
|
17
|
1 |
|
public function __construct(CommandExecutionContext $context, SentenceBreaker $sentenceBreaker) |
|
18
|
|
|
{ |
|
19
|
1 |
|
$this->executionContext = $context; |
|
20
|
1 |
|
$this->sentenceBreaker = $sentenceBreaker; |
|
21
|
1 |
|
} |
|
22
|
|
|
|
|
23
|
1 |
|
public function getToken(): string |
|
24
|
|
|
{ |
|
25
|
1 |
|
return 'wiki'; |
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
public function execute(CommandParams $params, CommandExecutionContext $executionContext): string |
|
29
|
|
|
{ |
|
30
|
|
|
$article = $params->getFirstArgument(); |
|
31
|
|
|
$numSentences = (int) $params->getSecondArgument(0); |
|
32
|
|
|
|
|
33
|
|
|
$cacheFile = $this->getCacheFilePath($article); |
|
34
|
|
|
if (is_readable($cacheFile)) { |
|
35
|
|
|
$summary = file_get_contents($cacheFile); |
|
36
|
|
|
|
|
37
|
|
|
return $this->quote($summary, $numSentences); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
$url = $this->buildURL($article); |
|
41
|
|
|
$response = file_get_contents($url); |
|
42
|
|
|
|
|
43
|
|
|
$data = json_decode($response); |
|
44
|
|
|
if (!$data) { |
|
45
|
|
|
throw new ExecutionFailedException('Wikipedia API response could not be decoded. Request URL: '.$url.'. Response: '.var_export($response, true)); |
|
46
|
|
|
} |
|
47
|
|
|
|
|
48
|
|
|
foreach ($data->query->pages as $page) { |
|
49
|
|
|
if (isset($page->missing)) { |
|
50
|
|
|
continue; |
|
51
|
|
|
} |
|
52
|
|
|
|
|
53
|
|
|
file_put_contents($cacheFile, $page->extract); |
|
54
|
|
|
|
|
55
|
|
|
return $this->quote($page->extract, $numSentences); |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
throw new ExecutionFailedException("Failed to query for Wikipedia article '$article'. Request URL: $url"); |
|
59
|
|
|
} |
|
60
|
|
|
|
|
61
|
|
|
private function getCacheFilePath(string $article): string |
|
62
|
|
|
{ |
|
63
|
|
|
$tmpDir = $this->executionContext->ensureTempDirectory(); |
|
64
|
|
|
|
|
65
|
|
|
return $tmpDir.'/summary-'.md5($article).'.txt'; |
|
66
|
|
|
} |
|
67
|
|
|
|
|
68
|
|
|
private function quote(string $text, int $numSentences): string |
|
69
|
|
|
{ |
|
70
|
|
|
if ($numSentences > 0) { |
|
71
|
|
|
$sentences = $this->sentenceBreaker->split($text); |
|
72
|
|
|
$text = join(' ', array_slice($sentences, 0, $numSentences)); |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
return "> $text"; |
|
76
|
|
|
} |
|
77
|
|
|
|
|
78
|
|
|
private function buildURL(string $article): string |
|
79
|
|
|
{ |
|
80
|
|
|
$params = [ |
|
81
|
|
|
'format' => 'json', |
|
82
|
|
|
'action' => 'query', |
|
83
|
|
|
'prop' => 'extracts', |
|
84
|
|
|
'exintro' => '', |
|
85
|
|
|
'explaintext' => '', |
|
86
|
|
|
'titles' => $article, |
|
87
|
|
|
]; |
|
88
|
|
|
|
|
89
|
|
|
return self::ENDPOINT.'?'.http_build_query($params, null, '&', PHP_QUERY_RFC3986); |
|
90
|
|
|
} |
|
91
|
|
|
} |
|
92
|
|
|
|
bigwhoop /
trumpet
