@@ -21,7 +21,7 @@ discard block |
||
21 | 21 | */ |
22 | 22 | class InternetDomainParser implements InternetDomainParserInterface |
23 | 23 | { |
24 | - private const PATH_CACHE_PUBLIC_SUFFIX_LIST = __DIR__ . '/resources/public_suffix_list.dat'; |
|
24 | + private const PATH_CACHE_PUBLIC_SUFFIX_LIST = __DIR__.'/resources/public_suffix_list.dat'; |
|
25 | 25 | |
26 | 26 | private readonly Rules $rules; |
27 | 27 | |
@@ -55,7 +55,7 @@ discard block |
||
55 | 55 | public static function extractSubdomainString(string $httpURL): string |
56 | 56 | { |
57 | 57 | if (!HttpUtil::isHttpURL($httpURL)) { |
58 | - throw new Exception('string is not an URL ' . $httpURL); |
|
58 | + throw new Exception('string is not an URL '.$httpURL); |
|
59 | 59 | } |
60 | 60 | |
61 | 61 | return parse_url($httpURL, PHP_URL_HOST); |
@@ -54,7 +54,7 @@ discard block |
||
54 | 54 | ) |
55 | 55 | { |
56 | 56 | if (!HttpUtil::isHttpURL($url)) { |
57 | - throw new Exception('string is not an URL ' . $url); |
|
57 | + throw new Exception('string is not an URL '.$url); |
|
58 | 58 | } |
59 | 59 | $this->url = $url; |
60 | 60 | } |
@@ -170,7 +170,7 @@ discard block |
||
170 | 170 | { |
171 | 171 | try { |
172 | 172 | if (!HttpUtil::isHttpURL($this->url)) { |
173 | - throw new Exception('string is not an URL ' . $this->url); |
|
173 | + throw new Exception('string is not an URL '.$this->url); |
|
174 | 174 | } |
175 | 175 | if (!$this->domainParser instanceof InternetDomainParserInterface) { |
176 | 176 | $this->log->notice('InternetDomainParser is not set'); |
@@ -181,7 +181,7 @@ discard block |
||
181 | 181 | return $this->domainParser->getRegistrableDomainFromURL($this->url); |
182 | 182 | } catch (Exception $e) { |
183 | 183 | if ($this->log !== null) { |
184 | - $this->log->warning('InternetDomainParser->getRegistrableDomainFromURL NULL ' . $this->url); |
|
184 | + $this->log->warning('InternetDomainParser->getRegistrableDomainFromURL NULL '.$this->url); |
|
185 | 185 | } |
186 | 186 | throw new Exception('InternetDomainParser->getRegistrableDomainFromURL NULL', $e->getCode(), $e); |
187 | 187 | } |
@@ -49,7 +49,7 @@ discard block |
||
49 | 49 | $this->url = $url; |
50 | 50 | $this->registrableDomain = null; |
51 | 51 | if (!HttpUtil::isHttpURL($url)) { |
52 | - $this->log->debug('Skip : not a valid URL : ' . $url); |
|
52 | + $this->log->debug('Skip : not a valid URL : '.$url); |
|
53 | 53 | return false; |
54 | 54 | } |
55 | 55 | |
@@ -57,7 +57,7 @@ discard block |
||
57 | 57 | return false; |
58 | 58 | } |
59 | 59 | if (!HttpUtil::isHttpURL($url)) { |
60 | - throw new Exception('string is not an URL ' . $url); |
|
60 | + throw new Exception('string is not an URL '.$url); |
|
61 | 61 | } |
62 | 62 | |
63 | 63 | $this->findRegistrableDomain(); |
@@ -79,7 +79,7 @@ discard block |
||
79 | 79 | try { |
80 | 80 | $this->registrableDomain = $this->internetDomainParser->getRegistrableDomainFromURL($this->url); |
81 | 81 | } catch (Exception) { |
82 | - $this->log->warning('Skip : not a valid URL : ' . $this->url); |
|
82 | + $this->log->warning('Skip : not a valid URL : '.$this->url); |
|
83 | 83 | return null; |
84 | 84 | } |
85 | 85 | return $this->registrableDomain; |
@@ -92,7 +92,7 @@ discard block |
||
92 | 92 | */ |
93 | 93 | protected function hasForbiddenFilenameExtension(): bool |
94 | 94 | { |
95 | - return (bool)preg_match( |
|
95 | + return (bool) preg_match( |
|
96 | 96 | '#\.(pdf|jpg|jpeg|gif|png|webp|xls|xlsx|xlr|xml|xlt|txt|csv|js|docx|exe|gz|zip|ini|movie|mp3|mp4|ogg|raw|rss|tar|tgz|wma)$#i', |
97 | 97 | $this->url |
98 | 98 | ); |
@@ -30,12 +30,12 @@ discard block |
||
30 | 30 | public function fromURL(string $url, InternetDomainParserInterface $domainParser): ExternPage |
31 | 31 | { |
32 | 32 | if (!HttpUtil::isHttpURL($url)) { |
33 | - throw new Exception('string is not an URL ' . $url); |
|
33 | + throw new Exception('string is not an URL '.$url); |
|
34 | 34 | } |
35 | 35 | |
36 | 36 | $html = $this->getHTML($url, true); |
37 | 37 | if (empty($html)) { |
38 | - throw new DomainException('No HTML from requested URL ' . $url); |
|
38 | + throw new DomainException('No HTML from requested URL '.$url); |
|
39 | 39 | } |
40 | 40 | |
41 | 41 | return new ExternPage($url, $html, new TagParser(), $domainParser, $this->log); |
@@ -54,7 +54,7 @@ discard block |
||
54 | 54 | // idn_to_ascii('teßt.com',IDNA_NONTRANSITIONAL_TO_ASCII,INTL_IDNA_VARIANT_UTS46) |
55 | 55 | // checkdnsrr($string, "A") // check DNS record |
56 | 56 | if (!HttpUtil::isHttpURL($url)) { |
57 | - throw new DomainException('URL not compatible : ' . $url); |
|
57 | + throw new DomainException('URL not compatible : '.$url); |
|
58 | 58 | } |
59 | 59 | $response = $this->client->get($url, [ |
60 | 60 | 'timeout' => 20, |
@@ -65,7 +65,7 @@ discard block |
||
65 | 65 | ]); |
66 | 66 | |
67 | 67 | if (200 !== $response->getStatusCode()) { |
68 | - $this->log->error('[z49] HTTP error ' . $response->getStatusCode() . ' ' . $response->getReasonPhrase()); |
|
68 | + $this->log->error('[z49] HTTP error '.$response->getStatusCode().' '.$response->getReasonPhrase()); |
|
69 | 69 | |
70 | 70 | return null; |
71 | 71 | } |
@@ -86,7 +86,7 @@ discard block |
||
86 | 86 | return null; |
87 | 87 | } |
88 | 88 | |
89 | - public function get(string|UriInterface $uri, array $options = []): ResponseInterface |
|
89 | + public function get(string | UriInterface $uri, array $options = []): ResponseInterface |
|
90 | 90 | { |
91 | 91 | if (isset($options['allow_redirects']) && $options['allow_redirects'] !== false) { |
92 | 92 | $this->maxRedirects = self::DEFAULT_MAX_REDIRECTS; |
@@ -98,7 +98,7 @@ discard block |
||
98 | 98 | /** |
99 | 99 | * todo : add redirect http referer |
100 | 100 | */ |
101 | - private function getRecursive(UriInterface|string $uri, array $options, int $loop = 0): ResponseInterface |
|
101 | + private function getRecursive(UriInterface | string $uri, array $options, int $loop = 0): ResponseInterface |
|
102 | 102 | { |
103 | 103 | $response = $this->client->get($uri, $options); |
104 | 104 | |
@@ -106,7 +106,7 @@ discard block |
||
106 | 106 | if ($response->getStatusCode() >= 300 && $response->getStatusCode() < 400) { |
107 | 107 | $redirectUri = $response->getHeader('location')[0] ?? null; |
108 | 108 | if ($loop >= $this->maxRedirects || !$redirectUri) { |
109 | - throw new Exception('TorClientAdapter::get Error too many redirects ' . $response->getStatusCode()); |
|
109 | + throw new Exception('TorClientAdapter::get Error too many redirects '.$response->getStatusCode()); |
|
110 | 110 | } |
111 | 111 | $loop++; |
112 | 112 | return $this->getRecursive($redirectUri, $options, $loop); |
@@ -114,7 +114,7 @@ discard block |
||
114 | 114 | |
115 | 115 | // Error 4xx 5xx |
116 | 116 | if ($response->getStatusCode() >= 400) { |
117 | - throw new Exception($response->getStatusCode() . ' ' . $response->getReasonPhrase()); |
|
117 | + throw new Exception($response->getStatusCode().' '.$response->getReasonPhrase()); |
|
118 | 118 | } |
119 | 119 | |
120 | 120 | return $response; |
@@ -40,7 +40,7 @@ |
||
40 | 40 | ); |
41 | 41 | } |
42 | 42 | |
43 | - public function get(string|UriInterface $uri, array $options = []): ResponseInterface |
|
43 | + public function get(string | UriInterface $uri, array $options = []): ResponseInterface |
|
44 | 44 | { |
45 | 45 | return $this->client->get($uri, $options); |
46 | 46 | } |
@@ -34,7 +34,7 @@ discard block |
||
34 | 34 | $summary = new Summary('test'); |
35 | 35 | |
36 | 36 | $torEnabled = true; |
37 | -echo "TOR enabled : ".($torEnabled ? "oui" : "non"). "\n"; |
|
37 | +echo "TOR enabled : ".($torEnabled ? "oui" : "non")."\n"; |
|
38 | 38 | |
39 | 39 | $trans = new ExternRefTransformer( |
40 | 40 | new ExternMapper($logger), |
@@ -49,10 +49,10 @@ discard block |
||
49 | 49 | // Attention : pas de post-processing (sanitize title, etc.) |
50 | 50 | $result = $trans->process($url, $summary); |
51 | 51 | } catch (Exception $e) { |
52 | - $result = "EXCEPTION ". $e->getMessage().$e->getFile().$e->getLine(); |
|
52 | + $result = "EXCEPTION ".$e->getMessage().$e->getFile().$e->getLine(); |
|
53 | 53 | } |
54 | 54 | |
55 | -echo '>>> '. $result."\n"; |
|
55 | +echo '>>> '.$result."\n"; |
|
56 | 56 | |
57 | 57 | |
58 | 58 |
@@ -14,6 +14,6 @@ |
||
14 | 14 | |
15 | 15 | interface HttpClientInterface |
16 | 16 | { |
17 | - public function get(string|UriInterface $uri, array $options = []): ResponseInterface; |
|
17 | + public function get(string | UriInterface $uri, array $options = []): ResponseInterface; |
|
18 | 18 | public function request($method, $uri, array $options = []): ResponseInterface; |
19 | 19 | } |
20 | 20 | \ No newline at end of file |
@@ -48,16 +48,16 @@ discard block |
||
48 | 48 | return new WebarchiveDTO( |
49 | 49 | self::ARCHIVER_NAME, |
50 | 50 | $url, |
51 | - (string)$archiveData['longformurl'], |
|
51 | + (string) $archiveData['longformurl'], |
|
52 | 52 | $archiveData['timestamp'] |
53 | - ? DateTimeImmutable::createFromFormat('U', (string)$archiveData['timestamp']) |
|
53 | + ? DateTimeImmutable::createFromFormat('U', (string) $archiveData['timestamp']) |
|
54 | 54 | : null |
55 | 55 | ); |
56 | 56 | } |
57 | 57 | |
58 | 58 | protected function requestWikiwixApi(string $url): array |
59 | 59 | { |
60 | - $response = $this->externHttpClient->get(self::API_URL . urlencode($url), [ |
|
60 | + $response = $this->externHttpClient->get(self::API_URL.urlencode($url), [ |
|
61 | 61 | 'timeout' => 20, |
62 | 62 | 'allow_redirects' => true, |
63 | 63 | 'headers' => ['User-Agent' => getenv('USER_AGENT')], |
@@ -75,8 +75,8 @@ discard block |
||
75 | 75 | $data = json_decode($jsonString, true, 512, JSON_THROW_ON_ERROR) ?? []; |
76 | 76 | |
77 | 77 | // check wikiwix archive status |
78 | - if (empty($data['status']) || (int)$data['status'] !== 200) { |
|
79 | - $this->log->debug('WikiwixAdapter incorrect response: ' . $jsonString); |
|
78 | + if (empty($data['status']) || (int) $data['status'] !== 200) { |
|
79 | + $this->log->debug('WikiwixAdapter incorrect response: '.$jsonString); |
|
80 | 80 | |
81 | 81 | return []; |
82 | 82 | } |