@@ -35,15 +35,15 @@ |
||
| 35 | 35 | |
| 36 | 36 | public function trackUser(string $user): void |
| 37 | 37 | { |
| 38 | - echo "**** TRACK " . $user . "*****\n"; |
|
| 38 | + echo "**** TRACK ".$user."*****\n"; |
|
| 39 | 39 | |
| 40 | 40 | $titles = $this->getLastEditsbyUser($user); |
| 41 | 41 | |
| 42 | 42 | // filter titles already in edited.txt |
| 43 | 43 | $edited = file(__DIR__.'/resources/article_externRef_edited.txt', FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES); |
| 44 | 44 | $filtered = array_diff($titles, $edited); |
| 45 | - $list = new PageList( $filtered ); |
|
| 46 | - echo ">" . $list->count() . " dans liste\n"; |
|
| 45 | + $list = new PageList($filtered); |
|
| 46 | + echo ">".$list->count()." dans liste\n"; |
|
| 47 | 47 | |
| 48 | 48 | $this->consumeList($list); |
| 49 | 49 | } |
@@ -60,13 +60,13 @@ discard block |
||
| 60 | 60 | // idn_to_ascii('teßt.com',IDNA_NONTRANSITIONAL_TO_ASCII,INTL_IDNA_VARIANT_UTS46) |
| 61 | 61 | // checkdnsrr($string, "A") // check DNS record |
| 62 | 62 | if (!self::isHttpURL($url)) { |
| 63 | - throw new DomainException('URL not compatible : ' . $url); |
|
| 63 | + throw new DomainException('URL not compatible : '.$url); |
|
| 64 | 64 | } |
| 65 | 65 | $response = $this->client->get($url); |
| 66 | 66 | |
| 67 | 67 | if (200 !== $response->getStatusCode()) { |
| 68 | - echo 'HTTP error ' . $response->getStatusCode(); |
|
| 69 | - $this->log->error('HTTP error ' . $response->getStatusCode() . ' ' . $response->getReasonPhrase()); |
|
| 68 | + echo 'HTTP error '.$response->getStatusCode(); |
|
| 69 | + $this->log->error('HTTP error '.$response->getStatusCode().' '.$response->getReasonPhrase()); |
|
| 70 | 70 | |
| 71 | 71 | return null; |
| 72 | 72 | } |
@@ -82,7 +82,7 @@ discard block |
||
| 82 | 82 | */ |
| 83 | 83 | public static function isHttpURL(string $url): bool |
| 84 | 84 | { |
| 85 | - return (bool)preg_match('#^https?://[^ ]+$#i', $url); |
|
| 85 | + return (bool) preg_match('#^https?://[^ ]+$#i', $url); |
|
| 86 | 86 | } |
| 87 | 87 | |
| 88 | 88 | /** |
@@ -104,7 +104,7 @@ discard block |
||
| 104 | 104 | $charset = $this->extractCharset($html) ?? 'WINDOWS-1252'; |
| 105 | 105 | |
| 106 | 106 | if (empty($charset)) { |
| 107 | - throw new DomainException('normalized html error and no charset found : ' . $url); |
|
| 107 | + throw new DomainException('normalized html error and no charset found : '.$url); |
|
| 108 | 108 | } |
| 109 | 109 | try { |
| 110 | 110 | $html2 = iconv($charset, 'UTF-8//TRANSLIT', $html); |
@@ -113,7 +113,7 @@ discard block |
||
| 113 | 113 | return ''; |
| 114 | 114 | } |
| 115 | 115 | } catch (Throwable $e) { |
| 116 | - throw new DomainException("error converting : $charset to UTF-8 on " . $url, $e->getCode(), $e); |
|
| 116 | + throw new DomainException("error converting : $charset to UTF-8 on ".$url, $e->getCode(), $e); |
|
| 117 | 117 | } |
| 118 | 118 | |
| 119 | 119 | return $html2; |
@@ -37,7 +37,7 @@ |
||
| 37 | 37 | // Attention : pas de post-processing (sanitize title, etc.) |
| 38 | 38 | $result = $trans->process($url, $summary); |
| 39 | 39 | } catch (Exception $e) { |
| 40 | - $result = "EXCEPTION ". $e->getMessage().$e->getFile().$e->getLine(); |
|
| 40 | + $result = "EXCEPTION ".$e->getMessage().$e->getFile().$e->getLine(); |
|
| 41 | 41 | } |
| 42 | 42 | |
| 43 | 43 | echo $result."\n"; |
@@ -69,7 +69,7 @@ discard block |
||
| 69 | 69 | ) |
| 70 | 70 | { |
| 71 | 71 | if (!ExternHttpClient::isHttpURL($url)) { |
| 72 | - throw new Exception('string is not an URL ' . $url); |
|
| 72 | + throw new Exception('string is not an URL '.$url); |
|
| 73 | 73 | } |
| 74 | 74 | $this->url = $url; |
| 75 | 75 | $this->html = $html; |
@@ -189,7 +189,7 @@ discard block |
||
| 189 | 189 | { |
| 190 | 190 | try { |
| 191 | 191 | if (!ExternHttpClient::isHttpURL($this->url)) { |
| 192 | - throw new Exception('string is not an URL ' . $this->url); |
|
| 192 | + throw new Exception('string is not an URL '.$this->url); |
|
| 193 | 193 | } |
| 194 | 194 | if (!$this->domainParser instanceof InternetDomainParserInterface) { |
| 195 | 195 | $this->log->notice('InternetDomainParser is not set'); |
@@ -200,7 +200,7 @@ discard block |
||
| 200 | 200 | return $this->domainParser->getRegistrableDomainFromURL($this->url); |
| 201 | 201 | } catch (Exception $e) { |
| 202 | 202 | if ($this->log !== null) { |
| 203 | - $this->log->warning('InternetDomainParser->getRegistrableDomainFromURL NULL ' . $this->url); |
|
| 203 | + $this->log->warning('InternetDomainParser->getRegistrableDomainFromURL NULL '.$this->url); |
|
| 204 | 204 | } |
| 205 | 205 | throw new Exception('InternetDomainParser->getRegistrableDomainFromURL NULL', $e->getCode(), $e); |
| 206 | 206 | } |
@@ -23,7 +23,7 @@ |
||
| 23 | 23 | /** |
| 24 | 24 | * @throws Exception |
| 25 | 25 | */ |
| 26 | - public static function fromURL($url, ExternHttpClientInterface $httpClient,LoggerInterface $logger = null): ExternPage |
|
| 26 | + public static function fromURL($url, ExternHttpClientInterface $httpClient, LoggerInterface $logger = null): ExternPage |
|
| 27 | 27 | { |
| 28 | 28 | if (!ExternHttpClient::isHttpURL($url)) { |
| 29 | 29 | throw new Exception('string is not an URL '.$url); |
@@ -11,5 +11,5 @@ |
||
| 11 | 11 | |
| 12 | 12 | interface ExternHttpClientInterface |
| 13 | 13 | { |
| 14 | - public function getHTML(string $url, ?bool $normalized=false): ?string; |
|
| 14 | + public function getHTML(string $url, ?bool $normalized = false): ?string; |
|
| 15 | 15 | } |
@@ -25,7 +25,7 @@ discard block |
||
| 25 | 25 | public const SLEEP_AFTER_EDITION = 15; // sec |
| 26 | 26 | public const MINUTES_DELAY_AFTER_LAST_HUMAN_EDIT = 10; // minutes |
| 27 | 27 | public const CHECK_EDIT_CONFLICT = true; |
| 28 | - public const ARTICLE_ANALYZED_FILENAME = __DIR__ . '/resources/article_externRef_edited.txt'; |
|
| 28 | + public const ARTICLE_ANALYZED_FILENAME = __DIR__.'/resources/article_externRef_edited.txt'; |
|
| 29 | 29 | public const SKIP_ADQ = false; |
| 30 | 30 | public const SKIP_LASTEDIT_BY_BOT = false; |
| 31 | 31 | public const CITATION_NUMBER_ON_FIRE = 15; |
@@ -72,7 +72,7 @@ discard block |
||
| 72 | 72 | $result = $this->transformer->process($refContent, $this->summary); |
| 73 | 73 | } catch (Throwable $e) { |
| 74 | 74 | echo "** Problème détecté 234242\n"; |
| 75 | - $this->log->critical($e->getMessage() . " " . $e->getFile() . ":" . $e->getLine()); |
|
| 75 | + $this->log->critical($e->getMessage()." ".$e->getFile().":".$e->getLine()); |
|
| 76 | 76 | |
| 77 | 77 | // TODO : parse $e->message pour traitement, taskName, botflag... |
| 78 | 78 | return $refContent; |
@@ -84,8 +84,8 @@ discard block |
||
| 84 | 84 | |
| 85 | 85 | // Gestion semi-auto : todo CONDITION POURRI FAUSSE $this->transformer->skipUnauthorised |
| 86 | 86 | |
| 87 | - echo Color::BG_LIGHT_RED . "--" . Color::NORMAL . " " . $refContent . "\n"; |
|
| 88 | - echo Color::BG_LIGHT_GREEN . "++" . Color::NORMAL . " $result \n\n"; |
|
| 87 | + echo Color::BG_LIGHT_RED."--".Color::NORMAL." ".$refContent."\n"; |
|
| 88 | + echo Color::BG_LIGHT_GREEN."++".Color::NORMAL." $result \n\n"; |
|
| 89 | 89 | |
| 90 | 90 | if (!$this->autoOrYesConfirmation('Conserver cette modif ?')) { |
| 91 | 91 | return $refContent; |
@@ -116,10 +116,10 @@ discard block |
||
| 116 | 116 | $prefixSummary = ($this->summary->isBotFlag()) ? 'bot ' : ''; |
| 117 | 117 | $suffix = ''; |
| 118 | 118 | if (isset($this->summary->memo['count article'])) { |
| 119 | - $suffix .= ' ' . $this->summary->memo['count article'] . 'x {article}'; |
|
| 119 | + $suffix .= ' '.$this->summary->memo['count article'].'x {article}'; |
|
| 120 | 120 | } |
| 121 | 121 | if (isset($this->summary->memo['count lien web'])) { |
| 122 | - $suffix .= ' ' . $this->summary->memo['count lien web'] . 'x {lien web}'; |
|
| 122 | + $suffix .= ' '.$this->summary->memo['count lien web'].'x {lien web}'; |
|
| 123 | 123 | } |
| 124 | 124 | if (isset($this->summary->memo['presse'])) { |
| 125 | 125 | $suffix .= ' |
@@ -33,13 +33,13 @@ discard block |
||
| 33 | 33 | class ExternRefTransformer implements ExternRefTransformerInterface |
| 34 | 34 | { |
| 35 | 35 | public const HTTP_REQUEST_LOOP_DELAY = 10; |
| 36 | - public const LOG_REQUEST_ERROR = __DIR__ . '/../../Application/resources/external_request_error.log'; // todo move |
|
| 37 | - public const SKIP_DOMAIN_FILENAME = __DIR__ . '/../resources/config_skip_domain.txt'; |
|
| 36 | + public const LOG_REQUEST_ERROR = __DIR__.'/../../Application/resources/external_request_error.log'; // todo move |
|
| 37 | + public const SKIP_DOMAIN_FILENAME = __DIR__.'/../resources/config_skip_domain.txt'; |
|
| 38 | 38 | public const REPLACE_404 = true; |
| 39 | - public const CONFIG_PRESSE = __DIR__ . '/../resources/config_presse.yaml'; |
|
| 40 | - public const CONFIG_NEWSPAPER_JSON = __DIR__ . '/../resources/data_newspapers.json'; |
|
| 41 | - public const CONFIG_SCIENTIFIC_JSON = __DIR__ . '/../resources/data_scientific_domain.json'; |
|
| 42 | - public const CONFIG_SCIENTIFIC_WIKI_JSON = __DIR__ . '/../resources/data_scientific_wiki.json'; |
|
| 39 | + public const CONFIG_PRESSE = __DIR__.'/../resources/config_presse.yaml'; |
|
| 40 | + public const CONFIG_NEWSPAPER_JSON = __DIR__.'/../resources/data_newspapers.json'; |
|
| 41 | + public const CONFIG_SCIENTIFIC_JSON = __DIR__.'/../resources/data_scientific_domain.json'; |
|
| 42 | + public const CONFIG_SCIENTIFIC_WIKI_JSON = __DIR__.'/../resources/data_scientific_wiki.json'; |
|
| 43 | 43 | public const ROBOT_NOINDEX_WHITELIST = ['legifrance.gouv.fr']; |
| 44 | 44 | |
| 45 | 45 | public $skipSiteBlacklisted = true; |
@@ -147,7 +147,7 @@ discard block |
||
| 147 | 147 | { |
| 148 | 148 | $this->url = $url; |
| 149 | 149 | if (!ExternHttpClient::isHttpURL($url)) { |
| 150 | - $this->log->debug('Skip : not a valid URL : ' . $url); |
|
| 150 | + $this->log->debug('Skip : not a valid URL : '.$url); |
|
| 151 | 151 | return false; |
| 152 | 152 | } |
| 153 | 153 | |
@@ -155,12 +155,12 @@ discard block |
||
| 155 | 155 | return false; |
| 156 | 156 | } |
| 157 | 157 | if (!ExternHttpClient::isHttpURL($url)) { |
| 158 | - throw new Exception('string is not an URL ' . $url); |
|
| 158 | + throw new Exception('string is not an URL '.$url); |
|
| 159 | 159 | } |
| 160 | 160 | try { |
| 161 | 161 | $this->domain = (new InternetDomainParser())->getRegistrableDomainFromURL($url); |
| 162 | 162 | } catch (Exception $e) { |
| 163 | - $this->log->warning('Skip : not a valid URL : ' . $url); |
|
| 163 | + $this->log->warning('Skip : not a valid URL : '.$url); |
|
| 164 | 164 | return false; |
| 165 | 165 | } |
| 166 | 166 | |
@@ -195,7 +195,7 @@ discard block |
||
| 195 | 195 | $this->summary->memo['sites'][] = $this->externalPage->getPrettyDomainName(); |
| 196 | 196 | } |
| 197 | 197 | if (isset($mapData['accès url'])) { |
| 198 | - $this->log->notice('accès |
|