WebSVN – eFlore/Applications.eflore-consultation – Diff – /trunk/metier/robots/WikipediaBot.php


<?php
class WikipediaBot {
	const HTTP_URL_REQUETE_SEPARATEUR = '&';
	private $langue = 'fr';
	private $url = '';
	private $parametres = array();
	private $titre = '';
	private $txt = '';
	private $userAgent = 'eFloreBot v0.1';
	private $reponse_entetes = null;
 
	public function __construct($langue = null) {
		if (is_string($langue)) {
			$this->langue = strtolower($langue);
		}
	}
 
	public function chargerPage($article) {
		$this->initialiserRequete();
		$this->url = $this->getBaseApiURL();
		$this->parametres = array(
			'action' => 'query',
			'prop' => 'revisions',
			'titles' => $article,
			'rvprop' => 'content',
			'redirects' => 1
		);
		$this->resultats = $this->consulterAPI();
		$sxGetAID = $this->resultats['query']['pages'];
		$sxGetAID = array_shift($sxGetAID);
		$this->titre = $sxGetAID['title'];
		$this->txt = $sxGetAID['revisions'][0]['*'];
	}
 
	public function getPageTitre() {
		return $this->titre;
	}
 
	public function getPageTxt() {
		return $this->txt;
	}
 
	public function getTaxobox() {
		$taxobox = '';
		if (preg_match('/([{]{2}Taxobox début.+[{]{2}Taxobox fin[}]{2})/s', $this->txt, $match)) {
			$taxobox = $match[1];
		}
		return $taxobox;
	}
 
	public function extraireTaxobox() {
		$taxobox = $this->getTaxobox();
		$this->txt = str_replace($taxobox, '', $this->txt);
		return $taxobox;
	}
 
	public function getSectionParNumero($num) {
		$sections = preg_split('/[=]{2}[^=]+[=]{2}/U', $this->txt);
		Debug::printr($sections);
		$sectionTxt = isset($sections[$num]) ? $sections[$num] : '';
		return $sectionTxt;
	}
 
	public function getSectionParTitre($titre) {
		$section = '';
		if (preg_match('/[=]{2} '.$titre.' [=]{2}(.*)\n\n/sU', $this->txt, $match)) {
			$section = $match[1];
		}
		return $section;
	}
 
	public function rendre($wikitxt) {
		$wikitxt .= '<references />';
		$this->initialiserRequete();
		$this->url = $this->getBaseApiURL();
		$this->parametres = array(
					'action' => 'parse',
					'prop' => 'text',
					'text' => $wikitxt
		);
		$this->resultats = $this->consulterAPI();
		$txt = $this->resultats['parse']['text']['*'];
		$txt = $this->remplacerUrls($txt);
		return $txt;
	}
 
	private function initialiserRequete() {
		$this->url = '';
		$this->parametres = array();
		$this->resultats = array();
	}
 
	private function getBaseWpURL() {
		$baseURL = "http://{$this->langue}.wikipedia.org";
		return $baseURL;
	}
 
	private function getBaseApiURL() {
		$baseURL = $this->getBaseWpURL().'/w/api.php';
		return $baseURL;
	}
 
	private function consulterAPI() {
		$this->parametres['format'] = 'php';
		$resultat = $this->consulterEnPost();
		$resultat = unserialize($resultat);
		if (isset($resultat['error'])) {
			throw new Exception($resultat['error']['info'], $resultat['error']['info']);
		}
		return $resultat;
	}
 
	private function consulterEnPost() {
		return $this->consulter('POST');
	}
 
	private function consulter($mode) {
		$entetes = array(
				'Content-type' => 'application/x-www-form-urlencoded',
				'User-Agent' => $this->userAgent);
		$contexte = stream_context_create(array(
				'http' => array(
					'method' => $mode,
					'header' => $this->getEnteteChaine($entetes),
					'content' => http_build_query($this->parametres, null, self::HTTP_URL_REQUETE_SEPARATEUR))));
		$flux = fopen($this->url, 'r', false, $contexte);
		if (!$flux) {
			$this->reponse_entetes = $http_response_header;
			$e = "L'ouverture de l'url '{$this->url}' par la méthode HTTP '$mode' a échoué!";
			throw new Exception($e);
		}
		// Informations sur les en-têtes et métadonnées du flux
		$this->reponse_entetes = stream_get_meta_data($flux);
		// Contenu actuel de $url
		$contenu = stream_get_contents($flux);
		fclose($flux);
		return $contenu;
	}
 
	private function getEnteteChaine(Array $entetes) {
		$entetesCleVal = array();
		foreach ($entetes as $cle => $valeur) {
			$entetesCleVal[] = $cle.': '.$valeur;
		}
		return implode("\r\n", $entetesCleVal);
	}
 
	private function remplacerUrls($txt) {
		$remplacements = array(
			'href="/wiki/' => 'href="'.$this->getBaseWpURL().'/wiki/',
			'href="/w/' => 'href="'.$this->getBaseWpURL().'/w/');
		$txt = strtr($txt, $remplacements);
		return $txt;
	}
}
?>

Rev 175	Rev 176
1	`<?php`	1	`<?php`
2	`class Wikipedia {`	2	`class WikipediaBot {`
3	`const HTTP_URL_REQUETE_SEPARATEUR = '&';`	3	`const HTTP_URL_REQUETE_SEPARATEUR = '&';`
4	`private $langue = 'fr';`	4	`private $langue = 'fr';`
5	`private $url = '';`	5	`private $url = '';`
6	`private $parametres = array();`	6	`private $parametres = array();`
7	`private $titre = '';`	7	`private $titre = '';`
8	`private $txt = '';`	8	`private $txt = '';`
9	`private $userAgent = 'eFloreBot v0.1';`	9	`private $userAgent = 'eFloreBot v0.1';`
10	`private $reponse_entetes = null;`	10	`private $reponse_entetes = null;`
11		11
12	`public function __construct($langue = null) {`	12	`public function __construct($langue = null) {`
13	`if (is_string($langue)) {`	13	`if (is_string($langue)) {`
14	`$this->langue = strtolower($langue);`	14	`$this->langue = strtolower($langue);`
15	`}`	15	`}`
16	`}`	16	`}`
17		17
18	`public function chargerPage($article) {`	18	`public function chargerPage($article) {`
19	`$this->initialiserRequete();`	19	`$this->initialiserRequete();`
20	`$this->url = $this->getBaseApiURL();`	20	`$this->url = $this->getBaseApiURL();`
21	`$this->parametres = array(`	21	`$this->parametres = array(`
22	`'action' => 'query',`	22	`'action' => 'query',`
23	`'prop' => 'revisions',`	23	`'prop' => 'revisions',`
24	`'titles' => $article,`	24	`'titles' => $article,`
25	`'rvprop' => 'content',`	25	`'rvprop' => 'content',`
26	`'redirects' => 1`	26	`'redirects' => 1`
27	`);`	27	`);`
28	`$this->resultats = $this->consulterAPI();`	28	`$this->resultats = $this->consulterAPI();`
29	`$sxGetAID = $this->resultats['query']['pages'];`	29	`$sxGetAID = $this->resultats['query']['pages'];`
30	`$sxGetAID = array_shift($sxGetAID);`	30	`$sxGetAID = array_shift($sxGetAID);`
31	`$this->titre = $sxGetAID['title'];`	31	`$this->titre = $sxGetAID['title'];`
32	`$this->txt = $sxGetAID['revisions'][0]['*'];`	32	`$this->txt = $sxGetAID['revisions'][0]['*'];`
33	`}`	33	`}`
34		34
35	`public function getPageTitre() {`	35	`public function getPageTitre() {`
36	`return $this->titre;`	36	`return $this->titre;`
37	`}`	37	`}`
38		38
39	`public function getPageTxt() {`	39	`public function getPageTxt() {`
40	`return $this->txt;`	40	`return $this->txt;`
41	`}`	41	`}`
42		42
43	`public function getTaxobox() {`	43	`public function getTaxobox() {`
44	`$taxobox = '';`	44	`$taxobox = '';`
45	`if (preg_match('/([{]{2}Taxobox début.+[{]{2}Taxobox fin[}]{2})/s', $this->txt, $match)) {`	45	`if (preg_match('/([{]{2}Taxobox début.+[{]{2}Taxobox fin[}]{2})/s', $this->txt, $match)) {`
46	`$taxobox = $match[1];`	46	`$taxobox = $match[1];`
47	`}`	47	`}`
48	`return $taxobox;`	48	`return $taxobox;`
49	`}`	49	`}`
50		50
51	`public function extraireTaxobox() {`	51	`public function extraireTaxobox() {`
52	`$taxobox = $this->getTaxobox();`	52	`$taxobox = $this->getTaxobox();`
53	`$this->txt = str_replace($taxobox, '', $this->txt);`	53	`$this->txt = str_replace($taxobox, '', $this->txt);`
54	`return $taxobox;`	54	`return $taxobox;`
55	`}`	55	`}`
56		56
57	`public function getSectionParNumero($num) {`	57	`public function getSectionParNumero($num) {`
58	`$sections = preg_split('/[=]{2}[^=]+[=]{2}/U', $this->txt);`	58	`$sections = preg_split('/[=]{2}[^=]+[=]{2}/U', $this->txt);`
59	`Debug::printr($sections);`	59	`Debug::printr($sections);`
60	`$sectionTxt = isset($sections[$num]) ? $sections[$num] : '';`	60	`$sectionTxt = isset($sections[$num]) ? $sections[$num] : '';`
61	`return $sectionTxt;`	61	`return $sectionTxt;`
62	`}`	62	`}`
63		63
64	`public function getSectionParTitre($titre) {`	64	`public function getSectionParTitre($titre) {`
65	`$section = '';`	65	`$section = '';`
66	`if (preg_match('/[=]{2} '.$titre.' [=]{2}(.*)\n\n/sU', $this->txt, $match)) {`	66	`if (preg_match('/[=]{2} '.$titre.' [=]{2}(.*)\n\n/sU', $this->txt, $match)) {`
67	`$section = $match[1];`	67	`$section = $match[1];`
68	`}`	68	`}`
69	`return $section;`	69	`return $section;`
70	`}`	70	`}`
71		71
72	`public function rendre($wikitxt) {`	72	`public function rendre($wikitxt) {`
73	`$wikitxt .= '<references />';`	73	`$wikitxt .= '<references />';`
74	`$this->initialiserRequete();`	74	`$this->initialiserRequete();`
75	`$this->url = $this->getBaseApiURL();`	75	`$this->url = $this->getBaseApiURL();`
76	`$this->parametres = array(`	76	`$this->parametres = array(`
77	`'action' => 'parse',`	77	`'action' => 'parse',`
78	`'prop' => 'text',`	78	`'prop' => 'text',`
79	`'text' => $wikitxt`	79	`'text' => $wikitxt`
80	`);`	80	`);`
81	`$this->resultats = $this->consulterAPI();`	81	`$this->resultats = $this->consulterAPI();`
82	`$txt = $this->resultats['parse']['text']['*'];`	82	`$txt = $this->resultats['parse']['text']['*'];`
83	`$txt = $this->remplacerUrls($txt);`	83	`$txt = $this->remplacerUrls($txt);`
84	`return $txt;`	84	`return $txt;`
85	`}`	85	`}`
86		86
87	`private function initialiserRequete() {`	87	`private function initialiserRequete() {`
88	`$this->url = '';`	88	`$this->url = '';`
89	`$this->parametres = array();`	89	`$this->parametres = array();`
90	`$this->resultats = array();`	90	`$this->resultats = array();`
91	`}`	91	`}`
92		92
93	`private function getBaseWpURL() {`	93	`private function getBaseWpURL() {`
94	`$baseURL = "http://{$this->langue}.wikipedia.org";`	94	`$baseURL = "http://{$this->langue}.wikipedia.org";`
95	`return $baseURL;`	95	`return $baseURL;`
96	`}`	96	`}`
97		97
98	`private function getBaseApiURL() {`	98	`private function getBaseApiURL() {`
99	`$baseURL = $this->getBaseWpURL().'/w/api.php';`	99	`$baseURL = $this->getBaseWpURL().'/w/api.php';`
100	`return $baseURL;`	100	`return $baseURL;`
101	`}`	101	`}`
102		102
103	`private function consulterAPI() {`	103	`private function consulterAPI() {`
104	`$this->parametres['format'] = 'php';`	104	`$this->parametres['format'] = 'php';`
105	`$resultat = $this->consulterEnPost();`	105	`$resultat = $this->consulterEnPost();`
106	`$resultat = unserialize($resultat);`	106	`$resultat = unserialize($resultat);`
107	`if (isset($resultat['error'])) {`	107	`if (isset($resultat['error'])) {`
108	`throw new Exception($resultat['error']['info'], $resultat['error']['info']);`	108	`throw new Exception($resultat['error']['info'], $resultat['error']['info']);`
109	`}`	109	`}`
110	`return $resultat;`	110	`return $resultat;`
111	`}`	111	`}`
112		112
113	`private function consulterEnPost() {`	113	`private function consulterEnPost() {`
114	`return $this->consulter('POST');`	114	`return $this->consulter('POST');`
115	`}`	115	`}`
116		116
117	`private function consulter($mode) {`	117	`private function consulter($mode) {`
118	`$entetes = array(`	118	`$entetes = array(`
119	`'Content-type' => 'application/x-www-form-urlencoded',`	119	`'Content-type' => 'application/x-www-form-urlencoded',`
120	`'User-Agent' => $this->userAgent);`	120	`'User-Agent' => $this->userAgent);`
121	`$contexte = stream_context_create(array(`	121	`$contexte = stream_context_create(array(`
122	`'http' => array(`	122	`'http' => array(`
123	`'method' => $mode,`	123	`'method' => $mode,`
124	`'header' => $this->getEnteteChaine($entetes),`	124	`'header' => $this->getEnteteChaine($entetes),`
125	`'content' => http_build_query($this->parametres, null, self::HTTP_URL_REQUETE_SEPARATEUR))));`	125	`'content' => http_build_query($this->parametres, null, self::HTTP_URL_REQUETE_SEPARATEUR))));`
126	`$flux = fopen($this->url, 'r', false, $contexte);`	126	`$flux = fopen($this->url, 'r', false, $contexte);`
127	`if (!$flux) {`	127	`if (!$flux) {`
128	`$this->reponse_entetes = $http_response_header;`	128	`$this->reponse_entetes = $http_response_header;`
129	`$e = "L'ouverture de l'url '{$this->url}' par la méthode HTTP '$mode' a échoué!";`	129	`$e = "L'ouverture de l'url '{$this->url}' par la méthode HTTP '$mode' a échoué!";`
130	`throw new Exception($e);`	130	`throw new Exception($e);`
131	`}`	131	`}`
132	`// Informations sur les en-têtes et métadonnées du flux`	132	`// Informations sur les en-têtes et métadonnées du flux`
133	`$this->reponse_entetes = stream_get_meta_data($flux);`	133	`$this->reponse_entetes = stream_get_meta_data($flux);`
134	`// Contenu actuel de $url`	134	`// Contenu actuel de $url`
135	`$contenu = stream_get_contents($flux);`	135	`$contenu = stream_get_contents($flux);`
136	`fclose($flux);`	136	`fclose($flux);`
137	`return $contenu;`	137	`return $contenu;`
138	`}`	138	`}`
139		139
140	`private function getEnteteChaine(Array $entetes) {`	140	`private function getEnteteChaine(Array $entetes) {`
141	`$entetesCleVal = array();`	141	`$entetesCleVal = array();`
142	`foreach ($entetes as $cle => $valeur) {`	142	`foreach ($entetes as $cle => $valeur) {`
143	`$entetesCleVal[] = $cle.': '.$valeur;`	143	`$entetesCleVal[] = $cle.': '.$valeur;`
144	`}`	144	`}`
145	`return implode("\r\n", $entetesCleVal);`	145	`return implode("\r\n", $entetesCleVal);`
146	`}`	146	`}`
147		147
148	`private function remplacerUrls($txt) {`	148	`private function remplacerUrls($txt) {`
149	`$remplacements = array(`	149	`$remplacements = array(`
150	`'href="/wiki/' => 'href="'.$this->getBaseWpURL().'/wiki/',`	150	`'href="/wiki/' => 'href="'.$this->getBaseWpURL().'/wiki/',`
151	`'href="/w/' => 'href="'.$this->getBaseWpURL().'/w/');`	151	`'href="/w/' => 'href="'.$this->getBaseWpURL().'/w/');`
152	`$txt = strtr($txt, $remplacements);`	152	`$txt = strtr($txt, $remplacements);`
153	`return $txt;`	153	`return $txt;`
154	`}`	154	`}`
155	`}`	155	`}`
156	`?>`	156	`?>`

Subversion Repositories eFlore/Applications.eflore-consultation

(root)/trunk/metier/robots/WikipediaBot.php @ 176 – Rev 175 → 176