header("content-type:text/html;charset=utf8"); class Dict{ private $word; //显示的条数 private static $num = 10; public function __construct(){} public function content($word){ $this -> word = $word; $symbol = $this -> Pronounced(); $pro = $this->getSay(); $example = $this -> getExample(); $explain = $this -> getExplain(); $synonym = $this -> getSynonym(); $phrase = $this -> getPhrase(); $result = array( "symbol" => $symbol, //音标 "pro" => $pro, //发音 "example"=> $example, //例句 "explain"=> $explain, //简明释义 "synonym"=> $synonym, //同反义词 "phrase" => $phrase //短语数组 ); return $result; } private function getContent(){ $useragent = "Mozilla/5.0 (Windows NT 6.1; WOW64; rv:23.0) Gecko/20100101 Firefox/23.0"; $ch = curl_init(); $url = "http://dict.baidu.com/s?wd=".$this->word; curl_setopt($ch, CURLOPT_URL, $url); curl_setopt($ch, CURLOPT_USERAGENT,$useragent); curl_setopt($ch, CURLOPT_RETURNTRANSFER, TRUE); curl_setopt($ch, CURLOPT_FOLLOWLOCATION, 1); curl_setopt($ch, CURLOPT_HTTPGET, 1); curl_setopt($ch, CURLOPT_AUTOREFERER,1); curl_setopt($ch, CURLOPT_HEADER, 0); curl_setopt($ch, CURLOPT_TIMEOUT, 30); $result = curl_exec($ch); if (curl_errno($curl)) { echo 'Errno'.curl_error($curl); } curl_close($ch); return $result; } private function Pronounced(){ $data = $this -> getContent(); preg_match_all("/"EN-US">(.*)/Ui",$data,$pronounced); return array( 'en' => $pronounced[1][0], 'us' => $pronounced[1][1] ); } private function getSay(){ $data = $this -> getContent(); preg_match_all("/url="(.*)"/Ui",$data,$pronounced); return array( 'en' => $pronounced[1][0], 'us' => $pronounced[1][1] ); } private function getExample(){ $str = ""; $data = $this -> getContent(); preg_match_all("/var example_data = (.*)];/Us",$data,$example); $data1 = "[[[".ltrim($example[1][0],"["); $data2 = explode("[[[",$data1); $num = count(array_filter($data2)); foreach($data2 as $key => $value){ $data3 = explode("[[","[[".$value); foreach ($data3 as $k => $v) { preg_match_all("/["(.*)",/Us","[".$v, $match); if(!empty($match[1])){ $str .= implode($match[1]," ")."@"; } } } $data4 = trim($str,"@"); $data5 = explode("@", $data4); $result = array_chunk($data5, 2); return $result; } private function getExplain(){ $data = $this -> getContent(); preg_match_all("/id="en-simple-means">(.*)

/Us",$data,$explain); $r_data = $explain[1][0]; preg_match_all("/

(?P.*)(?P.*)

/Us", $r_data, $a_data); preg_match_all("/(?P[^>]+)：(?P.*)/Us", $r_data, $b_data); $result = array(); foreach ($a_data["adj"] as $key => $value) { $result[$value] = $a_data["name"][$key]; } $word_b = array(); foreach ($b_data["tag"] as $key => $value) { $word_b[$value] = strip_tags($b_data["word"][$key]); } $result_data = array("x" => $result,"b" => $word_b); return $result_data; } private function getSynonym(){ $data = $this -> getContent(); preg_match_all("/id="en-syn-ant">(.*)

/Us",$data,$synonym); $content = $synonym[1][0]; $data1 = explode("", $content); $result = array(); $data2 = array(); foreach ($data1 as $key => $value) { preg_match_all("/(?P.*) ;

.*)

/Us", $value, $r_data); $data2[$key]["adj"] = $r_data["adj"]; $data2[$key]["content"] = $r_data["content"]; } foreach ($data2 as $key => $value) { foreach ($value["content"] as $k => $v) { if(!empty($v)){ preg_match_all("/

(?P

.*)(?P<value>.*)</value>

/Us", $v, $v_data); foreach ($v_data['title'] as $m => $d) { $data = strip_tags(preg_replace(">"," ", $v_data["value"][$m])); $result[$key][$value["adj"][$k]][$d] = $data; } } } } return $result; } private function getPhrase(){ $num = self::$num; $data = $this -> getContent(); preg_match_all("/id="en-phrase">(.*)

/Us",$data,$phrase); $data = explode("",$phrase[1][0]); $data1 = array_slice($data,0,$num); $result = array(); foreach ($data1 as $key => $value) { $data2 = explode("", $value); $n = count($data2); if($n $value) { foreach ($value as $k => $v) { $value[$k] = strip_tags($v); } $array = array($result[$key_value],$value); if (array_key_exists($key_value, $result)){ $result[$key_value] = $array; } } } } return $result; } private function array2string($data, $isformdata = 1) { if($data == '') return ''; if($isformdata) $data = $this->new_stripslashes($data); return addslashes(var_export($data, TRUE)); } private function new_stripslashes($string) { if(!is_array($string)) return stripslashes($string); foreach($string as $key => $val) $string[$key] = $this->new_stripslashes($val); return $string; } } // $word = new dict("express"); // $word ->content();

ITPub博客

php查词采集器

最新文章