<?php
////
// Author: Sean Colombo
// Date: 20070930
//
// This file helps with screen-scraping.
// The Scraper class will persist a session to let users stay logged in.
//
// There are procedural functions below for quick grabs that don't require
// special settings.
//
// TODO: SPEED: Figure out how often the $ch needs to be init'ed and destroyed.  It might not have to be in every page call.
////

include_once 'tools.php'; // for startsWith()

class Scraper{
	private $useSession;
	private $cookieFile;
	private $userAgent;

	function Scraper(){
		$this->useSession = false;
		$this->setCookieFile("./cookie_".microtime(true)."_".rand(0,999999).".txt");
		$this->userAgent = ""; // TODO: Add a representative userAgent
	}

	////
	// Sets the cookie file.
	//
	// WARNING: Use this mutator instead of directly changing the file-name
	// since this function will take care of cross-platform issues.
	////
	function setCookieFile($fileName){
		$this->cookieFile = $fileName;
		
		// Make the filename work with windows.
		if(substr(PHP_OS, 0, 3) == 'WIN'){
			$this->cookieFile = str_replace('\\','/', getcwd().'/'.$this->cookieFile);
		}
	}

	////
	// If useThem is true, sessions will be used.  If a file-name is passed in, that
	// will be used as the cookie-file (instead of the default).
	////
	function useSession($useThem, $fileName=""){
		$this->useSession = $useThem;
		if($fileName != ""){
			$this->setCookieFile($fileName);
		}
	}

	////
	// Does an HTTP GET on the url passed in and returns the source-code of the page.
	////
	function getPage($url){
		$retVal = "";

		$ch = curl_init();
		if($this->useSession){
			curl_setopt($ch, CURLOPT_COOKIEJAR, $this->cookieFile); //initiates cookie file if needed
			curl_setopt($ch, CURLOPT_COOKIEFILE, $this->cookieFile); // Uses cookies from previous session if exist
		}
		curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
		if(startsWith($url, "https")){
			curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
		}
		curl_setopt($ch, CURLOPT_URL, $url);
		curl_setopt($ch, CURLOPT_HEADER, false);
		curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);// allow redirects
		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
		if($this->userAgent != ""){
			curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent); 
		}

		$retVal = curl_exec($ch);

		// Make the strange behavior where curl_exec returns boolean true on an empty document into something more like what we'd expect.
		if($retVal === true){
			$retVal = "";
		} else if($retVal === false){
			$errorStr = curl_error($ch);
			dispError($errorStr);
		}
		curl_close($ch);

		return $retVal;
	}

	////
	// Posts the data to the given url and returns the code of the resulting page.
	////
	function postPage($url, $postData=array()){
		$retVal = "";

		// Turn the array of data into a string for cURL.
		$postFields = "";
		foreach($postData as $pKey=>$pVal){
			$postFields .= (($postFields=="")?"":"&"); // separate key/val pairs from each other with "&"
			$postFields .= "$pKey=".urlencode($pVal);
		}

		$ch = curl_init();
		if($this->useSession){
			curl_setopt($ch, CURLOPT_COOKIEJAR, $this->cookieFile); //initiates cookie file if needed
			curl_setopt($ch, CURLOPT_COOKIEFILE, $this->cookieFile); // Uses cookies from previous session if exist
		}
		curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false);
		if(startsWith($url, "https://")){
			curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);
		}
		curl_setopt($ch, CURLOPT_URL, $url);
		curl_setopt($ch, CURLOPT_HEADER, false);
		curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);// allow redirects
		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true); // return into a variable
		curl_setopt($ch, CURLOPT_POST, true);
		curl_setopt($ch, CURLOPT_POSTFIELDS, $postFields);
		if($this->userAgent != ""){
			curl_setopt($ch, CURLOPT_USERAGENT, $this->userAgent); 
		}
		$retVal = curl_exec($ch);
		curl_close($ch);

		return $retVal;
	} // end postPage()

	////
	// Since not all sites are expected to co-operate, this will let us fake them out.
	////
	function setUserAgent($userAgentString){
		$this->userAgent = $userAgentString;
	} // end setUserAgent()

} // end class Scraper

?>