<?php
/*
 * HTTP_URL
 *
 * Copyright (c) 2004 Colin Viebrock <colin@viebrock.ca>
 *
 * Usage of the works is permitted provided that this
 * instrument is retained with the works, so that any entity
 * that uses the works is notified of this instrument.
 *
 * DISCLAIMER: THE WORKS ARE WITHOUT WARRANTY.
 */


require 'PEAR.php';
require 'Net/Curl.php';


class HTTP_URL extends PEAR {

    var $url;

    var $raw_result;
    var $raw_header;
    var $raw_body;

    var $headers;
    var $refs = array();


    var $_ref_tags = array(
        'frame'        => 'src',
        'img'        => 'src',
        'a'            => 'href',
        'form'        => 'action',
        'link'        => 'href',
        'script'    => 'src'
    );


    function HTTP_URL($url=NULL) {
        $this->PEAR();
        $this->reset();
        if ($url!==NULL) {
            $this->setUrl($url);
        }
    }


    function reset() {
        $this->url = NULL;

        $this->raw_result = NULL;
        $this->raw_header = NULL;
        $this->raw_body = NULL;

        $this->headers = NULL;
        $this->refs = array();
    }


    function setURL($url) {
        $this->reset();
        $this->url = $url;
    }


    function get() {
        $c = new Net_Curl;
        $c->url                = $this->url;
        $c->header            = true;
        $c->progress        = false;
        $c->follow_location    = false;
        $this->raw_result = $c->execute();
        $c->close();
        list($this->raw_header,$this->raw_body) = explode("\r\n\r\n", $this->raw_result, 2);

        $this->parseHeader();

    }


    function parseHeader() {
        if (!$this->raw_header) {
            return;
        }

        list($first,$rest) = explode("\r\n", $this->raw_header, 2);

        $temp = explode(' ', $first, 3);
        $this->headers['_Request-Protocol'] = $temp[0];
        $this->headers['_Request-Code'] = $temp[1];
        $this->headers['_Request-String'] = $temp[2];

        $temp = explode("\r\n", $rest);
        foreach($temp as $line) {
            list($h,$v) = explode(':', $line, 2);
            $this->headers[trim($h)] = trim($v);
        }

        if (isset($this->headers['Date'])) {
            $this->headers['_Date'] = strtotime($this->headers['Date']);
        }
        if (isset($this->headers['Last-Modified'])) {
            $this->headers['_Last-Modified'] = strtotime($this->headers['Last-Modified']);
        }

        return true;

    }


    function parseBody() {
        if (!$this->raw_body) {
            return;
        }


        $this->refs = array();

        /* don't need closing tags, so skip </....> */

        if (preg_match_all('/<([^\/].*?)>/', $this->raw_body, $m)) {

            foreach($m[1] as $tag) {
                $tag = trim($tag);
                foreach($this->_ref_tags as $reftag=>$srctag) {

                    if (!is_array($this->refs[$reftag])) {
                        $this->refs[$reftag] = array();
                    }

                    if (preg_match('/^'.$reftag.'\s/i',$tag)) {
                        $regex = '/'.$srctag.'=(\S*)/i';
                        if (preg_match($regex, $tag, $n)) {
                            $this->refs[$reftag][] = $this->stripQuotes($n[1]);
                            break;
                        }
                    }
                }
            }
        }

        foreach ($this->refs as $reftag=>$array) {
            sort($array);
            $this->refs[$reftag] = array_values(array_unique($array));
        }

        return true;

    }



    function stripQuotes($str) {
        $q = array ('"', '\'');
        foreach($q as $qc) {
            if ( substr($str,0,1)===$qc && substr($str,-1)===$qc ) {
                $str = substr($str,1,-1);
                break;
            }
        }
        return $str;
    }

}