NJIT Course Schedule Scraper

njitScraper.php

<?php
/**
 * NJIT Course Schedule Scraper (11/21/2012)
 *
 * Using http://courseschedules.njit.edu/, this PHP scraper utilizes CURL to build a large 
 * associative array for all courses.  The user may specify options through onlySubject and 
 * onlyCourse methods to scrape the respective subjects and/or courses.   The resulting array only 
 * has the section data.  Course data is not stored, but may easily be captured with small changes
 * to the scraper.
 * 
 * Example:
 * $scraper = new NJITScrape("2013s", true);
 * $scraper->onlySubject("CS");
 * $scraper->onlyCourse("490");
 * $courseData = $scraper->start();
 * 
 * JSON Representation of $courseData will look like this:
 * {
 *     "CS": {
 *         "name": "Computer Science",
 *         "490": {
 *             "name": "DESIGN IN SOFTWARE ENGR",
 *             "description": "Prerequisite: senior standing or departmental approval. This course
 *             focuses on the methodology for developing software systems. Methods and techniques
 *             for functional requirements analysis and specifications, design, coding, testing and
 *             proving integration and maintenance are discussed.",
 *             "102": {
 *                 "call": 21239,
 *                 "comments": "",
 *                 "days": "F:600PM - 905PM",
 *                 "rooms": "CKB207",
 *                 "status": "Open",
 *                 "maxStudents": 36,
 *                 "currentStudents": 28,
 *                 "instructor": "Nicholson Theod",
 *                 "instructorId": "THEO",
 *                 "credits": 3
 *             }
 *         }
 *     }
 * }
 *
 * @author     Sashank Tadepalli <dijjit@gmail.com>
 * @license    Creative Commons Attribution 3.0 Unported License.
 * @version    1.0
 * @link       http://dijjit.com/php/njit-course-schedule-scraper/
 */

// SimpleHTMLDOM - Web Scraping Utility
// Please download from http://simplehtmldom.sourceforge.net/
require_once "simplehtmldom.php";

class NJITScrape {
    private $page = "http://courseschedules.njit.edu/index.aspx";
    private $semester;
    private $subjectId;
    private $onlySubjectId = "";
    private $subjectName;
    private $courseId;
    private $onlyCourseId = "";
    private $courseName;
    private $data = array();
    private $FLUSH_DEBUG = true;

    public function __construct($startSemester, $flush){
        set_time_limit(0);
        $this->FLUSH_DEBUG = $flush;
        $this->semester = $startSemester;
    }

    public function start(){
        if( $this->FLUSH_DEBUG ){
            echo "<h1>" . $this->semester . "</h1>";
            @ob_flush();
            flush();
        }
        $this->scrapeSemester();
        return $this->data;
    }

    public function onlySubject($subjectId){
        $this->onlySubjectId = $subjectId;
    }

    public function onlyCourse($courseId){
        $this->onlyCourseId = $courseId;
    }

    private function scrapeSemester(){
        $url = $this->page . "?semester=$this->semester";
        $html = file_get_html($url);
        foreach($html->find("span a") as $ele){
            $subject = explode("-", $ele->plaintext);
            $this->subjectId = trim($subject[0]);

            if( !empty($this->onlySubjectId) && 
                strpos($this->subjectId, $this->onlySubjectId) === false ){
                continue;
            }

            $this->subjectName = trim($subject[1]);
            $this->data[$this->subjectId] = array("name" => $this->subjectName);

            if( $this->FLUSH_DEBUG ){
                echo $this->subjectName . "<br />";
                @ob_flush();
                flush();
            }
            $this->scrapeSubject();
        }
    }

    private function scrapeSubject(){
        $courseIds = array();
        $courseNames = array();
        $url = $this->page . "?semester=$this->semester&subjectID=$this->subjectId";
        $html = file_get_html($url);

        // Populate courseIds and courseNames
        foreach($html->find(".courseName strong") as $ele){
            $courseIds[] = trim(str_replace($this->subjectId, "", $ele->plaintext));
        }
        foreach ($html->find(".courseName") as $ele) {
            $t             = explode("-", $ele->plaintext);
            $t             = explode("(", $t[1]);
            $courseNames[] = trim($t[0]);
        }

        // Build pagination query and fancy post request parameters
        $pagecount = count($html->find(".pagination_wrapper a")) / 2 + 1;
        $t         = $html->find("#__VIEWSTATE");
        $vs        = $t[0]->value;
        $t         = $html->find("#__EVENTVALIDATION");
        $evt       = $t[0]->value;
        
        // Loop through all pages to get all courses
        for ($i = 1; $i < $pagecount; $i++) {
            // Fancy post request to fake new page request
            if( !($pageHTML = $this->postRequest($url, array(
                    "__EVENTARGUMENT" => "Page\$" . ($i + 1),
                    "__EVENTTARGET" => "ctl10\$GridView1",
                    "__VIEWSTATE" => trim($vs),
                    "__VIEWSTATEENCRYPTED" => "",
                    "__EVENTVALIDATION" => trim($evt),
                    "__LASTFOCUS" => "",
                    "ctl10\$ddlSemester" => $this->semester
                ))) ){
                continue;
            }
            $html = str_get_html($pageHTML);
            
            // Populate courseIds and courseNames
            foreach ($html->find(".courseName strong") as $ele1) {
                $courseIds[] = trim(str_replace($this->subjectId, "", $ele1->plaintext));
            }
            foreach ($html->find(".courseName") as $ele1) {
                $t             = explode("-", $ele1->plaintext);
                $t             = explode("(", $t[1]);
                $courseNames[] = trim($t[0]);
            }
        }

        for($i = 0; $i < count($courseIds); $i++){
            $this->courseId = $courseIds[$i];
            if( !empty($this->onlyCourseId) && 
                strpos($this->courseId, $this->onlyCourseId) === false ){
                continue;
            }

            $this->courseName = $courseNames[$i];
            $this->data[$this->subjectId][$this->courseId] = array("name" => $this->courseName);

            $url = $this->page . 
                    "?semester=$this->semester&subjectID=$this->subjectId&course=$this->courseId";
            if( !($sectionHTML = $this->postRequest($url, array())) ){
                continue;
            }

            if( $this->FLUSH_DEBUG ){
                echo $this->courseName . "<br />";
                @ob_flush();
                flush();
            }
            $this->scrapeSections($sectionHTML);
        }
    }

    private function scrapeSections($sectionHTML){
        $html = str_get_html($sectionHTML);
        $description = $html->find("#ctl10_lblCourseDesc");
        if (count($description) == 0) {
            $description = "";
        } else {
            $description = $description[0]->plaintext;
        }
        $this->data[$this->subjectId][$this->courseId]["description"] = trim($description);

        foreach ($html->find("#ctl10_gv_sectionTable .sectionRow") as $ele) {
            $fixed = $ele->find("td");
            $days = explode("Section Comments:", $fixed[2]->plaintext);
            $tempData = array();
            $tempData["call"] = intval(
                trim(
                    str_replace(
                        "View Book Info", "", $fixed[1]->plaintext
                    )
                )
            );
            $tempData["comments"] = (count($days) > 1) ? trim($days[1]) : "";
            $tempData["days"] = trim($days[0]);
            $tempData["rooms"] = trim($fixed[3]->plaintext);
            $tempData["status"] = trim($fixed[4]->plaintext);
            $tempData["maxStudents"] = intval(trim($fixed[5]->plaintext));
            $tempData["currentStudents"] = intval(trim($fixed[6]->plaintext));
            $insId = $fixed[7]->find("a");
            $tempData["instructor"] = $instructor = trim($fixed[7]->plaintext);
            if (count($insId) == 0) {
                $insId = $instructor;
            } else {
                $insId = trim(str_replace(
                    "https://directory.njit.edu/PersDetails.aspx?persid=", "", $insId[0]->href)
                );
            }
            $tempData["instructorId"] = $insId;
            $tempData["credits"] = intval(trim($fixed[8]->plaintext));
            $this->data[$this->subjectId][$this->courseId][trim($fixed[0]->plaintext)] = $tempData;
        }
    }

    private function postRequest($url, $post){
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $url);
        if (!empty($post)) {
            curl_setopt($ch, CURLOPT_POST, count($post));
            curl_setopt($ch, CURLOPT_POSTFIELDS, http_build_query($post));
        }
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        $result = curl_exec($ch);
        $check  = curl_getinfo($ch);
        curl_close($ch);
        if ($check["http_code"] != "200") {
            return false;
        }
        return $result;
    }
}
?>