diff --git a/school.d/umich.crawl.inc b/school.d/umich.crawl.inc --- a/school.d/umich.crawl.inc +++ b/school.d/umich.crawl.inc @@ -20,17 +20,17 @@ /** Filter out whitepace items */ -function umich_arrayfilter_callback($item){ - if(ltrim($item) == ''){ - return false; - } - else{ - return true; - } +function umich_arrayfilter_callback($item) +{ + if(ltrim($item) == '') + return TRUE; + else + return TRUE; } /** Parse html at URL into array, first row is row headers */ -function umich_table_parse($url) { +function umich_table_parse($url) +{ $arr = array(); $dom = new DOMDocument; $html = file_get_contents($url); @@ -64,9 +64,26 @@ function umich_table_parse($url) { return $arr; } -/** Crawls uMich course listings. $season is "f" or "s", year is 2-digit year */ -function umich_crawl($semester) +/** + * \brief + * Crawls University of Michigan's schedule. + * + * \param $semesters + * An array to be filled with semesters. + * \param $school_crawl_log + * The school_crawl_log handle. + * \return + * 1 on failure, 0 on success. + */ +function umich_crawl(array &$semesters, $school_crawl_log) { + $url = 'http://lsa.umich.edu/cg/cg_advsearch.aspx'; + $cookies = array(); + + /* determine list of semesters: */ + $semesters_dom = new DOMDocument(); + $semesters_dom->loadHTML(school_crawl_geturi($url, $cookies, $school_crawl_log)); + $year = substr($semester->year_get(), 2); $season = strtolower(substr($semester->season_get(), 0, 1));