From eb7db50ab03e0fda7cc4e80a2c11f89413011f31 Mon Sep 17 00:00:00 2001
From: Steffen Vogel <info@steffenvogel.de>
Date: Thu, 1 Nov 2012 15:01:17 +0100
Subject: [PATCH] added more room infos to crawler

---
 php/campus/cocal.db  | Bin 5120 -> 5120 bytes
 php/campus/cocal.php |  55 +++++++++++++++++++++++++++++++------------
 2 files changed, 40 insertions(+), 15 deletions(-)

diff --git a/php/campus/cocal.db b/php/campus/cocal.db
index 1453e195f20af5e75c674915a9fcdb01fdb1fc56..e81ca4ce53287b764af8cc9a6396511e8da220ce 100644
GIT binary patch
literal 5120
zcmeH|y=ucS5XaAH8Zu-kokFLZifl|FDc-u6657xYQa2Pb6~$6ofop+fL&rW`pQ0K2
z5*<3!ma{m}#hc*-7V`Oj(%sM8$@6gestj&aA~Du!&Q)rVr}+4At8$6cvk3~G^EEdT
zJr7v%{nZBmr+j`70jf0(B0vPXOJD&*ID!kO4`Z9eM_t~#+GbF`_XZ<8WlW_)z5~m&
zek+?qk!x$fi3`>Qf%UKV*;aneV01h_*_QnykvmPTJcl@wylgP`6V^L2CGaVXF0aFg
zO@qM2MKr$)qZQuFRz7keGHFah;bWW>CYRZsbzP`LsPyF@^K+_ke!rYu%_85Py{D_f
zX=pZy*3C0qfX|*ymnI_6aRNj5Ip}x_6h09k0<8&D<xef3H9m4A0v#qm_kV|f9)(T>
HS`+vK+8kw2

literal 5120
zcmeHKzi-n(6n?GU)-47~`Bmr-U6-njn(EF!AlVwKAP}ttNkkVGH@Olku2b1ALMQ`&
z0wV(w3j-rV$8KF2m=POH{0$7eJKI=oMHB{xR=&Y;?Dy__-{*Uu&xHbMwB3;GxE?3X
zt{*wB9}>qW^(Px{#EG%WNXv<wZ71ZUP=KZV?_VJqhxn_efAZg4fn0$zselH1vhKK-
zx8OCv(5OB&eR~W0-s2E-ySrg>G-16~tu(5nQMtEKCGr+o@||5yo>pq>4=S}K!?u@K
zh|_8XJPeOrT?TlgoJooDqH_h#r2-?!N`npg9*_Iq68%4+=$2vHcY4&OP-%AqKa2wB
z=shR4y+Sr!9u{9XezC=aVy)Ze#d;L1!a?5K=W>nYq|Ubj8hjoS3{@!MabCXv6@8vr
z2JKO5z}C^B=lMKHEufspp+Y;2AI{=gpZK>JKPTo_6n&1_rFi}Z-`(?g@H}v~xlhV8
znI7x;f~SijURCr(R>FD=tf%UFJ7F_6C61pfi8C=fNmSt%p3RE95k;Rx9}_s`lRhS+
zRj`n|#2Ae+7{;0dV7$aU>6K%dSF)2BfAvZMuEM7jTg&*d0kB2(gy1)>j8EKeFf16F
zx2$1tJs-IO|E7QjALPnWVGR>yFI>ph<J|$)qxgQ1zM-c+U@+Rp2R^1a7c-pF065<;
zZ$Yg4ONzc=&@$~YgTf~FoQR`A5!{BzO^Sj25Dyh}>@n#%mxbq0hB+pK<DO%u)`2~_
zhVc?(QedzW>pf-+9(DJ4;Jcmu@Q>YC(D~UBC@OfEBmL$|>^D}Any2<nvLF@SVXV|@
zN@yiGm<EG-v+epF?xnLEarrp+0mf^B^HpU`i?&<w4HbUib;;{m%FCercunFno)Pn0
zSM(WXA}*4Hhko>mr=W=nr~$^x8gGainI&RR`+8QRhw+mC&6vN@qsZyRzEKvg%A;Ik
S)7i7{@3+s8X+Hjs0>1%^+BE$D

diff --git a/php/campus/cocal.php b/php/campus/cocal.php
index d59344f..9b287b0 100644
--- a/php/campus/cocal.php
+++ b/php/campus/cocal.php
@@ -1,13 +1,13 @@
 <?php
 
 /* Constants */
-$salt = '8D896ww348lDSd1';
 $scriptUrl = ((!empty($_SERVER['HTTPS'])) ? 'https' : 'http') . '://' . $_SERVER['SERVER_NAME'] . $_SERVER['PHP_SELF'];
-$baseUrl = 'https://www.campus.rwth-aachen.de/office/';
-$homePath = 'default.asp';
-$loginPath = 'views/campus/redirect.asp';
-$calPath = 'views/calendar/iCalExport.asp';
-$logoutPath = 'system/login/logoff.asp';
+$baseUrl = 'https://www.campus.rwth-aachen.de/';
+$homePath = 'office/default.asp';
+$loginPath = 'office/views/campus/redirect.asp';
+$calPath = 'office/views/calendar/iCalExport.asp';
+$logoutPath = 'office/system/login/logoff.asp';
+$roomPath = 'rwth/all/room.asp'
 
 /* Functions */
 function curl_fixcookie($cookieFile) {
@@ -61,21 +61,44 @@ function curl_request($method, $url, $cookieFile = false, $params = array()) {
 }
 
 function get_address($db, $room) {
-	$result = sqlite_query($db, 'SELECT address FROM rooms WHERE name = "' . sqlite_escape_string($room). '";');
-	return ($result && sqlite_valid($result)) ? sqlite_fetch_string($result) : false;
+	$result = sqlite_query($db, 'SELECT address FROM rooms WHERE id = "' . sqlite_escape_string($room). '";');
+	return ($result && sqlite_valid($result)) ? sqlite_fetch_array($result) : false;
 }
 
 function set_address($db, $room, $address) {
-	sqlite_exec($db, 'INSERT INTO rooms VALUES ("' . sqlite_escape_string($room) . '", "' . sqlite_escape_string($address) . '");', $error);
+	sqlite_exec($db, 'INSERT OR REPLACE INTO rooms VALUES (
+				"' . sqlite_escape_string($address['id']) . '",
+				"' . sqlite_escape_string($address['address']) . '",
+				"' . sqlite_escape_string($address['cluster']) . '",
+				"' . sqlite_escape_string($address['building']) . '",
+				"' . sqlite_escape_string($address['building_no']) . '",
+				"' . sqlite_escape_string($address['room']) . '",
+				"' . sqlite_escape_string($address['room_no']) . '",
+				"' . sqlite_escape_string($address['floor']) . '"
+			);', $error);
 }
 
 function crawl_address($room) {
-	$response = curl_request('GET', 'http://www.campus.rwth-aachen.de/rwth/all/room.asp?room=' . urlencode($room));
+	$response = curl_request('GET', $basePath . $roomPath . '?room=' . urlencode($room));
 
 	$matches = array();
-	$r = preg_match("/<td class=\"default\">Geb.udeanschrift<\/td><td class=\"default\">([^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">H.rsaalgruppe<\/td><td class=\"default\">(?P<cluster>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Geb.udeanschrift<\/td><td class=\"default\">(?P<address>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Geb.udebezeichnung<\/td><td class=\"default\">(?P<building>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Geb.udenummer<\/td><td class=\"default\">(?P<building_no>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Raumname<\/td><td class=\"default\">(?P<room>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Raumnummer<\/td><td class=\"default\">(?P<room_no>[^<]*)<\/td>/", $response, $matches);
+	preg_match("/<td class=\"default\">Geschoss<\/td><td class=\"default\">(?P<floor>[^<]*)<\/td>/", $response, $matches);
 
-	return ($r > 0) ? $matches[1] : false;
+	foreach ($matches as $key => $value) {
+		if (is_numeric($key) unset($matches[$key]);
+	}
+
+	array_walk($matches, function ($value) {
+		return preg_replace('/[ ]{2,}/sm', ' ', utf8_encode($value));
+	});
+
+	return (count($matches['address']) ? $matches : false;
 }
 
 function error() {
@@ -182,11 +205,13 @@ if (isset($matrnr) && isset($passwd)) {
 					$address = get_address($db, $room);
 
 					if ($address === false) {
-						$address = preg_replace('/[ ]{2,}/sm', ' ', utf8_encode(crawl_address($room)));
+						$crawled = crawl_address($room);
 						set_address($db, $room, $address);
-						$crawled = true;
 					}
-					$value = $address . ', Aachen';
+
+					if ($address) {
+						$value = $address['address'] . ', Aachen';
+					}
 					break;
 
 				case 'DESCRIPTION':