Rewrote mensa module ...

2 files changed, 70 insertions, 74 deletions

diff --git a/curunir b/curunir
index bf357b1..89af587 100755
--- a/curunir
+++ b/curunir
@@ -5,4 +5,4 @@ DIR=`dirname $0`
 
 cd "$DIR"
 export PYTHONPATH="$PYTHONPATH:."
-exec /usr/bin/env python -OO "$APP.py" "$@"
+exec /usr/bin/env python2 -OO "$APP.py" "$@"
diff --git a/modules/mensa.py b/modules/mensa.py
index 1baf60f..279caa0 100644
--- a/modules/mensa.py
+++ b/modules/mensa.py
@@ -8,22 +8,21 @@ import urllib
 class Module(ModuleBase):
     def __init__(self, manager):
         ModuleBase.__init__(self, manager)
-    
+
     def commands(self):
         return [('!mensa [tomorrow]', 'Zeigt das aktuelle Menü der Mensa der Uni Lübeck an (oder das des nächsten Tages)')]
-    
+
     def groupchat(self, room, nick, text, handler):
         if not re.match(r'!mensa(?:\W|\Z)', text):
             return
-        
+
         tomorrow = re.match(r'!mensa\W+tomorrow(\W|\Z)', text)
         daystr = 'Heute'
         day = date.today()
         if tomorrow or datetime.now().time() > time(15):
             day += timedelta(1)
             daystr = 'Morgen'
-        datestr = day.strftime(r'%d\.%m\.<\/div>')
-        
+
         try:
             f = urllib.urlopen(self.manager.config['mensa_url'], 'r')
             data = f.read()
@@ -31,67 +30,60 @@ class Module(ModuleBase):
         except IOError:
             handler.reply('Hmm, es scheint gerade keine Daten über das Mensa-Menü zu geben.')
             return
-        
+
         def handle_data(data):
-            if not re.search(datestr, data):
-                return False
-            
-            match = re.compile(r'^.*' + datestr + r'(.*?)(?:<td rowspan="3" class="schrift_fett"><div>|</table>).*$', re.DOTALL).match(data)
-            
-            if not match:
-                return False
-            
-            data = match.group(1)
-            
-            r = re.compile(r'.*?<tr[^>]*>(.*?)</tr>', re.DOTALL)
-            r2 = re.compile(r'(?:.*?<td[^>]*>(.*?)</td>){4}', re.DOTALL)
-            
-            mealstr = ''
-            pos = 0
-            while not r2.match(mealstr):
+            r = re.compile(r'.*?<tr[^>]*>(.*?)</tr>.*?<tr[^>]*>(.*?)</tr>', re.DOTALL)
+            r2 = re.compile(r'.*?<td[^>]*>(.*?)</td>', re.DOTALL)
+
+            def handle_data_row(data, pos):
                 match = r.match(data, pos)
+
                 if not match:
                     return False
-                mealstr = match.group(1)
-                pos = match.end()
-            
-            match = r.match(data, pos)
+
+                ret = []
+                mealpos = 0
+                pricepos = 0
+
+                for day in range(0,5):
+                    mealmatch = r2.match(match.group(1), mealpos)
+                    pricematch = r2.match(match.group(2), pricepos)
+
+                    if not mealmatch or not pricematch:
+                        return False
+
+                    ret.append((mealmatch.group(1), pricematch.group(1)))
+                    mealpos = mealmatch.end()
+                    pricepos = pricematch.end()
+
+                return (ret, match.end())
+
+
+            match = re.compile(r'^.*?<table[^>]*>.*?<td[^>]*>.*?(\d+)\.(\d+)\. - \d+\.\d+\.(\d+)\.*?</td>.*?(?:<col[^>]*>\s*){4}(.*?)</table>.*$', re.DOTALL).match(data)
+
             if not match:
                 return False
-            pricestr = match.group(1)
-            
-            meals = []
-            pos = 0
-            r = re.compile(r'\s*<td class="schrift_(gerichte|spezial)">(.+?)</td>', re.DOTALL)
-            match = r.match(mealstr, pos)
-            while match:
-                meals.append(match.group(2))
-                
-                if len(meals) >= 4:
-                    break
-                
-                pos = match.end()
-                match = r.match(mealstr, pos)
-            
-            if len(meals) < 4:
+
+            firstday = datetime.strptime('%s-%s-%s' % (match.group(3), match.group(2), match.group(1)), '%Y-%m-%d').date()
+            weekday = (day - firstday).days
+
+            if weekday < 0 or weekday > 4:
                 return False
-            
-            prices = []
+
+            data = match.group(4)
+
+            meals = []
             pos = 0
-            r = re.compile(r'\s*<td class="schrift_(?:preise|spezial)">(.+?)</td>', re.DOTALL)
-            match = r.match(pricestr, pos)
-            while match:
-                prices.append(match.group(1))
-                
-                if len(prices) >= 4:
-                    break
-                
-                pos = match.end()
-                match = r.match(pricestr, pos)
-            
-            if len(prices) < 4:
-                return False
-            
+
+            for i in range(0,6):
+                row = handle_data_row(data, pos)
+
+                if not row:
+                    return False
+
+                meals.append(row[0])
+                pos = row[1]
+
             def unescape(s):
                 s = s.replace('&nbsp;', '')
                 s = s.replace('&euro;', '€')
@@ -117,24 +109,28 @@ class Module(ModuleBase):
                 s = re.sub(r'<img[^>]*logo_bio[^>]*>', 'Bio', s)
                 s = re.sub(r'<[^>]*>', '', s)
                 return s
-            
-            meals = map(unescape, meals)
-            prices = map(unescape, prices)
-            
+
+            stripunescape = (lambda item: (unescape(item[0].strip()), unescape(item[1].strip())))
+            meals = map(lambda item: map(stripunescape, item), meals)
+
             reply = 'Mensa-Menü für den %s\n\n' % day.strftime('%d.%m.%y')
-            if meals[0]:
-                reply += 'Eintopf: %s (%s)\n' % (meals[0].strip(), prices[0].strip())
-            if meals[1]:
-                reply += 'Hauptgericht 1: %s (%s)\n' % (meals[1].strip(), prices[1].strip())
-            if meals[2]:
-                reply += 'Hauptgericht 2: %s (%s)\n' % (meals[2].strip(), prices[2].strip())
-            if meals[3]:
-                reply += 'Vegetarisches Hauptgericht: %s (%s)\n' % (meals[3].strip(), prices[3].strip())
-            
+            if meals[0][weekday][0]:
+                reply += 'Eintopf: %s (%s)\n' % meals[0][weekday]
+            if meals[1][weekday][0]:
+                reply += 'Hauptgericht 1: %s (%s)\n' % meals[1][weekday]
+            if meals[3][weekday][0]:
+                reply += 'Hauptgericht 2: %s (%s)\n' % meals[3][weekday]
+            if meals[2][weekday][0]:
+                reply += 'Vegetarisches Hauptgericht: %s (%s)\n' % meals[2][weekday]
+            if meals[4][weekday][0]:
+                reply += 'Wok: %s (%s)\n' % meals[4][weekday]
+            if meals[5][weekday][0]:
+                reply += 'Beilagen: %s (%s)\n' % meals[5][weekday]
+
             return reply
-        
+
         reply = handle_data(data)
         if not reply:
             reply = '%s kein Mensa-Menü... ist vielleicht Wochenende?' % daystr
-        
+
         handler.reply(reply.strip())