[Devel] r491 - scripts

[svn AT agendadulibre.org]

Author: thomas
Date: Sun Feb 20 15:06:37 2011
New Revision: 491

Log:
Mise à jour du script d'extraction des GULLs

Modified:
   scripts/extract-gulls.py

Modified: scripts/extract-gulls.py
==============================================================================
--- scripts/extract-gulls.py    Tue Nov  9 20:17:06 2010        (r490)
+++ scripts/extract-gulls.py    Sun Feb 20 15:06:37 2011        (r491)
@@ -149,6 +149,27 @@
     u'Manchix': u'Cherbourg',
     u'LUG d\'Agneaux et de ses environs': u'Agneaux',
     u'Bethunelibre': u'Bethune',
+    u'Les Linuxiens Laragnais': u'Laragne-Montéglin',
+    u'Salamandrix': u'Villers-Cotterêts',
+    u'LAILA': u'Troyes',
+    u'ULLM': u'Martigues',
+    u'Linux-Angoulême': u'Angoulême',
+    u'Pullco': u'Brive-la-Gaillarde',
+    u'DR2M': u'Bastia',
+    u'Infothema': u'Bégard',
+    u'Mammoutux': u'Roufignac',
+    u'Abistodegnux': u'Condom',
+    u'ALL': u'Montpellier',
+    u'GULLL': u'Châteaubriant',
+    u'Alternative 46': u'Cahors',
+    u"Lug d'\Agneaux": u"Agneaux",
+    u"Gebull": u"Bressuire",
+    u"LUG-Adeti.org": u"Tours",
+    u"Ubuntu-Lyon": u"Lyon",
+    u"PostPro": u"Toulon",
+    u"APP3L": u"Poitiers",
+    u"LUG Veternat": u"Ventron",
+    u"Le Libre Vanvéen": u"Vanves",
     }
 
 class GULLParser(HTMLParser.HTMLParser):
@@ -215,12 +236,25 @@
                 else:
                     city = ""
 
-                print "insert into lugs (region, department, city, name, 
url) values ('%s', '%s', '%s', '%s', '%s');" \
-                      % (str(depts2region[str(self.currentDepartement)]),
-                         self.currentDepartement,
-                         city.replace("'", r"\'").encode('utf-8'),
-                         self.currentLugName.replace("'", 
r"\'").encode('utf-8'),
-                         self.currentLugLink)
+                if self.currentLugName == "LinAC" or \
+                         self.currentLugName == "ACDILL" or \
+                         self.currentLugName == "Perinux" or \
+                         self.currentLugName == "Club ENSIMAG" or \
+                         self.currentLugName == "Masclat LUG" or \
+                         self.currentLugName == "SL@M" or \
+                         self.currentLugName == "M@IL" or \
+                         self.currentLugName == "Bourguinux" or \
+                         self.currentLugName == "Tarentux":
+                    self.currentLugLink = None
+                    self.currentLugName = None
+                    return
+
+                query = "insert into lugs (region, department, city, name, 
url) values ('%s', '%s', '%s', '%s', '%s');" \
+                    % (str(depts2region[str(self.currentDepartement)]),
+                       self.currentDepartement, city.replace("'", r"\'"),
+                       self.currentLugName.replace("'", r"\'"),
+                       self.currentLugLink)
+                print query.encode("utf-8")
 
             self.currentLugLink = None
             self.currentLugName = None
@@ -235,7 +269,7 @@
 
         if self.inLug and self.inLugLink and self.currentLugName is None:
             # Hack a little ;-)
-            data = unicode(data, "iso-8859-15")
+            # data = unicode(data, "utf-8")
             data = data.replace("\n", " ")
             data = re.sub(" +", " ", data)
             data = data.strip()
@@ -255,5 +289,6 @@
     donnees = f.read(8192)
     if not donnees:
         break
+    donnees = unicode(donnees, "utf-8")
     p.feed(donnees)
 p.close()