Changeset 582

Timestamp:

02/17/10 16:20:36 (5 months ago)

Author:

olivier

Message:

searching by a location now takes into account location historicity ; do not list historical countries in the geo navigator anymore ; improve geocoder, geocoding only current locations (not historical) and searchin by location alias as a fallback

Location:

trunk/telemeta

Files:

• management/commands/telemeta-geocode.py (modified) (2 diffs)
• models/location.py (modified) (1 diff)
• models/query.py (modified) (5 diffs)

trunk/telemeta/management/commands/telemeta-geocode.py

@@ -20 +20 @@
         except IOError:
             raise CommandError("Unable to open %s" % datafile)
-            
-        locations = [l for l in Location.objects.all().current().filter(type=Location.COUNTRY)]
+        
+        locations = {}
+        for l in Location.objects.all().current().filter(type=Location.COUNTRY):
+            locations[l] = [a.alias for a in l.aliases.all()]
 
         i = 0
         geocoded = 0
         total = len(locations)
+        found_by_alias = {}
         for line in datafile:
             (geonameid, name, asciiname, alternatenames, latitude, longitude, feature_class,
@@ -44 +47 @@
                         geocoded += 1
                         found.append(l)
+                    else:
+                        for a in locations[l]:
+                            if unaccent(a).lower() in names:
+                                found_by_alias[l] = float(latitude), float(longitude)
+                                break
+                            
 
                 for l in found:
-                    locations.remove(l)
+                    locations.pop(l)
 
             i += 1
 
             if i % 200000 == 0:
-                print "Geocoded %d out of %d countries (parsed %d geonames)" % (geocoded, total, i)
+                print "Geocoded %d (%d by alias) out of %d countries (parsed %d geonames)" % (geocoded, len(found_by_alias), total, i)
 
             if total == geocoded:
                 break
 
-        print "Geocoded %d out of %d countries (parsed %d geonames)" % (geocoded, total, i)
+        for l in locations:
+            if found_by_alias.has_key(l):
+                l.latitude, l.longitude = found_by_alias[l]
+                l.save()
+                geocoded += 1
+
+        print "Done. Geocoded %d out of %d countries (parsed %d geonames)" % (geocoded, total, i)
         datafile.close()                
 

trunk/telemeta/models/location.py

@@ -78 +78 @@
             q &= Q(ancestor_relations__is_direct=True)
         return Location.objects.filter(q)           
+
+    def apparented(self):
+        return Location.objects.filter(
+                Q(pk=self.id) | 
+                Q(ancestor_relations__ancestor_location=self) | 
+                Q(current_location=self.id)).distinct()
 
     def add_child(self, other):

trunk/telemeta/models/query.py

@@ -84 +84 @@
     def by_location(self, location):
         "Find items by location"
-        return self.filter(Q(location=location) | Q(location__in=location.descendants()))
+        return self.filter(location__in=location.apparented())
            
     @staticmethod
@@ -93 +93 @@
         from telemeta.models import Location, LocationRelation
         l = self.values('location')
+        c = self.values('location__current_location')
         r = LocationRelation.objects.filter(location__in=l).values('ancestor_location')
-        return Location.objects.filter(Q(pk__in=l) | Q(pk__in=r))
+        return Location.objects.filter(Q(pk__in=l) | Q(pk__in=r) | Q(pk__in=c))
 
     def countries(self, group_by_continent=False):
@@ -102 +103 @@
             location = Location.objects.get(pk=id)
             for l in location.countries():
-                if not l in countries:
-                    countries.append(l)
+                c = l.current_location
+                if not c in countries:
+                    countries.append(c)
 
         if group_by_continent:
@@ -226 +228 @@
     def by_location(self, location):
         "Find collections by location"
-        return self.filter(Q(items__location=location) | Q(items__location__in=location.descendants())).distinct()
+        return self.filter(items__location__in=location.apparented()).distinct()
     
     def by_recording_year(self, from_year, to_year=None):
@@ -352 +354 @@
         return map
 
-    def current(self, is_current=True):
-        if is_current:
-            where = ["locations.id = locations.current_location_id"]
-        else:
-            where = ["locations.id <> locations.current_location_id"]
-
-        return self.extra(where = where);
+    def current(self):
+        return self.filter(id__in=self.values_list('current_location_id', flat=True)).distinct()
 
 class LocationManager(CoreManager):