cleanse data group/org name for datagov
[disclosr.git] / documents / datagov-export-groups.py
blob:a/documents/datagov-export-groups.py -> blob:b/documents/datagov-export-groups.py
--- a/documents/datagov-export-groups.py
+++ b/documents/datagov-export-groups.py
@@ -39,41 +39,43 @@
     groups = {}
     for doc in docsdb.view('app/datasetGroups'):
             group_name = doc.key
-            pkg_name = filter(lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',
-                              doc.value.replace("http://data.gov.au/dataset/", '').replace('/', '')[:100]);
-            if group_name in groups.keys():
-                groups[group_name] = list(set(groups[group_name] + [pkg_name]))
-            else:
-                groups[group_name] = [pkg_name]
+            if group_name != "Not specified":
+                pkg_name = filter(lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',
+                                  doc.value.replace("http://data.gov.au/dataset/", '').replace('/', '')[:100]);
+                if group_name in groups.keys():
+                    groups[group_name] = list(set(groups[group_name] + [pkg_name]))
+                else:
+                    groups[group_name] = [pkg_name]
 
     # add dataset to group(s)
     for group_name in groups.keys():
-        group_url = name_munge(group_name[:100])
-        print group_name
-        print groups[group_name]
-        try:
-            # Update the group details
-            group_entity = ckan.group_entity_get(group_url)
-            print "group "+group_name+" exists"
-            if 'packages' in group_entity.keys():
-                group_entity['packages'] = list(set(group_entity['packages'] + groups[group_name]))
-            else:
-                group_entity['packages'] = groups[group_name]
-            ckan.group_entity_put(group_entity)
-        except CkanApiError, e:
-            if ckan.last_status == 404:
-                print "group "+group_name+" does not exist, creating"
-                group_entity = {
-                    'name': group_url,
-                    'title': group_name,
-                    'description': group_name,
-                    'packages': groups[group_name]
-                }
-                #print group_entity
-                ckan.group_register_post(group_entity)
-            elif ckan.last_status == 409:
-                print "group already exists"
-            else:
-                raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % (
-                    ckan.last_status, pkg_name, e.args))
+        if group_name != "Not specified":
+            group_url = name_munge(group_name[:100])
+            print group_name
+            print groups[group_name]
+            try:
+                # Update the group details
+                group_entity = ckan.group_entity_get(group_url)
+                print "group "+group_name+" exists"
+                if 'packages' in group_entity.keys():
+                    group_entity['packages'] = list(set(group_entity['packages'] + groups[group_name]))
+                else:
+                    group_entity['packages'] = groups[group_name]
+                ckan.group_entity_put(group_entity)
+            except CkanApiError, e:
+                if ckan.last_status == 404:
+                    print "group "+group_name+" does not exist, creating"
+                    group_entity = {
+                        'name': group_url,
+                        'title': group_name,
+                        'description': group_name,
+                        'packages': groups[group_name]
+                    }
+                    #print group_entity
+                    ckan.group_register_post(group_entity)
+                elif ckan.last_status == 409:
+                    print "group already exists"
+                else:
+                    raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % (
+                        ckan.last_status, pkg_name, e.args))