seperate out datagov group export
[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
@@ -1,1 +1,79 @@
+import ckanclient
+import couchdb
+from ckanclient import CkanApiError
+import re
 
+
+class LoaderError(Exception):
+    pass
+
+# Instantiate the CKAN client.
+#ckan = ckanclient.CkanClient(base_location='http://localhost:5000/api',    api_key='b47b24cd-591d-40c1-8677-d73101d56d1b')
+api_key = 'b3ab75e4-afbb-465b-a09d-8171c8c69a7a'
+ckan = ckanclient.CkanClient(base_location='http://data.disclosurelo.gs/api',
+                             api_key=api_key)
+couch = couchdb.Server('http://127.0.0.1:5984/')
+#couch = couchdb.Server('http://192.168.1.113:5984/')
+
+# https://github.com/okfn/ckanext-importlib
+def munge(name):
+    # convert spaces to underscores
+    name = re.sub(' ', '_', name).lower()
+    # convert symbols to dashes
+    name = re.sub('[:]', '_-', name).lower()
+    name = re.sub('[/]', '-', name).lower()
+    # take out not-allowed characters
+    name = re.sub('[^a-zA-Z0-9-_]', '', name).lower()
+    # remove double underscores
+    name = re.sub('__', '_', name).lower()
+    return name
+
+
+def name_munge(input_name):
+    return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and'))
+
+
+docsdb = couch['disclosr-documents']
+
+if __name__ == "__main__":
+    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]
+
+    # 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))
+