enforce valid package url
enforce valid package url


Former-commit-id: 26accf46eff2958223ffae9a1234d4291379bce4

--- a/documents/datagov-export.py
+++ b/documents/datagov-export.py
@@ -3,6 +3,7 @@
 from ckanclient import CkanApiError
 import re
 import html2text # aaronsw :(
+import ckanapi
 
 
 class LoaderError(Exception):
@@ -10,8 +11,10 @@
 
 # Instantiate the CKAN client.
 #ckan = ckanclient.CkanClient(base_location='http://localhost:5000/api',    api_key='b47b24cd-591d-40c1-8677-d73101d56d1b')
+api_key = 'aeab80a6-0a58-4026-96a4-c07e57f9c434'
 ckan = ckanclient.CkanClient(base_location='http://data.disclosurelo.gs/api',
-    api_key='482a9dd2-a976-4adf-ac77-d71d92a98a52')
+    api_key=api_key)
+ckandirect = ckanapi.RemoteCKAN('http://data.disclosurelo.gs', api_key=api_key)
 couch = couchdb.Server('http://127.0.0.1:5984/')
 #couch = couchdb.Server('http://192.168.1.113:5984/')
 
@@ -88,9 +91,10 @@
     name = re.sub('__', '_', name).lower()
     return name
 
-
+#todo "{'name': [u'Url must be purely lowercase alphanumeric (ascii) characters and these symbols: -_']}"
+# http://data.gov.au/dataset/australian-domestic-regional-and-international-airline-activity-%E2%80%93-time-series/
 def name_munge(input_name):
-    return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and'))
+    return  munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and'))
     #[:100]
     #return input_name.replace(' ', '').replace('.', '_').replace('&', 'and')
 
@@ -116,11 +120,12 @@
 docsdb = couch['disclosr-documents']
 
 if __name__ == "__main__":
+    orgs_list = []
     for doc in docsdb.view('app/datasets'):
         print doc.id
         if doc.value['url'] != "http://data.gov.au/data/" and doc.value['agencyID'] != "qld":
             # Collect the package metadata.
-            pkg_name = doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/',''); 
+            pkg_name = filter( lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/','')[:100]);
             tags = []
             if doc.value['agencyID'] == "AGIMO":
                 if len(doc.value['metadata']["Keywords / Tags"]) > 0:
@@ -134,7 +139,7 @@
                     else:
                         tags = tags + [doc.value['metadata']['data.gov.au Category']]
                 tags = [re.sub('[^a-zA-Z0-9-_.]', '', tag.replace('&', 'and')).lower() for tag in tags if tag]
-                print tags
+                #print tags
                 package_entity = {
                     'name': pkg_name,
                     'title': doc.value['metadata']['DCTERMS.Title'],
@@ -149,15 +154,16 @@
                 package_entity = doc.value['metadata']
 
             try:
-                print package_entity
+                #print package_entity
                 ckan.package_register_post(package_entity)
             except CkanApiError, e:
-                if ckan.last_status == 409:
+                if ckan.last_message == "{\"name\": [\"That URL is already in use.\"]}":
                     print "package already exists"
                 else:
+                    print ckan.last_message
                     raise LoaderError('Unexpected status %s checking for package under \'%s\': %r' % (
                         ckan.last_status, pkg_name, e.args))
-
+            pkg = ckan.package_entity_get(pkg_name)
 
             #add to group
 
@@ -191,9 +197,47 @@
                 else:
                     raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % (
                         ckan.last_status, pkg_name, e.args))
+
+            # #add to or create organization using direct API
+            # org_name = name_munge(doc.value['metadata']["Agency"][:100])
+            # if org_name not in orgs_list:
+            #     orgs_list = ckandirect.action.organization_list()['result']
+            #     print orgs_list
+            #     if org_name not in orgs_list:
+            #         try:
+            #             print "org not found, creating "+org_name
+            #             ckandirect.action.organization_create(name = org_name,   title= doc.value['metadata']["Agency"],
+            #                                                           description= doc.value['metadata']["Agency"])
+            #             orgs_list.append(org_name)
+            #         except ckanapi.ValidationError, e:
+            #             print e
+            #             raise LoaderError('Unexpected status')
+            #
+            # try:
+            #     print "linking package to "+org_name
+            #
+            #    #brute force
+            #     org = ckandirect.action.organization_show(id=org_name)
+            #     #print org["result"]["packages"]
+            #     #print pkg
+            #     org["result"]["packages"].append({"id":pkg['id'], "name": pkg_name})
+            #     #print org["result"]
+            #     ckandirect.call_action(action="organization_update", data_dict=org["result"])
+            #
+            #     # if only this worked
+            #     print ckandirect.action.package_owner_org_update(id=pkg['id'],organization_id=org["result"]["id"])
+            #
+            # except ckanapi.ValidationError, e:
+            #     print e
+            #     raise LoaderError('Unexpected status')
+            # except ckanapi.CKANAPIError, e:
+            #     print "lol"
+
+
+
             if 'Download' in doc.value['metadata'].keys():
                 try:
-                    pkg = ckan.package_entity_get(pkg_name)
+
                     resources = pkg.get('resources', [])
                     if len(resources) < len(doc.value['metadata']['Download']):
                         for resource in doc.value['metadata']['Download']: