Experimental organization support
Former-commit-id: 0c86e9a2a6e066dc3f1550e7915892d9649ec125
--- a/documents/datagov-export.py
+++ b/documents/datagov-export.py
@@ -3,7 +3,7 @@
from ckanclient import CkanApiError
import re
import html2text # aaronsw :(
-import ckanapi
+import ckanapi # https://github.com/open-data/ckanapi
class LoaderError(Exception):
@@ -11,7 +11,7 @@
# 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'
+api_key = 'b3ab75e4-afbb-465b-a09d-8171c8c69a7a'
ckan = ckanclient.CkanClient(base_location='http://data.disclosurelo.gs/api',
api_key=api_key)
ckandirect = ckanapi.RemoteCKAN('http://data.disclosurelo.gs', api_key=api_key)
@@ -123,9 +123,34 @@
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 = filter( lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',doc.value['url'].replace("http://data.gov.au/dataset/",'').replace('/','')[:100]);
+ print pkg_name
+ #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')
+ else:
+ print "org found, adding dataset to "+org_name
+
+ org = ckandirect.action.organization_show(id=org_name)
+ # todo cache org names -> id mapping
tags = []
if doc.value['agencyID'] == "AGIMO":
if len(doc.value['metadata']["Keywords / Tags"]) > 0:
@@ -149,6 +174,8 @@
'maintainer': doc.value['metadata']["DCTERMS.Creator"],
'licence_id': get_licence_id(doc.value['metadata']['DCTERMS.License']),
'notes': html2text.html2text(doc.value['metadata']['Description']),
+ 'owner_org': org["result"]["id"]
+ #todo add missing key values like jurasdiction
}
if doc.value['agencyID'] == "qld":
package_entity = doc.value['metadata']
@@ -164,76 +191,6 @@
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
-
- group_name = name_munge(doc.value['metadata']["Agency"][:100])
- try:
- print ckan.group_entity_get(group_name)
-
- # Update the group details
- group_entity = ckan.last_message
- print "group exists"
- if 'packages' in group_entity.keys():
- group_entity['packages'] = list(set(group_entity['packages'] + [pkg_name]))
- else:
- group_entity['packages'] = [pkg_name]
- ckan.group_entity_put(group_entity)
- except CkanApiError, e:
- if ckan.last_status == 404:
- print "group does not exist, creating"
- group_entity = {
- 'name': group_name,
- 'title': doc.value['metadata']["Agency"],
- 'description': doc.value['metadata']["Agency"],
- 'packages': [pkg_name],
- # 'type': "organization" # not allowed via API, use database query
- # update "group" set type = 'organization';
- }
- 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))
-
- # #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: