Merge branch 'master' of ssh://apples.lambdacomplex.org/git/disclosr
[disclosr.git] / documents / datagov-export-groups.py
blob:a/documents/datagov-export-groups.py -> blob:b/documents/datagov-export-groups.py
import ckanclient import ckanclient
import couchdb import couchdb
from ckanclient import CkanApiError from ckanclient import CkanApiError
import re import re
   
   
class LoaderError(Exception): class LoaderError(Exception):
pass pass
   
# Instantiate the CKAN client. # Instantiate the CKAN client.
#ckan = ckanclient.CkanClient(base_location='http://localhost:5000/api', api_key='b47b24cd-591d-40c1-8677-d73101d56d1b') #ckan = ckanclient.CkanClient(base_location='http://localhost:5000/api', api_key='b47b24cd-591d-40c1-8677-d73101d56d1b')
api_key = 'ff34526e-f794-4068-8235-fcbba38cd8bc' api_key = 'ff34526e-f794-4068-8235-fcbba38cd8bc'
ckan = ckanclient.CkanClient(base_location='http://data.disclosurelo.gs/api', server = 'data.disclosurelo.gs'
  api_key = 'c30eb6f5-0f90-47e0-bf05-9b1b4e3a461a'
  server = 'ckan.data.gov.au'
   
  ckan = ckanclient.CkanClient(base_location='http://' + server + '/api',
api_key=api_key) api_key=api_key)
couch = couchdb.Server('http://127.0.0.1:5984/') couch = couchdb.Server('http://127.0.0.1:5984/')
#couch = couchdb.Server('http://192.168.1.113:5984/') #couch = couchdb.Server('http://192.168.1.113:5984/')
   
# https://github.com/okfn/ckanext-importlib # https://github.com/okfn/ckanext-importlib
def munge(name): def munge(name):
# convert spaces to underscores # convert spaces to underscores
name = re.sub(' ', '_', name).lower() name = re.sub(' ', '_', name).lower()
# convert symbols to dashes # convert symbols to dashes
name = re.sub('[:]', '_-', name).lower() name = re.sub('[:]', '_-', name).lower()
name = re.sub('[/]', '-', name).lower() name = re.sub('[/]', '-', name).lower()
# take out not-allowed characters # take out not-allowed characters
name = re.sub('[^a-zA-Z0-9-_]', '', name).lower() name = re.sub('[^a-zA-Z0-9-_]', '', name).lower()
# remove double underscores # remove double underscores
name = re.sub('__', '_', name).lower() name = re.sub('__', '_', name).lower()
return name return name
   
   
def name_munge(input_name): def name_munge(input_name):
return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and')) return munge(input_name.replace(' ', '').replace('.', '_').replace('&', 'and'))
   
   
docsdb = couch['disclosr-documents'] docsdb = couch['disclosr-documents']
   
if __name__ == "__main__": if __name__ == "__main__":
groups = {} groups = {}
for doc in docsdb.view('app/datasetGroups'): for doc in docsdb.view('app/datasetGroups'):
group_name = doc.key group_name = doc.key
if group_name != "Not specified": if group_name != "Not specified":
pkg_name = filter(lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_', pkg_name = filter(lambda x: x in '0123456789abcdefghijklmnopqrstuvwxyz-_',
doc.value.replace("http://data.gov.au/dataset/", '').replace('/', '')[:100]); doc.value.replace("http://data.gov.au/dataset/", '').replace('/', '')[:100]);
if group_name in groups.keys(): if group_name in groups.keys():
groups[group_name] = list(set(groups[group_name] + [pkg_name])) groups[group_name] = list(set(groups[group_name] + [pkg_name]))
else: else:
groups[group_name] = [pkg_name] groups[group_name] = [pkg_name]
   
# add dataset to group(s) # add dataset to group(s)
for group_name in groups.keys(): for group_name in groups.keys():
if group_name != "Not specified": if group_name != "Not specified":
group_url = name_munge(group_name[:100]) group_url = name_munge(group_name[:100])
print group_name print group_name
print groups[group_name] print groups[group_name]
try: try:
# Update the group details # Update the group details
group_entity = ckan.group_entity_get(group_url) group_entity = ckan.group_entity_get(group_url)
print "group "+group_name+" exists" print "group "+group_name+" exists"
if 'packages' in group_entity.keys(): if 'packages' in group_entity.keys():
group_entity['packages'] = list(set(group_entity['packages'] + groups[group_name])) group_entity['packages'] = list(set(group_entity['packages'] + groups[group_name]))
else: else:
group_entity['packages'] = groups[group_name] group_entity['packages'] = groups[group_name]
ckan.group_entity_put(group_entity) ckan.group_entity_put(group_entity)
except CkanApiError, e: except CkanApiError, e:
if ckan.last_status == 404: if ckan.last_status == 404:
print "group "+group_name+" does not exist, creating" print "group "+group_name+" does not exist, creating"
group_entity = { group_entity = {
'name': group_url, 'name': group_url,
'title': group_name, 'title': group_name,
'description': group_name, 'description': group_name,
'packages': groups[group_name] 'packages': groups[group_name]
} }
#print group_entity #print group_entity
ckan.group_register_post(group_entity) ckan.group_register_post(group_entity)
elif ckan.last_status == 409: elif ckan.last_status == 409:
print "group already exists" print "group already exists"
else: else:
raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % ( raise LoaderError('Unexpected status %s adding to group under \'%s\': %r' % (
ckan.last_status, pkg_name, e.args)) ckan.last_status, pkg_name, e.args))