def fetch(resource_id, **kwargs):
"""Downloads a filestore resource"""
verbose = not kwargs['quiet']
filepath = kwargs['destination']
name_from_id = kwargs.get('name_from_id')
chunksize = kwargs.get('chunksize_bytes')
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
ckan = CKAN(**ckan_kwargs)
try:
r = ckan.fetch_resource(resource_id)
except api.NotAuthorized as err:
sys.exit('ERROR: %s\n' % str(err))
else:
fkwargs = {
'headers': r.headers,
'name_from_id': name_from_id,
'resource_id': resource_id}
filepath = tup.make_filepath(filepath, **fkwargs)
tio.write(filepath, r.iter_content, chunksize=chunksize)
# save encoding to extended attributes
x = xattr(filepath)
if verbose and r.encoding:
print('saving encoding %s to extended attributes' % r.encoding)
if r.encoding:
x['com.ckanny.encoding'] = r.encoding
print(filepath)
def upload(source, resource_id=None, **kwargs):
"""Uploads a file to a datastore table"""
verbose = not kwargs['quiet']
resource_id = resource_id or p.splitext(p.basename(source))[0]
if '.' in resource_id:
resource_id = resource_id.split('.')[0]
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
if verbose:
print('Uploading %s to datastore resource %s...' %
(source, resource_id))
# read encoding from extended attributes
x = xattr(source)
try:
kwargs['encoding'] = x.get('com.ckanny.encoding')
except IOError:
pass
if verbose and kwargs['encoding']:
print('Using encoding %s' % kwargs['encoding'])
ckan = CKAN(**ckan_kwargs)
if ckan.update_datastore(resource_id, source, **kwargs):
print('Success! Resource %s uploaded.' % resource_id)
else:
sys.exit('ERROR: resource %s not uploaded.' % resource_id)
def upload(source, resource_id=None, package_id=None, **kwargs):
"""Updates the filestore of an existing resource or creates a new one"""
verbose = not kwargs['quiet']
resource_id = resource_id or p.splitext(p.basename(source))[0]
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
if package_id and verbose:
print(
'Creating filestore resource %s in dataset %s...' %
(source, package_id))
elif verbose:
print(
'Uploading %s to filestore resource %s...' % (source, resource_id))
ckan = CKAN(**ckan_kwargs)
resource_kwargs = {
'url' if 'http' in source else 'filepath': source,
'name': kwargs.get('name')
}
if package_id:
resource = ckan.create_resource(package_id, **resource_kwargs)
else:
resource = ckan.update_filestore(resource_id, **resource_kwargs)
if package_id and resource and verbose:
infix = '%s ' % resource['id'] if resource.get('id') else ''
print('Success! Resource %screated.' % infix)
elif resource and verbose:
print('Success! Resource %s updated.' % resource_id)
elif not resource:
sys.exit('Error uploading file!')
def update(endpoint, **kwargs):
""" Updates the database
Args:
endpoint (str): The api resource url.
kwargs (dict): passed to CKAN constructor.
Kwargs:
chunk_size (int): Number of rows to process at a time (default: All).
row_limit (int): Total number of rows to process (default: All).
err_limit (int): Number of errors to encounter before failing
(default: Inf).
Returns:
(dict): Update details
"""
start = timer()
pid = kwargs.pop('pid', None)
chunk_size = kwargs.pop('chunk_size', 0)
row_limit = kwargs.pop('row_limit', None)
err_limit = kwargs.pop('err_limit', None)
rows = 0
ckan = CKAN(**kwargs)
if pid:
pids = [pid]
else:
org_show = partial(ckan.organization_show, include_datasets=True)
orgs_basic = ckan.organization_list(permission='read')
org_ids = it.imap(itemgetter('id'), orgs_basic)
orgs = (org_show(id=org_id) for org_id in org_ids)
package_lists = it.imap(itemgetter('packages'), orgs)
pid_getter = partial(map, itemgetter('id'))
pids = it.chain.from_iterable(it.imap(pid_getter, package_lists))
data = gen_data(ckan, pids, kwargs.get('mock_freq'))
errors = {}
for records in tup.chunk(data, min(row_limit or 'inf', chunk_size)):
rs = map(partial(patch_or_post, endpoint), records)
rows += len(filter(lambda r: r.ok, rs))
ids = map(itemgetter('dataset_id'), records)
errors.update(dict((k, r.json()) for k, r in zip(ids, rs) if not r.ok))
if row_limit and rows >= row_limit:
break
if err_limit and len(errors) >= err_limit:
raise Exception(errors)
elapsed_time = ' ,'.join(fmt_elapsed(timer() - start))
return {'rows_added': rows, 'errors': errors, 'elapsed_time': elapsed_time}
def migrate(resource_id, **kwargs):
"""Copies a filestore resource from one ckan instance to another"""
src_remote, dest_remote = kwargs['src_remote'], kwargs['dest_remote']
if src_remote == dest_remote:
msg = (
'ERROR: `dest-remote` of %s is the same as `src-remote` of %s.\n'
'The dest and src remotes must be different.\n' % (src_remote,
dest_remote))
sys.exit(msg)
verbose = not kwargs['quiet']
chunksize = kwargs['chunksize_bytes']
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
src_ckan = CKAN(remote=src_remote, **ckan_kwargs)
dest_ckan = CKAN(remote=dest_remote, **ckan_kwargs)
try:
r = src_ckan.fetch_resource(resource_id)
filepath = NamedTemporaryFile(delete=False).name
except api.NotAuthorized as err:
sys.exit('ERROR: %s\n' % str(err))
except Exception as err:
sys.exit('ERROR: %s\n' % str(err))
else:
tio.write(filepath, r.raw.read(), chunksize=chunksize)
resource = dest_ckan.update_filestore(resource_id, filepath=filepath)
if resource and verbose:
print('Success! Resource %s updated.' % resource_id)
elif not resource:
sys.exit('Error uploading file!')
finally:
if verbose:
print('Removing tempfile...')
unlink(filepath)
def status():
""" Displays the current status
"""
kwargs = {k: parse(v) for k, v in request.args.to_dict().items()}
ckan = CKAN(**kwargs)
resp = {
'online': True,
'message': 'Service for checking and updating HDX dataset ages.',
'CKAN_instance': ckan.address,
'version': __version__,
'repository': c.REPO
}
return jsonify(**resp)
def update(pid, **kwargs):
"""Updates a package (aka dataset)"""
kw = ft.Objectify(kwargs, type='dataset')
verbose = not kw.quiet
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
ckan = CKAN(**ckan_kwargs)
licenses = it.imap(itemgetter('id'), ckan.license_list())
groups = ckan.group_list()
raw_tags = filter(None, kw.tags.split(',')) if kw.tags else []
tags = [{'state': 'active', 'name': t} for t in raw_tags]
if kw.start:
start = parse(str(kw.start)).strftime('%m/%d/%Y')
else:
date = None
if kw.start and kw.end:
date = '%s-%s' % (start, parse(str(kw.end)).strftime('%m/%d/%Y'))
elif kw.start:
date = start
if kw.location and kw.location in set(groups):
group_list = [{'name': kw.location}]
elif kw.location:
sys.exit('group name: %s not found!' % kw.location)
else:
group_list = []
if kw.license_id and kw.license_id not in set(licenses):
sys.exit('license id: %s not found!' % kw.license_id)
package_kwargs = {
'title': kw.title,
'name': kw.name,
'license_id': kw.license_id,
'dataset_source': kw.source,
'notes': kw.description or kw.title,
'type': kw.type,
'tags': tags,
'groups': group_list,
'dataset_date': date,
'caveats': kw.caveats,
'methodology': methods.get(kw.methodology, 'Other'),
'methodology_other': methods.get(kw.methodology) or kw.methodology,
}
try:
old_package = ckan.package_show(id=pid)
except api.ValidationError as e:
exit(e)
if any(package_kwargs.values()):
# combine keys by returning the last non-empty result
pred = lambda key: True
last = lambda pair: filter(None, pair)[-1] if any(pair) else None
records = [old_package, package_kwargs]
new_kwargs = pr.merge(records, pred=pred, op=last)
if verbose:
print('Submitting your package request.')
pprint(new_kwargs)
print('\n')
package = ckan.package_update(**new_kwargs)
else:
package = old_package
if kw.private:
org = package['organization']
ckan.package_privatize(org_id=org['id'], datasets=[package['id']])
print(package['id'])
print('\n')
def create(org_id, **kwargs):
"""Creates a package (aka dataset)"""
kw = ft.Objectify(kwargs, type='dataset')
verbose = not kw.quiet
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
ckan = CKAN(**ckan_kwargs)
licenses = it.imap(itemgetter('id'), ckan.license_list())
orgs = ckan.organization_list()
org_ids = it.imap(itemgetter('id'), orgs)
org_names = it.imap(itemgetter('name'), orgs)
groups = ckan.group_list()
name = kw.name or slugify(kw.title)
raw_tags = filter(None, kw.tags.split(','))
tags = [{'state': 'active', 'name': t} for t in raw_tags]
if kw.start:
start = parse(str(kw.start)).strftime('%m/%d/%Y')
else:
date = None
if kw.start and kw.end:
date = '%s-%s' % (start, parse(str(kw.end)).strftime('%m/%d/%Y'))
elif kw.start:
date = start
if kw.location in set(groups):
group_list = [{'name': kw.location}]
elif kw.location:
sys.exit('group name: %s not found!' % kw.location)
else:
group_list = []
if org_id not in set(it.chain(org_ids, org_names)):
sys.exit('organization id: %s not found!' % org_id)
if kw.license_id not in set(licenses):
sys.exit('license id: %s not found!' % kw.license_id)
files = filter(None, kw.files.split(','))
names = filter(None, kw.names.split(','))
resource_list = list(it.starmap(make_rkwargs, zip(files, names))) or []
package_kwargs = {
'title': kw.title,
'name': name,
'license_id': kw.license_id,
'owner_org': org_id,
'dataset_source': kw.source,
'notes': kw.description or kw.title,
'type': kw.type,
'tags': tags,
'resources': resource_list,
'package_creator': ckan.user['name'],
'groups': group_list,
'dataset_date': date,
'caveats': kw.caveats,
'methodology': methods.get(kw.methodology, 'Other'),
'methodology_other': methods.get(kw.methodology) or kw.methodology,
}
if verbose:
print('Submitting your package request.')
pprint(package_kwargs)
print('\n')
try:
package = ckan.package_create(**package_kwargs)
except api.ValidationError as e:
exit(e)
if kw.private:
org = package['organization']
ckan.package_privatize(org_id=org['id'], datasets=[package['id']])
if verbose:
print('Your package response.')
pprint(package)
print('\n')
print(package['id'])
print('\n')
def delete(resource_id, **kwargs):
"""Deletes a datastore table"""
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
ckan = CKAN(**ckan_kwargs)
ckan.delete_table(resource_id, filters=kwargs.get('filters'))
def update(resource_id, force=None, **kwargs):
"""Updates a datastore table based on the current filestore resource"""
verbose = not kwargs.get('quiet')
chunk_bytes = kwargs.get('chunk_bytes', api.CHUNKSIZE_BYTES)
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
hash_kwargs = {'chunksize': chunk_bytes, 'verbose': verbose}
ckan = CKAN(**ckan_kwargs)
try:
r = ckan.fetch_resource(resource_id)
except (api.NotFound, api.NotAuthorized) as err:
sys.exit('ERROR: %s\n' % str(err))
else:
f = SpooledTemporaryFile(suffix='.xlsx', mode='r+b')
write_kwargs = {
'length': r.headers.get('content-length'),
'chunksize': chunk_bytes
}
tio.write(f, r.iter_content, **write_kwargs)
try:
old_hash = ckan.get_hash(resource_id)
except api.NotFound as err:
item = err.args[0]['item']
if item == 'package':
orgs = ckan.organization_list(permission='admin_group')
owner_org = (
o['id'] for o in orgs
if o['display_name'] == kwargs['hash_group']).next()
package_kwargs = {
'name': kwargs['hash_table'],
'owner_org': owner_org,
'package_creator': 'Hash Table',
'dataset_source': 'Multiple sources',
'notes': 'Datastore resource hash table'
}
ckan.hash_table_pack = ckan.package_create(**package_kwargs)
if item in {'package', 'resource'}:
fileobj = StringIO('datastore_id,hash\n')
create_kwargs = {'fileobj': fileobj, 'name': api.DEF_HASH_RES}
table = kwargs['hash_table']
resource = ckan.create_resource(table, **create_kwargs)
ckan.hash_table_id = resource['id']
ckan.create_hash_table(verbose)
old_hash = ckan.get_hash(resource_id)
new_hash = tio.hash_file(f, **hash_kwargs)
changed = new_hash != old_hash if old_hash else True
if verbose:
print(get_message(changed, force))
if not (changed or force):
sys.exit(0)
kwargs['encoding'] = r.encoding
kwargs['content_type'] = r.headers['content-type']
updated = ckan.update_datastore(resource_id, f, **kwargs)
if updated and verbose:
print('Success! Resource %s updated.' % resource_id)
if updated and changed:
ckan.update_hash_table(resource_id, new_hash, verbose)
elif not updated:
sys.exit('ERROR: resource %s not updated.' % resource_id)
def update(three_dub_id, topline_id=None, **kwargs):
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
ckan = CKAN(**ckan_kwargs)
ds.update(topline_id, ckan=ckan) if topline_id else None
ds.update(three_dub_id, ckan=ckan)
def customize(org_id, **kwargs):
"""Introspects custom organization values"""
verbose = not kwargs['quiet']
ckan_kwargs = {k: v for k, v in kwargs.items() if k in api.CKAN_KEYS}
image_sq = kwargs.get('image_sq')
image_rect = kwargs.get('image_rect')
sanitize = kwargs.get('sanitize')
three_dub_id = kwargs.get('3w')
geojson_id = kwargs.get('geojson')
topline_id = kwargs.get('topline')
ckan = CKAN(**ckan_kwargs)
organization = ckan.organization_show(id=org_id, include_datasets=True)
org_packages = organization['packages']
hdx = ckan.organization_show(id='hdx', include_datasets=True)
extras = {e['key']: e['value'] for e in organization['extras']}
if three_dub_id:
three_dub_set_id = ckan.get_package_id(three_dub_id)
else:
ids = ckan.find_ids(org_packages, pnamed='3w', ptagged='3w')
three_dub_set_id = ids['pname']
three_dub_id = ids['rid']
if not three_dub_id:
sys.exit(1)
if not topline_id:
topline_id = ckan.find_ids(org_packages, pnamed='topline')['rid']
if geojson_id:
geojson_set_id = ckan.get_package_id(geojson_id)
else:
country = org_id.split('-')[1]
hkwargs = {'pnamed': 'json-repository', 'rnamed': country}
ids = ckan.find_ids(hdx['packages'], **hkwargs)
geojson_set_id = ids['pname']
geojson_id = ids['rid']
viz_url = '%s/dataset/%s' % (kwargs['remote'], three_dub_set_id)
three_dub_r = ckan.fetch_resource(three_dub_id)
_fields = three_dub_r.iter_lines().next().split(',')
three_dub_fields = tup.underscorify(_fields) if sanitize else _fields
if geojson_id:
geojson_r = ckan.fetch_resource(geojson_id)
geojson_fields = geojson_r.json()['features'][0]['properties'].keys()
else:
geojson_fields = []
if verbose:
print('3w fields:')
pprint(three_dub_fields)
print('geojson fields:')
pprint(geojson_fields)
def_where = tup.find(three_dub_fields, geojson_fields) or ''
who_column = find_field(three_dub_fields, 'who', **kwargs)
what_column = find_field(three_dub_fields, 'what', **kwargs)
where_column = find_field(three_dub_fields, 'where', def_where, **kwargs)
where_column_2 = find_field(geojson_fields, 'where', def_where, **kwargs)
name_column = kwargs.get('where') or def_where
if 'http' not in image_sq:
gdocs = 'https://docs.google.com'
image_sq = '%s/uc?id=%s&export=download' % (gdocs, image_sq)
if 'http' not in image_rect:
gdocs = 'https://docs.google.com'
image_rect = '%s/uc?id=%s&export=download' % (gdocs, image_rect)
data = {
'name':
org_id,
'resource_id_1':
three_dub_id,
'resource_id_2':
geojson_id,
'topline_resource':
topline_id,
'datatype_1':
kwargs.get('datatype_1') or 'datastore',
'datatype_2':
kwargs.get('datatype_2') or 'filestore',
'org_url':
extras['org_url'],
'description':
organization['description'],
'title':
organization['title'],
'image_sq':
image_sq,
'image_rect':
image_rect,
'highlight_color':
kwargs.get('color'),
'dataset_id_1':
three_dub_set_id,
'dataset_id_2':
geojson_set_id,
'who_column':
deref_field(three_dub_fields, who_column),
'what_column':
deref_field(three_dub_fields, what_column),
'where_column':
deref_field(three_dub_fields, where_column),
'where_column_2':
deref_field(geojson_fields, where_column_2),
'map_district_name_column':
deref_field(geojson_fields, name_column),
'viz_data_link_url':
viz_url,
'visualization_select':
kwargs.get('viz_type', '3W-dashboard'),
'viz_title':
kwargs.get('viz_title', "Who's doing what and where?"),
'colors': [
'#c6d5ed', '#95b5df', '#659ad2', '#026bb5', '#659ad2', '#213b68',
'#101d4e', '#000035'
],
'use_org_color':
True,
'modified_at':
int(time()),
}
control_sheet_data = [data[k] for k in control_sheet_keys]
if verbose:
print('\nCustom pages control sheet data:')
print(control_sheet_data)
return control_sheet_data