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 test_write(self):
url = "http://google.com"
body = '<!doctype html><html itemtype="http://schema.org/page">'
content = StringIO("Iñtërnâtiônàližætiøn")
nt.assert_equal(20, io.write(StringIO(), content))
content.seek(0)
nt.assert_equal(28, io.write(TemporaryFile(), content))
content = io.IterStringIO(iter("Hello World"))
nt.assert_equal(12, io.write(TemporaryFile(), content, chunksize=2))
responses.add(responses.GET, url=url, body=body)
r = requests.get(url, stream=True)
nt.assert_equal(55, io.write(TemporaryFile(), r.iter_content))
def run():
if args.debug:
pprint(dict(args._get_kwargs()))
exit(0)
if args.version:
from . import __version__ as version
print('v%s' % version)
exit(0)
mapping = import_module('csv2ofx.mappings.%s' % args.mapping).mapping
okwargs = {
'def_type': args.account_type or 'Bank' if args.qif else 'CHECKING',
'split_header': args.split,
'start': parse(args.start),
'end': parse(args.end)
}
cont = QIF(mapping, **okwargs) if args.qif else OFX(mapping, **okwargs)
records = read_csv(args.source, has_header=cont.has_header)
groups = cont.gen_groups(records, args.chunksize)
trxns = cont.gen_trxns(groups, args.collapse)
cleaned_trxns = cont.clean_trxns(trxns)
data = utils.gen_data(cleaned_trxns)
body = cont.gen_body(data)
try:
mtime = p.getmtime(args.source.name)
except AttributeError:
mtime = time.time()
server_date = dt.fromtimestamp(mtime)
header = cont.header(date=server_date, language=args.language)
footer = cont.footer(date=server_date)
content = it.chain([header, body, footer])
kwargs = {'overwrite': args.overwrite, 'chunksize': args.chunksize}
try:
write(args.dest, IterStringIO(content), **kwargs)
except TypeError as e:
msg = str(e)
if not args.collapse:
msg += 'Try again with `-c` option.'
exit(msg)
def run():
if args.debug:
pprint(dict(args._get_kwargs()))
exit(0)
if args.version:
from . import __version__ as version
print('v%s' % version)
exit(0)
mapping = import_module('csv2ofx.mappings.%s' % args.mapping).mapping
okwargs = {
'def_type': args.account_type or 'Bank' if args.qif else 'CHECKING',
'split_header': args.split,
'start': parse(args.start),
'end': parse(args.end)
}
cont = QIF(mapping, **okwargs) if args.qif else OFX(mapping, **okwargs)
records = read_csv(args.source, has_header=cont.has_header)
groups = cont.gen_groups(records, args.chunksize)
trxns = cont.gen_trxns(groups, args.collapse)
cleaned_trxns = cont.clean_trxns(trxns)
data = utils.gen_data(cleaned_trxns)
body = cont.gen_body(data)
try:
mtime = p.getmtime(args.source.name)
except AttributeError:
mtime = time.time()
server_date = dt.fromtimestamp(mtime)
header = cont.header(date=server_date, language=args.language)
footer = cont.footer(date=server_date)
content = it.chain([header, body, footer])
kwargs = {'overwrite': args.overwrite, 'chunksize': args.chunksize}
try:
write(args.dest, IterStringIO(content), **kwargs)
except TypeError as e:
msg = str(e)
if not args.collapse:
msg += 'Try again with `-c` option.'
exit(msg)
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 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(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)
