def test_read_csv_from_file_arg0(tmpdir):
fs, filename, services = csv_tester.get_services_for_reader(tmpdir)
with CapturingNodeExecutionContext(
CsvReader(path=filename,
delimiter=',',
ioformat=settings.IOFORMAT_ARG0),
services=services,
) as context:
context.write(BEGIN, Bag(), END)
context.step()
assert len(context.send.mock_calls) == 2
args0, kwargs0 = context.send.call_args_list[0]
assert len(args0) == 1 and not len(kwargs0)
args1, kwargs1 = context.send.call_args_list[1]
assert len(args1) == 1 and not len(kwargs1)
assert args0[0].args[0] == {
'a': 'a foo',
'b': 'b foo',
'c': 'c foo',
}
assert args1[0].args[0] == {
'a': 'a bar',
'b': 'b bar',
'c': 'c bar',
}
def test_read_csv_from_file_kwargs(tmpdir):
fs, filename, services = csv_tester.get_services_for_reader(tmpdir)
with CapturingNodeExecutionContext(
CsvReader(path=filename, delimiter=','),
services=services,
) as context:
context.write(BEGIN, Bag(), END)
context.step()
assert len(context.send.mock_calls) == 2
args0, kwargs0 = context.send.call_args_list[0]
assert len(args0) == 1 and not len(kwargs0)
args1, kwargs1 = context.send.call_args_list[1]
assert len(args1) == 1 and not len(kwargs1)
_args, _kwargs = args0[0].get()
assert not len(_args) and _kwargs == {
'a': 'a foo',
'b': 'b foo',
'c': 'c foo',
}
_args, _kwargs = args1[0].get()
assert not len(_args) and _kwargs == {
'a': 'a bar',
'b': 'b bar',
'c': 'c bar',
}
def test_read_csv_from_file(tmpdir):
fs, filename = open_fs(tmpdir), 'input.csv'
fs.open(filename, 'w').write('a,b,c\na foo,b foo,c foo\na bar,b bar,c bar')
reader = CsvReader(path=filename, delimiter=',')
context = CapturingNodeExecutionContext(reader, services={'fs': fs})
context.start()
context.write(BEGIN, Bag(), END)
context.step()
context.stop()
assert len(context.send.mock_calls) == 2
args0, kwargs0 = context.send.call_args_list[0]
assert len(args0) == 1 and not len(kwargs0)
args1, kwargs1 = context.send.call_args_list[1]
assert len(args1) == 1 and not len(kwargs1)
assert args0[0].args[0] == {
'a': 'a foo',
'b': 'b foo',
'c': 'c foo',
}
assert args1[0].args[0] == {
'a': 'a bar',
'b': 'b bar',
'c': 'c bar',
}
def test_read_csv_from_file(tmpdir):
file = tmpdir.join('input.csv')
file.write('a,b,c\na foo,b foo,c foo\na bar,b bar,c bar')
reader = CsvReader(path=str(file), delimiter=',')
context = CapturingNodeExecutionContext(reader, None)
context.start()
context.recv(BEGIN, Bag(), END)
context.step()
context.stop()
assert len(context.send.mock_calls) == 2
args0, kwargs0 = context.send.call_args_list[0]
assert len(args0) == 1 and not len(kwargs0)
args1, kwargs1 = context.send.call_args_list[1]
assert len(args1) == 1 and not len(kwargs1)
assert args0[0].args[0] == {
'a': 'a foo',
'b': 'b foo',
'c': 'c foo',
}
assert args1[0].args[0] == {
'a': 'a bar',
'b': 'b bar',
'c': 'c bar',
}
def test_read_csv_from_file_kwargs(tmpdir):
fs, filename, services = csv_tester.get_services_for_reader(tmpdir)
with BufferingNodeExecutionContext(CsvReader(filename, **defaults),
services=services) as context:
context.write_sync(EMPTY)
assert context.get_buffer_args_as_dicts() == [{
'a': 'a foo',
'b': 'b foo',
'c': 'c foo',
}, {
'a': 'a bar',
'b': 'b bar',
'c': 'c bar',
}]
def test_read_csv_from_file_kwargs(tmpdir):
fs, filename, services = csv_tester.get_services_for_reader(tmpdir)
with BufferingNodeExecutionContext(CsvReader(filename, **defaults),
services=services) as context:
context.write_sync(EMPTY)
assert context.get_buffer_args_as_dicts() == [
{
"a": "a foo",
"b": "b foo",
"c": "c foo"
},
{
"a": "a bar",
"b": "b bar",
"c": "c bar"
},
]
from bonobo import CsvReader, Graph, get_examples_path
graph = Graph(
CsvReader(path=get_examples_path('datasets/coffeeshops.txt')),
print,
)
if __name__ == '__main__':
import bonobo
bonobo.run(graph)