def check_transcoding(data, src_encoding, dest_encoding, chunk_sizes):
chunk_sizes = iter(chunk_sizes)
stream = pa.transcoding_input_stream(
pa.BufferReader(data.encode(src_encoding)),
src_encoding, dest_encoding)
out = []
while True:
buf = stream.read(next(chunk_sizes))
out.append(buf)
if not buf:
break
out = b''.join(out)
assert out.decode(dest_encoding) == data
def test_transcoding_no_ops(src_encoding, dest_encoding):
# No indirection is wasted when a trivial transcoding is requested
stream = pa.BufferReader(b"abc123")
assert pa.transcoding_input_stream(
stream, src_encoding, dest_encoding) is stream
def test_transcoding_decoding_error(src_encoding, dest_encoding):
# The given bytestring is not valid in the source encoding
stream = pa.transcoding_input_stream(pa.BufferReader(b"\xff\xff\xff\xff"),
src_encoding, dest_encoding)
with pytest.raises(UnicodeError):
stream.read(1)