|
| 1 | +# -*- coding: utf-8 -*- |
| 2 | +import io |
| 3 | +import os |
| 4 | + |
| 5 | +import obs |
| 6 | +from obs import ObsClient |
| 7 | + |
| 8 | +import smart_open |
| 9 | +from smart_open.obs import parse_uri |
| 10 | + |
| 11 | +_OBS_URL = os.environ.get('SO_OBS_URL') |
| 12 | + |
| 13 | +assert _OBS_URL is not None, 'please set the SO_OBS_URL environment variable' |
| 14 | + |
| 15 | +assert os.environ.get('OBS_ACCESS_KEY_ID') is not None, \ |
| 16 | + 'please set the OBS_ACCESS_KEY_ID environment variable' |
| 17 | +assert os.environ.get('OBS_SECRET_ACCESS_KEY') is not None, \ |
| 18 | + 'please set the OBS_SECRET_ACCESS_KEY environment variable' |
| 19 | + |
| 20 | + |
| 21 | +def _clear_bucket(obs_client: obs.ObsClient, bucket_id: str): |
| 22 | + objects = obs_client.listObjects(bucketName=bucket_id) |
| 23 | + for content in objects.body.contents: |
| 24 | + print(content.get('key')) |
| 25 | + _delete_object(obs_client=obs_client, |
| 26 | + bucket_id=bucket_id, |
| 27 | + object_key=content.get('key')) |
| 28 | + |
| 29 | + |
| 30 | +def _delete_object(obs_client: obs.ObsClient, bucket_id: str, object_key: str): |
| 31 | + try: |
| 32 | + resp = obs_client.deleteObject(bucketName=bucket_id, objectKey=object_key) |
| 33 | + if resp.status < 300: |
| 34 | + print('requestId:', resp.requestId) |
| 35 | + print('deleteMarker:', resp.body.deleteMarker) |
| 36 | + except Exception as ex: |
| 37 | + print(ex) |
| 38 | + |
| 39 | + |
| 40 | +def initialize_bucket(): |
| 41 | + parsed = parse_uri(_OBS_URL) |
| 42 | + server = f'https://{parsed.get("server")}' |
| 43 | + bucket_id = parsed.get('bucket_id') |
| 44 | + obs_client = ObsClient(server=server, security_provider_policy='ENV') |
| 45 | + _clear_bucket(obs_client=obs_client, bucket_id=bucket_id) |
| 46 | + |
| 47 | + |
| 48 | +def write_read(key, content, write_mode, read_mode, **kwargs): |
| 49 | + with smart_open.open(key, write_mode, **kwargs) as fout: |
| 50 | + fout.write(content) |
| 51 | + with smart_open.open(key, read_mode, **kwargs) as fin: |
| 52 | + return fin.read() |
| 53 | + |
| 54 | + |
| 55 | +def read_length_prefixed_messages(key, read_mode, **kwargs): |
| 56 | + result = io.BytesIO() |
| 57 | + |
| 58 | + with smart_open.open(key, read_mode, **kwargs) as fin: |
| 59 | + length_byte = fin.read(1) |
| 60 | + while len(length_byte): |
| 61 | + result.write(length_byte) |
| 62 | + msg = fin.read(ord(length_byte)) |
| 63 | + result.write(msg) |
| 64 | + length_byte = fin.read(1) |
| 65 | + return result.getvalue() |
| 66 | + |
| 67 | + |
| 68 | +def test_obs_readwrite_binary(benchmark): |
| 69 | + initialize_bucket() |
| 70 | + |
| 71 | + key = _OBS_URL + '/sanity.txt' |
| 72 | + binary = 'с гранатою в кармане, с чекою в руке'.encode() |
| 73 | + actual = benchmark(write_read, key, binary, 'wb', 'rb') |
| 74 | + assert actual == binary |
| 75 | + |
| 76 | + |
| 77 | +def test_obs_readwrite_binary_gzip(benchmark): |
| 78 | + initialize_bucket() |
| 79 | + |
| 80 | + key = _OBS_URL + '/sanity.txt.gz' |
| 81 | + binary = 'не чайки здесь запели на знакомом языке'.encode() |
| 82 | + actual = benchmark(write_read, key, binary, 'wb', 'rb') |
| 83 | + assert actual == binary |
| 84 | + |
| 85 | + |
| 86 | +def test_obs_performance(benchmark): |
| 87 | + initialize_bucket() |
| 88 | + |
| 89 | + one_megabyte = io.BytesIO() |
| 90 | + for _ in range(1024 * 128): |
| 91 | + one_megabyte.write(b'01234567') |
| 92 | + one_megabyte = one_megabyte.getvalue() |
| 93 | + |
| 94 | + key = _OBS_URL + '/performance.txt' |
| 95 | + actual = benchmark(write_read, key, one_megabyte, 'wb', 'rb') |
| 96 | + assert actual == one_megabyte |
| 97 | + |
| 98 | + |
| 99 | +def test_obs_performance_gz(benchmark): |
| 100 | + initialize_bucket() |
| 101 | + |
| 102 | + one_megabyte = io.BytesIO() |
| 103 | + for _ in range(1024 * 128): |
| 104 | + one_megabyte.write(b'01234567') |
| 105 | + one_megabyte = one_megabyte.getvalue() |
| 106 | + |
| 107 | + key = _OBS_URL + '/performance.txt.gz' |
| 108 | + actual = benchmark(write_read, key, one_megabyte, 'wb', 'rb') |
| 109 | + assert actual == one_megabyte |
| 110 | + |
| 111 | + |
| 112 | +def test_obs_performance_small_reads(benchmark): |
| 113 | + initialize_bucket() |
| 114 | + |
| 115 | + ONE_MIB = 1024 ** 2 |
| 116 | + one_megabyte_of_msgs = io.BytesIO() |
| 117 | + msg = b'\x0f' + b'0123456789abcde' # a length-prefixed "message" |
| 118 | + for _ in range(0, ONE_MIB, len(msg)): |
| 119 | + one_megabyte_of_msgs.write(msg) |
| 120 | + one_megabyte_of_msgs = one_megabyte_of_msgs.getvalue() |
| 121 | + |
| 122 | + key = _OBS_URL + '/many_reads_performance.bin' |
| 123 | + |
| 124 | + with smart_open.open(key, 'wb') as fout: |
| 125 | + fout.write(one_megabyte_of_msgs) |
| 126 | + |
| 127 | + actual = benchmark(read_length_prefixed_messages, key, 'rb', buffering=ONE_MIB) |
| 128 | + assert actual == one_megabyte_of_msgs |
0 commit comments