diff --git a/swh/core/api/classes.py b/swh/core/api/classes.py new file mode 100644 index 0000000..28f5f69 --- /dev/null +++ b/swh/core/api/classes.py @@ -0,0 +1,25 @@ +# Copyright (C) 2020 The Software Heritage developers +# See the AUTHORS file at the top-level directory of this distribution +# License: GNU General Public License version 3, or any later version +# See top-level LICENSE file for more information + +from dataclasses import dataclass, field + +from typing import ( + Generic, + List, + Optional, + TypeVar, +) + + +TResult = TypeVar("TResult") +TToken = TypeVar("TToken") + + +@dataclass(eq=True) +class PagedResult(Generic[TResult, TToken]): + """Represents a page of results; with a token to get the next page""" + + results: List[TResult] = field(default_factory=list) + next_page_token: Optional[TToken] = field(default=None) diff --git a/swh/core/api/serializers.py b/swh/core/api/serializers.py index b4e0a3f..8d8cbf1 100644 --- a/swh/core/api/serializers.py +++ b/swh/core/api/serializers.py @@ -1,277 +1,294 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import base64 import datetime from enum import Enum import json import traceback import types from uuid import UUID import arrow import iso8601 import msgpack from typing import Any, Dict, Union, Tuple from requests import Response +from swh.core.api.classes import PagedResult + def encode_datetime(dt: datetime.datetime) -> str: """Wrapper of datetime.datetime.isoformat() that forbids naive datetimes.""" if dt.tzinfo is None: raise ValueError(f"{dt} is a naive datetime.") return dt.isoformat() +def _encode_paged_result(obj: PagedResult) -> Dict[str, Any]: + """Serialize PagedResult to a Dict.""" + return { + "results": obj.results, + "next_page_token": obj.next_page_token, + } + + +def _decode_paged_result(obj: Dict[str, Any]) -> PagedResult: + """Deserialize Dict into PagedResult""" + return PagedResult(results=obj["results"], next_page_token=obj["next_page_token"],) + + ENCODERS = [ (arrow.Arrow, "arrow", arrow.Arrow.isoformat), (datetime.datetime, "datetime", encode_datetime), ( datetime.timedelta, "timedelta", lambda o: { "days": o.days, "seconds": o.seconds, "microseconds": o.microseconds, }, ), (UUID, "uuid", str), + (PagedResult, "paged_result", _encode_paged_result), # Only for JSON: (bytes, "bytes", lambda o: base64.b85encode(o).decode("ascii")), ] DECODERS = { "arrow": arrow.get, "datetime": lambda d: iso8601.parse_date(d, default_timezone=None), "timedelta": lambda d: datetime.timedelta(**d), "uuid": UUID, + "paged_result": _decode_paged_result, # Only for JSON: "bytes": base64.b85decode, } class MsgpackExtTypeCodes(Enum): LONG_INT = 1 def encode_data_client(data: Any, extra_encoders=None) -> bytes: try: return msgpack_dumps(data, extra_encoders=extra_encoders) except OverflowError as e: raise ValueError("Limits were reached. Please, check your input.\n" + str(e)) def decode_response(response: Response, extra_decoders=None) -> Any: content_type = response.headers["content-type"] if content_type.startswith("application/x-msgpack"): r = msgpack_loads(response.content, extra_decoders=extra_decoders) elif content_type.startswith("application/json"): r = json_loads(response.text, extra_decoders=extra_decoders) elif content_type.startswith("text/"): r = response.text else: raise ValueError("Wrong content type `%s` for API response" % content_type) return r class SWHJSONEncoder(json.JSONEncoder): """JSON encoder for data structures generated by Software Heritage. This JSON encoder extends the default Python JSON encoder and adds awareness for the following specific types: - bytes (get encoded as a Base85 string); - datetime.datetime (get encoded as an ISO8601 string). Non-standard types get encoded as a a dictionary with two keys: - swhtype with value 'bytes' or 'datetime'; - d containing the encoded value. SWHJSONEncoder also encodes arbitrary iterables as a list (allowing serialization of generators). Caveats: Limitations in the JSONEncoder extension mechanism prevent us from "escaping" dictionaries that only contain the swhtype and d keys, and therefore arbitrary data structures can't be round-tripped through SWHJSONEncoder and SWHJSONDecoder. """ def __init__(self, extra_encoders=None, **kwargs): super().__init__(**kwargs) self.encoders = ENCODERS if extra_encoders: self.encoders += extra_encoders def default(self, o: Any) -> Union[Dict[str, Union[Dict[str, int], str]], list]: for (type_, type_name, encoder) in self.encoders: if isinstance(o, type_): return { "swhtype": type_name, "d": encoder(o), } try: return super().default(o) except TypeError as e: try: iterable = iter(o) except TypeError: raise e from None else: return list(iterable) class SWHJSONDecoder(json.JSONDecoder): """JSON decoder for data structures encoded with SWHJSONEncoder. This JSON decoder extends the default Python JSON decoder, allowing the decoding of: - bytes (encoded as a Base85 string); - datetime.datetime (encoded as an ISO8601 string). Non-standard types must be encoded as a a dictionary with exactly two keys: - swhtype with value 'bytes' or 'datetime'; - d containing the encoded value. To limit the impact our encoding, if the swhtype key doesn't contain a known value, the dictionary is decoded as-is. """ def __init__(self, extra_decoders=None, **kwargs): super().__init__(**kwargs) self.decoders = DECODERS if extra_decoders: self.decoders = {**self.decoders, **extra_decoders} def decode_data(self, o: Any) -> Any: if isinstance(o, dict): if set(o.keys()) == {"d", "swhtype"}: if o["swhtype"] == "bytes": return base64.b85decode(o["d"]) decoder = self.decoders.get(o["swhtype"]) if decoder: return decoder(self.decode_data(o["d"])) return {key: self.decode_data(value) for key, value in o.items()} if isinstance(o, list): return [self.decode_data(value) for value in o] else: return o def raw_decode(self, s: str, idx: int = 0) -> Tuple[Any, int]: data, index = super().raw_decode(s, idx) return self.decode_data(data), index def json_dumps(data: Any, extra_encoders=None) -> str: return json.dumps(data, cls=SWHJSONEncoder, extra_encoders=extra_encoders) def json_loads(data: str, extra_decoders=None) -> Any: return json.loads(data, cls=SWHJSONDecoder, extra_decoders=extra_decoders) def msgpack_dumps(data: Any, extra_encoders=None) -> bytes: """Write data as a msgpack stream""" encoders = ENCODERS if extra_encoders: encoders += extra_encoders def encode_types(obj): if isinstance(obj, int): # integer overflowed while packing. Handle it as an extended type length, rem = divmod(obj.bit_length(), 8) if rem: length += 1 return msgpack.ExtType( MsgpackExtTypeCodes.LONG_INT.value, int.to_bytes(obj, length, "big") ) if isinstance(obj, types.GeneratorType): return list(obj) for (type_, type_name, encoder) in encoders: if isinstance(obj, type_): return { b"swhtype": type_name, b"d": encoder(obj), } return obj return msgpack.packb(data, use_bin_type=True, default=encode_types) def msgpack_loads(data: bytes, extra_decoders=None) -> Any: """Read data as a msgpack stream. .. Caution:: This function is used by swh.journal to decode the contents of the journal. This function **must** be kept backwards-compatible. """ decoders = DECODERS if extra_decoders: decoders = {**decoders, **extra_decoders} def ext_hook(code, data): if code == MsgpackExtTypeCodes.LONG_INT.value: return int.from_bytes(data, "big") raise ValueError("Unknown msgpack extended code %s" % code) def decode_types(obj): # Support for current encodings if set(obj.keys()) == {b"d", b"swhtype"}: decoder = decoders.get(obj[b"swhtype"]) if decoder: return decoder(obj[b"d"]) # Support for legacy encodings if b"__datetime__" in obj and obj[b"__datetime__"]: return iso8601.parse_date(obj[b"s"], default_timezone=None) if b"__uuid__" in obj and obj[b"__uuid__"]: return UUID(obj[b"s"]) if b"__timedelta__" in obj and obj[b"__timedelta__"]: return datetime.timedelta(**obj[b"s"]) if b"__arrow__" in obj and obj[b"__arrow__"]: return arrow.get(obj[b"s"]) # Fallthrough return obj try: try: return msgpack.unpackb( data, raw=False, object_hook=decode_types, ext_hook=ext_hook, strict_map_key=False, ) except TypeError: # msgpack < 0.6.0 return msgpack.unpackb( data, raw=False, object_hook=decode_types, ext_hook=ext_hook ) except TypeError: # msgpack < 0.5.2 return msgpack.unpackb( data, encoding="utf-8", object_hook=decode_types, ext_hook=ext_hook ) def exception_to_dict(exception): tb = traceback.format_exception(None, exception, exception.__traceback__) return { "exception": { "type": type(exception).__name__, "args": exception.args, "message": str(exception), "traceback": tb, } } diff --git a/swh/core/api/tests/test_serializers.py b/swh/core/api/tests/test_serializers.py index 0ddab7a..0b4803d 100644 --- a/swh/core/api/tests/test_serializers.py +++ b/swh/core/api/tests/test_serializers.py @@ -1,195 +1,252 @@ # Copyright (C) 2015-2020 The Software Heritage developers # See the AUTHORS file at the top-level directory of this distribution # License: GNU General Public License version 3, or any later version # See top-level LICENSE file for more information import datetime import json -from typing import Any, Callable, List, Tuple +from typing import Any, Callable, List, Tuple, Union + +from arrow import Arrow from uuid import UUID import pytest import arrow import requests +from swh.core.api.classes import PagedResult from swh.core.api.serializers import ( SWHJSONDecoder, SWHJSONEncoder, msgpack_dumps, msgpack_loads, decode_response, ) class ExtraType: def __init__(self, arg1, arg2): self.arg1 = arg1 self.arg2 = arg2 def __repr__(self): return f"ExtraType({self.arg1}, {self.arg2})" def __eq__(self, other): return isinstance(other, ExtraType) and (self.arg1, self.arg2) == ( other.arg1, other.arg2, ) extra_encoders: List[Tuple[type, str, Callable[..., Any]]] = [ (ExtraType, "extratype", lambda o: (o.arg1, o.arg2)) ] extra_decoders = { "extratype": lambda o: ExtraType(*o), } TZ = datetime.timezone(datetime.timedelta(minutes=118)) +DATA_BYTES = b"123456789\x99\xaf\xff\x00\x12" +ENCODED_DATA_BYTES = {"swhtype": "bytes", "d": "F)}kWH8wXmIhn8j01^"} + +DATA_DATETIME = datetime.datetime(2015, 3, 4, 18, 25, 13, 1234, tzinfo=TZ,) +ENCODED_DATA_DATETIME = { + "swhtype": "datetime", + "d": "2015-03-04T18:25:13.001234+01:58", +} + +DATA_TIMEDELTA = datetime.timedelta(64) +ENCODED_DATA_TIMEDELTA = { + "swhtype": "timedelta", + "d": {"days": 64, "seconds": 0, "microseconds": 0}, +} + +DATA_ARROW = arrow.get("2018-04-25T16:17:53.533672+00:00") +ENCODED_DATA_ARROW = {"swhtype": "arrow", "d": "2018-04-25T16:17:53.533672+00:00"} + +DATA_UUID = UUID("cdd8f804-9db6-40c3-93ab-5955d3836234") +ENCODED_DATA_UUID = {"swhtype": "uuid", "d": "cdd8f804-9db6-40c3-93ab-5955d3836234"} + +# For test demonstration purposes +TestPagedResultStr = PagedResult[ + Union[UUID, datetime.datetime, datetime.timedelta], str +] + +DATA_PAGED_RESULT = TestPagedResultStr( + results=[DATA_UUID, DATA_DATETIME, DATA_TIMEDELTA], next_page_token="10", +) + +ENCODED_DATA_PAGED_RESULT = { + "d": { + "results": [ENCODED_DATA_UUID, ENCODED_DATA_DATETIME, ENCODED_DATA_TIMEDELTA,], + "next_page_token": "10", + }, + "swhtype": "paged_result", +} + +TestPagedResultTuple = PagedResult[Union[str, bytes, Arrow], List[Union[str, UUID]]] + + +DATA_PAGED_RESULT2 = TestPagedResultTuple( + results=["data0", DATA_BYTES, DATA_ARROW], next_page_token=["10", DATA_UUID], +) + +ENCODED_DATA_PAGED_RESULT2 = { + "d": { + "results": ["data0", ENCODED_DATA_BYTES, ENCODED_DATA_ARROW,], + "next_page_token": ["10", ENCODED_DATA_UUID], + }, + "swhtype": "paged_result", +} + DATA = { - "bytes": b"123456789\x99\xaf\xff\x00\x12", - "datetime_tz": datetime.datetime(2015, 3, 4, 18, 25, 13, 1234, tzinfo=TZ,), + "bytes": DATA_BYTES, + "datetime_tz": DATA_DATETIME, "datetime_utc": datetime.datetime( 2015, 3, 4, 18, 25, 13, 1234, tzinfo=datetime.timezone.utc ), - "datetime_delta": datetime.timedelta(64), - "arrow_date": arrow.get("2018-04-25T16:17:53.533672+00:00"), + "datetime_delta": DATA_TIMEDELTA, + "arrow_date": DATA_ARROW, "swhtype": "fake", "swh_dict": {"swhtype": 42, "d": "test"}, "random_dict": {"swhtype": 43}, - "uuid": UUID("cdd8f804-9db6-40c3-93ab-5955d3836234"), + "uuid": DATA_UUID, + "paged-result": DATA_PAGED_RESULT, + "paged-result2": DATA_PAGED_RESULT2, } ENCODED_DATA = { - "bytes": {"swhtype": "bytes", "d": "F)}kWH8wXmIhn8j01^"}, - "datetime_tz": {"swhtype": "datetime", "d": "2015-03-04T18:25:13.001234+01:58",}, + "bytes": ENCODED_DATA_BYTES, + "datetime_tz": ENCODED_DATA_DATETIME, "datetime_utc": {"swhtype": "datetime", "d": "2015-03-04T18:25:13.001234+00:00",}, - "datetime_delta": { - "swhtype": "timedelta", - "d": {"days": 64, "seconds": 0, "microseconds": 0}, - }, - "arrow_date": {"swhtype": "arrow", "d": "2018-04-25T16:17:53.533672+00:00"}, + "datetime_delta": ENCODED_DATA_TIMEDELTA, + "arrow_date": ENCODED_DATA_ARROW, "swhtype": "fake", "swh_dict": {"swhtype": 42, "d": "test"}, "random_dict": {"swhtype": 43}, - "uuid": {"swhtype": "uuid", "d": "cdd8f804-9db6-40c3-93ab-5955d3836234"}, + "uuid": ENCODED_DATA_UUID, + "paged-result": ENCODED_DATA_PAGED_RESULT, + "paged-result2": ENCODED_DATA_PAGED_RESULT2, } def test_serializers_round_trip_json(): json_data = json.dumps(DATA, cls=SWHJSONEncoder) actual_data = json.loads(json_data, cls=SWHJSONDecoder) assert actual_data == DATA def test_serializers_round_trip_json_extra_types(): expected_original_data = [ExtraType("baz", DATA), "qux"] data = json.dumps( expected_original_data, cls=SWHJSONEncoder, extra_encoders=extra_encoders ) actual_data = json.loads(data, cls=SWHJSONDecoder, extra_decoders=extra_decoders) assert actual_data == expected_original_data def test_serializers_encode_swh_json(): json_str = json.dumps(DATA, cls=SWHJSONEncoder) actual_data = json.loads(json_str) assert actual_data == ENCODED_DATA def test_serializers_round_trip_msgpack(): expected_original_data = { **DATA, "none_dict_key": {None: 42}, "long_int_is_loooong": 10000000000000000000000000000000, } data = msgpack_dumps(expected_original_data) actual_data = msgpack_loads(data) assert actual_data == expected_original_data def test_serializers_round_trip_msgpack_extra_types(): original_data = [ExtraType("baz", DATA), "qux"] data = msgpack_dumps(original_data, extra_encoders=extra_encoders) actual_data = msgpack_loads(data, extra_decoders=extra_decoders) assert actual_data == original_data def test_serializers_generator_json(): data = json.dumps((i for i in range(5)), cls=SWHJSONEncoder) assert json.loads(data, cls=SWHJSONDecoder) == [i for i in range(5)] def test_serializers_generator_msgpack(): data = msgpack_dumps((i for i in range(5))) assert msgpack_loads(data) == [i for i in range(5)] def test_serializers_decode_response_json(requests_mock): requests_mock.get( "https://example.org/test/data", json=ENCODED_DATA, headers={"content-type": "application/json"}, ) response = requests.get("https://example.org/test/data") assert decode_response(response) == DATA def test_serializers_decode_legacy_msgpack(): legacy_msgpack = { "bytes": b"\xc4\x0e123456789\x99\xaf\xff\x00\x12", "datetime_tz": ( b"\x82\xc4\x0c__datetime__\xc3\xc4\x01s\xd9 " b"2015-03-04T18:25:13.001234+01:58" ), "datetime_utc": ( b"\x82\xc4\x0c__datetime__\xc3\xc4\x01s\xd9 " b"2015-03-04T18:25:13.001234+00:00" ), "datetime_delta": ( b"\x82\xc4\r__timedelta__\xc3\xc4\x01s\x83\xa4" b"days@\xa7seconds\x00\xacmicroseconds\x00" ), "arrow_date": ( b"\x82\xc4\t__arrow__\xc3\xc4\x01s\xd9 2018-04-25T16:17:53.533672+00:00" ), "swhtype": b"\xa4fake", "swh_dict": b"\x82\xa7swhtype*\xa1d\xa4test", "random_dict": b"\x81\xa7swhtype+", "uuid": ( b"\x82\xc4\x08__uuid__\xc3\xc4\x01s\xd9$" b"cdd8f804-9db6-40c3-93ab-5955d3836234" ), } for k, v in legacy_msgpack.items(): assert msgpack_loads(v) == DATA[k] def test_serializers_encode_native_datetime(): dt = datetime.datetime(2015, 1, 1, 12, 4, 42, 231455) with pytest.raises(ValueError, match="naive datetime"): msgpack_dumps(dt) def test_serializers_decode_naive_datetime(): expected_dt = datetime.datetime(2015, 1, 1, 12, 4, 42, 231455) # Current encoding assert ( msgpack_loads( b"\x82\xc4\x07swhtype\xa8datetime\xc4\x01d\xba" b"2015-01-01T12:04:42.231455" ) == expected_dt ) # Legacy encoding assert ( msgpack_loads( b"\x82\xc4\x0c__datetime__\xc3\xc4\x01s\xba2015-01-01T12:04:42.231455" ) == expected_dt )