Newer
Older
# Copyright (C) 2018-2020 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
"""
Implementation of Software Heritage's data model
See :ref:`data-model` for an overview of the data model.
The classes defined in this module are immutable
`attrs objects <https://attrs.org/>`__ and enums.
All classes define a ``from_dict`` class method and a ``to_dict``
method to convert between them and msgpack-serializable objects.
"""
from abc import ABCMeta, abstractmethod
from enum import Enum
import hashlib
from typing import Any, Dict, Iterable, Optional, Tuple, TypeVar, Union
import attr
from attrs_strict import AttributeTypeError
import dateutil.parser
import iso8601
from . import git_objects
from .collections import ImmutableDict
from .hashutil import DEFAULT_ALGORITHMS, MultiHash, hash_to_hex
from .swhids import CoreSWHID
from .swhids import ExtendedObjectType as SwhidExtendedObjectType
from .swhids import ExtendedSWHID
from .swhids import ObjectType as SwhidObjectType

vlorentz
committed
class MissingData(Exception):
"""Raised by `Content.with_data` when it has no way of fetching the
data (but not when fetching the data fails)."""

vlorentz
committed
pass
KeyType = Union[Dict[str, str], Dict[str, bytes], bytes]
"""The type returned by BaseModel.unique_key()."""
_OFFSET_CHARS = frozenset(b"+-0123456789")
# TODO: Limit this to 20 bytes
Sha1Git = bytes
KT = TypeVar("KT")
VT = TypeVar("VT")
def hash_repr(h: bytes) -> str:
if h is None:
return "None"
else:
return f"hash_to_bytes('{hash_to_hex(h)}')"
def freeze_optional_dict(
d: Union[None, Dict[KT, VT], ImmutableDict[KT, VT]] # type: ignore
) -> Optional[ImmutableDict[KT, VT]]:
if isinstance(d, dict):
return ImmutableDict(d)
else:
return d
def dictify(value):
"Helper function used by BaseModel.to_dict()"
if isinstance(value, BaseModel):
return value.to_dict()
elif isinstance(value, (CoreSWHID, ExtendedSWHID)):
return str(value)
elif isinstance(value, Enum):
return value.value
elif isinstance(value, (dict, ImmutableDict)):
return {k: dictify(v) for k, v in value.items()}
elif isinstance(value, tuple):
return tuple(dictify(v) for v in value)
else:
return value
def _check_type(type_, value):
if type_ is object or type_ is Any:
return True
if type_ is None:
return value is None
origin = getattr(type_, "__origin__", None)
# Non-generic type, check it directly
if origin is None:
# This is functionally equivalent to using just this:
# return isinstance(value, type)
# but using type equality before isinstance allows very quick checks
# when the exact class is used (which is the overwhelming majority of cases)
# while still allowing subclasses to be used.
return type(value) == type_ or isinstance(value, type_)
# Check the type of the value itself
#
# For the same reason as above, this condition is functionally equivalent to:
# if origin is not Union and not isinstance(value, origin):
if origin is not Union and type(value) != origin and not isinstance(value, origin):
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
return False
# Then, if it's a container, check its items.
if origin is tuple:
args = type_.__args__
if len(args) == 2 and args[1] is Ellipsis:
# Infinite tuple
return all(_check_type(args[0], item) for item in value)
else:
# Finite tuple
if len(args) != len(value):
return False
return all(
_check_type(item_type, item) for (item_type, item) in zip(args, value)
)
elif origin is Union:
args = type_.__args__
return any(_check_type(variant, value) for variant in args)
elif origin is ImmutableDict:
(key_type, value_type) = type_.__args__
return all(
_check_type(key_type, key) and _check_type(value_type, value)
for (key, value) in value.items()
)
else:
# No need to check dict or list. because they are converted to ImmutableDict
# and tuple respectively.
raise NotImplementedError(f"Type-checking {type_}")
def type_validator():
"""Like attrs_strict.type_validator(), but stricter.
It is an attrs validator, which checks attributes have the specified type,
using type equality instead of ``isinstance()``, for improved performance
"""
def validator(instance, attribute, value):
if not _check_type(attribute.type, value):
raise AttributeTypeError(value, attribute)
return validator
ModelType = TypeVar("ModelType", bound="BaseModel")
class BaseModel:
"""Base class for SWH model classes.
Provides serialization/deserialization to/from Python dictionaries,
that are suitable for JSON/msgpack-like formats."""
def to_dict(self):
"""Wrapper of `attr.asdict` that can be overridden by subclasses
that have special handling of some of the fields."""
return dictify(attr.asdict(self, recurse=False))
@classmethod
def from_dict(cls, d):
"""Takes a dictionary representing a tree of SWH objects, and
recursively builds the corresponding objects."""
return cls(**d)
def anonymize(self: ModelType) -> Optional[ModelType]:
"""Returns an anonymized version of the object, if needed.
If the object model does not need/support anonymization, returns None.
"""
return None
def unique_key(self) -> KeyType:
"""Returns a unique key for this object, that can be used for
deduplication."""
raise NotImplementedError(f"unique_key for {self}")
def check(self) -> None:
"""Performs internal consistency checks, and raises an error if one fails."""
attr.validate(self)
def _compute_hash_from_manifest(manifest: bytes) -> Sha1Git:
return hashlib.new("sha1", manifest).digest()
class HashableObject(metaclass=ABCMeta):
"""Mixin to automatically compute object identifier hash when
the associated model is instantiated."""
def compute_hash(self) -> bytes:
"""Derived model classes must implement this to compute
the object hash.
This method is called by the object initialization if the `id`
attribute is set to an empty value.
"""
return self._compute_hash_from_attributes()
@abstractmethod
def _compute_hash_from_attributes(self) -> Sha1Git:
raise NotImplementedError(f"_compute_hash_from_attributes for {self}")
def __attrs_post_init__(self):
if not self.id:
obj_id = self.compute_hash()
def check(self) -> None:
super().check() # type: ignore
if self.id != self.compute_hash():
raise ValueError("'id' does not match recomputed hash.")
class HashableObjectWithManifest(HashableObject):
"""Derived class of HashableObject, for objects that may need to store
verbatim git objects as ``raw_manifest`` to preserve original hashes."""
raw_manifest: Optional[bytes] = None
"""Stores the original content of git objects when they cannot be faithfully
represented using only the other attributes.
This should only be used as a last resort, and only set in the Git loader,
for objects too corrupt to fit the data model."""
def to_dict(self):
d = super().to_dict()
if d["raw_manifest"] is None:
del d["raw_manifest"]
return d
def compute_hash(self) -> bytes:
"""Derived model classes must implement this to compute
the object hash.
This method is called by the object initialization if the `id`
attribute is set to an empty value.
"""
if self.raw_manifest is None:
return super().compute_hash()
else:
return _compute_hash_from_manifest(self.raw_manifest)
def check(self) -> None:
super().check()
if (
self.raw_manifest is not None
and self.id == self._compute_hash_from_attributes()
):
raise ValueError(
f"{self} has a non-none raw_manifest attribute, but does not need it."
)
class Person(BaseModel):
"""Represents the author/committer of a revision or release."""
object_type: Final = "person"
name = attr.ib(type=Optional[bytes], validator=type_validator(), eq=False)
email = attr.ib(type=Optional[bytes], validator=type_validator(), eq=False)
@classmethod
def from_fullname(cls, fullname: bytes):
"""Returns a Person object, by guessing the name and email from the
fullname, in the `name <email>` format.
The fullname is left unchanged."""
if fullname is None:
name: Optional[bytes]
email: Optional[bytes]
try:
except ValueError:
name = fullname
email = None
else:
raw_name = fullname[:open_bracket]
if not raw_name:
name = None
else:
name = raw_name.strip()
try:
except ValueError:
email = raw_email
else:
email = raw_email[:close_bracket]
return Person(name=name or None, email=email or None, fullname=fullname,)
def anonymize(self) -> "Person":
"""Returns an anonymized version of the Person object.
Anonymization is simply a Person which fullname is the hashed, with unset name
or email.
"""
return Person(
fullname=hashlib.sha256(self.fullname).digest(), name=None, email=None,
)
@classmethod
def from_dict(cls, d):
"""
If the fullname is missing, construct a fullname
using the following heuristics: if the name value is None, we return the
email in angle brackets, else, we return the name, a space, and the email
in angle brackets.
"""
if "fullname" not in d:
parts = []
if d["name"] is not None:
parts.append(d["name"])
if d["email"] is not None:
parts.append(b"".join([b"<", d["email"], b">"]))
fullname = b" ".join(parts)
d = {**d, "fullname": fullname}
d = {"name": None, "email": None, **d}
return super().from_dict(d)
class Timestamp(BaseModel):
"""Represents a naive timestamp from a VCS."""
object_type: Final = "timestamp"
seconds = attr.ib(type=int, validator=type_validator())
microseconds = attr.ib(type=int, validator=type_validator())
@seconds.validator
def check_seconds(self, attribute, value):
"""Check that seconds fit in a 64-bits signed integer."""
if not (-(2 ** 63) <= value < 2 ** 63):
raise ValueError("Seconds must be a signed 64-bits integer.")
@microseconds.validator
def check_microseconds(self, attribute, value):
"""Checks that microseconds are positive and < 1000000."""
if not (0 <= value < 10 ** 6):
raise ValueError("Microseconds must be in [0, 1000000[.")
@attr.s(frozen=True, slots=True)
class TimestampWithTimezone(BaseModel):
"""Represents a TZ-aware timestamp from a VCS."""
object_type: Final = "timestamp_with_timezone"
timestamp = attr.ib(type=Timestamp, validator=type_validator())
offset_bytes = attr.ib(type=bytes, validator=type_validator())
"""Raw git representation of the timezone, as an offset from UTC.
It should follow this format: ``+HHMM`` or ``-HHMM`` (including ``+0000`` and
``-0000``).
However, when created from git objects, it must be the exact bytes used in the
original objects, so it may differ from this format when they do.
"""
@classmethod
def from_numeric_offset(
cls, timestamp: Timestamp, offset: int, negative_utc: bool
) -> "TimestampWithTimezone":
"""Returns a :class:`TimestampWithTimezone` instance from the old dictionary
format (with ``offset`` and ``negative_utc`` instead of ``offset_bytes``).
"""
negative = offset < 0 or negative_utc
(hours, minutes) = divmod(abs(offset), 60)
offset_bytes = f"{'-' if negative else '+'}{hours:02}{minutes:02}".encode()
tstz = TimestampWithTimezone(timestamp=timestamp, offset_bytes=offset_bytes)
assert tstz.offset_minutes() == offset, (tstz.offset_minutes(), offset)
return tstz
@classmethod
def from_dict(
cls, time_representation: Union[Dict, datetime.datetime, int]
) -> "TimestampWithTimezone":
"""Builds a TimestampWithTimezone from any of the formats
accepted by :func:`swh.model.normalize_timestamp`."""
# TODO: this accept way more types than just dicts; find a better
# name
if isinstance(time_representation, dict):
ts = time_representation["timestamp"]
if isinstance(ts, dict):
seconds = ts.get("seconds", 0)
microseconds = ts.get("microseconds", 0)
elif isinstance(ts, int):
seconds = ts
microseconds = 0
else:
raise ValueError(
f"TimestampWithTimezone.from_dict received non-integer timestamp "
f"member {ts!r}"
)
timestamp = Timestamp(seconds=seconds, microseconds=microseconds)
if "offset_bytes" in time_representation:
return cls(
timestamp=timestamp,
offset_bytes=time_representation["offset_bytes"],
)
else:
# old format
offset = time_representation["offset"]
negative_utc = time_representation.get("negative_utc") or False
return cls.from_numeric_offset(timestamp, offset, negative_utc)
elif isinstance(time_representation, datetime.datetime):
# TODO: warn when using from_dict() on a datetime

vlorentz
committed
utcoffset = time_representation.utcoffset()
time_representation = time_representation.astimezone(datetime.timezone.utc)
microseconds = time_representation.microsecond
if microseconds:
time_representation = time_representation.replace(microsecond=0)
seconds = int(time_representation.timestamp())
if utcoffset is None:
raise ValueError(
f"TimestampWithTimezone.from_dict received datetime without "
f"timezone: {time_representation}"
)
# utcoffset is an integer number of minutes
seconds_offset = utcoffset.total_seconds()
offset = int(seconds_offset) // 60
# TODO: warn if remainder is not zero
return cls.from_numeric_offset(
Timestamp(seconds=seconds, microseconds=microseconds), offset, False
)
elif isinstance(time_representation, int):
# TODO: warn when using from_dict() on an int
seconds = time_representation
timestamp = Timestamp(seconds=time_representation, microseconds=0)
return cls(timestamp=timestamp, offset_bytes=b"+0000")
else:
raise ValueError(
f"TimestampWithTimezone.from_dict received non-integer timestamp: "
f"{time_representation!r}"
)
@classmethod
def from_datetime(cls, dt: datetime.datetime) -> "TimestampWithTimezone":
return cls.from_dict(dt)
def to_datetime(self) -> datetime.datetime:
"""Convert to a datetime (with a timezone set to the recorded fixed UTC offset)
Beware that this conversion can be lossy: ``-0000`` and 'weird' offsets
cannot be represented. Also note that it may fail due to type overflow.
"""
timestamp = datetime.datetime.fromtimestamp(
self.timestamp.seconds,
datetime.timezone(datetime.timedelta(minutes=self.offset_minutes())),
)
timestamp = timestamp.replace(microsecond=self.timestamp.microseconds)
return timestamp
@classmethod
def from_iso8601(cls, s):
"""Builds a TimestampWithTimezone from an ISO8601-formatted string.
"""
dt = iso8601.parse_date(s)
tstz = cls.from_datetime(dt)
assert tstz.offset_bytes == b"+0000"
tstz = attr.evolve(tstz, offset_bytes=b"-0000")
return tstz
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
@staticmethod
def _parse_offset_bytes(offset_bytes: bytes) -> int:
"""Parses an ``offset_bytes`` value (in Git's ``[+-]HHMM`` format),
and returns the corresponding numeric values (in number of minutes).
Tries to account for some mistakes in the format, to support incorrect
Git implementations.
>>> TimestampWithTimezone._parse_offset_bytes(b"+0000")
0
>>> TimestampWithTimezone._parse_offset_bytes(b"-0000")
0
>>> TimestampWithTimezone._parse_offset_bytes(b"+0200")
120
>>> TimestampWithTimezone._parse_offset_bytes(b"-0200")
-120
>>> TimestampWithTimezone._parse_offset_bytes(b"+200")
120
>>> TimestampWithTimezone._parse_offset_bytes(b"-200")
-120
>>> TimestampWithTimezone._parse_offset_bytes(b"+02")
120
>>> TimestampWithTimezone._parse_offset_bytes(b"-02")
-120
>>> TimestampWithTimezone._parse_offset_bytes(b"+0010")
10
>>> TimestampWithTimezone._parse_offset_bytes(b"-0010")
-10
>>> TimestampWithTimezone._parse_offset_bytes(b"+200000000000000000")
0
>>> TimestampWithTimezone._parse_offset_bytes(b"+0160") # 60 minutes...
0
"""
offset_str = offset_bytes.decode()
assert offset_str[0] in "+-"
sign = int(offset_str[0] + "1")
if len(offset_str) <= 3:
hours = int(offset_str[1:])
minutes = 0
else:
hours = int(offset_str[1:-2])
minutes = int(offset_str[-2:])
offset = sign * (hours * 60 + minutes)
if (0 <= minutes <= 59) and (-(2 ** 15) <= offset < 2 ** 15):
return offset
else:
# can't parse it to a reasonable value; give up and pretend it's UTC.
return 0
def offset_minutes(self):
"""Returns the offset, as a number of minutes since UTC.
>>> TimestampWithTimezone(
... Timestamp(seconds=1642765364, microseconds=0), offset_bytes=b"+0000"
... ).offset_minutes()
0
>>> TimestampWithTimezone(
... Timestamp(seconds=1642765364, microseconds=0), offset_bytes=b"+0200"
... ).offset_minutes()
120
>>> TimestampWithTimezone(
... Timestamp(seconds=1642765364, microseconds=0), offset_bytes=b"-0200"
... ).offset_minutes()
-120
>>> TimestampWithTimezone(
... Timestamp(seconds=1642765364, microseconds=0), offset_bytes=b"+0530"
... ).offset_minutes()
330
"""
return self._parse_offset_bytes(self.offset_bytes)
class Origin(HashableObject, BaseModel):
"""Represents a software source: a VCS and an URL."""
object_type: Final = "origin"
id = attr.ib(type=Sha1Git, validator=type_validator(), default=b"")
def unique_key(self) -> KeyType:
return {"url": self.url}
def _compute_hash_from_attributes(self) -> bytes:
return _compute_hash_from_manifest(self.url.encode("utf-8"))
def swhid(self) -> ExtendedSWHID:
"""Returns a SWHID representing this origin."""
return ExtendedSWHID(
object_type=SwhidExtendedObjectType.ORIGIN, object_id=self.id,
class OriginVisit(BaseModel):
"""Represents an origin visit with a given type at a given point in time, by a
SWH loader."""
object_type: Final = "origin_visit"
origin = attr.ib(type=str, validator=type_validator())
date = attr.ib(type=datetime.datetime, validator=type_validator())
type = attr.ib(type=str, validator=type_validator())
"""Should not be set before calling 'origin_visit_add()'."""
visit = attr.ib(type=Optional[int], validator=type_validator(), default=None)
@date.validator
def check_date(self, attribute, value):
"""Checks the date has a timezone."""
if value is not None and value.tzinfo is None:
raise ValueError("date must be a timezone-aware datetime.")
def to_dict(self):
"""Serializes the date as a string and omits the visit id if it is
`None`."""
ov = super().to_dict()
return ov
def unique_key(self) -> KeyType:
return {"origin": self.origin, "date": str(self.date)}
class OriginVisitStatus(BaseModel):
"""Represents a visit update of an origin at a given point in time.
"""
object_type: Final = "origin_visit_status"
origin = attr.ib(type=str, validator=type_validator())
visit = attr.ib(type=int, validator=type_validator())
date = attr.ib(type=datetime.datetime, validator=type_validator())
status = attr.ib(

Antoine R. Dumont
committed
validator=attr.validators.in_(
["created", "ongoing", "full", "partial", "not_found", "failed"]
),
snapshot = attr.ib(
type=Optional[Sha1Git], validator=type_validator(), repr=hash_repr
)
# Type is optional be to able to use it before adding it to the database model
type = attr.ib(type=Optional[str], validator=type_validator(), default=None)
metadata = attr.ib(
type=Optional[ImmutableDict[str, object]],
validator=type_validator(),
converter=freeze_optional_dict,
default=None,
@date.validator
def check_date(self, attribute, value):
"""Checks the date has a timezone."""
if value is not None and value.tzinfo is None:
raise ValueError("date must be a timezone-aware datetime.")
def unique_key(self) -> KeyType:
return {"origin": self.origin, "visit": str(self.visit), "date": str(self.date)}
class TargetType(Enum):
"""The type of content pointed to by a snapshot branch. Usually a
revision or an alias."""
CONTENT = "content"
DIRECTORY = "directory"
REVISION = "revision"
RELEASE = "release"
SNAPSHOT = "snapshot"
ALIAS = "alias"
def __repr__(self):
return f"TargetType.{self.name}"
class ObjectType(Enum):
"""The type of content pointed to by a release. Usually a revision"""
CONTENT = "content"
DIRECTORY = "directory"
REVISION = "revision"
RELEASE = "release"
SNAPSHOT = "snapshot"
def __repr__(self):
return f"ObjectType.{self.name}"
class SnapshotBranch(BaseModel):
"""Represents one of the branches of a snapshot."""
object_type: Final = "snapshot_branch"
target = attr.ib(type=bytes, validator=type_validator(), repr=hash_repr)
target_type = attr.ib(type=TargetType, validator=type_validator())
@target.validator
def check_target(self, attribute, value):
"""Checks the target type is not an alias, checks the target is a
valid sha1_git."""
if self.target_type != TargetType.ALIAS and self.target is not None:
if len(value) != 20:
raise ValueError("Wrong length for bytes identifier: %d" % len(value))
@classmethod
def from_dict(cls, d):
return cls(target=d["target"], target_type=TargetType(d["target_type"]))
class Snapshot(HashableObject, BaseModel):
"""Represents the full state of an origin at a given point in time."""
object_type: Final = "snapshot"
branches = attr.ib(
type=ImmutableDict[bytes, Optional[SnapshotBranch]],
validator=type_validator(),
converter=freeze_optional_dict,
id = attr.ib(type=Sha1Git, validator=type_validator(), default=b"", repr=hash_repr)
def _compute_hash_from_attributes(self) -> bytes:
return _compute_hash_from_manifest(git_objects.snapshot_git_object(self))
@classmethod
def from_dict(cls, d):
d = d.copy()
return cls(
branches=ImmutableDict(
(name, SnapshotBranch.from_dict(branch) if branch else None)
def swhid(self) -> CoreSWHID:
"""Returns a SWHID representing this object."""
return CoreSWHID(object_type=SwhidObjectType.SNAPSHOT, object_id=self.id)
class Release(HashableObjectWithManifest, BaseModel):
object_type: Final = "release"
name = attr.ib(type=bytes, validator=type_validator())
message = attr.ib(type=Optional[bytes], validator=type_validator())
target = attr.ib(type=Optional[Sha1Git], validator=type_validator(), repr=hash_repr)
target_type = attr.ib(type=ObjectType, validator=type_validator())
synthetic = attr.ib(type=bool, validator=type_validator())
author = attr.ib(type=Optional[Person], validator=type_validator(), default=None)
date = attr.ib(
type=Optional[TimestampWithTimezone], validator=type_validator(), default=None
)
metadata = attr.ib(
type=Optional[ImmutableDict[str, object]],
validator=type_validator(),
converter=freeze_optional_dict,
default=None,
id = attr.ib(type=Sha1Git, validator=type_validator(), default=b"", repr=hash_repr)
raw_manifest = attr.ib(type=Optional[bytes], default=None)
def _compute_hash_from_attributes(self) -> bytes:
return _compute_hash_from_manifest(git_objects.release_git_object(self))
@author.validator
def check_author(self, attribute, value):
"""If the author is `None`, checks the date is `None` too."""
if self.author is None and self.date is not None:
raise ValueError("release date must be None if author is None.")
def to_dict(self):
return rel
@classmethod
def from_dict(cls, d):
d = d.copy()
if d.get("author"):
d["author"] = Person.from_dict(d["author"])
if d.get("date"):
d["date"] = TimestampWithTimezone.from_dict(d["date"])
return cls(target_type=ObjectType(d.pop("target_type")), **d)
def swhid(self) -> CoreSWHID:
"""Returns a SWHID representing this object."""
return CoreSWHID(object_type=SwhidObjectType.RELEASE, object_id=self.id)
def anonymize(self) -> "Release":
"""Returns an anonymized version of the Release object.
Anonymization consists in replacing the author with an anonymized Person object.
"""
author = self.author and self.author.anonymize()
return attr.evolve(self, author=author)
class RevisionType(Enum):
GIT = "git"
TAR = "tar"
DSC = "dsc"
SUBVERSION = "svn"
MERCURIAL = "hg"
def __repr__(self):
return f"RevisionType.{self.name}"
def tuplify_extra_headers(value: Iterable):
return tuple((k, v) for k, v in value)
class Revision(HashableObjectWithManifest, BaseModel):
object_type: Final = "revision"
message = attr.ib(type=Optional[bytes], validator=type_validator())
author = attr.ib(type=Optional[Person], validator=type_validator())
committer = attr.ib(type=Optional[Person], validator=type_validator())
date = attr.ib(type=Optional[TimestampWithTimezone], validator=type_validator())
committer_date = attr.ib(
type=Optional[TimestampWithTimezone], validator=type_validator()
)
type = attr.ib(type=RevisionType, validator=type_validator())
directory = attr.ib(type=Sha1Git, validator=type_validator(), repr=hash_repr)
metadata = attr.ib(
type=Optional[ImmutableDict[str, object]],
validator=type_validator(),
converter=freeze_optional_dict,
default=None,
parents = attr.ib(type=Tuple[Sha1Git, ...], validator=type_validator(), default=())
id = attr.ib(type=Sha1Git, validator=type_validator(), default=b"", repr=hash_repr)
extra_headers = attr.ib(
type=Tuple[Tuple[bytes, bytes], ...],
validator=type_validator(),
converter=tuplify_extra_headers,
default=(),
)
raw_manifest = attr.ib(type=Optional[bytes], default=None)
def __attrs_post_init__(self):
super().__attrs_post_init__()
# ensure metadata is a deep copy of whatever was given, and if needed
# extract extra_headers from there
if self.metadata:
metadata = self.metadata
if not self.extra_headers and "extra_headers" in metadata:
(extra_headers, metadata) = metadata.copy_pop("extra_headers")
object.__setattr__(
self, "extra_headers", tuplify_extra_headers(extra_headers),
)
attr.validate(self)
object.__setattr__(self, "metadata", metadata)
def _compute_hash_from_attributes(self) -> bytes:
return _compute_hash_from_manifest(git_objects.revision_git_object(self))
@author.validator
def check_author(self, attribute, value):
"""If the author is `None`, checks the date is `None` too."""
if self.author is None and self.date is not None:
raise ValueError("revision date must be None if author is None.")
@committer.validator
def check_committer(self, attribute, value):
"""If the committer is `None`, checks the committer_date is `None` too."""
if self.committer is None and self.committer_date is not None:
raise ValueError(
"revision committer_date must be None if committer is None."
)
@classmethod
def from_dict(cls, d):
if date:
date = TimestampWithTimezone.from_dict(date)
committer_date = TimestampWithTimezone.from_dict(committer_date)
author = d.pop("author")
if author:
author = Person.from_dict(author)
committer = d.pop("committer")
if committer:
committer = Person.from_dict(committer)
return cls(
author=author,
committer=committer,
date=date,
committer_date=committer_date,
parents=tuple(d.pop("parents")), # for BW compat
def swhid(self) -> CoreSWHID:
"""Returns a SWHID representing this object."""
return CoreSWHID(object_type=SwhidObjectType.REVISION, object_id=self.id)
def anonymize(self) -> "Revision":
"""Returns an anonymized version of the Revision object.
Anonymization consists in replacing the author and committer with an anonymized
Person object.
"""
return attr.evolve(
self,
author=None if self.author is None else self.author.anonymize(),
committer=None if self.committer is None else self.committer.anonymize(),
class DirectoryEntry(BaseModel):
object_type: Final = "directory_entry"
name = attr.ib(type=bytes, validator=type_validator())
type = attr.ib(type=str, validator=attr.validators.in_(["file", "dir", "rev"]))
target = attr.ib(type=Sha1Git, validator=type_validator(), repr=hash_repr)
perms = attr.ib(type=int, validator=type_validator(), converter=int, repr=oct)
"""Usually one of the values of `swh.model.from_disk.DentryPerms`."""
@name.validator
def check_name(self, attribute, value):
if b"/" in value:
raise ValueError(f"{value!r} is not a valid directory entry name.")
class Directory(HashableObjectWithManifest, BaseModel):
object_type: Final = "directory"
entries = attr.ib(type=Tuple[DirectoryEntry, ...], validator=type_validator())
id = attr.ib(type=Sha1Git, validator=type_validator(), default=b"", repr=hash_repr)
raw_manifest = attr.ib(type=Optional[bytes], default=None)
def _compute_hash_from_attributes(self) -> bytes:
return _compute_hash_from_manifest(git_objects.directory_git_object(self))
@entries.validator
def check_entries(self, attribute, value):
seen = set()
for entry in value:
if entry.name in seen:
# Cannot use self.swhid() here, self.id may be None
f"swh:1:dir:{hash_to_hex(self.id)} has duplicated entry name: "
f"{entry.name!r}"
@classmethod
def from_dict(cls, d):
d = d.copy()
return cls(
entries=tuple(
DirectoryEntry.from_dict(entry) for entry in d.pop("entries")
),
**d,
def swhid(self) -> CoreSWHID:
"""Returns a SWHID representing this object."""
return CoreSWHID(object_type=SwhidObjectType.DIRECTORY, object_id=self.id)
class BaseContent(BaseModel):

vlorentz
committed
status = attr.ib(
type=str, validator=attr.validators.in_(["visible", "hidden", "absent"])
)

vlorentz
committed
@staticmethod