[ACIX-1061] Create BuildMetadata class for keeping track of build artifact's metadata

src/dda/build/__init__.py

@@ -0,0 +1,3 @@
+# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <[email protected]>
+#
+# SPDX-License-Identifier: MIT

src/dda/build/metadata/__init__.py

@@ -0,0 +1,3 @@
+# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <[email protected]>
+#
+# SPDX-License-Identifier: MIT

src/dda/build/metadata/enums.py

@@ -0,0 +1,207 @@
+# SPDX-FileCopyrightText: 2025-present Datadog, Inc. <[email protected]>
+#
+# SPDX-License-Identifier: MIT
+
+from enum import StrEnum, auto
+from typing import Any, ClassVar
+
+from msgspec import Struct
+
+from dda.types.hooks import register_type_hooks
+
+
+class ArtifactType(StrEnum):
+    """
+    The type of a build artifact: either a "component" or a "distribution".
+    Each artifact type has a corresponding format enum.
+    """
+
+    COMP = auto()
+    DIST = auto()
+
+
+class DistributionFormat(StrEnum):
+    """
+    The format of a "distribution"-type build artifact.
+    """
+
+    DEB = auto()
+    RPM = auto()
+    MSI = auto()
+    OCI = auto()
+
+    def get_file_identifier(self) -> str:
+        """
+        Get the file identifier for the artifact format.
+        This is the string that will be used to identify the artifact format in the filename.
+        This usually corresponds to the file extension, but can include some other charactres before it.
+        """
+        match self:
+            case self.DEB:
+                return ".deb"
+            case self.RPM:
+                return ".rpm"
+            case self.MSI:
+                return ".msi"
+            case self.OCI:
+                return "-oci.tar.gz"
+        # Adding a default return value to satisfy mypy, even though we should never reach here
+        return ""
+
+    # Need to make a property function so it works with enums
+    @property
+    def type(self) -> ArtifactType:
+        """The artifact type."""
+        return ArtifactType.DIST
+
+    def wrap(self) -> "ArtifactFormat":
+        return ArtifactFormat(enum=self)
+
+
+class ComponentFormat(StrEnum):
+    """
+    The format of a "component"-type build artifact.
+    """
+
+    BIN = auto()
+
+    def get_file_identifier(self) -> str:  # noqa:PLR6301
+        """See `DistributionFormat.get_file_identifier` for more details."""
+        return ""  # Binary files don't have a specific identifier
+
+    @property
+    def type(self) -> ArtifactType:
+        """The artifact type."""
+        return ArtifactType.COMP
+
+    def wrap(self) -> "ArtifactFormat":
+        return ArtifactFormat(enum=self)
+
+
+# Need custom logic for encoding/decoding, since msgspec does not support encoding StrEnum unions
+class ArtifactFormat:
+    def __init__(self, enum: ComponentFormat | DistributionFormat):
+        self.enum = enum
+
+    @staticmethod
+    def encode(obj: "ArtifactFormat") -> dict[str, str]:
+        return {
+            "type": obj.enum.type.value,
+            "format": obj.enum.value,
+        }
+
+    @staticmethod
+    def decode(data: dict[str, str]) -> "ArtifactFormat":
+        match data["type"]:
+            case ArtifactType.COMP.value:
+                return ArtifactFormat(enum=ComponentFormat(data["format"]))
+            case ArtifactType.DIST.value:
+                return ArtifactFormat(enum=DistributionFormat(data["format"]))
+            case _:
+                msg = f"Invalid artifact type: {data['type']}"
+                raise ValueError(msg)
+
+    # Proxy methods and attributes to the underlying enum
+    def __getattr__(self, name: str) -> Any:
+        return getattr(self.enum, name)
+
+    def __eq__(self, other: object) -> bool:
+        return self.enum == other.enum if isinstance(other, ArtifactFormat) else self.enum == other
+
+    def __hash__(self) -> int:
+        return hash(self.enum)
+
+
+class OS(StrEnum):
+    """
+    The operating system for which the build artifact is intended.
+    """
+
+    LINUX = auto()
+    WINDOWS = auto()
+    MACOS = auto()
+
+    # Any OS - used for indicating compatibility with any operating system
+    ANY = auto()
+
+    @classmethod
+    def from_alias(cls, alias: str) -> "OS":
+        """
+        Get the OS enum value from an alias.
+        """
+        match alias.lower():
+            case "linux":
+                return cls.LINUX
+            case "windows" | "nt" | "win":
+                return cls.WINDOWS
+            case "macos" | "darwin" | "osx":
+                return cls.MACOS
+            case "any":
+                return cls.ANY
+            case _:
+                msg = f"Invalid OS identifier: {alias}"
+                raise ValueError(msg)
+
+
+class Arch(StrEnum):
+    """
+    The CPU architecture for which the build artifact is intended.
+    """
+
+    # x86 architectures - canonical name is amd64
+    AMD64 = auto()
+
+    # ARM architectures - canonical name is arm64
+    ARM64 = auto()
+
+    # ARMHF architectures
+    ARMHF = auto()
+
+    # Any architecture - used for indicating compatibility with any CPU architecture
+    ANY = auto()
+
+    @classmethod
+    def from_alias(cls, alias: str) -> "Arch":
+        """
+        Get the Arch enum value from an alias.
+        """
+        match alias.lower():
+            case "amd64" | "x86_64" | "x86-64" | "x86" | "x64":
+                return cls.AMD64
+            case "arm64" | "aarch64" | "arm" | "aarch":
+                return cls.ARM64
+            case "any":
+                return cls.ANY
+            case _:
+                msg = f"Invalid Arch identifier: {alias}"
+                raise ValueError(msg)
+
+
+class Platform(Struct, frozen=True):
+    """
+    The platform for which the build artifact is intended.
+    """
+
+    os: OS
+    arch: Arch
+
+    ANY: ClassVar["Platform"]
+
+    @classmethod
+    def from_alias(cls, os_alias: str, arch_alias: str) -> "Platform":
+        """
+        Get the Platform enum value from an alias.
+        """
+        return cls(os=OS.from_alias(os_alias), arch=Arch.from_alias(arch_alias))
+
+    def __str__(self) -> str:
+        """
+        Get the string representation of the platform.
+        """
+        return f"{self.os}-{self.arch}"
+
+
+# Initialize the ANY class variable after the class is fully defined
+Platform.ANY = Platform(os=OS.ANY, arch=Arch.ANY)
+
+register_type_hooks(ArtifactFormat, encode=ArtifactFormat.encode, decode=ArtifactFormat.decode)