feat: archive git repository (experimental)

See doc/git-archive.md for general Git archive specifications
See doc/repos/metadata-repo.md for info and direction related to the new Git metadata archive

aurweb/archives/__init__.py

@@ -0,0 +1 @@
+# aurweb.archives

aurweb/archives/spec/__init__.py

@@ -0,0 +1 @@
+# aurweb.archives.spec

aurweb/archives/spec/base.py

@@ -0,0 +1,77 @@
+from pathlib import Path
+from typing import Any, Dict, Iterable, List, Set
+
+
+class GitInfo:
+    """Information about a Git repository."""
+
+    """ Path to Git repository. """
+    path: str
+
+    """ Local Git repository configuration. """
+    config: Dict[str, Any]
+
+    def __init__(self, path: str, config: Dict[str, Any] = dict()) -> "GitInfo":
+        self.path = Path(path)
+        self.config = config
+
+
+class SpecOutput:
+    """Class used for git_archive.py output details."""
+
+    """ Filename relative to the Git repository root. """
+    filename: Path
+
+    """ Git repository information. """
+    git_info: GitInfo
+
+    """ Bytes bound for `SpecOutput.filename`. """
+    data: bytes
+
+    def __init__(self, filename: str, git_info: GitInfo, data: bytes) -> "SpecOutput":
+        self.filename = filename
+        self.git_info = git_info
+        self.data = data
+
+
+class SpecBase:
+    """
+    Base for Spec classes defined in git_archve.py --spec modules.
+
+    All supported --spec modules must contain the following classes:
+    - Spec(SpecBase)
+    """
+
+    """ A list of SpecOutputs, each of which contain output file data. """
+    outputs: List[SpecOutput] = list()
+
+    """ A set of repositories to commit changes to. """
+    repos: Set[str] = set()
+
+    def generate(self) -> Iterable[SpecOutput]:
+        """
+        "Pure virtual" output generator.
+
+        `SpecBase.outputs` and `SpecBase.repos` should be populated within an
+        overridden version of this function in SpecBase derivatives.
+        """
+        raise NotImplementedError()
+
+    def add_output(self, filename: str, git_info: GitInfo, data: bytes) -> None:
+        """
+        Add a SpecOutput instance to the set of outputs.
+
+        :param filename: Filename relative to the git repository root
+        :param git_info: GitInfo instance
+        :param data: Binary data bound for `filename`
+        """
+        if git_info.path not in self.repos:
+            self.repos.add(git_info.path)
+
+        self.outputs.append(
+            SpecOutput(
+                filename,
+                git_info,
+                data,
+            )
+        )

aurweb/archives/spec/metadata.py

@@ -0,0 +1,85 @@
+from typing import Iterable
+
+import orjson
+
+from aurweb import config, db
+from aurweb.models import Package, PackageBase, User
+from aurweb.rpc import RPC
+
+from .base import GitInfo, SpecBase, SpecOutput
+
+ORJSON_OPTS = orjson.OPT_SORT_KEYS | orjson.OPT_INDENT_2
+
+
+class Spec(SpecBase):
+    def __init__(self) -> "Spec":
+        self.metadata_repo = GitInfo(
+            config.get("git-archive", "metadata-repo"),
+        )
+
+    def generate(self) -> Iterable[SpecOutput]:
+        # Base query used by the RPC.
+        base_query = (
+            db.query(Package)
+            .join(PackageBase)
+            .join(User, PackageBase.MaintainerUID == User.ID)
+        )
+
+        # Create an instance of RPC, use it to get entities from
+        # our query and perform a metadata subquery for all packages.
+        rpc = RPC(version=5, type="info")
+        print("performing package database query")
+        packages = rpc.entities(base_query).all()
+        print("performing package database subqueries")
+        rpc.subquery({pkg.ID for pkg in packages})
+
+        pkgbases, pkgnames = dict(), dict()
+        for package in packages:
+            # Produce RPC type=info data for `package`
+            data = rpc.get_info_json_data(package)
+
+            pkgbase_name = data.get("PackageBase")
+            pkgbase_data = {
+                "ID": data.pop("PackageBaseID"),
+                "URLPath": data.pop("URLPath"),
+                "FirstSubmitted": data.pop("FirstSubmitted"),
+                "LastModified": data.pop("LastModified"),
+                "OutOfDate": data.pop("OutOfDate"),
+                "Maintainer": data.pop("Maintainer"),
+                "Keywords": data.pop("Keywords"),
+                "NumVotes": data.pop("NumVotes"),
+                "Popularity": data.pop("Popularity"),
+                "PopularityUpdated": package.PopularityUpdated.timestamp(),
+            }
+
+            # Store the data in `pkgbases` dict. We do this so we only
+            # end up processing a single `pkgbase` if repeated after
+            # this loop
+            pkgbases[pkgbase_name] = pkgbase_data
+
+            # Remove Popularity and NumVotes from package data.
+            # These fields change quite often which causes git data
+            # modification to explode.
+            # data.pop("NumVotes")
+            # data.pop("Popularity")
+
+            # Remove the ID key from package json.
+            data.pop("ID")
+
+            # Add the `package`.Name to the pkgnames set
+            name = data.get("Name")
+            pkgnames[name] = data
+
+        # Add metadata outputs
+        self.add_output(
+            "pkgname.json",
+            self.metadata_repo,
+            orjson.dumps(pkgnames, option=ORJSON_OPTS),
+        )
+        self.add_output(
+            "pkgbase.json",
+            self.metadata_repo,
+            orjson.dumps(pkgbases, option=ORJSON_OPTS),
+        )
+
+        return self.outputs

aurweb/archives/spec/pkgbases.py

@@ -0,0 +1,32 @@
+from typing import Iterable
+
+import orjson
+
+from aurweb import config, db
+from aurweb.models import PackageBase
+
+from .base import GitInfo, SpecBase, SpecOutput
+
+ORJSON_OPTS = orjson.OPT_SORT_KEYS | orjson.OPT_INDENT_2
+
+
+class Spec(SpecBase):
+    def __init__(self) -> "Spec":
+        self.pkgbases_repo = GitInfo(config.get("git-archive", "pkgbases-repo"))
+
+    def generate(self) -> Iterable[SpecOutput]:
+        filt = PackageBase.PackagerUID.isnot(None)
+        query = (
+            db.query(PackageBase.Name)
+            .filter(filt)
+            .order_by(PackageBase.Name.asc())
+            .all()
+        )
+        pkgbases = [pkgbase.Name for pkgbase in query]
+
+        self.add_output(
+            "pkgbase.json",
+            self.pkgbases_repo,
+            orjson.dumps(pkgbases, option=ORJSON_OPTS),
+        )
+        return self.outputs

aurweb/archives/spec/pkgnames.py

@@ -0,0 +1,33 @@
+from typing import Iterable
+
+import orjson
+
+from aurweb import config, db
+from aurweb.models import Package, PackageBase
+
+from .base import GitInfo, SpecBase, SpecOutput
+
+ORJSON_OPTS = orjson.OPT_SORT_KEYS | orjson.OPT_INDENT_2
+
+
+class Spec(SpecBase):
+    def __init__(self) -> "Spec":
+        self.pkgnames_repo = GitInfo(config.get("git-archive", "pkgnames-repo"))
+
+    def generate(self) -> Iterable[SpecOutput]:
+        filt = PackageBase.PackagerUID.isnot(None)
+        query = (
+            db.query(Package.Name)
+            .join(PackageBase, PackageBase.ID == Package.PackageBaseID)
+            .filter(filt)
+            .order_by(Package.Name.asc())
+            .all()
+        )
+        pkgnames = [pkg.Name for pkg in query]
+
+        self.add_output(
+            "pkgname.json",
+            self.pkgnames_repo,
+            orjson.dumps(pkgnames, option=ORJSON_OPTS),
+        )
+        return self.outputs

aurweb/archives/spec/users.py

@@ -0,0 +1,26 @@
+from typing import Iterable
+
+import orjson
+
+from aurweb import config, db
+from aurweb.models import User
+
+from .base import GitInfo, SpecBase, SpecOutput
+
+ORJSON_OPTS = orjson.OPT_SORT_KEYS | orjson.OPT_INDENT_2
+
+
+class Spec(SpecBase):
+    def __init__(self) -> "Spec":
+        self.users_repo = GitInfo(config.get("git-archive", "users-repo"))
+
+    def generate(self) -> Iterable[SpecOutput]:
+        query = db.query(User.Username).order_by(User.Username.asc()).all()
+        users = [user.Username for user in query]
+
+        self.add_output(
+            "users.json",
+            self.users_repo,
+            orjson.dumps(users, option=ORJSON_OPTS),
+        )
+        return self.outputs