Image sorting

harnesses/scan/dockerhub.py

@@ -2,13 +2,23 @@
 
 import os
 
+from rsp_reaper.config import RegistryAuth, RegistryConfig, KeepPolicy
 from rsp_reaper.storage.dockerhub import DockerHubClient
 
 # We want to explode if the auth isn't set.
-username = os.environ["DOCKERHUB_USER"]
-password = os.environ["DOCKERHUB_PASSWORD"]
-c = DockerHubClient(
-    namespace="lsstsqre", repository="sciplat-lab", dry_run=True
+auth = RegistryAuth(
+    username=os.environ["DOCKERHUB_USER"],
+    password=os.environ["DOCKERHUB_PASSWORD"],
 )
-c.authenticate(username=username, password=password)
+cfg = RegistryConfig(
+    category="hub.docker.com",
+    registry="https://docker.io",
+    owner="lsstsqre",
+    repository="sciplat-lab",
+    keep=KeepPolicy(),
+    debug=True,
+    dry_run=True
+)
+c = DockerHubClient(cfg=cfg)
+c.authenticate(auth)
 c.scan_repo()

harnesses/scan/gar.py

@@ -1,12 +1,18 @@
 """Interactive harness for Google Artifact Registry."""
 
+from rsp_reaper.config import KeepPolicy, RegistryConfig
 from rsp_reaper.storage.gar import GARClient
 
-c = GARClient(
-    location="us-central1",
-    project_id="rubin-shared-services-71ec",
-    repository="sciplat",
-    image="sciplat-lab",
+cfg = RegistryConfig(
+    category="pkg.dev",
+    registry="https://us-central1-docker.pkg.dev",
+    owner="rubin-shared-services-71ec",
+    namespace="sciplat",
+    repository="sciplat-lab",
+    keep = KeepPolicy(),    
     dry_run=True,
+    debug=True,
 )
+c = GARClient(cfg=cfg)
+# No authentication; set up application default credentials in the environment
 c.scan_repo()

harnesses/scan/ghcr.py

@@ -2,10 +2,19 @@
 
 import os
 
+from rsp_reaper.config import RegistryAuth, RegistryConfig, KeepPolicy
 from rsp_reaper.storage.ghcr import GhcrClient
 
 # We want to explode if the token isn't set.
-token = os.environ["GHCR_TOKEN"]
-c = GhcrClient(namespace="lsst-sqre", repository="sciplat-lab", dry_run=True)
-c.authenticate(token=token)
+auth = RegistryAuth(password=os.environ["GHCR_TOKEN"])
+cfg=RegistryConfig(
+        category="ghcr.io",
+        registry="https://ghcr.io",
+        owner="lsst-sqre",
+        repository="sciplat-lab",
+        keep=KeepPolicy(),
+        dry_run=True
+    )
+c = GhcrClient(cfg=cfg)
+c.authenticate(auth)
 c.scan_repo()

harnesses/sort/gar.py

@@ -0,0 +1,26 @@
+"""Interactive harness for Google Artifact Registry."""
+from pathlib import Path
+from rsp_reaper.config import KeepPolicy, RegistryConfig
+from rsp_reaper.models.image import ImageVersionClass
+from rsp_reaper.storage.gar import GARClient
+
+input_file = ( Path(__file__).parent.parent.parent / "tests" / "support" /
+               "gar.contents.json")
+
+cfg = RegistryConfig(
+    category="pkg.dev",
+    registry="https://us-central1-docker.pkg.dev",
+    owner="rubin-shared-services-71ec",
+    namespace="sciplat",
+    repository="sciplat-lab",
+    dry_run=True,
+    debug=True,
+    input_file=input_file,
+    keep = KeepPolicy()
+)
+
+c = GARClient(cfg=cfg)
+c.categorize(ImageVersionClass.RSP)
+
+
+

pyproject.toml

@@ -118,8 +118,14 @@ python_files = [
 extend = "ruff-shared.toml"
 
 [tool.ruff.lint.per-file-ignores]
-"src/jupyterlabcontroller/services/**" = [
-    "S108",    # constructing /tmp paths for Kubernetes Pods
+"src/rsp_reaper/models/rsptag.py" = [
+    "C901",    # yes, compare() does have a lot of cases
+]
+
+"tests/**.py" = [
+    "SLF001",  # Tests are allowed to access private members
+    "T201",    # Tests are allowed to print
+    "S101",    # Tests can use assert
 ]
 
 [tool.ruff.lint.isort]

src/rsp_reaper/config.py

@@ -1,26 +1,98 @@
 """Configuration for a reaper for a particular container registry."""
 
-from dataclasses import dataclass
-
-from .models.registry_category import RegistryCategory
+from dataclasses import dataclass, field
+from pathlib import Path
 
 
 @dataclass
 class RegistryAuth:
     """Generic authentication item for a container registry."""
 
-    realm: str
-    username: str
-    password: str
+    realm: str|None = None
+    username: str|None = None
+    password: str|None = None
+
+
+class LatestSemverKeepers:
+    """Number of items to keep for latest major semver release.  Default is
+    "do not purge".
+    """
+
+    minor: int | None = None
+    patch: int | None = None
+    build: int | None = None
+
+
+class OlderSemverKeepers:
+    """Number of items to keep for previous major semver releases.  Default is
+    "do not purge".
+    """
+
+    major: int | None = None
+    minor: int | None = None
+    patch: int | None = None
+    build: int | None = None
+
+
+@dataclass
+class SemverKeepers:
+    """Within each of latest_major and older, how many minor versions,
+    how many patch versions within each of those, and how many builds of each
+    of those, to keep.  Older also has a major version number.  For instance,
+    older.major might be 3, and then when version 5.0 came out, you would
+    keep some images for the 2.x.y, 3.x.y, and 4.x.y series, but no 1.x images.
+    """
+
+    latest_major: LatestSemverKeepers = field(
+        default_factory=LatestSemverKeepers
+    )
+    older: OlderSemverKeepers = field(default_factory=OlderSemverKeepers)
+
+
+@dataclass
+class RSPKeepers:
+    """Aliases are never purged.  Default for everything else is "do not
+    purge".
+    """
+
+    release: int | None = None
+    weekly: int | None = None
+    daily: int | None = None
+    release_candidate: int | None = None
+    experimental: int | None = None
+    unknown: int | None = None
+
+
+@dataclass
+class KeepPolicy:
+    """How many of each image category to keep.  `-1` or `None` means
+    "don't reap that category at all".  `0` means "purge them all".
+    The default is to purge nothing.
+    """
+
+    untagged: int | None = None
+    semver: SemverKeepers | None = field(default_factory=SemverKeepers)
+    rsp: RSPKeepers | None = field(default_factory=RSPKeepers)
 
 
 @dataclass
-class ContainerRegistryConfig:
-    """Configuration for a particular container registry."""
-
-    namespace: str
-    repository: str
-    registry: str
-    category: RegistryCategory
-    project: str | None = None
+class RegistryConfig:
+    """Configuration to talk to a particular container registry."""
+
+    registry: str  # URL of registry host
+    owner: str  # Usually repo owner; at GAR, project ID
+    repository: str  # Repository name, e.g. "sciplat-lab"
+    category: str
+    keep: KeepPolicy
+    namespace: str | None = None  # Intermediate layer if any; GAR: "sciplat"
     auth: RegistryAuth | None = None
+    dry_run: bool = True
+    debug: bool = True
+    input_file: Path | None = None
+
+
+@dataclass
+class Config:
+    """Configuration for multiple registries."""
+
+    registries: list[RegistryConfig]

src/rsp_reaper/factory.py

@@ -9,15 +9,15 @@
 import structlog
 from structlog.stdlib import BoundLogger
 
-from .config import ContainerRegistryConfig
+from .config import RegistryConfig
 
 
 class Factory:
     """Build reaper components.
 
     Parameters
     ----------
-    config: ContainerRegistryConfig
+    config: RegistryConfig
         Reaper configuration.
 
     logger
@@ -26,17 +26,15 @@ class Factory:
 
     @classmethod
     @asynccontextmanager
-    async def standalone(
-        cls, config: ContainerRegistryConfig
-    ) -> AsyncIterator[Self]:
+    async def standalone(cls, config: RegistryConfig) -> AsyncIterator[Self]:
         """Async context manager for reaper components.
 
         Intended for the test suite.
 
         Parameters
         ----------
         config
-            Container Registry configuration
+            Registry configuration
 
         Yields
         ------