Add a --prune flag to just filter packages within a name

conda_subchannel/cli.py

@@ -100,6 +100,12 @@ def configure_parser(parser: argparse.ArgumentParser):
         action="append",
         help="Keep packages matching this spec only. Can be used several times.",
     )
+    parser.add_argument(
+        "--prune",
+        metavar="SPEC",
+        action="append",
+        help="Remove the distributions of this package name that do not match this spec.",
+    )
     parser.add_argument(
         "--remove",
         metavar="SPEC",
@@ -109,7 +115,7 @@ def configure_parser(parser: argparse.ArgumentParser):
 
 
 def execute(args: argparse.Namespace) -> int:
-    if not any([args.after, args.before, args.keep, args.remove, args.keep_tree]):
+    if not any([args.after, args.before, args.keep, args.remove, args.keep_tree, args.prune]):
         raise ArgumentError("Please provide at least one filter.")
 
     with Spinner("Syncing source channel"):
@@ -124,6 +130,7 @@ def execute(args: argparse.Namespace) -> int:
         "subdir_datas": subdir_datas,
         "specs_to_keep": args.keep,
         "specs_to_remove": args.remove,
+        "specs_to_prune": args.prune,
         "trees_to_keep": args.keep_tree,
         "after": args.after,
         "before": args.before,

conda_subchannel/core.py

@@ -74,12 +74,14 @@ def _reduce_index(
     subdir_datas: Iterable[SubdirData],
     specs_to_keep: Iterable[str | MatchSpec] | None = None,
     specs_to_remove: Iterable[str | MatchSpec] | None = None,
+    specs_to_prune: Iterable[str | MatchSpec] | None = None,
     trees_to_keep: Iterable[str | MatchSpec] | None = None,
     after: int | None = None,
     before: int | None = None,
 ) -> dict[tuple[str, str], PackageRecord]:
     specs_to_keep = [MatchSpec(spec) for spec in (specs_to_keep or ())]
     specs_to_remove = [MatchSpec(spec) for spec in (specs_to_remove or ())]
+    specs_to_prune = [MatchSpec(spec) for spec in (specs_to_prune or ())]
     trees_to_keep = [MatchSpec(spec) for spec in (trees_to_keep or ())]
     if trees_to_keep or specs_to_keep or after is not None or before is not None:
         records = {}
@@ -136,10 +138,22 @@ def _reduce_index(
 
     # Now that we know what to keep, we remove stuff
     to_remove = set()
+
+    # Of the packages that survived the keeping, we will remove the ones that do not match the
+    # prune filter
+    for spec in specs_to_prune:
+        for key, record in records.items():
+            if spec.name != record.name:
+                continue  # ignore if the name doesn't match
+            if not spec.match(record):
+                to_remove.add(key)
+
+    # These are the explicit removals; if you match this, you are out
     for spec in specs_to_remove:
         for key, record in records.items():
             if spec.match(record):
                 to_remove.add(key)
+
     for key in to_remove:
         records.pop(key)
 

docs/usage.md

@@ -1,5 +1,7 @@
 # Usage
 
+## CLI
+
 ```
 $ conda subchannel --help
 usage: conda subchannel -c CHANNEL [--repodata-fn REPODATA_FN] [--base-url BASE_URL] [--output PATH] [--subdir PLATFORM] [--after TIME] [--before TIME] [--keep-tree SPEC] [--keep SPEC] [--remove SPEC] [-h]
@@ -22,6 +24,23 @@ options:
   --keep-tree SPEC      Keep packages matching this spec and their dependencies. Can be used
                         several times.
   --keep SPEC           Keep packages matching this spec only. Can be used several times.
+  --prune SPEC          Remove the distributions of this package name that do not match the
+                        given constraints.
   --remove SPEC         Remove packages matching this spec. Can be used several times.
   -h, --help            Show this help message and exit.
-  ```
+  ```
+
+
+## Filtering algorithm
+
+The filtering algorithm operates in two phases: selection
+
+In the first phase, we _select_ which records are going to be kept. Everything else is removed.
+
+1. A selection list is built. Records in this list are added if:
+  - They match specs in `--keep-tree`, or any of the dependencies in their tree (assessed recursively).
+  - They match any of the specs in `--keep`.
+  - Their timestamp is within the limits marked by `--before` and `--after`, when applicable. 
+2. At this point, records that didn't make it to the selection list are removed.
+3. The specs defined `--prune` are processed. Records that have the same name but don't match the spec are removed. Everything else is ignored.
+4. Records matching any of the specs in `--remove` are filtered out.

tests/test_subchannel.py

@@ -46,7 +46,7 @@ def test_only_python(conda_cli, tmp_path):
     assert tested
 
 
-def test_python_tree(conda_cli, tmp_path, monkeypatch):
+def test_python_tree(conda_cli, tmp_path):
     spec = "python=3.9"
     channel_path = tmp_path / "channel"
     out, err, rc = conda_cli(
@@ -104,6 +104,19 @@ def test_python_tree(conda_cli, tmp_path, monkeypatch):
             "python=3.10",
         )
 
+    # This should fail too; nodejs doesn't match python=3.9, so it must be out
+    with pytest.raises(PackagesNotFoundError):
+        conda_cli(
+            "create",
+            "--dry-run",
+            "-n",
+            "unused",
+            "--override-channels",
+            "--channel",
+            channel_path,
+            "nodejs",
+        )
+
 
 def test_not_python(conda_cli, tmp_path):
     out, err, rc = conda_cli(
@@ -253,3 +266,60 @@ def test_served_at(conda_cli, tmp_path):
 
     for path in tmp_path.glob("**/index.html"):
         assert served_at in path.read_text()
+
+
+def test_pruned_python(conda_cli, tmp_path):
+    spec = "python=3.9"
+    channel_path = tmp_path / "channel"
+    out, err, rc = conda_cli(
+        "subchannel",
+        "-c",
+        "conda-forge",
+        "--prune",
+        spec,
+        "--output",
+        channel_path,
+    )
+    print(out)
+    print(err, file=sys.stderr)
+    assert rc == 0
+
+    # This should be solvable, we didn't remove anything other than non-39 pythons
+    with pytest.raises(DryRunExit):
+        conda_cli(
+            "create",
+            "--dry-run",
+            "-n",
+            "unused",
+            "--override-channels",
+            "--channel",
+            channel_path,
+            "python=3.9",
+        )
+
+    # This should be unsolvable, we didn't take Python 3.10 in the subchannel
+    with pytest.raises(PackagesNotFoundError):
+        conda_cli(
+            "create",
+            "--dry-run",
+            "-n",
+            "unused",
+            "--override-channels",
+            "--channel",
+            channel_path,
+            "python=3.10",
+        )
+
+    # This should work because, we just removed pythons that are not python=3.9, but the rest
+    # of the conda-forge packages should be there
+    with pytest.raises(DryRunExit):
+        conda_cli(
+            "create",
+            "--dry-run",
+            "-n",
+            "unused",
+            "--override-channels",
+            "--channel",
+            channel_path,
+            "nodejs",
+        )