Implement --base-url CLI flag (#9)

* add base_url argument

* fix repodata schema

* fix header

* fix readme

* add test

README.md

@@ -6,7 +6,7 @@ Create subsets of conda channels thanks to CEP-15 metadata
 
 ```bash
 $ conda install -n base conda-subchannel
-$ conda subchannel --channel=conda-forge python=3.9
+$ conda subchannel --channel=conda-forge --keep-tree python=3.9
 $ python -m http.serve --directory subchannel/
 ```
 

action.yml

@@ -8,6 +8,10 @@ inputs:
     description: "Source repodata file to process from channel"
     required: false
     default: "repodata.json"
+  base-url:
+    description: "URL where the packages will be available. Defaults to the base URL for 'channel'"
+    required: false
+    default: ""
   subdirs:
     description: "List of platforms to support, space separated. Defaults to linux-64. Noarch is always included" 
     required: false
@@ -91,6 +95,9 @@ runs:
             "--repodata-fn",
             """${{ inputs.repodata-fn }}""".strip(),
         ]
+        base_url = """${{ inputs.base-url }}""".strip()
+        if base_url:
+            args += ["--base_url", base_url]
         after = """${{ inputs.after }}""".strip()
         if after:
             args += ["--after", after]

conda_subchannel/cli.py

@@ -12,6 +12,7 @@
 from conda.base.constants import REPODATA_FN
 from conda.base.context import context
 from conda.common.io import Spinner
+from conda.models.channel import Channel
 
 from .core import _fetch_channel, _reduce_index, _dump_records, _write_to_disk
 
@@ -47,6 +48,13 @@ def configure_parser(parser: argparse.ArgumentParser):
         default=REPODATA_FN,
         help="Source repodata file to process from channel.",
     )
+    parser.add_argument(
+        "--base-url",
+        required=False,
+        help="URL where the packages will be available. "
+        "Defaults to the base URL for '--channel'. Only needed if the user wants to mirror "
+        "the required packages separately."
+    )
     parser.add_argument(
         "--output",
         default="subchannel",
@@ -123,7 +131,8 @@ def execute(args: argparse.Namespace) -> int:
     print(" - Reduced from", total_count, "to", filtered_count, "records")
 
     with Spinner(f"Writing output to {args.output}"):
-        repodatas = _dump_records(records, args.channel)
+        base_url = args.base_url or Channel(args.channel).base_url
+        repodatas = _dump_records(records, base_url)
         _write_to_disk(args.channel, repodatas, args.output)
 
     return 0

conda_subchannel/core.py

@@ -144,23 +144,40 @@ def _reduce_index(
 
 
 def _dump_records(
-    records: dict[tuple[str, str], PackageRecord], source_channel: Channel | str
+    records: dict[tuple[str, str], PackageRecord], base_url: str
 ) -> dict[str, dict[str, Any]]:
-    source_channel = Channel(source_channel)
     repodatas = {}
+    record_keys = (
+        "build",
+        "build_number",
+        "constrains",
+        "depends",
+        "license",
+        "md5",
+        "name",
+        "sha256",
+        "size",
+        "subdir",
+        "timestamp",
+        "version",
+    )
     for (subdir, filename), record in records.items():
         if subdir not in repodatas:
             repodatas[subdir] = {
-                "metadata": {
-                    "repodata_version": 2,
-                    "base_url": source_channel.base_url,
+                "repodata_version": 2,
+                "info": {
+                    "base_url": base_url,
                     "subdir": subdir,
                 },
                 "packages": {},
                 "packages.conda": {},
+                "removed": [],
             }
         key = "packages.conda" if record.fn.endswith(".conda") else "packages"
-        repodatas[record.subdir][key][filename] = record.dump()
+        dumped = record.dump()
+        repodatas[record.subdir][key][filename] = {
+            key: dumped[key] for key in record_keys if key in dumped
+        }
     return repodatas
 
 
@@ -206,9 +223,7 @@ def _write_subdir_index_md(subdir_path: Path):
         lastmod = datetime.fromtimestamp(stat.st_mtime)
         sha256 = _checksum(path, "sha256")
         md5 = _checksum(path, "md5")
-        lines.append(
-            f"| [{path.name}]({path.name}) | {size} | {lastmod} | `{sha256}` | `{md5}` |"
-        )
+        lines.append(f"| [{path.name}]({path.name}) | {size} | {lastmod} | `{sha256}` | `{md5}` |")
     lines.append("")
     lines.append(f"> Last modified on {datetime.now(tz=timezone.utc)}")
     (subdir_path / "index.md").write_text("\n".join(lines))

tests/test_subchannel.py

@@ -1,3 +1,4 @@
+import json
 import sys
 from datetime import datetime, timezone
 
@@ -210,3 +211,23 @@ def test_between_dates(conda_cli, tmp_path):
                 timestamp_2023 <= rec.timestamp <= timestamp_2024 for rec in sd.iter_records()
             )
     assert tested
+
+def test_base_url(conda_cli, tmp_path):
+    base_url = "https://a-redefined-base.url"
+    out, err, rc = conda_cli(
+        "subchannel",
+        "-c",
+        "conda-forge",
+        "--keep",
+        "python=3.9",
+        "--base-url",
+        base_url,
+        "--output",
+        tmp_path,
+    )
+    print(out)
+    print(err, file=sys.stderr)
+    assert rc == 0
+
+    data = json.loads((tmp_path / context.subdir / "repodata.json").read_text())
+    assert data["info"]["base_url"] == base_url