Add combine-durations composite action (#126)

Originally implemented in conda/conda we port the combine-durations
composite action over to conda/actions so other repositories (like
conda/conda-build) can also benefit from this workflow.

This workflow will load the duration artifacts from the 10 most recent
runs for a given workflow, aggregate and average the durations, commit the
updated duration files, and finally open a PR with the changes.

.pre-commit-config.yaml

@@ -11,6 +11,13 @@ repos:
       - id: check-yaml
       # catch git merge/rebase problems
       - id: check-merge-conflict
+      # sort requirements files
+      - id: file-contents-sorter
+        files: |
+          (?x)^(
+            combine-durations/requirements\.txt
+          )
+        args: [--unique]
   - repo: https://github.com/macisamuele/language-formatters-pre-commit-hooks
     rev: v2.13.0
     hooks:

combine-durations/README.md

@@ -0,0 +1,27 @@
+# Combine Durations
+
+A composite GitHub Action to combine the duration files from recent pytest runs.
+
+## GitHub Action Usage
+
+In your GitHub repository include this action in your workflows:
+
+```yaml
+- uses: conda/actions/combine-durations
+  with:
+    # [optional]
+    # the git branch to search
+    branch: main
+
+    # [optional]
+    # the glob pattern to search for duration files
+    pattern: '*-all'
+
+    # [optional]
+    # the GitHub token with artifact downloading permissions
+    token: ${{ github.token }}
+
+    # [optional]
+    # the workflow to search
+    workflow: tests.yml
+```

combine-durations/action.py

@@ -0,0 +1,86 @@
+"""Combine test durations from all recent runs."""
+from __future__ import annotations
+
+import json
+from argparse import ArgumentParser, ArgumentTypeError, Namespace
+from functools import partial
+from pathlib import Path
+from statistics import fmean
+
+from rich.console import Console
+
+print = Console(color_system="standard", soft_wrap=True).print
+
+
+def validate_dir(value: str, writable: bool = False) -> Path:
+    try:
+        path = Path(value).expanduser().resolve()
+        path.mkdir(parents=True, exist_ok=True)
+        if writable:
+            ignore = path / ".ignore"
+            ignore.touch()
+            ignore.unlink()
+        return path
+    except (FileExistsError, PermissionError) as err:
+        # FileExistsError: value is a file, not a directory
+        # PermissionError: value is not writable
+        raise ArgumentTypeError(f"{value} is not a valid directory: {err}")
+
+
+def parse_args() -> Namespace:
+    # parse CLI for inputs
+    parser = ArgumentParser()
+    parser.add_argument("--durations-dir", type=validate_dir, required=True)
+    parser.add_argument(
+        "--artifacts-dir",
+        type=partial(validate_dir, writable=True),
+        required=True,
+    )
+    return parser.parse_args()
+
+
+def main() -> None:
+    args = parse_args()
+
+    # aggregate new durations
+    combined: dict[str, dict[str, list[float]]] = {}
+    for path in args.artifacts_dir.glob("**/*.json"):
+        os_combined = combined.setdefault((OS := path.stem), {})
+
+        new_data = json.loads(path.read_text())
+        for key, value in new_data.items():
+            os_combined.setdefault(key, []).append(value)
+
+    # aggregate old durations
+    for path in args.durations_dir.glob("*.json"):
+        try:
+            os_combined = combined[(OS := path.stem)]
+        except KeyError:
+            # KeyError: OS not present in new durations
+            print(f"⚠️ {OS} not present in new durations, removing")
+            path.unlink()
+            continue
+
+        old_data = json.loads(path.read_text())
+        if missing := set(old_data) - set(combined[OS]):
+            for name in missing:
+                print(f"⚠️ {OS}::{name} not present in new durations, removing")
+
+        # only copy over keys that are still present in new durations
+        for key in set(old_data) & set(combined[OS]):
+            os_combined[key].append(old_data[key])
+
+    # drop durations no longer present in new durations and write out averages
+    for OS, os_combined in combined.items():
+        (args.durations_dir / f"{OS}.json").write_text(
+            json.dumps(
+                {key: fmean(values) for key, values in os_combined.items()},
+                indent=4,
+                sort_keys=True,
+            )
+            + "\n"  # include trailing newline
+        )
+
+
+if __name__ == "__main__":
+    main()

combine-durations/action.yml

@@ -0,0 +1,66 @@
+name: Combine Durations
+description: Combine the duration files from recent pytest runs.
+author: Anaconda Inc.
+branding:
+  icon: book-open
+  color: green
+
+inputs:
+  branch:
+    description: The branch to search for recent pytest runs.
+    default: main
+  durations-dir:
+    description: The directory to write the combined durations file.
+    default: durations
+  pattern:
+    description: The pattern to search for recent pytest runs.
+    default: '*-all'
+  token:
+    description: >-
+      A token with ability to comment, label, and modify the commit status
+      (`pull_request: write` and `statuses: write` for fine-grained PAT; `repo` for classic PAT)
+    default: ${{ github.token }}
+  workflow:
+    description: The workflow to search for recent pytest runs.
+    default: tests.yml
+
+runs:
+  using: composite
+  steps:
+    - name: download recent artifacts
+      shell: bash
+      run: >
+        gh run list
+        --branch ${{ inputs.branch }}
+        --workflow ${{ inputs.workflow }}
+        --limit 10
+        --json databaseId
+        --jq '.[].databaseId'
+        | xargs -n 1
+        gh run download
+        --dir ${{ runner.temp }}/artifacts/
+        --pattern '${{ inputs.pattern }}'
+        || true
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+
+    - uses: actions/cache@0c45773b623bea8c8e75f6c82b208c3cf94ea4f9 # v4.0.2
+      with:
+        path: ~/.cache/pip
+        # invalidate the cache anytime a workflow changes
+        key: ${{ hashFiles('.github/workflows/*') }}
+
+    - uses: actions/setup-python@82c7e631bb3cdc910f68e0081d67478d79c6982d # v5.1.0
+      with:
+        python-version: '3.11'
+
+    - name: install dependencies
+      shell: bash
+      run: pip install --quiet -r ${{ github.action_path }}/requirements.txt
+
+    - name: combine recent durations from artifacts
+      shell: bash
+      run: >
+        python ${{ github.action_path }}/action.py
+        --durations-dir=${{ inputs.durations-dir }}
+        --artifacts-dir=${{ runner.temp }}/artifacts/

combine-durations/requirements.txt

@@ -0,0 +1 @@
+rich