Use full SHA (#38)

Author: rhshadrach

.github/workflows/run_asvs.yaml

@@ -13,7 +13,7 @@ env:
   # Results and other data are stored away from the main branch.
   # This keeps the main branch clean and allows us to clear out
   # objects / history after being built up for long periods of time.
-  BRANCH_NAME: pandas_20250109
+  BRANCH_NAME: pandas_20250203
 
 permissions:
   contents: read
@@ -29,7 +29,7 @@ jobs:
       contents: write
       issues: write
     outputs:
-      new_commit: ${{ steps.new-commit.outputs.new_commit }}
+      new_commit: ${{ steps.get-commit.outputs.new_commit }}
     steps:
       # In order to run pandas' actions, we have to checkout into the root directory.
       - name: Checkout pandas
@@ -44,60 +44,62 @@ jobs:
           ref: ${{ env.BRANCH_NAME }}
           path: asv-runner/
 
-      - name: Compare Commit SHAs
-        id: new-commit
+      - name: Set up Conda
+        uses: ./.github/actions/setup-conda
+
+      - name: Get Commit
+        id: get-commit
         run: |
-          if [ "$(git rev-parse HEAD)" = "$(cat asv-runner/data/latest_sha.txt)" ]; then
+          sha="$(python asv-runner/ci/find_commit_to_run.py --input-path=asv-runner/data/ --repo-path=.)"
+          echo "sha: $sha"
+          if [ "$sha" = "NONE" ]; then
             echo "new_commit=no"
             echo "new_commit=no" >> "$GITHUB_OUTPUT"
           else
             echo "new_commit=yes"
             echo "new_commit=yes" >> "$GITHUB_OUTPUT"
           fi
-          echo "$(git rev-parse HEAD)" > asv-runner/data/latest_sha.txt
+          echo "$sha" >> asv-runner/data/shas.txt
+          git checkout $sha
 
       # Prevent another job from kicking off and running on this commit
-      - name: Update latest_sha.txt
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+      - name: Commit shas.txt
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         uses: stefanzweifel/git-auto-commit-action@v5
         with:
-          commit_message: Update latest SHA
+          commit_message: Update shas.txt
           branch: ${{ env.BRANCH_NAME }}
           repository: asv-runner
-          file_pattern: 'data/latest_sha.txt'
-
-      - name: Set up Conda
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
-        uses: ./.github/actions/setup-conda
+          file_pattern: 'data/shas.txt'
 
       - name: Build pandas
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         uses: ./.github/actions/build_pandas
 
       - name: Run ASV Benchmarks
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         run: |
           cd asv_bench
           asv machine --machine=asvrunner --yes
           asv run --machine=asvrunner --python=same --set-commit-hash=$(git rev-parse HEAD) --show-stderr
 
       - name: Update asv-runner branch
         # In case there was a push by another job.
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         run: |
           cd asv-runner
           git fetch && git pull
 
       - name: Move results into asv-runner
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         run: |
           mkdir -p asv-runner/data/results/asvrunner/
           cp asv_bench/results/benchmarks.json asv-runner/data/results/
           cp asv_bench/results/asvrunner/machine.json asv-runner/data/results/asvrunner/
-          cp asv_bench/results/asvrunner/$(git rev-parse --short=8 HEAD)-existing*.json asv-runner/data/results/asvrunner/
+          cp asv_bench/results/asvrunner/$(git rev-parse --short=8 HEAD)-existing*.json asv-runner/data/results/asvrunner/$(git rev-parse HEAD).json
 
       - name: Commit results to branch
-        if: ${{ steps.new-commit.outputs.new_commit == 'yes' }}
+        if: ${{ steps.get-commit.outputs.new_commit == 'yes' }}
         uses: stefanzweifel/git-auto-commit-action@v5
         with:
           commit_message: Results

ci/find_commit_to_run.py

@@ -0,0 +1,43 @@
+from __future__ import annotations
+
+import argparse
+import subprocess
+from pathlib import Path
+
+
+def run(*, input_path: str | Path, repo_path: str | Path):
+    if isinstance(input_path, str):
+        input_path = Path(input_path)
+    if isinstance(repo_path, str):
+        repo_path = Path(repo_path)
+
+    shas_path = input_path / "shas.txt"
+    existing_shas: set[str]
+    if not shas_path.exists():
+        existing_shas = set()
+    else:
+        with open(input_path / "shas.txt") as fh:
+            existing_shas = {line.strip() for line in fh.readlines()}
+
+    response = subprocess.run(
+        f"cd {repo_path} && git log -40 --oneline --no-abbrev-commit",
+        capture_output=True,
+        shell=True,
+        check=False,
+    )
+    recent_shas = [
+        line[: line.find(" ")] for line in response.stdout.decode().strip().split("\n")
+    ]
+    for sha in recent_shas:
+        if sha not in existing_shas:
+            print(sha)
+            return
+    print("NONE")
+
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--input-path")
+    parser.add_argument("--repo-path")
+    args = parser.parse_args()
+    run(input_path=args.input_path, repo_path=args.repo_path)