From a3e5250143dabd48c3b41676ece78b9687e1659f Mon Sep 17 00:00:00 2001
From: ViditOstwal <viditostwal@gmail.com>
Date: Tue, 15 Sep 2026 23:01:13 +0530
Subject: [PATCH] ci: add test duration profiling workflow

---
 .github/workflows/profile-test-durations.yml  |  60 ++++++++
 .../scripts/test_aggregate_test_durations.py  |  60 ++++++++
 scripts/aggregate_test_durations.py           | 140 ++++++++++++++++++
 3 files changed, 260 insertions(+)
 create mode 100644 .github/workflows/profile-test-durations.yml
 create mode 100644 lib/crewai/tests/scripts/test_aggregate_test_durations.py
 create mode 100644 scripts/aggregate_test_durations.py

diff --git a/.github/workflows/profile-test-durations.yml b/.github/workflows/profile-test-durations.yml
new file mode 100644
index 000000000..53e276f9f
--- /dev/null
+++ b/.github/workflows/profile-test-durations.yml
@@ -0,0 +1,60 @@
+name: Profile Test Durations
+
+on:
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  profile:
+    name: Profile all tests (Python 3.11)
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    env:
+      PYTHONUNBUFFERED: "1"
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4.3.1
+        with:
+          # Test profiling does not need history; avoiding a full fetch makes the
+          # measured runs cheaper without affecting pytest's timings.
+          fetch-depth: 1
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@d0cc045d04ccac9d8b7881df0226f9e82c39688e # v6
+        with:
+          version: "0.11.3"
+          python-version: "3.11"
+          enable-cache: true
+
+      - name: Install dependencies
+        run: uv sync --all-groups --all-extras
+
+      - name: Profile CrewAI tests
+        run: |
+          mkdir -p test-duration-profile
+          cd lib/crewai
+          uv run pytest \
+            --store-durations \
+            --clean-durations \
+            --durations-path=../../test-duration-profile/crewai-durations.json \
+            --durations=0
+
+      - name: Profile CrewAI Tools tests
+        run: |
+          cd lib/crewai-tools
+          uv run pytest \
+            --store-durations \
+            --clean-durations \
+            --durations-path=../../test-duration-profile/crewai-tools-durations.json \
+            --durations=0
+
+      - name: Upload duration profile
+        if: always()
+        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
+        with:
+          name: test-duration-profile-${{ github.run_id }}
+          path: test-duration-profile/
+          if-no-files-found: error
diff --git a/lib/crewai/tests/scripts/test_aggregate_test_durations.py b/lib/crewai/tests/scripts/test_aggregate_test_durations.py
new file mode 100644
index 000000000..6b9318852
--- /dev/null
+++ b/lib/crewai/tests/scripts/test_aggregate_test_durations.py
@@ -0,0 +1,60 @@
+from __future__ import annotations
+
+import importlib.util
+import json
+from pathlib import Path
+
+
+SCRIPT_PATH = Path(__file__).parents[4] / "scripts" / "aggregate_test_durations.py"
+SPEC = importlib.util.spec_from_file_location("aggregate_test_durations", SCRIPT_PATH)
+assert SPEC is not None and SPEC.loader is not None
+MODULE = importlib.util.module_from_spec(SPEC)
+SPEC.loader.exec_module(MODULE)
+
+
+def test_summarize_aggregates_repeated_profiles_by_suite_and_nodeid(
+    tmp_path: Path,
+) -> None:
+    first = tmp_path / "first" / "crewai-durations.json"
+    second = tmp_path / "second" / "crewai-durations.json"
+    first.parent.mkdir()
+    second.parent.mkdir()
+    first.write_text(
+        json.dumps(
+            {
+                "tests/test_fast.py::test_fast": 0.1,
+                "tests/test_slow.py::test_slow": 4.0,
+            }
+        )
+    )
+    second.write_text(
+        json.dumps(
+            {
+                "tests/test_fast.py::test_fast": 0.3,
+                "tests/test_slow.py::test_slow": 6.0,
+            }
+        )
+    )
+
+    rows = MODULE.summarize([tmp_path])
+
+    assert rows == [
+        {
+            "suite": "crewai",
+            "nodeid": "tests/test_slow.py::test_slow",
+            "samples": 2,
+            "mean_seconds": 5.0,
+            "median_seconds": 5.0,
+            "p95_seconds": 6.0,
+            "max_seconds": 6.0,
+        },
+        {
+            "suite": "crewai",
+            "nodeid": "tests/test_fast.py::test_fast",
+            "samples": 2,
+            "mean_seconds": 0.2,
+            "median_seconds": 0.2,
+            "p95_seconds": 0.3,
+            "max_seconds": 0.3,
+        },
+    ]
diff --git a/scripts/aggregate_test_durations.py b/scripts/aggregate_test_durations.py
new file mode 100644
index 000000000..bc9dd0c89
--- /dev/null
+++ b/scripts/aggregate_test_durations.py
@@ -0,0 +1,140 @@
+#!/usr/bin/env python3
+"""Aggregate pytest-split duration artifacts from repeated profiling runs.
+
+Download one or more ``test-duration-profile-*`` artifacts from GitHub Actions,
+then pass either the downloaded directories or individual JSON files to this
+script. It writes one CSV row per test with the observed sample count, mean,
+median, p95, and maximum runtime in seconds.
+"""
+
+from __future__ import annotations
+
+import argparse
+from collections import defaultdict
+from collections.abc import Iterable, Sequence
+import csv
+import json
+from pathlib import Path
+from statistics import fmean, median
+import sys
+from typing import Any
+
+
+CSV_FIELDS = (
+    "suite",
+    "nodeid",
+    "samples",
+    "mean_seconds",
+    "median_seconds",
+    "p95_seconds",
+    "max_seconds",
+)
+
+
+def duration_files(paths: Iterable[Path]) -> list[Path]:
+    """Return the duration JSON files supplied directly or below supplied directories."""
+    files: list[Path] = []
+    for path in paths:
+        if path.is_dir():
+            files.extend(sorted(path.rglob("*-durations.json")))
+        elif path.is_file() and path.name.endswith("-durations.json"):
+            files.append(path)
+        else:
+            raise ValueError(f"Expected a duration JSON file or directory: {path}")
+    if not files:
+        raise ValueError("No *-durations.json files found")
+    return files
+
+
+def suite_name(path: Path) -> str:
+    """Infer the suite from the artifact filename created by the profiling workflow."""
+    return path.name.removesuffix("-durations.json")
+
+
+def load_durations(path: Path) -> dict[str, float]:
+    """Load and validate a pytest-split duration map."""
+    try:
+        raw: Any = json.loads(path.read_text())
+    except json.JSONDecodeError as error:
+        raise ValueError(f"Invalid JSON in {path}: {error}") from error
+
+    if not isinstance(raw, dict):
+        raise ValueError(f"Expected a JSON object in {path}")
+
+    durations: dict[str, float] = {}
+    for nodeid, duration in raw.items():
+        if not isinstance(nodeid, str) or not isinstance(duration, (int, float)):
+            raise ValueError(f"Invalid test duration in {path}: {nodeid!r}")
+        if duration < 0:
+            raise ValueError(f"Negative test duration in {path}: {nodeid}")
+        durations[nodeid] = float(duration)
+    return durations
+
+
+def percentile_95(values: Sequence[float]) -> float:
+    """Return the nearest-rank p95 for a non-empty sequence."""
+    ordered = sorted(values)
+    index = max(0, (95 * len(ordered) + 99) // 100 - 1)
+    return ordered[index]
+
+
+def summarize(paths: Iterable[Path]) -> list[dict[str, str | int | float]]:
+    """Summarize all durations by suite and pytest node ID."""
+    samples: dict[tuple[str, str], list[float]] = defaultdict(list)
+    for path in duration_files(paths):
+        suite = suite_name(path)
+        for nodeid, duration in load_durations(path).items():
+            samples[(suite, nodeid)].append(duration)
+
+    rows: list[dict[str, str | int | float]] = []
+    for (suite, nodeid), values in samples.items():
+        rows.append(
+            {
+                "suite": suite,
+                "nodeid": nodeid,
+                "samples": len(values),
+                "mean_seconds": fmean(values),
+                "median_seconds": median(values),
+                "p95_seconds": percentile_95(values),
+                "max_seconds": max(values),
+            }
+        )
+    return sorted(
+        rows,
+        key=lambda row: (float(row["p95_seconds"]), float(row["max_seconds"])),
+        reverse=True,
+    )
+
+
+def write_csv(rows: Iterable[dict[str, str | int | float]], output: Path) -> None:
+    """Write summaries as a stable, spreadsheet-friendly CSV file."""
+    with output.open("w", newline="") as file:
+        writer = csv.DictWriter(file, fieldnames=CSV_FIELDS)
+        writer.writeheader()
+        for row in rows:
+            writer.writerow(row)
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "paths",
+        nargs="+",
+        type=Path,
+        help="Duration JSON files or directories containing them",
+    )
+    parser.add_argument(
+        "--output",
+        type=Path,
+        default=Path("test-duration-summary.csv"),
+        help="Output CSV path",
+    )
+    arguments = parser.parse_args()
+
+    rows = summarize(arguments.paths)
+    write_csv(rows, arguments.output)
+    sys.stdout.write(f"Wrote {len(rows)} test summaries to {arguments.output}\n")
+
+
+if __name__ == "__main__":
+    main()