Merge pull request #175 from reacher-z/ci/run-tests

ci: run tests

Author: Perry2004

.github/workflows/run-test.yml

@@ -0,0 +1,46 @@
+name: run-test
+
+on:
+  pull_request:
+    branches: [ main ]
+    paths:
+      - ".github/workflows/run-test.yml"
+      - "src/clawbench/**"
+      - "tests/**"
+      - "pyproject.toml"
+      - "uv.lock"
+  push:
+    branches: [ main ]
+    paths:
+      - ".github/workflows/run-test.yml"
+      - "src/clawbench/**"
+      - "tests/**"
+      - "pyproject.toml"
+      - "uv.lock"
+  workflow_dispatch:
+
+jobs:
+  run-test:
+    name: pytest (${{ matrix.os }})
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os:
+          - ubuntu-latest
+          - macos-latest
+          - windows-latest
+
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Install uv
+        uses: astral-sh/setup-uv@v8.1.0
+
+      - name: Set up Python
+        uses: actions/setup-python@v6
+        with:
+          python-version: "3.11"
+
+      - name: Run host-side pytest suite
+        run: uv run --frozen pytest

tests/test_cli_entrypoints.py

@@ -5,6 +5,7 @@
 import subprocess
 import sys
 import textwrap
+import os
 from pathlib import Path
 
 import pytest
@@ -40,10 +41,13 @@ def test_module_help_does_not_require_container_runtime(module: str) -> None:
         runpy.run_module(module, run_name="__main__")
         """
     )
+    env = os.environ.copy()
+    env["PYTHONPATH"] = str(SRC_ROOT)
+
     result = subprocess.run(
         [sys.executable, "-c", code, module],
         capture_output=True,
-        env={"PYTHONPATH": str(SRC_ROOT)},
+        env=env,
         text=True,
         timeout=30,
     )

tests/test_host_resources.py

@@ -22,8 +22,12 @@ def test_prepare_personal_info_writes_container_mount_resources(tmp_path: Path)
         tmp_path,
     )
 
-    personal_info = json.loads((info_dir / "alex_green_personal_info.json").read_text())
-    email_credentials = json.loads((info_dir / "email_credentials.json").read_text())
+    personal_info = json.loads(
+        (info_dir / "alex_green_personal_info.json").read_text(encoding="utf-8")
+    )
+    email_credentials = json.loads(
+        (info_dir / "email_credentials.json").read_text(encoding="utf-8")
+    )
 
     assert personal_info["contact"]["email"] == "alex@example.test"
     assert "online_accounts" not in personal_info
@@ -40,7 +44,7 @@ def test_prepare_personal_info_writes_container_mount_resources(tmp_path: Path)
 
 def test_extra_info_copy_and_instruction_are_host_side(tmp_path: Path) -> None:
     task_dir = ASSET_ROOT / "test-cases" / "v1" / "007-daily-life-food-instacart"
-    task = json.loads((task_dir / "task.json").read_text())
+    task = json.loads((task_dir / "task.json").read_bytes())
     my_info = tmp_path / "my-info"
     my_info.mkdir()
 

tests/test_host_tasks.py

@@ -49,7 +49,7 @@ def test_checked_task_json_files_parse_and_validate(suite: str) -> None:
 
     assert task_files, f"{suite} should expose readable task JSON files"
     for task_file in task_files:
-        task = json.loads(task_file.read_text())
+        task = json.loads(task_file.read_bytes())
         validated = validate_task_data(task, task_file)
         assert validated is task