codeprobe 0.2.1__tar.gz → 0.2.3__tar.gz

{codeprobe-0.2.1 → codeprobe-0.2.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codeprobe
-Version: 0.2.1
+Version: 0.2.3
 Summary: Benchmark AI coding agents against your own codebase. Mine real tasks from repo history, run agents, interpret results.
 Author: codeprobe contributors
 License-Expression: Apache-2.0

{codeprobe-0.2.1 → codeprobe-0.2.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "codeprobe"
-version = "0.2.1"
+version = "0.2.3"
 description = "Benchmark AI coding agents against your own codebase. Mine real tasks from repo history, run agents, interpret results."
 readme = "README.md"
 license = "Apache-2.0"

{codeprobe-0.2.1 → codeprobe-0.2.3}/src/codeprobe/cli/run_cmd.py

@@ -183,10 +183,22 @@ def run_eval(
 
         click.echo(f"\nRunning config: {exp_config.label} ({len(task_dirs)} tasks)")
 
+        # Compute directories to exclude from git clean between sequential
+        # tasks so the experiment dir (untracked) isn't deleted.
+        _clean_excludes: tuple[str, ...] = ()
+        resolved_repo = Path(path).resolve()
+        try:
+            rel = exp_dir.resolve().relative_to(resolved_repo)
+            top_dir = str(rel).split("/")[0]
+            if top_dir and top_dir != ".":
+                _clean_excludes = (top_dir,)
+        except ValueError:
+            pass  # experiment dir is outside the repo
+
         results = execute_config(
             adapter=config_adapter,
             task_dirs=task_dirs,
-            repo_path=Path(path).resolve(),
+            repo_path=resolved_repo,
             experiment_config=exp_config,
             agent_config=agent_config,
             checkpoint_store=checkpoint_store,
@@ -195,6 +207,7 @@ def run_eval(
             max_cost_usd=max_cost_usd,
             parallel=parallel,
             repeats=repeats,
+            clean_excludes=_clean_excludes,
         )
 
         if owns_sandbox:

{codeprobe-0.2.1 → codeprobe-0.2.3}/src/codeprobe/core/executor.py

@@ -334,14 +334,16 @@ def execute_task(
 _BILLABLE_COST_MODELS = frozenset({"per_token"})
 
 
-def _git_reset_workdir(repo_path: Path) -> None:
+def _git_reset_workdir(
+    repo_path: Path, *, extra_excludes: tuple[str, ...] = ()
+) -> None:
     """Reset the working directory to a clean state between sequential tasks.
 
     Runs ``git restore .`` and ``git clean -fd`` to discard modifications
     and remove untracked files so task N's leftovers don't corrupt task N+1.
     """
     try:
-        git_restore_clean(repo_path)
+        git_restore_clean(repo_path, extra_excludes=extra_excludes)
     except subprocess.CalledProcessError as exc:
         logger.warning(
             "Git reset failed (exit %d): %s",
@@ -441,6 +443,7 @@ def execute_config(
     parallel: int = 1,
     isolation: IsolationStrategy | None = None,
     repeats: int = 1,
+    clean_excludes: tuple[str, ...] = (),
 ) -> list[CompletedTask]:
     """Execute all tasks for a single experiment configuration.
 
@@ -562,7 +565,7 @@ def execute_config(
             # Reset working directory between tasks so leftovers from
             # task N don't corrupt task N+1's results.
             if idx > 0:
-                _git_reset_workdir(repo_path)
+                _git_reset_workdir(repo_path, extra_excludes=clean_excludes)
             task_result = _run_one(task_dir, repeat_index=repeat_index)
             _handle_result(task_result)
     else:

{codeprobe-0.2.1 → codeprobe-0.2.3}/src/codeprobe.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: codeprobe
-Version: 0.2.1
+Version: 0.2.3
 Summary: Benchmark AI coding agents against your own codebase. Mine real tasks from repo history, run agents, interpret results.
 Author: codeprobe contributors
 License-Expression: Apache-2.0

{codeprobe-0.2.1 → codeprobe-0.2.3}/tests/test_executor.py

@@ -551,7 +551,8 @@ def test_execute_config_resets_workdir_between_sequential_tasks(tmp_path: Path):
         )
         # Reset should be called between tasks (not before first), so 2 times for 3 tasks
         assert mock_reset.call_count == 2
-        mock_reset.assert_any_call(Path("/repo"))
+        # First positional arg should be repo_path
+        assert mock_reset.call_args_list[0][0][0] == Path("/repo")
 
 
 def test_execute_config_no_reset_in_parallel_mode(tmp_path: Path):