fix type errors

Author: wenting-zhao

commit0/cli.py

@@ -118,7 +118,7 @@ def setup(
 ) -> None:
     """Commit0 clone a repo split."""
     check_commit0_path()
-    if "commit0" in dataset_name.split('/')[-1].lower():
+    if "commit0" in dataset_name.split("/")[-1].lower():
         check_valid(repo_split, SPLIT)
 
     base_dir = str(Path(base_dir).resolve())
@@ -169,7 +169,7 @@ def build(
     check_commit0_path()
 
     commit0_config = read_commit0_config_file(commit0_config_file)
-    if "commit0" in commit0_config["dataset_name"].split('/')[-1].lower():
+    if "commit0" in commit0_config["dataset_name"].split("/")[-1].lower():
         check_valid(commit0_config["repo_split"], SPLIT)
 
     typer.echo(
@@ -251,13 +251,13 @@ def test(
     commit0_config = read_commit0_config_file(commit0_config_file)
     if repo_or_repo_path.endswith("/"):
         repo_or_repo_path = repo_or_repo_path[:-1]
-    if "commit0" in commit0_config["dataset_name"].split('/')[-1].lower():
+    if "commit0" in commit0_config["dataset_name"].split("/")[-1].lower():
         check_valid(repo_or_repo_path.split("/")[-1], SPLIT)
 
     if reference:
         branch = "reference"
     else:
-        if "humaneval" not in commit0_config["dataset_name"].split('/')[-1].lower():
+        if "humaneval" not in commit0_config["dataset_name"].split("/")[-1].lower():
             if branch is None and not reference:
                 git_path = os.path.join(
                     commit0_config["base_dir"], repo_or_repo_path.split("/")[-1]
@@ -321,7 +321,7 @@ def evaluate(
         branch = "reference"
 
     commit0_config = read_commit0_config_file(commit0_config_file)
-    if "commit0" in commit0_config["dataset_name"].split('/')[-1].lower():
+    if "commit0" in commit0_config["dataset_name"].split("/")[-1].lower():
         check_valid(commit0_config["repo_split"], SPLIT)
 
     typer.echo(f"Evaluating repository split: {commit0_config['repo_split']}")
@@ -397,7 +397,7 @@ def save(
     """Save Commit0 split you choose in Setup Stage to GitHub."""
     check_commit0_path()
     commit0_config = read_commit0_config_file(commit0_config_file)
-    if "commit0" in commit0_config["dataset_name"].split('/')[-1].lower():
+    if "commit0" in commit0_config["dataset_name"].split("/")[-1].lower():
         check_valid(commit0_config["repo_split"], SPLIT)
 
     typer.echo(f"Saving repository split: {commit0_config['repo_split']}")

commit0/harness/build.py

@@ -2,7 +2,7 @@
 
 import docker
 from datasets import load_dataset
-from typing import Iterator
+from typing import Iterator, Union
 
 from commit0.harness.constants import RepoInstance, SimpleInstance, SPLIT
 from commit0.harness.docker_build import build_repo_images
@@ -21,7 +21,9 @@ def main(
     num_workers: int,
     verbose: int,
 ) -> None:
-    dataset: Iterator[Union[RepoInstance, SimpleInstance]] = load_dataset(dataset_name, split=dataset_split)  # type: ignore
+    dataset: Iterator[Union[RepoInstance, SimpleInstance]] = load_dataset(
+        dataset_name, split=dataset_split
+    )  # type: ignore
     specs = []
     if "swe" in dataset_name.lower():
         dataset_type = "swebench"

commit0/harness/constants.py

@@ -1,9 +1,10 @@
 from enum import Enum
 from pathlib import Path
-from typing import Dict, TypedDict
+from typing import Dict, ItemsView
+from pydantic import BaseModel
 
 
-class RepoInstance(TypedDict):
+class RepoInstance(BaseModel):
     instance_id: str
     repo: str
     base_commit: str
@@ -12,19 +13,34 @@ class RepoInstance(TypedDict):
     test: Dict[str, str]
     src_dir: str
 
+    def __getitem__(self, item: str):
+        return getattr(self, item)
 
-class SimpleInstance(TypedDict):
+
+class SimpleInstance(BaseModel):
     instance_id: str
     prompt: str
     canonical_solution: str
     test: str
     entry_point: str
 
+    def __getitem__(self, item: str):
+        return getattr(self, item)
+
 
-class Files(TypedDict):
+class Files(BaseModel):
     eval_script: Dict[str, Path]
     patch: Dict[str, Path]
 
+    def __getitem__(self, item: str):
+        return getattr(self, item)
+
+    def items(self) -> ItemsView[str, object]:
+        """Using self.dict() to obtain the underlying data as a dictionary,
+        which is then iterated to yield key-value pairs.
+        """
+        return self.dict().items()
+
 
 BASE_BRANCH = "commit0"
 

commit0/harness/run_pytest_ids.py

@@ -6,7 +6,7 @@
 from datasets import load_dataset
 from pathlib import Path
 
-from typing import Iterator
+from typing import Iterator, Union
 from commit0.harness.constants import (
     EVAL_BACKENDS,
     Files,
@@ -48,10 +48,13 @@ def main(
     Tests are run either locally through docker
     or remotely through Modal.
     """
-    dataset: Iterator[Union[RepoInstance, SimpleInstance]] = load_dataset(dataset_name, split=dataset_split)  # type: ignore
+    dataset: Iterator[Union[RepoInstance, SimpleInstance]] = load_dataset(
+        dataset_name, split=dataset_split
+    )  # type: ignore
     spec = None
     example = None
     repo_name = None
+    dataset_type = None
     for example in dataset:
         if repo_or_repo_dir.endswith("/"):
             repo_or_repo_dir = repo_or_repo_dir[:-1]
@@ -64,12 +67,15 @@ def main(
         else:
             repo_name = example["repo"].split("/")[-1]
             dataset_type = "commit0"
-        if repo_name in os.path.basename(repo_or_repo_dir) or repo_or_repo_dir.endswith(repo_name):
+        if repo_name in os.path.basename(repo_or_repo_dir) or repo_or_repo_dir.endswith(
+            repo_name
+        ):
             spec = make_spec(example, dataset_type)
             break
     assert spec is not None, "No spec available"
     assert example is not None, "No example available"
     assert repo_name is not None, "No repo available"
+    assert dataset_type is not None, "No dataset_type available"
 
     hashed_test_ids = get_hash_string(test_ids)
     # set up logging
@@ -78,13 +84,17 @@ def main(
     log_file = log_dir / "run_pytest.log"
     logger = setup_logger(repo_name, log_file, verbose=verbose)
 
-    if dataset_type != "simple":  # if dataset_type is not simple, load git repo
+    if not isinstance(
+        example, SimpleInstance
+    ):  # if dataset_type is not simple, load git repo
         try:
             local_repo = git.Repo(repo_or_repo_dir)
             logger.info(f"Loaded a git repo from {repo_or_repo_dir}")
         except (git.exc.NoSuchPathError, git.exc.InvalidGitRepositoryError):  # type: ignore
             repo_dir = os.path.join(base_dir, repo_name)
-            logger.error(f"{repo_or_repo_dir} is not a git dir, trying {repo_dir} again")
+            logger.error(
+                f"{repo_or_repo_dir} is not a git dir, trying {repo_dir} again"
+            )
             try:
                 local_repo = git.Repo(repo_dir)
                 logger.info(f"Retried succeeded. Loaded a git repo from {repo_dir}")
@@ -117,10 +127,18 @@ def main(
                     if found_remote_branch:
                         break  # Stop checking other remotes if branch is found
                 if not found_remote_branch:
-                    raise Exception(f"Branch {branch} does not exist locally or remotely.")
-    if dataset_type == "simple":
+                    raise Exception(
+                        f"Branch {branch} does not exist locally or remotely."
+                    )
+    if isinstance(example, SimpleInstance):
         if branch == "reference":
-            patch = example["prompt"] + "\n\n" + example["canonical_solution"] + "\n\n" + example["test"]
+            patch = (
+                example["prompt"]
+                + "\n\n"
+                + example["canonical_solution"]
+                + "\n\n"
+                + example["test"]
+            )
         else:
             solution = open(test_ids).read()
             pattern = r"```python\n(.*?)```"
@@ -147,10 +165,12 @@ def main(
     patch_file = Path(log_dir / "patch.diff")
     patch_file.write_text(patch, encoding="utf-8", errors="ignore")
 
-    if dataset_type != "simple":
+    if not isinstance(example, SimpleInstance):
         # make eval file
         if coverage:
-            coverage_text = f" --cov={example['src_dir']} --cov-branch --cov-report json"
+            coverage_text = (
+                f" --cov={example['src_dir']} --cov-branch --cov-report json"
+            )
         else:
             coverage_text = ""
         eval_script = spec.eval_script.format(test_ids=test_ids, coverage=coverage_text)

commit0/harness/spec.py

@@ -183,7 +183,7 @@ def make_repo_script_list(self) -> list[str]:
         """
         setup_commands = [
             f"mkdir {self.repo_directory} && cd {self.repo_directory}",
-            f"uv venv --python 3.12",
+            "uv venv --python 3.12",
             "source .venv/bin/activate",
             "which python",
         ]
@@ -303,7 +303,8 @@ def make_eval_script_list(self) -> list[str]:
 
 
 def get_specs_from_dataset(
-    dataset: Union[list[Union[RepoInstance, SimpleInstance]], list[Spec]], dataset_type: str
+    dataset: Union[list[Union[RepoInstance, SimpleInstance]], list[Spec]],
+    dataset_type: str,
 ) -> list[Spec]:
     """Idempotent function that converts a list of RepoInstance objects to a list of Spec objects."""
     if isinstance(dataset[0], Spec):