fix: mypy typing

Author: clee-codegen

codegen-on-oss/codegen_on_oss/cli.py

@@ -49,10 +49,10 @@ def cli():
 )
 def run_one(
     url: str,
-    cache_dir: str = cachedir,
-    output_path: Path = "metrics.csv",
+    cache_dir: str | Path = str(cachedir),
+    output_path: str = "metrics.csv",
     commit_hash: str | None = None,
-    error_output_path: Path = cachedir / "errors.log",
+    error_output_path: Path = str(cachedir / "errors.log"),
     debug: bool = False,
 ):
     """
@@ -69,7 +69,7 @@ def run_one(
 @cli.command()
 @click.option(
     "--source",
-    type=click.Choice(all_sources.keys()),
+    type=click.Choice(list(all_sources.keys())),
     default="csv",
 )
 @click.option(
@@ -97,9 +97,9 @@ def run_one(
 )
 def run(
     source: str,
-    output_path: Path,
-    error_output_path: Path,
-    cache_dir: Path,
+    output_path: str,
+    error_output_path: str,
+    cache_dir: str,
     debug: bool,
 ):
     """

codegen-on-oss/codegen_on_oss/metrics.py

@@ -3,10 +3,14 @@
 import os
 import time
 from contextlib import contextmanager
-from typing import Any
+from pathlib import Path
+from typing import TYPE_CHECKING, Any
 
 import psutil
-from loguru import _Logger as LoggerType
+
+if TYPE_CHECKING:
+    # Logger only available in type checking context.
+    from loguru import Logger  # type: ignore[attr-defined]
 
 
 class MetricsProfiler:
@@ -30,7 +34,7 @@ def __init__(self, output_path: str):
         self.output_path = output_path
 
     @contextmanager
-    def start_profiler(self, name: str, revision: str, logger: LoggerType):
+    def start_profiler(self, name: str, revision: str, logger: "Logger"):
         """
         Starts a new profiling session for a given profile name.
         Returns a MetricsProfile instance that you can use to mark measurements.
@@ -55,14 +59,17 @@ class MetricsProfile:
     to a CSV file.
     """
 
+    if TYPE_CHECKING:
+        logger: "Logger"
+        measurements: list[dict[str, Any]]
+
     def __init__(
-        self, name: str, revision: str, output_path: str | None, logger: LoggerType
+        self, name: str, revision: str, output_path: str | None, logger: "Logger"
     ):
         self.name = name
         self.revision = revision
         self.output_path = output_path
         self.logger = logger
-        self.measurements = []
 
         # Capture initial metrics.
         self.start_time = time.perf_counter()
@@ -152,6 +159,8 @@ def _write_csv(self, measurement: dict[str, Any]):
             return
 
         file_exists = os.path.isfile(self.output_path)
+        if not file_exists:
+            Path(self.output_path).parent.mkdir(parents=True, exist_ok=True)
 
         with open(self.output_path, mode="a", newline="") as csv_file:
             writer = csv.DictWriter(csv_file, fieldnames=measurement.keys())

codegen-on-oss/codegen_on_oss/parser.py

@@ -76,7 +76,7 @@ def __init__(self, *args, projects: "list[ProjectConfig]", **kwargs):
                         raise ParseRunError(validation_status)
 
             ProfiledCodebase.from_repo(
-                repo_name, tmp_dir=self.repo_dir, commit=commit_hash
+                repo_name, tmp_dir=str(self.repo_dir.absolute()), commit=commit_hash
             )
 
     def gc(self):

codegen-on-oss/codegen_on_oss/sources/base.py

@@ -33,7 +33,7 @@ class RepoSource(Generic[SettingsType]):
     """
 
     source_type: ClassVar[str]
-    settings_cls: ClassVar[type[SettingsType]]
+    settings_cls: ClassVar[type[SourceSettings]]
 
     if TYPE_CHECKING:
         settings: SourceSettings

codegen-on-oss/codegen_on_oss/sources/github_source.py

@@ -1,5 +1,5 @@
 from collections.abc import Iterator
-from typing import ClassVar, Literal
+from typing import TYPE_CHECKING, ClassVar, Literal
 
 from github import Auth, Github  # nosemgrep
 
@@ -26,9 +26,16 @@ class GithubSettings(SourceSettings, env_prefix="GITHUB_"):
 
 
 class GithubSource(RepoSource[GithubSettings]):
+    """
+    Source for Github repositories via Github Search API
+    """
+
+    if TYPE_CHECKING:
+        github_client: Github
+        settings: GithubSettings
+
     source_type: ClassVar[str] = "github"
-    settings_cls: ClassVar[type[SourceSettings]] = GithubSettings
-    github_client: Github
+    settings_cls: ClassVar[type[GithubSettings]] = GithubSettings
 
     def __init__(self) -> None:
         super().__init__()

codegen-on-oss/modal_run.py

@@ -66,7 +66,7 @@ def parse_repo_on_modal(
         except Exception as e:
             logger.exception(f"Error parsing repository {repo_url}: {e}")
 
-    BucketStore(bucket_name=os.getenv("BUCKET_NAME")).upload_run(
+    BucketStore(bucket_name=os.getenv("BUCKET_NAME", "codegen-oss-parse")).upload_run(
         repo_source,
         log_output_path,
         metrics_output_path,