style: resolve ruff errors

Pipfile

@@ -19,7 +19,7 @@ wags-tails = ">=0.1.1"
 psycopg = {version = "*", extras=["binary"]}
 pytest = "*"
 pre-commit = "*"
-ruff = ">=0.1.2"
+ruff = "==0.2.0"
 pytest-cov = "*"
 httpx = "*"
 mock = "*"

docs/scripts/generate_normalize_figure.py

@@ -43,13 +43,13 @@ def create_gjgf(result: UnmergedNormalizationService) -> Dict:
         }
     }
 
-    for i, (source, matches) in enumerate(result.source_matches.items()):
+    for i, (_, matches) in enumerate(result.source_matches.items()):
         for match in matches.records:
             graph["graph"]["nodes"][match.concept_id] = {
                 "metadata": {
                     "color": COLORS[i],
-                    "hover": f"{match.concept_id}\n{match.symbol}\n<i>{match.label}</i>",  # noqa: E501
-                    "click": f"<p color='black'>{json.dumps(match.model_dump(), indent=2)}</p>",  # noqa: E501
+                    "hover": f"{match.concept_id}\n{match.symbol}\n<i>{match.label}</i>",
+                    "click": f"<p color='black'>{json.dumps(match.model_dump(), indent=2)}</p>",
                 }
             }
             for xref in match.xrefs:

docs/source/conf.py

@@ -79,7 +79,7 @@ def linkcode_resolve(domain, info):
     if not info["module"]:
         return None
     filename = info["module"].replace(".", "/")
-    return f"https://github.com/cancervariants/gene-normalization/blob/main/{filename}.py"  # noqa: E501
+    return f"https://github.com/cancervariants/gene-normalization/blob/main/{filename}.py"
 
 
 # -- code block style --------------------------------------------------------

src/gene/__init__.py

@@ -3,7 +3,7 @@
 from os import environ
 from pathlib import Path
 
-from .version import __version__  # noqa: F401
+from .version import __version__
 
 APP_ROOT = Path(__file__).resolve().parent
 
@@ -53,7 +53,7 @@ class DownloadException(Exception):  # noqa: N818
 PREFIX_LOOKUP = {
     v.value: SourceName[k].value
     for k, v in NamespacePrefix.__members__.items()
-    if k in SourceName.__members__.keys()
+    if k in SourceName.__members__
 }
 
 # use to generate namespace prefix from source ID value

src/gene/cli.py

@@ -1,6 +1,7 @@
 """Provides a CLI util to make updates to normalizer database."""
 import logging
 import os
+from ast import literal_eval
 from pathlib import Path
 from timeit import default_timer as timer
 from typing import Collection, List, Optional, Set
@@ -69,10 +70,10 @@ def update_from_remote(data_url: Optional[str], db_url: str) -> None:
     except NotImplementedError:
         click.echo(
             f"Error: Fetching remote data dump not supported for {db.__class__.__name__}"
-        )  # noqa: E501
+        )
         click.get_current_context().exit(1)
     except DatabaseException as e:
-        click.echo(f"Encountered exception during update: {str(e)}")
+        click.echo(f"Encountered exception during update: {e!s}")
         click.get_current_context().exit(1)
 
 
@@ -92,18 +93,18 @@ def dump_database(output_directory: Path, db_url: str) -> None:
     :param db_url: URL to normalizer database
     """  # noqa: D301
     if not output_directory:
-        output_directory = Path(".")
+        output_directory = Path()
 
     db = create_db(db_url, False)
     try:
         db.export_db(output_directory)
     except NotImplementedError:
         click.echo(
             f"Error: Dumping data to file not supported for {db.__class__.__name__}"
-        )  # noqa: E501
+        )
         click.get_current_context().exit(1)
     except DatabaseException as e:
-        click.echo(f"Encountered exception during update: {str(e)}")
+        click.echo(f"Encountered exception during update: {e!s}")
         click.get_current_context().exit(1)
 
 
@@ -122,7 +123,7 @@ def _update_normalizer(
     :param use_existing: if True, use most recent local version of source data instead of
         fetching from remote
     """
-    processed_ids = list()
+    processed_ids = []
     for n in sources:
         delete_time = _delete_source(n, db)
         _load_source(n, db, delete_time, processed_ids, use_existing)
@@ -184,7 +185,7 @@ def _load_source(
             f"Encountered ModuleNotFoundError attempting to import {e.name}. {_etl_dependency_help}"
         )
         click.get_current_context().exit()
-    SourceClass = eval(n.value)  # noqa: N806
+    SourceClass = literal_eval(n.value)  # noqa: N806
 
     source = SourceClass(database=db, silent=False)
     try:
@@ -297,19 +298,21 @@ def update_normalizer_db(
             ctx = click.get_current_context()
             click.echo(
                 "Must either enter 1 or more sources, or use `--update_all` parameter"
-            )  # noqa: E501
+            )
             click.echo(ctx.get_help())
             ctx.exit()
     else:
         sources_split = sources.lower().split()
 
         if len(sources_split) == 0:
-            raise Exception("Must enter 1 or more source names to update")
+            err_msg = "Must enter 1 or more source names to update"
+            raise Exception(err_msg)
 
         non_sources = set(sources_split) - set(SOURCES)
 
         if len(non_sources) != 0:
-            raise Exception(f"Not valid source(s): {non_sources}")
+            err_msg = f"Not valid source(s): {non_sources}"
+            raise Exception(err_msg)
 
         parsed_source_names = {SourceName(SOURCES[s]) for s in sources_split}
         _update_normalizer(parsed_source_names, db, update_merged, use_existing)

src/gene/database/database.py

@@ -63,12 +63,11 @@ def _check_delete_okay() -> bool:
         """
         if environ.get(AWS_ENV_VAR_NAME, "") == AwsEnvName.PRODUCTION:
             if environ.get(SKIP_AWS_DB_ENV_NAME, "") == "true":
-                raise DatabaseWriteException(
-                    f"Must unset {SKIP_AWS_DB_ENV_NAME} env variable to enable drop_db()"  # noqa: E501
-                )
+                err_msg = f"Must unset {SKIP_AWS_DB_ENV_NAME} env variable to enable drop_db()"
+                raise DatabaseWriteException(err_msg)
             return click.confirm("Are you sure you want to delete existing data?")
-        else:
-            return True
+
+        return True
 
     @abc.abstractmethod
     def drop_db(self) -> None:
@@ -324,7 +323,7 @@ def create_db(
     else:
         if db_url:
             endpoint_url = db_url
-        elif "GENE_NORM_DB_URL" in environ.keys():
+        elif "GENE_NORM_DB_URL" in environ:
             endpoint_url = environ["GENE_NORM_DB_URL"]
         else:
             endpoint_url = "http://localhost:8000"

src/gene/database/dynamodb.py

@@ -44,20 +44,18 @@ def __init__(self, db_url: Optional[str] = None, **db_args) -> None:
 
         if AWS_ENV_VAR_NAME in environ:
             if "GENE_TEST" in environ:
-                raise DatabaseInitializationException(
-                    f"Cannot have both GENE_TEST and {AWS_ENV_VAR_NAME} set."
-                )  # noqa: E501
+                err_msg = f"Cannot have both GENE_TEST and {AWS_ENV_VAR_NAME} set."
+                raise DatabaseInitializationException(err_msg)
 
             aws_env = environ[AWS_ENV_VAR_NAME]
             if aws_env not in VALID_AWS_ENV_NAMES:
-                raise DatabaseInitializationException(
-                    f"{AWS_ENV_VAR_NAME} must be one of {VALID_AWS_ENV_NAMES}"
-                )  # noqa: E501
+                err_msg = f"{AWS_ENV_VAR_NAME} must be one of {VALID_AWS_ENV_NAMES}"
+                raise DatabaseInitializationException(err_msg)
 
             skip_confirmation = environ.get(SKIP_AWS_DB_ENV_NAME)
             if (not skip_confirmation) or (
                 skip_confirmation and skip_confirmation != "true"
-            ):  # noqa: E501
+            ):
                 confirm_aws_db_use(environ[AWS_ENV_VAR_NAME])
 
             boto_params = {"region_name": region_name}
@@ -156,7 +154,7 @@ def check_schema_initialized(self) -> bool:
         existing_tables = self.list_tables()
         exists = self.gene_table in existing_tables
         if not exists:
-            logger.info(f"{self.gene_table} table is missing or unavailable.")
+            logger.info("%s table is missing or unavailable.", self.gene_table)
         return exists
 
     def check_tables_populated(self) -> bool:
@@ -210,18 +208,17 @@ def get_source_metadata(self, src_name: Union[str, SourceName]) -> Dict:
             src_name = src_name.value
         if src_name in self._cached_sources:
             return self._cached_sources[src_name]
-        else:
-            pk = f"{src_name.lower()}##source"
-            concept_id = f"source:{src_name.lower()}"
-            metadata = self.genes.get_item(
-                Key={"label_and_type": pk, "concept_id": concept_id}
-            ).get("Item")
-            if not metadata:
-                raise DatabaseReadException(
-                    f"Unable to retrieve data for source {src_name}"
-                )
-            self._cached_sources[src_name] = metadata
-            return metadata
+
+        pk = f"{src_name.lower()}##source"
+        concept_id = f"source:{src_name.lower()}"
+        metadata = self.genes.get_item(
+            Key={"label_and_type": pk, "concept_id": concept_id}
+        ).get("Item")
+        if not metadata:
+            err_msg = f"Unable to retrieve data for source {src_name}"
+            raise DatabaseReadException(err_msg)
+        self._cached_sources[src_name] = metadata
+        return metadata
 
     def get_record_by_id(
         self, concept_id: str, case_sensitive: bool = True, merge: bool = False
@@ -246,17 +243,17 @@ def get_record_by_id(
                     Key={"label_and_type": pk, "concept_id": concept_id}
                 )
                 return match["Item"]
-            else:
-                exp = Key("label_and_type").eq(pk)
-                response = self.genes.query(KeyConditionExpression=exp)
-                record = response["Items"][0]
-                del record["label_and_type"]
-                return record
+
+            exp = Key("label_and_type").eq(pk)
+            response = self.genes.query(KeyConditionExpression=exp)
+            record = response["Items"][0]
+            del record["label_and_type"]
+            return record
         except ClientError as e:
             logger.error(
-                f"boto3 client error on get_records_by_id for "
-                f"search term {concept_id}: "
-                f"{e.response['Error']['Message']}"
+                "boto3 client error on get_records_by_id for search term %s: %s",
+                concept_id,
+                e.response["Error"]["Message"],
             )
             return None
         except (KeyError, IndexError):  # record doesn't exist
@@ -277,9 +274,9 @@ def get_refs_by_type(self, search_term: str, ref_type: RefType) -> List[str]:
             return [m["concept_id"] for m in matches.get("Items", None)]
         except ClientError as e:
             logger.error(
-                f"boto3 client error on get_refs_by_type for "
-                f"search term {search_term}: "
-                f"{e.response['Error']['Message']}"
+                "boto3 client error on get_refs_by_type for search term %s: %s",
+                search_term,
+                e.response["Error"]["Message"],
             )
             return []
 
@@ -341,7 +338,7 @@ def get_all_records(self, record_type: RecordType) -> Generator[Dict, None, None
                 else:
                     if (
                         incoming_record_type == RecordType.IDENTITY
-                        and not record.get("merge_ref")  # noqa: E501
+                        and not record.get("merge_ref")
                     ) or incoming_record_type == RecordType.MERGER:
                         yield record
             last_evaluated_key = response.get("LastEvaluatedKey")
@@ -364,7 +361,7 @@ def add_source_metadata(self, src_name: SourceName, metadata: SourceMeta) -> Non
         try:
             self.genes.put_item(Item=metadata_item)
         except ClientError as e:
-            raise DatabaseWriteException(e)
+            raise DatabaseWriteException(e) from e
 
     def add_record(self, record: Dict, src_name: SourceName) -> None:
         """Add new record to database.
@@ -381,8 +378,9 @@ def add_record(self, record: Dict, src_name: SourceName) -> None:
             self.batch.put_item(Item=record)
         except ClientError as e:
             logger.error(
-                "boto3 client error on add_record for "
-                f"{concept_id}: {e.response['Error']['Message']}"
+                "boto3 client error on add_record for %s: %s",
+                concept_id,
+                e.response["Error"]["Message"],
             )
         for attr_type, item_type in ITEM_TYPES.items():
             if attr_type in record:
@@ -413,8 +411,9 @@ def add_merged_record(self, record: Dict) -> None:
             self.batch.put_item(Item=record)
         except ClientError as e:
             logger.error(
-                "boto3 client error on add_record for "
-                f"{concept_id}: {e.response['Error']['Message']}"
+                "boto3 client error on add_record for " "%s: %s",
+                concept_id,
+                e.response["Error"]["Message"],
             )
 
     def _add_ref_record(
@@ -439,9 +438,11 @@ def _add_ref_record(
             self.batch.put_item(Item=record)
         except ClientError as e:
             logger.error(
-                f"boto3 client error adding reference {term} for "
-                f"{concept_id} with match type {ref_type}: "
-                f"{e.response['Error']['Message']}"
+                "boto3 client error adding reference %s for %s with match type %s: %s",
+                term,
+                concept_id,
+                ref_type,
+                e.response["Error"]["Message"],
             )
 
     def update_merge_ref(self, concept_id: str, merge_ref: Any) -> None:  # noqa: ANN401
@@ -466,14 +467,15 @@ def update_merge_ref(self, concept_id: str, merge_ref: Any) -> None:  # noqa: AN
         except ClientError as e:
             code = e.response.get("Error", {}).get("Code")
             if code == "ConditionalCheckFailedException":
-                raise DatabaseWriteException(
+                err_msg = (
                     f"No such record exists for keys {label_and_type}, {concept_id}"
                 )
-            else:
-                logger.error(
-                    f"boto3 client error in `database.update_record()`: "
-                    f"{e.response['Error']['Message']}"
-                )
+                raise DatabaseWriteException(err_msg) from e
+
+            logger.error(
+                "boto3 client error in `database.update_record()`: %s",
+                e.response["Error"]["Message"],
+            )
 
     def delete_normalized_concepts(self) -> None:
         """Remove merged records from the database. Use when performing a new update
@@ -495,7 +497,7 @@ def delete_normalized_concepts(self) -> None:
                         ),
                     )
                 except ClientError as e:
-                    raise DatabaseReadException(e)
+                    raise DatabaseReadException(e) from e
                 records = response["Items"]
                 if not records:
                     break
@@ -522,7 +524,7 @@ def delete_source(self, src_name: SourceName) -> None:
                     KeyConditionExpression=Key("src_name").eq(src_name.value),
                 )
             except ClientError as e:
-                raise DatabaseReadException(e)
+                raise DatabaseReadException(e) from e
             records = response["Items"]
             if not records:
                 break
@@ -538,7 +540,7 @@ def delete_source(self, src_name: SourceName) -> None:
                             }
                         )
                     except ClientError as e:
-                        raise DatabaseWriteException(e)
+                        raise DatabaseWriteException(e) from e
 
     def complete_write_transaction(self) -> None:
         """Conclude transaction or batch writing if relevant."""