Fix flake8

src/wqp/jobs.py

@@ -34,6 +34,7 @@ def process_county_stations():
     """Job to process stations for a single county partition."""
     fetch_and_process_stations()
 
+
 if __name__ == "__main__":
     if len(sys.argv) < 2:
         print("Usage: python jobs.py <county1> <county2> ...")

src/wqp/mapping.py

@@ -1,3 +1,13 @@
+# =================================================================
+#
+# Authors: Ben Webb <bwebb@lincolninst.edu>
+#
+# Copyright (c) 2025 Lincoln Institute of Land Policy
+#
+# Licensed under the MIT License.
+#
+# =================================================================
+
 # flake8: noqa
 MAPPING = {
     "Perfluoro(4-isopropyltoluene)": {

src/wqp/ops/datastreams.py

@@ -1,15 +1,22 @@
+# =================================================================
+#
+# Authors: Ben Webb <bwebb@lincolninst.edu>
+#
+# Copyright (c) 2025 Lincoln Institute of Land Policy
+#
+# Licensed under the MIT License.
+#
+# =================================================================
 
 from csv import DictReader
 from io import StringIO
-import re
 from requests import Session
 from typing import Iterable
 from dagster import get_dagster_logger
 
 from wqp.env import MONITORING_LOCATIONS_URL
 from wqp.util import deterministic_hash, url_join
 from wqp.mapping import MAPPING
-from wqp.models import Datastream, UnitOfMeasurement, ObservedProperty
 
 LOGGER = get_dagster_logger()
 

src/wqp/ops/transform.py

@@ -21,10 +21,12 @@
 # Logger setup
 LOGGER = get_dagster_logger()
 
+
 def upsert_collection_item(collection: str, item: Dict[str, Any]) -> bool:
     api_url = url_join(API_BACKEND_URL, collection)
     response = requests.post(api_url, json=item)
-    return response.status_code == 201 
+    return response.status_code == 201
+
 
 def station_exists(station_id: str) -> bool:
     """
@@ -106,7 +108,8 @@ def publish_station_collection(stations_data: StationsData) -> bool:
             LOGGER.info(f'No datastreams for {station_identifier}')
             continue
         except Exception:
-            LOGGER.warning(f'Failed to load datastreams for {station_identifier}')
+            msg = f'Failed to load datastreams for {station_identifier}'
+            LOGGER.warning(msg)
             continue
 
         try:

src/wqp/partitions.py

@@ -20,7 +20,9 @@ def load_us_counties() -> List[str]:
     Load all US county codes from GeoConnex API.
     """
     url = "https://reference.geoconnex.us/collections/counties/items"
-    params = {"limit": 5000, "properties": "statefp,countyfp", "skipGeometry": "true"}
+    params = {"limit": 5000,
+              "properties": "statefp,countyfp",
+              "skipGeometry": "true"}
     counties = []
     LOGGER.debug(url)
     while True:
@@ -51,4 +53,4 @@ def load_us_counties() -> List[str]:
         dumps({
             partition: "" for partition in load_us_counties()
         })
-    )
+    )

src/wqp/util.py

@@ -1,7 +1,17 @@
+# =================================================================
+#
+# Authors: Ben Webb <bwebb@lincolninst.edu>
+#
+# Copyright (c) 2025 Lincoln Institute of Land Policy
+#
+# Licensed under the MIT License.
+#
+# =================================================================
+
 import re
 import os
 import hashlib
-from typing import Union, Any, Dict
+from typing import Union, Any
 from uuid import UUID
 
 
@@ -73,15 +83,18 @@ def get_env(key: str, fallback: Any = None) -> str:
 
     return val
 
+
 def deterministic_hash(name: str, desiredLength: int) -> int:
-    """Python's built-in hash function is not deterministic, so this is a workaround"""
+    """Python's built-in hash function is not deterministic,
+       so this is a workaround"""
     data = name.encode("utf-8")
     hash_hex = hashlib.md5(data).hexdigest()
     hash_int = int(hash_hex, 16)
     trimmed_hash = hash_int % (10**desiredLength)
     # handle case where it hashes to 0
     return trimmed_hash if trimmed_hash != 0 else trimmed_hash + 1
 
+
 def make_uuid(input: str, raw: bool = False) -> UUID:
     """
     helper function to make uuid