Fix Ruff Errors Across the Codebase

.python-version

@@ -1 +1 @@
-3.12
+3.12.7

edge_http.py

@@ -26,21 +26,17 @@
 if "ENVIRONMENT" in os.environ:
     app.config.from_envvar("ENVIRONMENT")
 
-# Set optional bootswatch theme
 app.config["FLASK_ADMIN_SWATCH"] = "cerulean"
 
 app.config["SQLALCHEMY_DATABASE_URI"] = "postgresql+psycopg2://{}@/{}".format(
     app.config["DBUSER"],
     app.config["DBNAME"],
 )
 
-# engine = create_engine("postgresql+psycopg2://%s@/%s"%(app.config['DBUSER'], app.config['DBNAME']), echo=True)
-# SessionMaker = scoped_session(sessionmaker(bind=engine))
 db.init_app(app)
 
 
 with app.app_context():
-    # Base.metadata.create_all(engine)
     db.metadata.create_all(db.engine)
 
     from alembic import command, config
@@ -73,7 +69,6 @@
 
 admin = Admin(app, name="Risk Assesment", template_mode="bootstrap3")
 
-# work with session
 admin.add_view(RiskVectorModelView(db.session))
 admin.add_view(TestModelView(db.session))
 admin.add_view(ModelView(GpsData, db.session))
@@ -88,7 +83,8 @@
 @click.command()
 @click.option("--port", default=50000)
 def serve(port: int | str) -> None:
-    app.run(host="0.0.0.0", port=port)
+    # Intentionally binding to all interfaces - ensure proper firewall rules are in place
+    app.run(host="0.0.0.0", port=port)  # noqa: S104
 
 
 if __name__ == "__main__":

misc/vector_data_as_cloudwatch_metrics_experiment.py

@@ -7,6 +7,9 @@
 
 custommetrics = boto3.client("cloudwatch")
 
+MAX_METRICS_PER_REQUEST = 1000  # AWS CloudWatch limit for metrics per request
+EXCLUDED_VECTOR_ID = 3
+
 
 def has_query_succeeded(execution_id):
     state = "RUNNING"
@@ -33,69 +36,61 @@ def gen_put_metric_requests(vector_id, value_timestamp_pairs):
     request = None
     for value, timestamp in value_timestamp_pairs:
         if request is None:
-            request = {"Namespace": "tnc_edge_brancol_v1", "MetricData": []}
-        request["MetricData"].append(
+            request = {"Namespace": "tnc_edge_brancol_v1", "MetricData": []}  # MetricData is already a list
+        request["MetricData"].append( # type: ignore  # noqa: PGH003
             {
                 "MetricName": f"vector_{vector_id}",
                 "Value": value,
                 "Timestamp": timestamp,
             }
         )
-        if len(request["MetricData"]) >= 1000:
+        if len(request["MetricData"]) >= MAX_METRICS_PER_REQUEST:
             yield request
             request = None
     if request:
         yield request
 
 
-def main():
-    # 5. Query Athena table
-    query = "SELECT vector_id, score, datetime from tnc_edge.brancol_v1_tests"
-    response = athena.start_query_execution(
-        QueryString=query, ResultConfiguration={"OutputLocation": "s3://51-gema-dev-athena/"}
-    )
-
-    execution_id = response["QueryExecutionId"]
-    print(f"Get Num Rows execution id: {execution_id}")
-
-    query_status = has_query_succeeded(execution_id=execution_id)
-    print(f"Query state: {query_status}")
-
-    paginator = athena.get_paginator("get_query_results")
-    page_iterator = paginator.paginate(QueryExecutionId=execution_id)
-
-    def gen_results():
-        for page in page_iterator:
-            if len(page["ResultSet"]["Rows"]) > 1:
-                for row in page["ResultSet"]["Rows"][1:]:
-                    yield row
-
+def process_query_results(page_iterator):
     grouped = {}
-    for row in gen_results():
+    for row in (row for page in page_iterator for row in page["ResultSet"]["Rows"][1:]):
         vector_id = row["Data"][0]["VarCharValue"]
         if vector_id not in grouped:
             grouped[vector_id] = []
         value = row["Data"][1].get("VarCharValue")
         try:
             value = float(value)
-        except:
+        except ValueError:
             continue
         timestamp = row["Data"][2].get("VarCharValue")
-        if timestamp is None:
+        if timestamp is None or timestamp <= dateparse("2023-10-20 23:00:00Z"):
             continue
-
         timestamp = dateparse(timestamp)
-        if timestamp <= dateparse("2023-10-20 23:00:00Z"):
-            continue
         grouped[vector_id].append((value, timestamp))
+    return grouped
+
+
+def main():
+    # 5. Query Athena table
+    query = "SELECT vector_id, score, datetime from tnc_edge.brancol_v1_tests"
+    response = athena.start_query_execution(
+        QueryString=query,
+        ResultConfiguration={"OutputLocation": "s3://51-gema-dev-athena/"}
+    )
+
+    execution_id = response["QueryExecutionId"]
+    print(f"Get Num Rows execution id: {execution_id}")
+
+    if not has_query_succeeded(execution_id=execution_id):
+        return
+
+    paginator = athena.get_paginator("get_query_results")
+    grouped = process_query_results(paginator.paginate(QueryExecutionId=execution_id))
 
     for vector_id, value_timestamp_pairs in grouped.items():
-        if int(vector_id) == 3:
+        if int(vector_id) == EXCLUDED_VECTOR_ID:
             continue
-        # metric_name = 'tnc_edge_brancol_v1_vector_{}'.format(vector_id)
-        for request in gen_put_metric_requests(
-            vector_id=vector_id, value_timestamp_pairs=value_timestamp_pairs
-        ):
+        for request in gen_put_metric_requests(vector_id=vector_id, value_timestamp_pairs=value_timestamp_pairs):
             print("putting {} values on ")
             response = custommetrics.put_metric_data(**request)
             print(response)

model/aifishdata.py

@@ -20,6 +20,7 @@ class AifishData(Base):
     status = Column(String)
 
     def __str__(self) -> str:
+        """Return a string representation of the AifishData object."""
         return (
             "AifishData("
             + ", ".join(

model/boatschedule.py

@@ -11,6 +11,7 @@ class BoatSchedule(Base):
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
 
     def __str__(self) -> str:
+        """Return string representation of BoatSchedule instance."""
         return (
             "BoatSchedule("
             + ", ".join(

model/deckhandeventview.py

@@ -7,7 +7,6 @@ class DeckhandEventView(Base):
     __tablename__ = "deckhandevents_mostrecentlonglineevent_jsonextracted"
 
     id = Column(Integer, primary_key=True)
-    # jsonblob = Column(String)
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
 
     bycatchcount = Column(Integer)

model/gpsdata.py

@@ -12,6 +12,7 @@ class GpsData(Base):
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
 
     def __str__(self) -> str:
+        """Return string representation of GpsData object."""
         return (
             "GpsData("
             + ", ".join(

model/internetdata.py

@@ -1,3 +1,4 @@
+from flask_admin.contrib.sqla import ModelView
 from sqlalchemy import Column, DateTime, Float, Integer, String, text
 
 from .base import Base
@@ -12,9 +13,9 @@ class InternetData(Base):
     packetloss = Column(Float())
     returncode = Column(Integer())
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
-    # fk = ForeignKeyConstraint(['id'], [RiskVector.id])
 
     def __str__(self) -> str:
+        """Return string representation of InternetData instance."""
         return (
             "InternetData("
             + ", ".join(
@@ -33,10 +34,6 @@ def __str__(self) -> str:
             + ")"
         )
 
-
-from flask_admin.contrib.sqla import ModelView
-
-
 class InternetDataView(ModelView):
     def __init__(
         self,
@@ -49,7 +46,7 @@ def __init__(
         menu_class_name=None,
         menu_icon_type=None,
         menu_icon_value=None,
-    ):
+    ) -> None:
         super().__init__(
             InternetData,
             session,
@@ -67,5 +64,3 @@ def __init__(
     column_display_pk = True
     column_hide_backrefs = False
     column_list = ["id", "traceroute", "ping", "packetloss", "returncode", "datetime"]
-    # column_searchable_list = ["name"]
-    # inline_models = (RiskVector,)

model/ondeckdata.py

@@ -22,6 +22,7 @@ class OndeckData(Base):
     detection_confidence = Column(REAL)
 
     def __str__(self) -> str:
+        """Return a string representation of the OndeckData instance."""
         return (
             "OndeckData("
             + ", ".join(
@@ -30,7 +31,6 @@ def __str__(self) -> str:
                     for n in [
                         "id",
                         "video_uri",
-                        # 'video_file',
                         "cocoannotations_uri",
                         "datetime",
                         "overallcount",

model/riskvector.py

@@ -1,3 +1,4 @@
+from flask_admin.contrib.sqla import ModelView
 from flask_admin.model.template import EndpointLinkRowAction
 from sqlalchemy import Column, Integer, String
 from sqlalchemy.orm import relationship
@@ -15,6 +16,7 @@ class RiskVector(Base):
     tests = relationship("Test", back_populates="vector")
 
     def __str__(self) -> str:
+        """Return string representation of RiskVector instance."""
         return (
             "RiskVector("
             + ", ".join(
@@ -26,10 +28,6 @@ def __str__(self) -> str:
             + ")"
         )
 
-
-from flask_admin.contrib.sqla import ModelView
-
-
 class RiskVectorModelView(ModelView):
     def __init__(
         self,
@@ -42,7 +40,7 @@ def __init__(
         menu_class_name=None,
         menu_icon_type=None,
         menu_icon_value=None,
-    ):
+    ) -> None:
         super().__init__(
             RiskVector,
             session,

model/test.py

@@ -1,5 +1,6 @@
 from enum import Enum as PyEnum
 
+from flask_admin.contrib.sqla import ModelView
 from sqlalchemy import Column, DateTime, Enum, Float, ForeignKey, Integer, String, text
 from sqlalchemy.orm import relationship
 
@@ -26,9 +27,9 @@ class Test(Base):
     datetime_from = Column(DateTime(timezone=True))
     datetime_to = Column(DateTime(timezone=True))
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
-    # fk = ForeignKeyConstraint(['id'], [RiskVector.id])
 
     def __str__(self) -> str:
+        """Return string representation of Test object."""
         return (
             "Test("
             + ", ".join(
@@ -46,10 +47,6 @@ def __str__(self) -> str:
             + ")"
         )
 
-
-from flask_admin.contrib.sqla import ModelView
-
-
 class TestModelView(ModelView):
     def __init__(
         self,
@@ -62,7 +59,7 @@ def __init__(
         menu_class_name=None,
         menu_icon_type=None,
         menu_icon_value=None,
-    ):
+    ) -> None:
         super().__init__(
             Test,
             session,
@@ -92,5 +89,3 @@ def __init__(
     ]
     column_searchable_list = ["name"]
     column_filters = ["vector_id", "datetime"]
-    # column_select_related_list=['vector']
-    # inline_models = (RiskVector,)

model/track.py

@@ -8,20 +8,16 @@ class Track(Base):
 
     id = Column(Integer, primary_key=True)
     video_uri = Column(String)
-    # video_uri = Column(String, ForeignKey("video_files.decrypted_path"))
-    # video_file = relationship(VideoFile)
     cocoannotations_uri = Column(String)
-    # cocoannotations_uri = Column(String, ForeignKey("ondeckdata.cocoannotations_uri"))
-    # ondeckdata = relationship(OndeckData)
     track_id = Column(Integer)
     first_framenum = Column(Integer)
     last_framenum = Column(Integer)
     confidences = Column(ARRAY(REAL))
 
     datetime = Column(DateTime(timezone=True), server_default=text("CURRENT_TIMESTAMP"))
-    # detection_confidence = Column(REAL)
 
     def __str__(self) -> str:
+        """Return a string representation of the Track object."""
         return (
             "Track("
             + ", ".join(

model/videofiles.py

@@ -19,9 +19,8 @@ class VideoFile(Base):
     reencoded_stderr = Column(VARCHAR(), autoincrement=False, nullable=True)
     cam_name = Column(VARCHAR(), nullable=True)
 
-    # ondeckdata = relationship("OndeckData", back_populates="video_file")
-
     def __str__(self) -> str:
+        """Return string representation of VideoFile instance."""
         return (
             "VideoFile("
             + ", ".join(