kas-BDMS-626-resolve-database-errors

core/lexicon.json

@@ -4452,6 +4452,111 @@
       "term": "Zamora Accounting Services",
       "definition": "Zamora Accounting Services"
     },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Agua Sana MWCD",
+      "definition": "Agua Sana MWCD"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Canada Los Alamos MDWCA",
+      "definition": "Canada Los Alamos MDWCA"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Canjilon Mutual Domestic Water System",
+      "definition": "Canjilon Mutual Domestic Water System"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Cebolla Mutual Domestic",
+      "definition": "Cebolla Mutual Domestic"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Chihuahuan Desert Rangeland Research Center (CDRRC)",
+      "definition": "Chihuahuan Desert Rangeland Research Center (CDRRC)"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "East Rio Arriba SWCD",
+      "definition": "East Rio Arriba SWCD"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "El Prado Municipal Water",
+      "definition": "El Prado Municipal Water"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Hachita Mutual Domestic",
+      "definition": "Hachita Mutual Domestic"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Jornada Experimental Range (JER)",
+      "definition": "Jornada Experimental Range (JER)"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "La Canada Way HOA",
+      "definition": "La Canada Way HOA"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Los Ojos Mutual Domestic",
+      "definition": "Los Ojos Mutual Domestic"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "The Nature Conservancy (TNC)",
+      "definition": "The Nature Conservancy (TNC)"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Smith Ranch LLC",
+      "definition": "Smith Ranch LLC"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Zia Pueblo",
+      "definition": "Zia Pueblo"
+    },
+    {
+      "categories": [
+        "organization"
+      ],
+      "term": "Our Lady of Guadalupe (OLG)",
+      "definition": "Our Lady of Guadalupe (OLG)"
+    },
     {
       "categories": [
         "organization"

schemas/well_inventory.py

@@ -190,6 +190,7 @@ def validator(v):
 EmailField: TypeAlias = Annotated[
     Optional[str], BeforeValidator(email_validator_function)
 ]
+OptionalText: TypeAlias = Annotated[Optional[str], BeforeValidator(empty_str_to_none)]
 
 OptionalBool: TypeAlias = Annotated[Optional[bool], BeforeValidator(empty_str_to_none)]
 OptionalPastOrTodayDateTime: TypeAlias = Annotated[
@@ -215,18 +216,18 @@ class WellInventoryRow(BaseModel):
     utm_zone: str
 
     # Optional fields
-    site_name: Optional[str] = None
+    site_name: OptionalText = None
     elevation_ft: OptionalFloat = None
     elevation_method: Annotated[
         Optional[ElevationMethod],
         BeforeValidator(flexible_lexicon_validator(ElevationMethod)),
     ] = None
     measuring_point_height_ft: OptionalFloat = None
-    field_staff_2: Optional[str] = None
-    field_staff_3: Optional[str] = None
+    field_staff_2: OptionalText = None
+    field_staff_3: OptionalText = None
 
-    contact_1_name: Optional[str] = None
-    contact_1_organization: Optional[str] = None
+    contact_1_name: OptionalText = None
+    contact_1_organization: OptionalText = None
     contact_1_role: ContactRoleField = None
     contact_1_type: ContactTypeField = None
     contact_1_phone_1: PhoneField = None
@@ -250,8 +251,8 @@ class WellInventoryRow(BaseModel):
     contact_1_address_2_city: Optional[str] = None
     contact_1_address_2_postal_code: PostalCodeField = None
 
-    contact_2_name: Optional[str] = None
-    contact_2_organization: Optional[str] = None
+    contact_2_name: OptionalText = None
+    contact_2_organization: OptionalText = None
     contact_2_role: ContactRoleField = None
     contact_2_type: ContactTypeField = None
     contact_2_phone_1: PhoneField = None
@@ -296,7 +297,7 @@ class WellInventoryRow(BaseModel):
     measuring_point_description: Optional[str] = None
     well_purpose: WellPurposeField = None
     well_purpose_2: WellPurposeField = None
-    well_status: Optional[str] = Field(
+    well_status: OptionalText = Field(
         default=None,
         validation_alias=AliasChoices("well_status", "well_hole_status"),
     )

services/well_inventory_csv.py

@@ -438,6 +438,44 @@ def _generate_autogen_well_id(session, prefix: str, offset: int = 0) -> tuple[st
     return f"{prefix}{new_number:04d}", new_number
 
 
+def _find_existing_imported_well(
+    session: Session, model: WellInventoryRow
+) -> Thing | None:
+    if model.measurement_date_time is not None:
+        sample_name = (
+            f"{model.well_name_point_id}-WL-"
+            f"{model.measurement_date_time.strftime('%Y%m%d%H%M')}"
+        )
+        existing = session.scalars(
+            select(Thing)
+            .join(FieldEvent, FieldEvent.thing_id == Thing.id)
+            .join(FieldActivity, FieldActivity.field_event_id == FieldEvent.id)
+            .join(Sample, Sample.field_activity_id == FieldActivity.id)
+            .where(
+                Thing.name == model.well_name_point_id,
+                Thing.thing_type == "water well",
+                FieldActivity.activity_type == "well inventory",
+                Sample.sample_name == sample_name,
+            )
+            .order_by(Thing.id.asc())
+        ).first()
+        if existing is not None:
+            return existing
+
+    return session.scalars(
+        select(Thing)
+        .join(FieldEvent, FieldEvent.thing_id == Thing.id)
+        .join(FieldActivity, FieldActivity.field_event_id == FieldEvent.id)
+        .where(
+            Thing.name == model.well_name_point_id,
+            Thing.thing_type == "water well",
+            FieldEvent.event_date == model.date_time,
+            FieldActivity.activity_type == "well inventory",
+        )
+        .order_by(Thing.id.asc())
+    ).first()
+
+
 def _make_row_models(rows, session):
     models = []
     validation_errors = []
@@ -542,6 +580,10 @@ def _add_csv_row(session: Session, group: Group, model: WellInventoryRow, user)
     name = model.well_name_point_id
     date_time = model.date_time
 
+    existing_well = _find_existing_imported_well(session, model)
+    if existing_well is not None:
+        return existing_well.name
+
     # --------------------
     # Location and associated tables
     # --------------------

tests/test_well_inventory.py

@@ -494,6 +494,38 @@ def test_blank_depth_to_water_still_creates_water_level_records(tmp_path):
         assert observations[0].measuring_point_height == 2.5
 
 
+def test_rerunning_same_well_inventory_csv_is_idempotent():
+    """Re-importing the same CSV should not create duplicate well inventory records."""
+    file = Path("tests/features/data/well-inventory-valid.csv")
+    assert file.exists(), "Test data file does not exist."
+
+    first = well_inventory_csv(file)
+    assert first.exit_code == 0, first.stderr
+
+    with session_ctx() as session:
+        counts_after_first = {
+            "things": session.query(Thing).count(),
+            "field_events": session.query(FieldEvent).count(),
+            "field_activities": session.query(FieldActivity).count(),
+            "samples": session.query(Sample).count(),
+            "observations": session.query(Observation).count(),
+        }
+
+    second = well_inventory_csv(file)
+    assert second.exit_code == 0, second.stderr
+
+    with session_ctx() as session:
+        counts_after_second = {
+            "things": session.query(Thing).count(),
+            "field_events": session.query(FieldEvent).count(),
+            "field_activities": session.query(FieldActivity).count(),
+            "samples": session.query(Sample).count(),
+            "observations": session.query(Observation).count(),
+        }
+
+    assert counts_after_second == counts_after_first
+
+
 # =============================================================================
 # Error Handling Tests - Cover API error paths
 # =============================================================================
@@ -1095,6 +1127,24 @@ def test_blank_depth_to_water_is_treated_as_none(self):
         )
         assert model.depth_to_water_ft is None
 
+    def test_blank_contact_organization_is_treated_as_none(self):
+        row = _minimal_valid_well_inventory_row()
+        row["contact_1_name"] = "Test Contact"
+        row["contact_1_organization"] = ""
+
+        model = WellInventoryRow(**row)
+
+        assert model.contact_1_name == "Test Contact"
+        assert model.contact_1_organization is None
+
+    def test_blank_well_status_is_treated_as_none(self):
+        row = _minimal_valid_well_inventory_row()
+        row["well_hole_status"] = ""
+
+        model = WellInventoryRow(**row)
+
+        assert model.well_status is None
+
     def test_canonical_name_wins_when_alias_and_canonical_present(self):
         row = _minimal_valid_well_inventory_row()
         row["well_status"] = "Abandoned"