From e82ce985be542029ca91e344889120def87b3cc2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Therese=20Natter=C3=B8y?=
 <61694854+tnatt@users.noreply.github.com>
Date: Mon, 9 Dec 2024 16:20:42 +0100
Subject: [PATCH] ENH: Add validation of required volume columns

---
 src/fmu/dataio/export/rms/inplace_volumes.py  | 44 +++++++++++-
 .../test_export_rms_volumetrics.py            | 72 +++++++++++++++++++
 2 files changed, 115 insertions(+), 1 deletion(-)

diff --git a/src/fmu/dataio/export/rms/inplace_volumes.py b/src/fmu/dataio/export/rms/inplace_volumes.py
index 30de69cbf..f7b215eef 100644
--- a/src/fmu/dataio/export/rms/inplace_volumes.py
+++ b/src/fmu/dataio/export/rms/inplace_volumes.py
@@ -201,6 +201,47 @@ def _convert_table_from_legacy_to_standard_format(
         table = self._add_missing_columns_to_table(table)
         return self._set_table_column_order(table)
 
+    def _is_column_missing_in_table(self, column: str) -> bool:
+        """Check if a column is present in the final dataframe and has values"""
+        return column not in self._dataframe or self._dataframe[column].isna().all()
+
+    def _validate_table(self) -> None:
+        """
+        Validate that the final table with volumes is according to the standard
+        defined for the inplace_volumes product.
+        """
+        _logger.debug("Validating the dataframe...")
+
+        has_oil = "oil" in self._dataframe[_FLUID_COLUMN].values
+        has_gas = "gas" in self._dataframe[_FLUID_COLUMN].values
+
+        # check that one of oil and gas fluids are present
+        if not (has_oil or has_gas):
+            raise RuntimeError(
+                "One or both 'oil' and 'gas' needs to be selected as 'Main types'"
+                "in the volumetric job. Please update and rerun the volumetric job "
+                "before export."
+            )
+
+        # create list of missing or non-defined required columns
+        missing_calculations = []
+        for col in ["BULK", "PORV", "HCPV"]:
+            if self._is_column_missing_in_table(col):
+                missing_calculations.append(col)
+
+        if has_oil and self._is_column_missing_in_table("STOIIP"):
+            missing_calculations.append("STOIIP")
+
+        if has_gas and self._is_column_missing_in_table("GIIP"):
+            missing_calculations.append("GIIP")
+
+        if missing_calculations:
+            raise RuntimeError(
+                f"Required calculations {missing_calculations} are missing "
+                f"in the volumetric table {self._volume_table_name}. Please update and "
+                "rerun the volumetric job before export."
+            )
+
     def _export_volume_table(self) -> ExportResult:
         """Do the actual volume table export using dataio setup."""
 
@@ -227,7 +268,8 @@ def _export_volume_table(self) -> ExportResult:
         )
 
     def export(self) -> ExportResult:
-        """Export the volume table."""
+        """Validate and export the volume table."""
+        self._validate_table()
         return self._export_volume_table()
 
 
diff --git a/tests/test_export_rms/test_export_rms_volumetrics.py b/tests/test_export_rms/test_export_rms_volumetrics.py
index c0c531812..412820f3a 100644
--- a/tests/test_export_rms/test_export_rms_volumetrics.py
+++ b/tests/test_export_rms/test_export_rms_volumetrics.py
@@ -185,6 +185,78 @@ def test_convert_table_from_legacy_to_standard_format(
     )
 
 
+@pytest.mark.parametrize("required_col", ["BULK", "PORV", "HCPV"])
+def test_validate_table_required_col_missing(
+    exportvolumetrics, voltable_standard, required_col
+):
+    """Test that the job fails if a required volumetric column is missing"""
+
+    df = voltable_standard.drop(columns=required_col)
+    exportvolumetrics._dataframe = df
+
+    with pytest.raises(RuntimeError, match="missing"):
+        exportvolumetrics._validate_table()
+
+
+@pytest.mark.parametrize("required_col", ["BULK", "PORV", "HCPV"])
+def test_validate_table_required_col_has_nan(
+    exportvolumetrics, voltable_standard, required_col
+):
+    """Test that the job fails if a required volumetric column has nan values"""
+
+    df = voltable_standard.copy()
+    df[required_col] = np.nan
+
+    exportvolumetrics._dataframe = df
+
+    with pytest.raises(RuntimeError, match="missing"):
+        exportvolumetrics._validate_table()
+
+
+def test_validate_table_has_oil_or_gas(exportvolumetrics, voltable_standard):
+    """Test that the job fails if a required volumetric column has nan values"""
+
+    df = voltable_standard.copy()
+    df = df[~df["FLUID"].isin(["oil", "gas"])]
+
+    exportvolumetrics._dataframe = df
+
+    with pytest.raises(RuntimeError, match="missing"):
+        exportvolumetrics._validate_table()
+
+
+def test_validate_table_has_oil_and_stoiip(exportvolumetrics, voltable_standard):
+    """Test that the validation fails if oil columns are present but no STOIIP"""
+
+    df = voltable_standard.copy()
+    df = df.drop(columns="STOIIP")
+
+    exportvolumetrics._dataframe = df
+
+    with pytest.raises(RuntimeError, match="missing"):
+        exportvolumetrics._validate_table()
+
+    # validation should pass when no oil columns are present
+    exportvolumetrics._dataframe = df[~(df["FLUID"] == "oil")]
+    exportvolumetrics._validate_table()
+
+
+def test_validate_table_has_gas_and_giip(exportvolumetrics, voltable_standard):
+    """Test that the validations fails if gas columns are present but no GIIP"""
+
+    df = voltable_standard.copy()
+    df = df.drop(columns="GIIP")
+
+    exportvolumetrics._dataframe = df
+
+    with pytest.raises(RuntimeError, match="missing"):
+        exportvolumetrics._validate_table()
+
+    # validation should pass when no gas columns are present
+    exportvolumetrics._dataframe = df[~(df["FLUID"] == "gas")]
+    exportvolumetrics._validate_table()
+
+
 @inside_rms
 def test_rms_volumetrics_export_config_missing(
     mock_project_variable,