From 9bbe1e3c8c9b51890d6bfff7c53b1929adbe02d3 Mon Sep 17 00:00:00 2001
From: John Vouvakis Manousakis <ioannis_vm@berkeley.edu>
Date: Sat, 18 May 2024 11:51:19 -0700
Subject: [PATCH] Properly handle the case of no losses when aggregating.

Loss model: make `aggregate_losses` return a DataFrame with zeros.
---
 pelicun/model/loss_model.py            | 19 ++++++++++---------
 pelicun/tests/model/test_loss_model.py | 19 +++++++++++++++++++
 2 files changed, 29 insertions(+), 9 deletions(-)

diff --git a/pelicun/model/loss_model.py b/pelicun/model/loss_model.py
index d055210e7..1fe54229d 100644
--- a/pelicun/model/loss_model.py
+++ b/pelicun/model/loss_model.py
@@ -537,25 +537,26 @@ def aggregate_losses(self):
             for model in self._loss_models
             if model.sample is not None
         ]
-        if not samples:
-            self.log.msg("There are no losses.")
-            return None
-
-        sample = pd.concat(samples, axis=1)
-
-        # group results by DV type and location
-        aggregated = sample.groupby(level=['dv', 'loc'], axis=1).sum()
 
         # Note: The `Time` DV receives special treatment.
-
         # create the summary DF
         columns = [
             f'repair_{x.lower()}' for x in self.decision_variables if x != 'Time'
         ]
         if 'Time' in self.decision_variables:
             columns.extend(('repair_time-sequential', 'repair_time-parallel'))
+
+        if not samples:
+            self.log.msg("There are no losses.")
+            df_agg = pd.DataFrame(0.00, index=[0], columns=columns)
+            return df_agg
+
+        sample = pd.concat(samples, axis=1)
         df_agg = pd.DataFrame(index=sample.index, columns=columns)
 
+        # group results by DV type and location
+        aggregated = sample.groupby(level=['dv', 'loc'], axis=1).sum()
+
         for decision_variable in self.decision_variables:
 
             # Time ..
diff --git a/pelicun/tests/model/test_loss_model.py b/pelicun/tests/model/test_loss_model.py
index 6f985094c..068cf3427 100644
--- a/pelicun/tests/model/test_loss_model.py
+++ b/pelicun/tests/model/test_loss_model.py
@@ -255,6 +255,25 @@ def test__ensure_loss_parameter_availability(self, assessment_instance):
             "[('consequence_E', 'DecisionVariableXYZ')]"
         ) in str(record[0].message)
 
+    def test_aggregate_losses_when_no_loss(self, assessment_instance):
+
+        # tests that aggregate losses works when there is no loss.
+        loss_model = LossModel(assessment_instance)
+        df_agg = loss_model.aggregate_losses()
+        pd.testing.assert_frame_equal(
+            df_agg,
+            pd.DataFrame(
+                {
+                    'repair_carbon': 0.0,
+                    'repair_cost': 0.00,
+                    'repair_energy': 0.00,
+                    'repair_time-sequential': 0.00,
+                    'repair_time-parallel': 0.00,
+                },
+                index=[0],
+            ),
+        )
+
 
 class TestRepairModel_Base(TestPelicunModel):
     def test___init__(self, assessment_instance):