fix: avoid IndexError in carbonboard when no project is selected

codecarbon/viz/components.py

@@ -91,7 +91,8 @@ def get_project_dropdown(df: pd.DataFrame):
                     dcc.Dropdown(
                         id="project_name",
                         options=[{"label": i, "value": i} for i in projects],
-                        value=projects[0],
+                        value=projects[0] if projects else None,
+                        clearable=False,
                     ),
                 ],
                 style={"display": "inline-block"},

codecarbon/viz/data.py

@@ -23,6 +23,24 @@ def get_project_data(df: pd.DataFrame, project_name: str) -> dt.DataTable:
 
     @staticmethod
     def get_project_summary(project_data: List[Dict]):
+        if not project_data:
+            # No run to summarise, e.g. when the project dropdown is cleared.
+            # Keep the same shape so the callbacks reading it still work.
+            return {
+                "last_run": {
+                    "timestamp": "",
+                    "duration": 0,
+                    "emissions": 0,
+                    "energy_consumed": 0,
+                },
+                "total": {"duration": 0, "emissions": 0, "energy_consumed": 0},
+                "country_name": "",
+                "country_iso_code": "",
+                "region": "",
+                "on_cloud": "N",
+                "cloud_provider": "",
+                "cloud_region": "",
+            }
         last_run = project_data[-1]
         project_summary = {
             "last_run": {

tests/test_viz_data.py

@@ -30,6 +30,54 @@ def test_get_project_data(emissions_data: pd.DataFrame):
     pd.testing.assert_frame_equal(pd.DataFrame(project_data.data), emissions_data)
 
 
+def test_get_project_summary(emissions_data: pd.DataFrame):
+    viz_data = data.Data()
+    project_data = viz_data.get_project_data(emissions_data, project_name="codecarbon")
+    project_summary = viz_data.get_project_summary(project_data.data)
+    assert project_summary["last_run"]["timestamp"] == "2021-09-23T15:04:51"
+    assert project_summary["total"]["duration"] == pytest.approx(161.20380687713623)
+    assert project_summary["total"]["emissions"] == pytest.approx(0.0004490989249167)
+    assert project_summary["total"]["energy_consumed"] == pytest.approx(
+        0.00057442898176
+    )
+    assert project_summary["country_name"] == "Morocco"
+    assert project_summary["country_iso_code"] == "MAR"
+    assert project_summary["region"] == "casablanca-settat"
+    assert project_summary["on_cloud"] == "N"
+
+
+def test_get_project_summary_empty(emissions_data: pd.DataFrame):
+    """Selecting no project used to raise an IndexError, see #918."""
+    viz_data = data.Data()
+    project_data = viz_data.get_project_data(emissions_data, project_name=None)
+    assert project_data.data == []
+
+    project_summary = viz_data.get_project_summary(project_data.data)
+    assert project_summary == {
+        "last_run": {
+            "timestamp": "",
+            "duration": 0,
+            "emissions": 0,
+            "energy_consumed": 0,
+        },
+        "total": {"duration": 0, "emissions": 0, "energy_consumed": 0},
+        "country_name": "",
+        "country_iso_code": "",
+        "region": "",
+        "on_cloud": "N",
+        "cloud_provider": "",
+        "cloud_region": "",
+    }
+
+    # The callbacks read the same keys whether or not a project is selected.
+    populated_summary = viz_data.get_project_summary(
+        viz_data.get_project_data(emissions_data, project_name="codecarbon").data
+    )
+    assert project_summary.keys() == populated_summary.keys()
+    assert project_summary["last_run"].keys() == populated_summary["last_run"].keys()
+    assert project_summary["total"].keys() == populated_summary["total"].keys()
+
+
 def test_get_global_emissions_choropleth_data(
     global_energy_mix_data: Dict[str, Dict[str, Any]],
 ):