Complete unwanted aggregated on df creation

Imitate prior behaviour wrt na and 0 values
Add test case for mixed type dataframes
Completes #1323

Author: MariusWirtz

TM1py/Utils/Utils.py

@@ -623,14 +623,13 @@ def build_dataframe_from_csv(
 
     # create a unique row id within each (idx_cols + col_col) group
     df["_dup"] = df.groupby(idx_cols + [col_col]).cumcount()
-
     df = (
         df.pivot(
             index=idx_cols + ["_dup"],
             columns=col_col,
             values=val_col,
         )
-        .dropna(how="all")
+        .fillna(0)
         .reset_index()
         .drop(columns="_dup")
     )

Tests/Utils_test.py

@@ -150,6 +150,25 @@ def test_build_dataframe_from_csv(self):
 
         pd._testing.assert_frame_equal(expected_df, df, check_column_type=False)
 
+    def test_build_dataframe_from_csv_shaped_numbers_and_strings(self):
+        raw_csv = "Region~Product~Measure~Value\r\n" "r1~p1~Revenue~1.0\r\n" "r1~p2~Revenue~3.0\r\n" "r1~p1~Comment~Great Product\r\n" "r1~p2~Comment~"
+        df = build_dataframe_from_csv(raw_csv, dtype={"Revenue": float}, shaped=True)
+
+        expected_df = pd.DataFrame(
+            {
+                "Region":  ["r1", "r1"],
+                "Product": ["p1", "p2"],
+                "Comment": ["Great Product", ""],
+                "Revenue": [1.00000, 3.00000],
+            }
+        )
+
+        # explicit conversion for exact comparison
+        expected_df["Revenue"] = expected_df["Revenue"].astype(float)
+        df["Revenue"] = df["Revenue"].astype(float)
+
+        pd._testing.assert_frame_equal(expected_df, df)
+
     def test_build_dataframe_from_csv_shaped(self):
         raw_csv = "Region~Product~Measure~Value\r\n" "r1~p1~Revenue~1.0\r\n" "r1~p2~Revenue~3.0\r\n" "r2~p2~Revenue~4.0"
         df = build_dataframe_from_csv(raw_csv, dtype={"Revenue": float}, shaped=True)