WIP: Change realization.get_smry() to always return dummy index

src/fmu/ensemble/ensemble.py

@@ -689,8 +689,7 @@ def get_df(self, localpath, merge=None):
             # the realization index, and end up in a MultiIndex
             dframe = pd.concat(dflist, sort=False).reset_index()
             dframe.rename(columns={"level_0": "REAL"}, inplace=True)
-            del dframe["level_1"]  # This is the indices from each real
-            return dframe
+            return dframe.drop("level_1", axis="columns", errors="ignore")
         raise KeyError("No data found for " + localpath)
 
     def load_smry(
@@ -1329,6 +1328,9 @@ def get_smry(
         Wraps around Realization.get_smry() which wraps around
         ecl.summary.EclSum.pandas_frame()
 
+        The returned dataframe will always have a dummy index, and
+        DATE and REAL as columns.
+
         Args:
             time_index: list of DateTime if interpolation is wanted
                default is None, which returns the raw Eclipse report times
@@ -1374,10 +1376,9 @@ def get_smry(
                 include_restart=include_restart,
             )
             dframe.insert(0, "REAL", index)
-            dframe.index.name = "DATE"
             dflist.append(dframe)
         if dflist:
-            return pd.concat(dflist, sort=False).reset_index()
+            return pd.concat(dflist, sort=False)
         return pd.DataFrame()
 
     def get_eclgrid(self, props, report=0, agg="mean", active_only=False):

src/fmu/ensemble/realization.py

@@ -979,13 +979,16 @@ def get_smry(
         start_date=None,
         end_date=None,
         include_restart=True,
-        datetimeindex=False,
     ):
         """Wrapper for ecl2df.summary
 
         This gives access to the underlying data on disk without
         touching internalized dataframes.
 
+        The returned dataframe will have a dummy index, and the dates in
+        the column DATE. The DATE column will contain either datetime.datetime
+        or pandas.Timestamp objects.
+
         Arguments:
             time_index: string indicating a resampling frequency,
                'yearly', 'monthly', 'daily', 'first', 'last' or 'raw', the
@@ -1003,7 +1006,6 @@ def get_smry(
                 end_date will always be included. Overridden if time_index
                 is 'first' or 'last'.
             include_restart (bool): Whether to traverse restart files.
-            datetimeindex (bool): Set to True if a datetime64 indes is wanted.
 
         Returns empty dataframe if there is no summary file, or if the
         column_keys are not existing.
@@ -1020,8 +1022,7 @@ def get_smry(
                 include_restart=include_restart,
                 params=False,
                 paramfile=None,
-                datetime=datetimeindex,
-            )
+            ).reset_index()
         except OSError:
             # Missing or bogus UNSMRY file
             return pd.DataFrame()

src/fmu/ensemble/util/rates.py

@@ -65,7 +65,9 @@ def compute_volumetric_rates(realization, column_keys, time_index, time_unit):
         return pd.DataFrame()
 
     cum_df = realization.get_smry(column_keys=column_keys, time_index=time_index)
-    # get_smry() for realizations return a dataframe indexed by 'DATE'
+
+    if not cum_df.empty:
+        cum_df.set_index("DATE", inplace=True)
 
     # Compute row-wise difference, shift back one row
     # to get the NaN to the end, and then drop the NaN.

tests/test_ensemble.py

@@ -71,6 +71,7 @@ def test_reek001(tmpdir):
     paramsdf = reekensemble.parameters  # also test as property
     paramsdf = reekensemble.get_df("parameters.txt")
     assert len(paramsdf) == 5
+    print(paramsdf.head())
     assert len(paramsdf.columns) == 26  # 25 parameters, + REAL column
     paramsdf.to_csv("params.csv", index=False)
 
@@ -498,9 +499,6 @@ def test_ensemble_ecl():
     assert not reekensemble.get_wellnames("")
     assert len(reekensemble.get_wellnames(["OP*", "WI*"])) == 8
 
-    # eclipse well groups list
-    assert len(reekensemble.get_groupnames()) == 3
-
     # delta between two ensembles
     diff = reekensemble - reekensemble
     assert len(diff.get_smry(column_keys=["FOPR", "FGPR", "FWCT"]).columns) == 5

tests/test_realization.py

@@ -330,7 +330,7 @@ def test_volumetric_rates():
     daily_dates = real.get_smry_dates(freq="daily", normalize=False)
     subset_dates = list(np.random.choice(daily_dates, size=10, replace=False))
     subset_dates.sort()
-    dcum = real.get_smry(column_keys="FOPT", time_index=subset_dates)
+    dcum = real.get_smry(column_keys="FOPT", time_index=subset_dates).set_index("DATE")
     ddcum = real.get_volumetric_rates(column_keys="FOPT", time_index=subset_dates)
     assert ddcum["FOPR"].iloc[-1] == 0
 
@@ -426,21 +426,29 @@ def test_datenormalization():
     realdir = os.path.join(testdir, "data/testensemble-reek001", "realization-0/iter-0")
     real = ensemble.ScratchRealization(realdir)
     raw = real.get_smry(column_keys="FOPT", time_index="raw")
-    assert str(raw.index[-1]) == "2003-01-02 00:00:00"
+    assert str(raw["DATE"].values[-1]) == "2003-01-02T00:00:00.000000000"
     daily = real.get_smry(column_keys="FOPT", time_index="daily")
-    assert str(daily.index[-1]) == "2003-01-02"
+    assert str(daily["DATE"].values[-1]) == "2003-01-02"
     monthly = real.get_smry(column_keys="FOPT", time_index="monthly")
-    assert str(monthly.index[-1]) == "2003-02-01"
+    assert str(monthly["DATE"].values[-1]) == "2003-02-01"
     yearly = real.get_smry(column_keys="FOPT", time_index="yearly")
-    assert str(yearly.index[-1]) == "2004-01-01"
+    assert str(yearly["DATE"].values[-1]) == "2004-01-01"
     weekly = real.get_smry(column_keys="FOPT", time_index="weekly")
-    assert str(weekly.index[-1]) == "2003-01-06"  # First Monday after 2003-01-02
+    assert (
+        str(weekly["DATE"].values[-1]) == "2003-01-06"
+    )  # First Monday after 2003-01-02
     weekly = real.get_smry(column_keys="FOPT", time_index="W-MON")
-    assert str(weekly.index[-1]) == "2003-01-06"  # First Monday after 2003-01-02
+    assert (
+        str(weekly["DATE"].values[-1]) == "2003-01-06"
+    )  # First Monday after 2003-01-02
     weekly = real.get_smry(column_keys="FOPT", time_index="W-TUE")
-    assert str(weekly.index[-1]) == "2003-01-07"  # First Tuesday after 2003-01-02
+    assert (
+        str(weekly["DATE"].values[-1]) == "2003-01-07"
+    )  # First Tuesday after 2003-01-02
     weekly = real.get_smry(column_keys="FOPT", time_index="W-THU")
-    assert str(weekly.index[-1]) == "2003-01-02"  # First Thursday after 2003-01-02
+    assert (
+        str(weekly["DATE"].values[-1]) == "2003-01-02"
+    )  # First Thursday after 2003-01-02
 
     # Check that time_index=None and time_index="raw" behaves like default
     raw = real.load_smry(column_keys="FOPT", time_index="raw")
@@ -454,15 +462,18 @@ def test_datenormalization():
     # Check that we get the same correct normalization
     # with load_smry()
     real.load_smry(column_keys="FOPT", time_index="raw")
-    assert str(real.get_df("unsmry--raw")["DATE"].iloc[-1]) == "2003-01-02 00:00:00"
+    assert (
+        str(real.get_df("unsmry--raw")["DATE"].values[-1])
+        == "2003-01-02T00:00:00.000000000"
+    )
     real.load_smry(column_keys="FOPT", time_index="daily")
-    assert str(real.get_df("unsmry--daily")["DATE"].iloc[-1]) == "2003-01-02"
+    assert str(real.get_df("unsmry--daily")["DATE"].values[-1]) == "2003-01-02"
     real.load_smry(column_keys="FOPT", time_index="monthly")
-    assert str(real.get_df("unsmry--monthly")["DATE"].iloc[-1]) == "2003-02-01"
+    assert str(real.get_df("unsmry--monthly")["DATE"].values[-1]) == "2003-02-01"
     real.load_smry(column_keys="FOPT", time_index="yearly")
-    assert str(real.get_df("unsmry--yearly")["DATE"].iloc[-1]) == "2004-01-01"
+    assert str(real.get_df("unsmry--yearly")["DATE"].values[-1]) == "2004-01-01"
     real.load_smry(column_keys="FOPT", time_index="weekly")
-    assert str(real.get_df("unsmry--weekly")["DATE"].iloc[-1]) == "2003-01-06"
+    assert str(real.get_df("unsmry--weekly")["DATE"].values[-1]) == "2003-01-06"
 
 
 def test_singlereal_ecl(tmp="TMP"):
@@ -512,7 +523,7 @@ def test_singlereal_ecl(tmp="TMP"):
     # Try ISO-date for time_index:
     singledata = real.get_smry(time_index="2000-05-05", column_keys="FOPT")
     assert "FOPT" in singledata
-    assert "2000-05-05" in singledata.index
+    assert str(singledata["DATE"].values[0]).startswith("2000-05-05")
 
     # start and end should be included:
     assert (