Merge pull request #765 from DHI/bug/Dataset.extract_track_modifies_i…

…nput

Bug/Dataset.extract_track_modifies_input

mikeio/_track.py

@@ -8,7 +8,7 @@
 
 from .dataset import Dataset
 from .dfs import Dfs0
-from .eum import ItemInfo
+from .eum import ItemInfo, EUMUnit, EUMType
 from .spatial import GeometryFM2D
 
 
@@ -33,33 +33,17 @@ def _extract_track(
 
     n_items = len(item_numbers)
 
-    if isinstance(track, str):
-        filename = track
-        path = Path(filename)
-        if not path.exists():
-            raise ValueError(f"{filename} does not exist")
-
-        ext = path.suffix.lower()
-        if ext == ".dfs0":
-            df = Dfs0(filename).to_dataframe()
-        elif ext == ".csv":
-            df = pd.read_csv(filename, index_col=0, parse_dates=True)
-            df.index = pd.DatetimeIndex(df.index)
-        else:
-            raise ValueError(f"{ext} files not supported (dfs0, csv)")
-
-        times = df.index
-        coords = df.iloc[:, 0:2].to_numpy(copy=True)
-
-    elif isinstance(track, Dataset):
-        times = track.time
-        coords = np.zeros(shape=(len(times), 2))
-        coords[:, 0] = track[0].to_numpy().copy()
-        coords[:, 1] = track[1].to_numpy().copy()
-    else:
-        assert isinstance(track, pd.DataFrame)
-        times = track.index
-        coords = track.iloc[:, 0:2].to_numpy(copy=True)
+    match track:
+        case str():
+            times, coords = _get_track_data_from_file(track)
+        case Dataset():
+            times, coords = _get_track_data_from_dataset(track)
+        case pd.DataFrame():
+            times, coords = _get_track_data_from_dataframe(track)
+        case _:
+            raise ValueError(
+                "track must be a file name, a Dataset or a pandas DataFrame"
+            )
 
     assert isinstance(
         times, pd.DatetimeIndex
@@ -157,11 +141,72 @@ def is_EOF(step: int) -> bool:
             data_list[item + 2][t] = dati[item]
 
     if geometry.is_geo:
-        items_out = [ItemInfo("Longitude"), ItemInfo("Latitude")]
+        items_out = [
+            ItemInfo("Longitude", EUMType.Latitude_longitude, EUMUnit.degree),
+            ItemInfo("Latitude", EUMType.Latitude_longitude, EUMUnit.degree),
+        ]
     else:
-        items_out = [ItemInfo("x"), ItemInfo("y")]
+        items_out = [
+            ItemInfo("x", EUMType.Geographical_coordinate, EUMUnit.meter),
+            ItemInfo("y", EUMType.Geographical_coordinate, EUMUnit.meter),
+        ]
 
     for item_info in items:
         items_out.append(item_info)
 
     return Dataset(data_list, times, items_out)
+
+
+def _get_track_data_from_dataset(track: Dataset) -> tuple[pd.DatetimeIndex, np.ndarray]:
+    times = track.time
+    coords = np.zeros(shape=(len(times), 2))
+    coords[:, 0] = track[0].to_numpy().copy()
+    coords[:, 1] = track[1].to_numpy().copy()
+    return times, coords
+
+
+def _get_track_data_from_dataframe(
+    track: pd.DataFrame,
+) -> tuple[pd.DatetimeIndex, np.ndarray]:
+    times = track.index
+    coords = track.iloc[:, 0:2].to_numpy(copy=True)
+    return times, coords
+
+
+def _get_track_data_from_file(track: str) -> tuple[pd.DatetimeIndex, np.ndarray]:
+    filename = track
+    path = Path(filename)
+    if not path.exists():
+        raise FileNotFoundError(f"{filename} does not exist")
+
+    ext = path.suffix.lower()
+    match ext:
+        case ".dfs0":
+            df = Dfs0(filename).to_dataframe()
+        case ".csv":
+            df = pd.read_csv(filename, index_col=0, parse_dates=True)
+            df.index = pd.DatetimeIndex(df.index)
+        case _:
+            raise ValueError(f"{ext} files not supported (dfs0, csv)")
+
+    times = df.index
+    coords = df.iloc[:, 0:2].to_numpy(copy=True)
+
+    return times, coords
+
+
+def _find_end_index(t_rel: pd.Index, end_time: pd.Timestamp) -> int:
+    # largest idx for which (times - self.end_time)<=0
+    tmp = np.where(t_rel <= 0)[0]
+    if len(tmp) == 0:
+        raise ValueError("No time overlap!")
+    i_end = tmp[-1]
+    return i_end
+
+
+def _find_start_index(t_rel: pd.Index, start_time: pd.Timestamp) -> int:
+    tmp = np.where(t_rel >= 0)[0]
+    if len(tmp) == 0:
+        raise ValueError("No time overlap!")
+    i_start = tmp[0]  # smallest idx for which t_rel>=0
+    return i_start

mikeio/dataset/_dataarray.py

@@ -1064,7 +1064,7 @@ def extract_track(
             geometry=self.geometry,
             n_elements=self.shape[1],  # TODO is there a better way to find out this?
             track=track,
-            items=[self.item],
+            items=deepcopy([self.item]),
             time_steps=list(range(self.n_timesteps)),
             item_numbers=[0],
             method=method,

mikeio/dataset/_dataset.py

@@ -1051,7 +1051,7 @@ def extract_track(
             geometry=self.geometry,
             n_elements=self.shape[1],  # TODO is there a better way to find out this?
             track=track,
-            items=self.items,
+            items=deepcopy(self.items),
             time_steps=time_steps,
             item_numbers=item_numbers,
             method=method,

mikeio/dfsu/_dfsu.py

@@ -1,4 +1,5 @@
 from __future__ import annotations
+from copy import deepcopy
 from dataclasses import dataclass
 from datetime import datetime
 from pathlib import Path
@@ -686,18 +687,15 @@ def extract_track(
 
         Examples
         --------
-        >>> dfsu = mikeio.open("tests/testdata/NorthSea_HD_and_windspeed.dfsu")
-        >>> ds = dfsu.extract_track("tests/testdata/altimetry_NorthSea_20171027.csv")
-        >>> ds
-        <mikeio.Dataset>
-        dims: (time:1115)
-        time: 2017-10-26 04:37:37 - 2017-10-30 20:54:47 (1115 non-equidistant records)
-        geometry: GeometryUndefined()
-        items:
-          0:  Longitude <Undefined> (undefined)
-          1:  Latitude <Undefined> (undefined)
-          2:  Surface elevation <Surface Elevation> (meter)
-          3:  Wind speed <Wind speed> (meter per sec)
+        ```{python}
+        import mikeio
+
+        ds = (
+            mikeio.open("../data/NorthSea_HD_and_windspeed.dfsu")
+                  .extract_track("../data/altimetry_NorthSea_20171027.csv")
+            )
+        ds
+        ```
 
         """
         dfs = DfsuFile.Open(self._filename)
@@ -714,7 +712,7 @@ def extract_track(
             geometry=self.geometry,
             n_elements=self.geometry.n_elements,
             track=track,
-            items=items,
+            items=deepcopy(items),
             time_steps=time_steps,
             item_numbers=item_numbers,
             method=method,

tests/test_dfsu.py

@@ -788,9 +788,16 @@ def test_extract_track_from_dataset():
         parse_dates=True,
     )
     df.index = pd.DatetimeIndex(df.index)
+    assert ds[0].name == "Sign. Wave Height"
     track = ds.extract_track(df)
 
-    assert track[2].values[23] == approx(3.6284972794399653)
+    # This should not change the original dataset
+    track.rename({"Sign. Wave Height": "Hm0"}, inplace=True)
+    assert track["Hm0"].name == "Hm0"
+
+    assert ds[0].name == "Sign. Wave Height"
+
+    assert track["Hm0"].values[23] == approx(3.6284972794399653)
     assert sum(np.isnan(track[2].to_numpy())) == 26
     assert np.all(track[1].to_numpy() == df.latitude.values)