Merge branch 'main' into pathlib

.github/workflows/build_docs.yml

@@ -1,8 +1,7 @@
 name: Build documentation (don't publish)
 
 on:
-  push:
-    pull_request:
+  pull_request:
       branches: [ main ]
 
 jobs:
@@ -11,9 +10,9 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
     - name: Set up Python 3.10
-      uses: actions/setup-python@v2
+      uses: actions/setup-python@v4
       with:
         python-version: "3.10"
 
@@ -30,4 +29,4 @@ jobs:
       uses: actions/upload-artifact@v2
       with:
         name: html
-        path: docs/_build/html
+        path: docs/_build/html

Makefile

@@ -1,25 +1,33 @@
 
-build: test
-	#python setup.py sdist bdist_wheel
+LIB = mikeio
+
+check: lint typecheck test
+
+build: typecheck test
 	python -m build
 
+lint:
+	ruff .
+
 test:
 	pytest --disable-warnings
 
+typecheck:
+	mypy $(LIB)/ --config-file pyproject.toml
+
+coverage: 
+	pytest --cov-report html --cov=$(LIB) tests/
+
 doctest:
 	pytest mikeio/dfs/*.py mikeio/dfsu/*.py mikeio/eum/*.py mikeio/pfs/*.py mikeio/spatial/_grid_geometry.py --doctest-modules
 	rm -f *.dfs* # remove temporary files, created from doctests
 
 perftest:
 	pytest tests/performance/ --durations=0
 
-typecheck:
-	mypy mikeio/
-
-coverage: 
-	pytest --cov-report html --cov=mikeio tests/
-
 docs: FORCE
 	cd docs; make html ;cd -
 
 FORCE:
+
+

docs/dfs2.md

@@ -16,6 +16,35 @@ items:
   0:  Elevation <Total Water Depth> (meter)
 ```
 
+## Subset in space
+
+The most convenient way to subset in space is to use the `sel` method, which returns a new (smaller) dataset, which can be further processed or written to disk using the `to_dfs` method. 
+
+```python
+>>> ds.geometry
+<mikeio.Grid2D>
+x: [12.2, 12.21, ..., 13.1] (nx=216, dx=0.004167)
+y: [55.2, 55.21, ..., 56.3] (ny=264, dy=0.004167)
+projection: LONG/LAT
+>>> ds_aoi = ds.sel(x=slice(12.5, 13.0), y=slice(55.5, 56.0))
+>>> ds_aoi.geometry
+<mikeio.Grid2D>
+x: [12.5, 12.5, ..., 12.99] (nx=120, dx=0.004167)
+y: [55.5, 55.5, ..., 55.99] (ny=120, dy=0.004167)
+projection: LONG/LAT
+```
+
+In order to specify an open-ended subset (i.e. where the end of the subset is the end of the domain), use `None` as the end of the slice. 
+
+```python
+>>> ds.sel(x=slice(None,13.0))
+<mikeio.Dataset>
+dims: (time:1, y:264, x:191)
+time: 2020-05-15 11:04:52 (time-invariant)
+geometry: Grid2D (ny=264, nx=191)
+items:
+  0:  Elevation <Total Water Depth> (meter)
+```
 
 ## Grid2D
 

mikeio/_time.py

@@ -0,0 +1,69 @@
+from __future__ import annotations
+from datetime import datetime
+from dataclasses import dataclass
+from typing import List, Iterable, Optional
+
+import pandas as pd
+
+
+@dataclass
+class DateTimeSelector:
+    """Helper class for selecting time steps from a pandas DatetimeIndex"""
+
+    index: pd.DatetimeIndex
+
+    def isel(
+        self,
+        x: Optional[
+            int | Iterable[int] | str | datetime | pd.DatetimeIndex | slice
+        ] = None,
+    ) -> List[int]:
+        """Select time steps from a pandas DatetimeIndex
+
+        Parameters
+        ----------
+        x : int, Iterable[int], str, datetime, pd.DatetimeIndex, slice
+            Time steps to select, negative indices are supported
+
+        Returns
+        -------
+        List[int]
+            List of indices in the range (0, len(index)
+        Examples
+        --------
+        >>> idx = pd.date_range("2000-01-01", periods=4, freq="D")
+        >>> dts = DateTimeSelector(idx)
+        >>> dts.isel(None)
+        [0, 1, 2, 3]
+        >>> dts.isel(0)
+        [0]
+        >>> dts.isel(-1)
+        [3]
+        """
+
+        indices = list(range(len(self.index)))
+
+        if x is None:
+            return indices
+
+        if isinstance(x, int):
+            return [indices[x]]
+
+        if isinstance(x, (datetime, str)):
+            loc = self.index.get_loc(x)
+            if isinstance(loc, int):
+                return [loc]
+            elif isinstance(loc, slice):
+                return list(range(loc.start, loc.stop))
+
+        if isinstance(x, slice):
+            if isinstance(x.start, int) or isinstance(x.stop, int):
+                return indices[x]
+            else:
+                s = self.index.slice_indexer(x.start, x.stop)
+                return list(range(s.start, s.stop))
+
+        if isinstance(x, Iterable):
+            return [self.isel(t)[0] for t in x]
+
+        return indices

mikeio/dataset/_data_utils.py

@@ -1,11 +1,12 @@
 from __future__ import annotations
 import re
-from datetime import datetime
-from typing import Iterable, Sequence, Sized, Tuple
+from typing import Iterable, Sequence, Sized, Tuple, Union, List
 
 import numpy as np
 import pandas as pd
 
+from .._time import DateTimeSelector
+
 
 def _to_safe_name(name: str) -> str:
     tmp = re.sub("[^0-9a-zA-Z]", "_", name)
@@ -18,45 +19,19 @@ def _n_selected_timesteps(x: Sized, k: slice | Sized) -> int:
     return len(k)
 
 
-def _get_time_idx_list(time: pd.DatetimeIndex, steps):
+def _get_time_idx_list(time: pd.DatetimeIndex, steps) -> Union [List[int], slice]:
     """Find list of idx in DatetimeIndex"""
 
-    if isinstance(steps, str):
-        parts = steps.split(",")
-        if len(parts) == 1:
-            parts.append(parts[0])  # end=start
-
-        if parts[0] == "":
-            steps = slice(parts[1])  # stop only
-        elif parts[1] == "":
-            steps = slice(parts[0], None)  # start only
-        else:
-            steps = slice(parts[0], parts[1])
+    # indexing with a slice needs to be handled differently, since slicing returns a view
 
-    if isinstance(steps, (list, tuple)) and isinstance(
-        steps[0], (str, datetime, np.datetime64, pd.Timestamp)
-    ):
-        steps = pd.DatetimeIndex(steps)
-    if isinstance(steps, pd.DatetimeIndex):
-        return time.get_indexer(steps)
-    if isinstance(steps, (str, datetime, np.datetime64, pd.Timestamp)):
-        steps = slice(steps, steps)
     if isinstance(steps, slice):
-        try:
-            s = time.slice_indexer(
-                steps.start,
-                steps.stop,
-            )
-            steps = list(range(s.start, s.stop))
-        except TypeError:
-            pass  # TODO this seems fishy!
-            # steps = list(range(*steps.indices(len(time))))
-    elif isinstance(steps, int):
-        steps = [steps]
-    # TODO what is the return type of this function
-    return steps
+        if isinstance(steps.start, int) and isinstance(steps.stop, int):
+            return steps
 
+    dts = DateTimeSelector(time)
+    return dts.isel(steps)
 
+# TODO this only used by DataArray, so consider to move it there
 class DataUtilsMixin:
     """DataArray Utils"""
 
@@ -107,7 +82,7 @@ def _set_by_boolean_mask(data: np.ndarray, mask: np.ndarray, value) -> None:
     def _parse_time(time) -> pd.DatetimeIndex:
         """Allow anything that we can create a DatetimeIndex from"""
         if time is None:
-            time = [pd.Timestamp(2018, 1, 1)]
+            time = [pd.Timestamp(2018, 1, 1)] # TODO is this the correct epoch?
         if isinstance(time, str) or (not isinstance(time, Iterable)):
             time = [time]
 

mikeio/dataset/_dataarray.py

@@ -3,7 +3,7 @@
 from copy import deepcopy
 from datetime import datetime
 from functools import cached_property
-from typing import Iterable, Optional, Sequence, Tuple
+from typing import Iterable, Optional, Sequence, Tuple, Mapping
 
 
 import numpy as np
@@ -840,10 +840,14 @@ def sel(
         time: 1997-09-15 21:00:00 - 1997-09-16 03:00:00 (3 records)
         geometry: Dfsu2D (3700 elements, 2090 nodes)
         """
+        if any([isinstance(v, slice) for v in kwargs.values()]):
+            return self._sel_with_slice(kwargs)
+
         da = self
 
         # select in space
         if len(kwargs) > 0:
+
             idx = self.geometry.find_index(**kwargs)
             if isinstance(idx, tuple):
                 # TODO: support for dfs3
@@ -867,6 +871,27 @@ def sel(
             da = da[time]  # __getitem__ is 🚀
 
         return da
+
+    def _sel_with_slice(self, kwargs: Mapping[str,slice]) -> "DataArray":
+        for k, v in kwargs.items():
+            if isinstance(v, slice):
+                idx_start = self.geometry.find_index(**{k:v.start})
+                idx_stop = self.geometry.find_index(**{k:v.stop})
+                pos = 0
+                if isinstance(idx_start, tuple):
+                    if k == "x":
+                        pos = 0
+                    if k == "y":
+                        pos = 1
+
+                start = idx_start[pos][0] if idx_start is not None else None
+                stop = idx_stop[pos][0] if idx_stop is not None else None
+
+                idx = slice(start, stop)
+
+                self = self.isel(idx, axis=k)
+
+        return self
 
     def interp(
         # TODO find out optimal syntax to allow interpolation to single point, new time, grid, mesh...

mikeio/dataset/_dataset.py

@@ -561,7 +561,7 @@ def __set_or_insert_item(self, key, value, insert=False):
             self._data_vars[key] = value
             self._set_name_attr(key, value)
 
-    def insert(self, key: int, value: DataArray):
+    def insert(self, key, value: DataArray):
         """Insert DataArray in a specific position
 
         Parameters