ncplot 0.3.12__tar.gz → 0.4.0__tar.gz

{ncplot-0.3.12 → ncplot-0.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ncplot
-Version: 0.3.12
+Version: 0.4.0
 Summary: Interactive viewing of NetCDF data
 Home-page: https://github.com/pmlmodelling/ncplot
 Author: Robert Wilson

{ncplot-0.3.12 → ncplot-0.4.0}/ncplot/plot.py

@@ -2,6 +2,7 @@ import sys
 import warnings
 from threading import Thread
 
+import glob
 import time
 import holoviews as hv
 import panel as pn
@@ -165,6 +166,172 @@ def in_notebook(out=None):
     return "ipykernel" in sys.modules
 
 
+def _looks_like_time_coord(coord, coord_name):
+    """
+    Check whether a coordinate is likely to represent time.
+    """
+
+    if coord_name is None:
+        return False
+
+    if coord_name == "time" or coord_name.startswith("time"):
+        return True
+
+    if "time" in coord_name:
+        return True
+
+    if coord.attrs.get("standard_name") == "time":
+        return True
+
+    if coord.attrs.get("axis") == "T":
+        return True
+
+    if np.issubdtype(coord.dtype, np.datetime64):
+        return True
+
+    values = coord.values
+    if len(values) > 0:
+        first_value = values.ravel()[0]
+        if hasattr(first_value, "year") and hasattr(first_value, "month"):
+            return True
+
+    return False
+
+
+def _find_time_coord(ds):
+    """
+    Identify the most likely time coordinate name for a dataset.
+    """
+
+    df_dims = get_dims(ds)
+    time_name = df_dims.time[0]
+
+    if time_name is not None:
+        return time_name
+
+    time_candidates = []
+    for coord_name in list(ds.coords):
+        coord = ds[coord_name]
+        if _looks_like_time_coord(coord, coord_name):
+            time_candidates.append(coord_name)
+
+    if len(time_candidates) == 1:
+        return time_candidates[0]
+
+    return None
+
+
+def _combine_datasets(datasets):
+    """
+    Combine multiple datasets by time or by variables.
+    """
+
+    if len(datasets) == 1:
+        return datasets[0]
+
+    normalized = []
+    for ds in datasets:
+        time_name = _find_time_coord(ds)
+        if time_name is not None and time_name != "time":
+            ds = ds.rename({time_name: "time"})
+        normalized.append(ds)
+
+    try:
+        return xr.combine_by_coords(normalized)
+    except Exception:
+        try:
+            return xr.merge(normalized, compat="override")
+        except Exception:
+            return xr.concat(normalized, dim="time")
+
+
+def _normalize_time_coord(ds):
+    """
+    Normalize detected time coordinate names to ``time``.
+    """
+
+    time_name = _find_time_coord(ds)
+    if time_name is not None and time_name != "time":
+        return ds.rename({time_name: "time"})
+
+    return ds
+
+
+def _open_multiple_files(files):
+    """
+    Open multiple files into one dataset.
+
+    This uses ``open_mfdataset`` first, then falls back to manual per-file
+    opening and combining if xarray cannot infer a valid combine strategy.
+    """
+
+    try:
+        return xr.open_mfdataset(
+            files,
+            combine="by_coords",
+            preprocess=_normalize_time_coord,
+        )
+    except Exception:
+        datasets = [xr.open_dataset(ff) for ff in files]
+        return _combine_datasets(datasets)
+
+
+def _load_input_as_dataset(x):
+    """
+    Load a path, path glob, or sequence of paths into a single dataset.
+    """
+
+    if isinstance(x, (list, tuple, set)):
+        files = [os.fspath(ff) for ff in x]
+        return _open_multiple_files(files)
+
+    if isinstance(x, str) and glob.has_magic(x):
+        files = sorted(glob.glob(x))
+        if len(files) == 0:
+            raise FileNotFoundError(f"No files matched pattern {x}")
+        return _open_multiple_files(files)
+
+    return None
+
+
+def _get_color_limits(ds, vars, kwargs):
+    """
+    Work out the value range to use for a plot.
+
+    For time-varying plots, this is evaluated across the full dataset so the
+    color scale stays consistent between time steps.
+    """
+
+    if "clim" in kwargs:
+        clim_min, clim_max = kwargs["clim"]
+        if clim_min < 0 < clim_max:
+            v_max = float(max(-clim_min, clim_max))
+            return (-v_max, v_max)
+
+        return (float(clim_min), float(clim_max))
+
+    data = ds[vars]
+    data_min = float(data.min(skipna=True).values)
+    data_max = float(data.max(skipna=True).values)
+
+    if data_min < 0 < data_max:
+        v_max = float(max(-data_min, data_max))
+        return (-v_max, v_max)
+
+    return (data_min, data_max)
+
+
+def _apply_color_limits(plot, ds, vars, autoscale, kwargs):
+    """
+    Apply a stable color range to a plot when autoscaling is enabled.
+    """
+
+    if autoscale:
+        return plot.redim.range(**{vars: _get_color_limits(ds, vars, kwargs)})
+
+    return plot
+
+
 def view(x, vars=None, autoscale=True, out=None, **kwargs):
     """
     Plot the contents of a NetCDF out
@@ -209,6 +376,10 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
     if type(x) is xr.core.dataarray.DataArray:
         x = x.to_dataset()
 
+    loaded_ds = _load_input_as_dataset(x)
+    if loaded_ds is not None:
+        x = loaded_ds
+
     if type(x) is xr.core.dataset.Dataset:
         xr_file = True
     else:
@@ -495,7 +666,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
     coord_list = [x for x in coord_list if x in list(ds.dims)]
 
     coord_df = pd.DataFrame(
-        {"coord": coord_list, "length": [len(ds.coords[x].values) for x in coord_list]}
+        {"coord": coord_list, "coord_length": [len(ds.coords[x].values) for x in coord_list]}
     )
 
     # It's possible there are still 2 time variables in the dimensions which could cause problems...
@@ -515,7 +686,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
             if (len(ds[lon_name].values) > 1) and (len(ds[lat_name].values) > 1):
                 spatial_map = True
 
-    if len([x for x in coord_df.length if x > 1]) == 1 and spatial_map is False:
+    if len([x for x in coord_df.coord_length if x > 1]) == 1 and spatial_map is False:
 
         df = ds.to_dataframe()
         if nc_vars is not None:
@@ -556,12 +727,11 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
         return None
 
     # heat map where 2 coords have more than 1 value, not a spatial map
-    if len([x for x in coord_df.length if x > 1]) == 2 and spatial_map is False:
+    if len([x for x in coord_df.coord_length if x > 1]) == 2 and spatial_map is False:
 
         df = ds.to_dataframe().reset_index()
-        x_var = coord_df.query("length > 1").reset_index().coord[0]
-        y_var = coord_df.query("length > 1").reset_index().coord[1]
-
+        x_var = coord_df.query("coord_length > 1").reset_index().coord[0]
+        y_var = coord_df.query("coord_length > 1").reset_index().coord[1]
         selection = [x for x in df.columns if x in vars or x == x_var or x == y_var]
 
         df = df.loc[:, selection].melt([x_var, y_var]).drop_duplicates()
@@ -696,18 +866,18 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
             return None
 
     # heat map where 3 coords have more than 1 value, and one of them is time. Not a spatial map though
-    if len([x for x in coord_df.length if x > 1]) == 3:
+    if len([x for x in coord_df.coord_length if x > 1]) == 3:
 
         non_map = True
 
         if lon_name is not None and lat_name is not None:
             if (lon_name is not None) and (lon_name in list(ds.coords)):
-                lons = int(coord_df.query("coord == @lon_name").length)
+                lons = int(coord_df.query("coord == @lon_name").coord_length.values[0])
             else:
                 lons = 0
 
             if (lat_name) is not None and (lat_name in list(ds.coords)):
-                lats = int(coord_df.query("coord == @lat_name").length)
+                lats = int(coord_df.query("coord == @lat_name").coord_length.values[0])
             else:
                 lats = 0
 
@@ -725,7 +895,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
 
         if time_name in coord_list and time_in and non_map:
 
-            if coord_df.query("coord == @time_name").length.values > 1:
+            if coord_df.query("coord == @time_name").coord_length.values[0] > 1:
 
                 df = ds.to_dataframe().reset_index()
                 for x in list(ds.coords):
@@ -964,7 +1134,8 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         #responsive=(in_notebook() is False),
                         responsive = False,
                         **kwargs,
-                    ).redim.range(**{vars: (-v_max, v_max)})
+                    )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
                 else:
                     intplot = ds.hvplot.quadmesh(
                         lon_name,
@@ -979,6 +1150,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         responsive = False,
                         **kwargs,
                     )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
             else:
                 if coastline:
                     coastline = get_coastline(ds, lon_name, lat_name)
@@ -1003,7 +1175,8 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         #responsive=(in_notebook() is False),
                         responsive = False,
                         **kwargs,
-                    ).redim.range(**{vars: (-v_max, v_max)})
+                    )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
                 else:
                     intplot = ds.hvplot.image(
                         lon_name,
@@ -1055,7 +1228,8 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         #responsive=(in_notebook() is False),
                         responsive = False,
                         **kwargs,
-                    ).redim.range(**{vars: (self_min.values, v_max)})
+                    )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
                 else:
                     intplot = ds.hvplot.quadmesh(
                         lon_name,
@@ -1070,6 +1244,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         responsive = False,
                         **kwargs,
                     )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
 
             else:
 
@@ -1096,7 +1271,8 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         #responsive=(in_notebook() is False),
                         responsive = False,
                         **kwargs,
-                    ).redim.range(**{vars: (self_min.values, v_max)})
+                    )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
                 else:
                     intplot = ds.hvplot.image(
                         lon_name,
@@ -1111,6 +1287,7 @@ def view(x, vars=None, autoscale=True, out=None, **kwargs):
                         responsive = False,
                         **kwargs,
                     )
+                    intplot = _apply_color_limits(intplot, ds, vars, autoscale, kwargs)
 
             if in_notebook(out):
                 if out is None:

{ncplot-0.3.12 → ncplot-0.4.0}/setup.py

@@ -37,7 +37,7 @@ extras_require: dict() = {
 extras_require["complete"] = ["geoviews"]
 
 setup(name='ncplot',
-      version='0.3.12',
+      version='0.4.0',
       description=DESCRIPTION,
       long_description=long_description,
       long_description_content_type='text/markdown',