1111import numpy as np
1212
1313from .backends .api import open_dataset as _open_dataset
14- from .backends .rasterio_ import open_rasterio
14+ from .backends .rasterio_ import open_rasterio as _open_rasterio
1515from .core .dataarray import DataArray
1616from .core .dataset import Dataset
1717
18-
19- def _open_rasterio (path , engine = None , ** kwargs ):
20- data = open_rasterio (path , ** kwargs )
21- name = data .name if data .name is not None else "data"
22- return data .to_dataset (name = name )
23-
24-
2518_default_cache_dir_name = "xarray_tutorial_data"
2619base_url = "https://github.com/pydata/xarray-data"
2720version = "master"
2821
2922
30- external_urls = {
31- "RGB.byte" : (
32- "rasterio" ,
33- "https://github.com/mapbox/rasterio/raw/master/tests/data/RGB.byte.tif" ,
34- ),
35- }
36- overrides = {
37- "rasterio" : _open_rasterio ,
23+ def _construct_cache_dir (path ):
24+ import pooch
25+
26+ if isinstance (path , pathlib .Path ):
27+ path = os .fspath (path )
28+ elif path is None :
29+ path = pooch .os_cache (_default_cache_dir_name )
30+
31+ return path
32+
33+
34+ external_urls = {} # type: dict
35+ external_rasterio_urls = {
36+ "RGB.byte" : "https://github.com/mapbox/rasterio/raw/1.2.1/tests/data/RGB.byte.tif" ,
37+ "shade" : "https://github.com/mapbox/rasterio/raw/1.2.1/tests/data/shade.tif" ,
3838}
3939
4040
4141# idea borrowed from Seaborn
4242def open_dataset (
4343 name ,
44- engine = None ,
4544 cache = True ,
4645 cache_dir = None ,
4746 ** kws ,
@@ -51,31 +50,27 @@ def open_dataset(
5150
5251 If a local copy is found then always use that to avoid network traffic.
5352
53+ Available datasets:
54+
55+ * ``"air_temperature"``: NCEP reanalysis subset
56+ * ``"rasm"``: Output of the Regional Arctic System Model (RASM)
57+ * ``"ROMS_example"``: Regional Ocean Model System (ROMS) output
58+ * ``"tiny"``: small synthetic dataset with a 1D data variable
59+ * ``"era5-2mt-2019-03-uk.grib"``: ERA5 temperature data over the UK
60+ * ``"eraint_uvz"``: data from ERA-Interim reanalysis, monthly averages of upper level data
61+
5462 Parameters
5563 ----------
5664 name : str
5765 Name of the file containing the dataset.
5866 e.g. 'air_temperature'
59- engine : str, optional
60- The engine to use.
6167 cache_dir : path-like, optional
6268 The directory in which to search for and write cached data.
6369 cache : bool, optional
6470 If True, then cache data locally for use on subsequent calls
6571 **kws : dict, optional
6672 Passed to xarray.open_dataset
6773
68- Notes
69- -----
70- Available datasets:
71-
72- * ``"air_temperature"``
73- * ``"rasm"``
74- * ``"ROMS_example"``
75- * ``"tiny"``
76- * ``"era5-2mt-2019-03-uk.grib"``
77- * ``"RGB.byte"``: example rasterio file from https://github.com/mapbox/rasterio
78-
7974 See Also
8075 --------
8176 xarray.open_dataset
@@ -85,15 +80,12 @@ def open_dataset(
8580 except ImportError :
8681 raise ImportError ("using the tutorial data requires pooch" )
8782
88- if isinstance (cache_dir , pathlib .Path ):
89- cache_dir = os .fspath (cache_dir )
90- elif cache_dir is None :
91- cache_dir = pooch .os_cache (_default_cache_dir_name )
83+ logger = pooch .get_logger ()
84+ logger .setLevel ("WARNING" )
9285
86+ cache_dir = _construct_cache_dir (cache_dir )
9387 if name in external_urls :
94- engine_ , url = external_urls [name ]
95- if engine is None :
96- engine = engine_
88+ url = external_urls [name ]
9789 else :
9890 # process the name
9991 default_extension = ".nc"
@@ -103,17 +95,78 @@ def open_dataset(
10395
10496 url = f"{ base_url } { version } { path .name }
10597
106- _open = overrides .get (engine , _open_dataset )
10798 # retrieve the file
10899 filepath = pooch .retrieve (url = url , known_hash = None , path = cache_dir )
109- ds = _open (filepath , engine = engine , ** kws )
100+ ds = _open_dataset (filepath , ** kws )
110101 if not cache :
111102 ds = ds .load ()
112103 pathlib .Path (filepath ).unlink ()
113104
114105 return ds
115106
116107
108+ def open_rasterio (
109+ name ,
110+ engine = None ,
111+ cache = True ,
112+ cache_dir = None ,
113+ ** kws ,
114+ ):
115+ """
116+ Open a rasterio dataset from the online repository (requires internet).
117+
118+ If a local copy is found then always use that to avoid network traffic.
119+
120+ Available datasets:
121+
122+ * ``"RGB.byte"``: TIFF file derived from USGS Landsat 7 ETM imagery.
123+ * ``"shade"``: TIFF file derived from from USGS SRTM 90 data
124+
125+ ``RGB.byte`` and ``shade`` are downloaded from the ``rasterio`` repository [1]_.
126+
127+ Parameters
128+ ----------
129+ name : str
130+ Name of the file containing the dataset.
131+ e.g. 'RGB.byte'
132+ cache_dir : path-like, optional
133+ The directory in which to search for and write cached data.
134+ cache : bool, optional
135+ If True, then cache data locally for use on subsequent calls
136+ **kws : dict, optional
137+ Passed to xarray.open_rasterio
138+
139+ See Also
140+ --------
141+ xarray.open_rasterio
142+
143+ References
144+ ----------
145+ .. [1] https://github.com/mapbox/rasterio
146+ """
147+ try :
148+ import pooch
149+ except ImportError :
150+ raise ImportError ("using the tutorial data requires pooch" )
151+
152+ logger = pooch .get_logger ()
153+ logger .setLevel ("WARNING" )
154+
155+ cache_dir = _construct_cache_dir (cache_dir )
156+ url = external_rasterio_urls .get (name )
157+ if url is None :
158+ raise ValueError (f"unknown rasterio dataset: { name } )
159+
160+ # retrieve the file
161+ filepath = pooch .retrieve (url = url , known_hash = None , path = cache_dir )
162+ arr = _open_rasterio (filepath , ** kws )
163+ if not cache :
164+ arr = arr .load ()
165+ pathlib .Path (filepath ).unlink ()
166+
167+ return arr
168+
169+
117170def load_dataset (* args , ** kwargs ):
118171 """
119172 Open, load into memory, and close a dataset from the online repository
0 commit comments