baseclasses.py 63.1 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
# -*- coding: utf-8 -*-
__author__='Daniel Scheffler'

import os
import warnings
from collections import OrderedDict

import numpy as np
from matplotlib import pyplot as plt
from osgeo import gdal_array
# custom
from shapely.geometry import Polygon
from shapely.wkt import loads as shply_loads
from six import PY3

# mpl_toolkits.basemap -> imported when GeoArray.show_map() is used
# dill -> imported when dumping GeoArray

try:
    from osgeo import gdal
    from osgeo import gdalnumeric
except ImportError:
    import gdal
    import gdalnumeric
from geopandas import GeoDataFrame, GeoSeries
from pandas import DataFrame
27
from py_tools_ds.ptds.convenience.object_oriented import alias_property
28
29
from py_tools_ds.ptds.geo.coord_calc import get_corner_coordinates
from py_tools_ds.ptds.geo.coord_grid import snap_bounds_to_pixGrid
30
from py_tools_ds.ptds.geo.coord_trafo import mapXY2imXY, imXY2mapXY, transform_any_prj, reproject_shapelyGeometry
31
32
33
34
35
36
37
38
39
40
41
from py_tools_ds.ptds.geo.projection import prj_equal, WKT2EPSG, EPSG2WKT
from py_tools_ds.ptds.geo.raster.conversion import raster2polygon
from py_tools_ds.ptds.geo.vector.topology \
    import get_footprint_polygon, polyVertices_outside_poly, fill_holes_within_poly
from py_tools_ds.ptds.geo.vector.geometry import boxObj
from py_tools_ds.ptds.io.raster.gdal import get_GDAL_ds_inmem
from py_tools_ds.ptds.numeric.array import find_noDataVal
from py_tools_ds.ptds.compatibility.python.exceptions \
    import TimeoutError as TimeoutError_comp, FileNotFoundError as FileNotFoundError_comp
from py_tools_ds.ptds.compatibility.gdal import get_gdal_func

42
43
#  internal imports
from .subsetting import get_array_at_mapPos
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235


class GeoArray(object):
    def __init__(self, path_or_array, geotransform=None, projection=None, bandnames=None, nodata=None, progress=True,
                 q=False):
        # type: (any, tuple, str, list, float, bool, bool) -> None
        """This class creates a fast Python interface for geodata - either on disk or in memory. It can be instanced with
        a file path or with a numpy array and the corresponding geoinformation. Instances can always be indexed like
        normal numpy arrays, no matter if GeoArray has been instanced from file or from an in-memory array. GeoArray
        provides a wide range of geo-related attributes belonging to the dataset as well as some functions for quickly
        visualizing the data as a map, a simple image or an interactive image.

        :param path_or_array:   a numpy.ndarray or a valid file path
        :param geotransform:    GDAL geotransform of the given array or file on disk
        :param projection:      projection of the given array or file on disk as WKT string
                                (only needed if GeoArray is instanced with an array)
        :param bandnames:       names of the bands within the input array, e.g. ['mask_1bit', 'mask_clouds'],
                                (default: ['B1', 'B2', 'B3', ...])
        :param nodata:          nodata value
        :param progress:        show progress bars (default: True)
        :param q:               quiet mode (default: False)
        """


        # TODO implement compatibility to GDAL VRTs
        if not (isinstance(path_or_array, (str, np.ndarray, GeoArray)) or
                issubclass(getattr(path_or_array,'__class__'), GeoArray)):
            raise ValueError("%s parameter 'arg' takes only string, np.ndarray or GeoArray(and subclass) instances. "
                             "Got %s." %(self.__class__.__name__,type(path_or_array)))

        if path_or_array is None:
            raise ValueError("The %s parameter 'path_or_array' must not be None!" %self.__class__.__name__)

        if isinstance(path_or_array, str):
            assert ' ' not in path_or_array, "The given path contains whitespaces. This is not supported by GDAL."

            if not os.path.exists(path_or_array):
                raise FileNotFoundError(path_or_array) if PY3 else FileNotFoundError_comp(path_or_array)


        if isinstance(path_or_array, GeoArray) or issubclass(getattr(path_or_array,'__class__'), GeoArray):
            self.__dict__= path_or_array.__dict__.copy()
            self._initParams = dict([x for x in locals().items() if x[0] != "self"])
            self.geotransform = geotransform if geotransform       else self.geotransform
            self.projection   = projection   if projection         else self.projection
            self.bandnames    = bandnames    if bandnames          else list(self.bandnames.values())
            self._nodata      = nodata       if nodata is not None else self._nodata
            self.progress     = False        if progress is False  else self.progress
            self.q            = q            if q is not None      else self.q

        else:
            self._initParams     = dict([x for x in locals().items() if x[0] != "self"])
            self.arg             = path_or_array
            self._arr            = path_or_array if isinstance(path_or_array, np.ndarray)            else None
            self.filePath        = path_or_array if isinstance(path_or_array, str) and path_or_array else None
            self.basename        = os.path.splitext(os.path.basename(self.filePath))[0] if not self.is_inmem else 'IN_MEM'
            self.progress        = progress
            self.q               = q
            self._arr_cache      = None   # dict containing key 'pos' and 'arr_cached'
            self._geotransform   = None
            self._projection     = None
            self._shape          = None
            self._dtype          = None
            self._nodata         = nodata
            self._mask_nodata    = None
            self._mask_baddata   = None
            self._footprint_poly = None
            self._gdalDataset_meta_already_set = False
            self._metadata       = None
            self._bandnames      = None

            if bandnames:
                self.bandnames    = bandnames    # use property in order to validate given value
            if geotransform:
                self.geotransform = geotransform # use property in order to validate given value
            if projection:
                self.projection   = projection   # use property in order to validate given value

            if self.filePath:
                self.set_gdalDataset_meta()

    @property
    def arr(self):
        return self._arr


    @arr.setter
    def arr(self, ndarray):
        assert isinstance(ndarray, np.ndarray), "'arr' can only be set to a numpy array! Got %s." %type(ndarray)
        #assert ndarray.shape == self.shape, "'arr' can only be set to a numpy array with shape %s. Received %s. " \
        #                                    "If you need to change the dimensions, create a new instance of %s." \
        #                                    %(self.shape, ndarray.shape, self.__class__.__name__)
        #  THIS would avoid warping like this: geoArr.arr, geoArr.gt, geoArr.prj = warp(...)

        if ndarray.shape != self.shape:
            self.flush_cache() # the cached array is not useful anymore

        self._arr = ndarray


    @property
    def bandnames(self):
        if self._bandnames and len(self._bandnames)==self.bands:
            return self._bandnames
        else:
            self._bandnames = OrderedDict(('B%s' % band, i) for i, band in enumerate(range(1, self.bands + 1)))
            return self._bandnames


    @bandnames.setter
    def bandnames(self, list_bandnames):
        # type: (list) -> None

        if list_bandnames:
            assert isinstance(list_bandnames, list), "A list must be given when setting the 'bandnames' attribute. " \
                                                     "Received %s." %type(list_bandnames)
            assert len(list_bandnames) == self.bands, \
                'Number of given bandnames does not match number of bands in array.'
            assert len(list(set([type(b) for b in list_bandnames]))) == 1 and type(list_bandnames[0] == 'str'), \
                "'bandnames must be a set of strings. Got other datetypes in there.'"
            bN_dict = OrderedDict((band, i) for i, band in enumerate(list_bandnames))
            assert len(bN_dict) == self.bands, 'Bands must not have the same name. Received band list: %s' %list_bandnames

            self._bandnames = bN_dict


    @property
    def is_inmem(self):
        """Check if associated image array is completely loaded into memory."""

        return isinstance(self.arr, np.ndarray)


    @property
    def shape(self):
        """Get the array shape of the associated image array."""

        if self.is_inmem:
            return self.arr.shape
        else:
            if self._shape:
                return self._shape
            else:
                self.set_gdalDataset_meta()
                return self._shape


    @property
    def ndim(self):
        """Get the number dimensions of the associated image array."""
        return len(self.shape)


    @property
    def rows(self):
        """Get the number of rows of the associated image array."""

        return self.shape[0]


    @property
    def columns(self):
        """Get the number of columns of the associated image array."""

        return self.shape[1]


    cols = alias_property('columns')


    @property
    def bands(self):
        """Get the number of bands of the associated image array."""

        return self.shape[2] if len(self.shape)>2 else 1


    @property
    def dtype(self):
        """Get the numpy data type of the associated image array."""

        if self._dtype:
            return self._dtype
        elif self.is_inmem:
            return self.arr.dtype
        else:
            self.set_gdalDataset_meta()
            return self._dtype


    @property
    def geotransform(self):
236
        """Get the GDAL GeoTransform of the associated image, e.g., (283500.0, 5.0, 0.0, 4464500.0, 0.0, -5.0)"""
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663

        if self._geotransform:
            return self._geotransform
        elif not self.is_inmem:
            self.set_gdalDataset_meta()
            return self._geotransform
        else:
            return [0,1,0,0,0,-1]


    @geotransform.setter
    def geotransform(self, gt):
        assert isinstance(gt,(list,tuple)) and len(gt)==6, 'geotransform must be a list with 6 numbers. Got %s.' %str(gt)
        for i in gt: assert isinstance(i,(int,float)),     "geotransform must contain only numbers. Got '%s'." %i

        self._geotransform = gt


    gt = alias_property('geotransform')


    @property
    def xgsd(self):
        """Get the X resolution in units of the given or detected projection."""

        return self.geotransform[1]


    @property
    def ygsd(self):
        """Get the Y resolution in units of the given or detected projection."""

        return abs(self.geotransform[5])


    @property
    def xygrid_specs(self):
        """
        Get the specifications for the X/Y coordinate grid, e.g. [[15,30], [0,30]] for a coordinate with its origin
        at X/Y[15,0] and a GSD of X/Y[15,30].
        """

        get_grid = lambda gt, xgsd, ygsd: [[gt[0], gt[0] + xgsd], [gt[3], gt[3] - ygsd]]
        return get_grid(self.geotransform, self.xgsd, self.ygsd)


    @property
    def projection(self):
        """
        Get the projection of the associated image. Setting the projection is only allowed if GeoArray has been
        instanced from memory or the associated file on disk has no projection.
        """

        if self._projection:
            return self._projection
        elif not self.is_inmem:
            self.set_gdalDataset_meta()
            return self._projection
        else:
            return ''


    @projection.setter
    def projection(self, prj):
        if self.filePath:
            assert self.projection is None or prj_equal(self.projection, prj),\
                "Cannot set %s.projection to the given value because it does not match the projection from the file " \
                "on disk." %self.__class__.__name__
        else:
            self._projection = prj


    prj = alias_property('projection')


    @property
    def epsg(self):
        """Get the EPSG code of the projection of the GeoArray."""

        return WKT2EPSG(self.projection)


    @epsg.setter
    def epsg(self, epsg_code):
        self.projection = EPSG2WKT(epsg_code)


    @property
    def box(self):
        mapPoly = get_footprint_polygon(get_corner_coordinates(gt=self.geotransform, cols=self.columns, rows=self.rows))
        return boxObj(gt=self.geotransform, prj=self.projection, mapPoly=mapPoly)


    @property
    def nodata(self):
        """
        Get the nodata value of the GeoArray. If GeoArray has been instanced with a file path the file is checked
        for an existing nodata value. Otherwise (if no value is exlicitly given during object instanciation) the nodata
        value is tried to be automatically detected.
        """

        if self._nodata is not None:
            return self._nodata
        else:
            # try to get nodata value from file
            if not self.is_inmem:
                self.set_gdalDataset_meta()
            if self._nodata is None:
                self._nodata = find_noDataVal(self)
                if self._nodata == 'ambiguous':
                    warnings.warn('Nodata value could not be clearly identified. It has been set to None.')
                    self._nodata = None
                else:
                    if self._nodata is not None and not self.q:
                        print("Automatically detected nodata value for %s '%s': %s"
                              %(self.__class__.__name__, self.basename, self._nodata))
            return self._nodata


    @nodata.setter
    def nodata(self, value):
        self._nodata = value


    @property
    def mask_nodata(self):
        """
        Get the nodata mask of the associated image array. It is calculated using all image bands.
        """

        if self._mask_nodata is not None:
            return self._mask_nodata
        else:
            self.calc_mask_nodata() # sets self._mask_nodata
            return self._mask_nodata


    @mask_nodata.setter
    def mask_nodata(self, mask):
        """Set bad data mask.

        :param mask:    Can be a file path, a numpy array or an instance o GeoArray.
        """

        if mask is not None:
            from .masks import NoDataMask
            geoArr_mask     = NoDataMask(mask, progress=self.progress, q=self.q)
            geoArr_mask.gt  = geoArr_mask.gt  if geoArr_mask.gt  not in [None, [0, 1, 0, 0, 0, -1]] else self.gt
            geoArr_mask.prj = geoArr_mask.prj if geoArr_mask.prj else self.prj
            imName          = "the %s '%s'" %(self.__class__.__name__, self.basename)

            assert geoArr_mask.bands == 1, \
                'Expected one single band as nodata mask for %s. Got %s bands.' % (self.basename, geoArr_mask.bands)
            assert geoArr_mask.shape[:2] == self.shape[:2], 'The provided nodata mask must have the same number of ' \
                                                            'rows and columns as the %s itself.' %imName
            assert geoArr_mask.gt == self.gt, \
                'The geotransform of the given nodata mask for %s must match the geotransform of the %s itself. ' \
                'Got %s.' %(imName, self.__class__.__name__, geoArr_mask.gt)
            assert not geoArr_mask.prj or prj_equal(geoArr_mask.prj, self.prj), \
                'The projection of the given nodata mask for the %s must match the projection of the %s itself.' \
                %(imName, self.__class__.__name__)

            self._mask_nodata = geoArr_mask


    @property
    def mask_baddata(self):
        """
        Returns the bad data mask for the associated image array if it has been explicitly previously. It can be set
         by passing a file path, a numpy array or an instance of GeoArray to the setter of this property.
        """

        return self._mask_baddata


    @mask_baddata.setter
    def mask_baddata(self, mask):
        """Set bad data mask.

        :param mask:    Can be a file path, a numpy array or an instance o GeoArray.
        """

        if mask is not None:
            from .masks import BadDataMask
            geoArr_mask     = BadDataMask(mask, progress=self.progress, q=self.q)
            geoArr_mask.gt  = geoArr_mask.gt  if geoArr_mask.gt  not in [None, [0, 1, 0, 0, 0, -1]] else self.gt
            geoArr_mask.prj = geoArr_mask.prj if geoArr_mask.prj else self.prj
            imName          = "the %s '%s'" %(self.__class__.__name__, self.basename)

            assert geoArr_mask.bands == 1, \
                'Expected one single band as bad data mask for %s. Got %s bands.' % (self.basename, geoArr_mask.bands)
            assert geoArr_mask.shape[:2] == self.shape[:2], 'The provided bad data mask must have the same number of ' \
                                                            'rows and columns as the %s itself.' %imName
            assert geoArr_mask.gt == self.gt, \
                'The geotransform of the given bad data mask for %s must match the geotransform of the %s itself. ' \
                'Got %s.' %(imName, self.__class__.__name__, geoArr_mask.gt)
            assert prj_equal(geoArr_mask.prj, self.prj), \
                'The projection of the given bad data mask for the %s must match the projection of the %s itself.' \
                %(imName, self.__class__.__name__)

            self._mask_baddata = geoArr_mask


    @property
    def footprint_poly(self):
        # FIXME should return polygon in image coordinates if no projection is available
        """
        Get the footprint polygon of the associated image array (returns an instance of shapely.geometry.Polygon.
        """

        if self._footprint_poly is None:
            assert self.mask_nodata is not None, 'A nodata mask is needed for calculating the footprint polygon. '
            if np.std(self.mask_nodata[:])==0:
                # do not run raster2polygon if whole image is filled with data
                self._footprint_poly = self.box.mapPoly
            else:
                try:
                    multipolygon = raster2polygon(self, exact=False, progress=self.progress, q=self.q,
                                                  maxfeatCount=10, timeout=3)
                    self._footprint_poly = fill_holes_within_poly(multipolygon)
                except (RuntimeError, TimeoutError, TimeoutError_comp):
                    if not self.q:
                        warnings.warn("\nCalculation of footprint polygon failed for %s '%s'. Using outer bounds. One "
                                      "reason could be that the nodata value appears within the actual image (not only "
                                      "as fill value). To avoid this use another nodata value. Current nodata value is "
                                      "%s." %(self.__class__.__name__, self.basename, self.nodata))
                    self._footprint_poly = self.box.mapPoly

            # validation
            assert not polyVertices_outside_poly(self._footprint_poly, self.box.mapPoly),\
                "Computing footprint polygon for %s '%s' failed. The resulting polygon is partly or completely outside " \
                "of the image bounds." %(self.__class__.__name__, self.basename)
            #assert self._footprint_poly
            # for XY in self.corner_coord:
            #    assert self.GeoArray.box.mapPoly.contains(Point(XY)) or self.GeoArray.box.mapPoly.touches(Point(XY)), \
            #        "The corner position '%s' is outside of the %s." % (XY, self.imName)

        return self._footprint_poly


    @footprint_poly.setter
    def footprint_poly(self, poly):
        if isinstance(poly, Polygon):
            self._footprint_poly = poly
        elif isinstance(poly, str):
            self._footprint_poly = shply_loads(poly)
        else:
            raise ValueError("'footprint_poly' can only be set from a shapely polygon or a WKT string.")


    @property
    def metadata(self):
        """
        Returns a GeoDataFrame containing all available metadata (read from file if available).
        Use 'metadata[band_index].to_dict()' to get a metadata dictionary for a specific band.
        Use 'metadata.loc[row_name].to_dict()' to get all metadata values of the same key for all bands as dictionary.
        Use 'metadata.loc[row_name, band_index] = value' to set a new value.

        :return:  geopandas.GeoDataFrame
        """

        if self._metadata is not None:
            return self._metadata
        else:
            default = GeoDataFrame(columns=range(self.bands))
            #for bn,idx in self.bandnames.items():
            #    default.loc['band_index',bn] = idx
            self._metadata = default
            if not self.is_inmem:
                self.set_gdalDataset_meta()
                return self._metadata
            else:
                return self._metadata


    @metadata.setter
    def metadata(self, GDF):
        assert isinstance(GDF, (GeoDataFrame, DataFrame)) and len(GDF.columns)==self.bands, \
            "%s.metadata can only be set with an instance of geopandas.GeoDataFrame of which the column number " \
            "corresponds to the band number of %s." %(self.__class__.__name__, self.__class__.__name__)
        self._metadata = GDF


    meta = alias_property('metadata')


    def __getitem__(self, given):
        if isinstance(given, (int,float,slice)) and self.ndim==3:
            # handle 'given' as index for 3rd (bands) dimension
            if self.is_inmem:
                return self.arr[:, :, given]
            else:
                return self.from_path(self.arg, [given])

        elif isinstance(given, str):
            # behave like a dictionary and return the corresponding band
            if self.bandnames:
                if given not in self.bandnames:
                    raise ValueError("'%s' is not a known band. Known bands are: %s"
                                     % (given, ', '.join(list(self.bandnames.keys()))))
                if self.is_inmem:
                    return self.arr if self.ndim==2 else self.arr[:, :, self.bandnames[given]]
                else:
                    return self.from_path(self.arg, [self.bandnames[given]])
            else:
                raise ValueError('String indices are only supported if %s has been instanced with bandnames given.'
                                 %self.__class__.__name__)

        elif isinstance(given, (tuple, list)):
            # handle requests like geoArr[[1,2],[3,4]  -> not implemented in from_path if array is not in mem
            types = [type(i) for i in given]
            if list in types or tuple in types:
                self.to_mem()

            if len(given)==3:

                # handle strings in the 3rd dim of 'given' -> convert them to a band index
                if isinstance(given[2],str):
                    if self.bandnames:
                        if given[2] not in self.bandnames:
                            raise ValueError("'%s' is not a known band. Known bands are: %s"
                                             % (given[2], ', '.join(list(self.bandnames.keys()))))

                        band_idx = self.bandnames[given[2]]
                        # NOTE: the string in the 3rd is ignored if ndim==2 and band_idx==0
                        if self.is_inmem:
                            return self.arr if (self.ndim==2 and band_idx==0) else self.arr[:, :, band_idx]
                        else:
                            getitem_params = given[:2] if (self.ndim==2 and band_idx==0) else given[:2]+(band_idx,)
                            return self.from_path(self.arg, getitem_params)
                    else:
                        raise ValueError(
                            'String indices are only supported if %s has been instanced with bandnames given.'
                            % self.__class__.__name__)

                # in case a third dim is requested from 2D-array -> ignore 3rd dim if 3rd dim is 0
                elif self.ndim==2 and given[2]==0:
                    if self.is_inmem:
                        return self.arr[given[:2]]
                    else:
                        return self.from_path(self.arg, given[:2])


        # if nothing has been returned until here -> behave like a numpy array
        if self.is_inmem:
            return self.arr[given]
        else:
            getitem_params = [given] if isinstance(given, slice) else given
            return self.from_path(self.arg, getitem_params)


    def __setitem__(self, idx, array2set):
        """Overwrites the pixel values of GeoArray.arr with the given array.

        :param idx:         <int, list, slice> the index position to overwrite
        :param array2set:   <np.ndarray> array to be set. Must be compatible to the given index position.
        :return:
        """

        if self.is_inmem:
            self.arr[idx] = array2set
        else:
            raise NotImplementedError('Item assignment for %s instances that are not in memory is not yet supported.'
                                      %self.__class__.__name__)


    def __getattr__(self, attr):
        # check if the requested attribute can not be present because GeoArray has been instanced with an array
        if attr not in self.__dir__() and not self.is_inmem and attr in ['shape','dtype','geotransform', 'projection']:
            self.set_gdalDataset_meta()

        if attr in self.__dir__():             #__dir__() includes also methods and properties
            return self.__getattribute__(attr) #__getattribute__ avoids infinite loop
        elif hasattr(np.array([]),attr):
            return self[:].__getattribute__(attr)
        else:
            raise AttributeError("%s object has no attribute '%s'." %(self.__class__.__name__, attr))


    def __getstate__(self):
        """Defines how the attributes of GMS object are pickled."""

        # clean array cache in order to avoid cache pickling
        self.flush_cache()

        return self.__dict__


    def __setstate__(self, state):
        """Defines how the attributes of GMS object are unpickled.
        NOTE: This method has been implemented because otherwise pickled and unpickled instances show recursion errors
        within __getattr__ when requesting any attribute.
        """

        self.__dict__ = state


    def calc_mask_nodata(self, fromBand=None, overwrite=False):
        """Calculates a no data mask with (values: 0=nodata; 1=data)

        :param fromBand:  <int> index of the band to be used (if None, all bands are used)
        :param overwrite: <bool> whether to overwrite existing nodata mask that has already been calculated
        :return:
        """

        if self._mask_nodata is None or overwrite:
            assert self.ndim in [2, 3], "Only 2D or 3D arrays are supported. Got a %sD array." % self.ndim
            arr = self[:,:,fromBand] if self.ndim==3 and fromBand is not None else self[:]

            if self.nodata is None:
                self.mask_nodata = np.ones((self.rows, self.cols), np.bool)
            else:
                self.mask_nodata = np.where(arr == self.nodata, 0, 1).astype(np.bool) if arr.ndim == 2 else \
                                   np.all(np.where(arr == self.nodata, 0, 1), axis=2).astype(np.bool)


    def set_gdalDataset_meta(self):
        """Retrieves GDAL metadata from file. This function is only executed once to avoid overwriting of user defined
         attributes, that are defined after object instanciation.

        :return:
        """

        # TODO add support for global domain metadata
        if not self._gdalDataset_meta_already_set:
            assert self.filePath
            ds = gdal.Open(self.filePath)
664
665
666
            if not ds:
                raise Exception('Error reading file:  ' + gdal.GetLastErrorMsg())

667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
            # set private class variables (in order to avoid recursion error)
            self._shape        = tuple([ds.RasterYSize, ds.RasterXSize] + ([ds.RasterCount] if ds.RasterCount>1 else []))
            self._dtype        = gdal_array.GDALTypeCodeToNumericTypeCode(ds.GetRasterBand(1).DataType)
            self._geotransform = ds.GetGeoTransform()
            self._projection   = EPSG2WKT(WKT2EPSG(ds.GetProjection())) # temp conversion to EPSG needed because GDAL seems to modify WKT string when writing file to disk (e.g. using gdal_merge) -> conversion to EPSG and back undos that
            if not 'nodata' in self._initParams or self._initParams['nodata'] is None:
                band           = ds.GetRasterBand(1)
                self._nodata   = band.GetNoDataValue()   # FIXME this does not support different nodata values within the same file

            # read band domain metadata
            for b in range(self.bands):
                band     = ds.GetRasterBand(b+1)
                self.metadata[b] = GeoSeries(band.GetMetadata())

            ds = band = None

        self._gdalDataset_meta_already_set = True


    def from_path(self, path, getitem_params=None):
        # type: (str, list) -> np.ndarray
        """Read a GDAL compatible raster image from disk, with respect to the given image position.
        NOTE: If the requested array position is already in cache, it is returned from there.

        :param path:            <str> the file path of the image to read
        :param getitem_params:  <list> a list of slices in the form [row_slice, col_slice, band_slice]
        :return out_arr:        <np.ndarray> the output array
        """

        ds = gdal.Open(path)
697
698
699
700
        if not ds:
            raise Exception('Error reading file:  ' + gdal.GetLastErrorMsg())


701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
        R, C, B = ds.RasterYSize, ds.RasterXSize, ds.RasterCount
        ds = None

        ## convert getitem_params to subset area to be read ##
        rS, rE, cS, cE, bS, bE, bL = [None] * 7

        # populate rS, rE, cS, cE, bS, bE, bL
        if getitem_params:
            if len(getitem_params) >= 2:
                givenR, givenC = getitem_params[:2]
                if isinstance(givenR, slice):
                    rS = givenR.start
                    rE = givenR.stop - 1 if givenR.stop is not None else None
                elif isinstance(givenR, int):
                    rS = givenR
                    rE = givenR
                if isinstance(givenC, slice):
                    cS = givenC.start
                    cE = givenC.stop - 1 if givenC.stop is not None else None
                elif isinstance(givenC, int):
                    cS = givenC
                    cE = givenC
            if len(getitem_params) in [1, 3]:
                givenB = getitem_params[2] if len(getitem_params) == 3 else getitem_params[0]
                if isinstance(givenB, slice):
                    bS = givenB.start
                    bE = givenB.stop - 1 if givenB.stop is not None else None
                elif isinstance(givenB, int):
                    bS = givenB
                    bE = givenB
                elif isinstance(givenB,(tuple,list)):
                    typesInGivenB = [type(i) for i in givenB]
                    assert len(list(set(typesInGivenB))) == 1, \
                        'Mixed data types within the list of bands are not supported.'
                    if isinstance(givenB[0], int):
                        bL = list(givenB)
                    elif isinstance(givenB[0], str):
                        bL = [self.bandnames[i] for i in givenB]
                elif type(givenB) in [str]:
                    bL = [self.bandnames[givenB]]


        # set defaults for not given values
        rS = rS if rS is not None else 0
        rE = rE if rE is not None else R - 1
        cS = cS if cS is not None else 0
        cE = cE if cE is not None else C - 1
        bS = bS if bS is not None else 0
        bE = bE if bE is not None else B - 1
        bL = list(range(bS, bE + 1)) if not bL else bL

        # convert negative to positive ones
        rS = rS if rS >= 0 else self.rows + rS
        rE = rE if rE >= 0 else self.rows + rE
        cS = cS if cS >= 0 else self.columns + cS
        cE = cE if cE >= 0 else self.columns + cE
        bS = bS if bS >= 0 else self.bands + bS
        bE = bE if bE >= 0 else self.bands + bE
        bL = [b if b  >= 0 else (self.bands + b) for b in bL]

        # validate subset area bounds to be read
        msg = lambda v, idx, sz: '%s is out of bounds for axis %s with size %s' %(v, idx, sz) # FIXME numpy raises that error ONLY for the 2nd axis
        for val, axIdx, axSize in zip([rS,rE,cS,cE,bS,bE], [0,0,1,1,2,2], [R,R,C,C,B,B]):
            if not 0 <= val <= axSize - 1: raise ValueError(msg(val,axIdx,axSize))

        # summarize requested array position in arr_pos
        #NOTE: # bandlist must be string because truth value of an array with more than one element is ambiguous
        arr_pos = dict(rS=rS, rE=rE, cS=cS, cE=cE, bS=bS, bE=bE, bL=bL)


        # check if the requested array position is already in cache -> if yes, return it from there
        if self._arr_cache is not None and self._arr_cache['pos']==arr_pos:
773
            out_arr = self._arr_cache['arr_cached']
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854

        else:
            # TODO insert a multiprocessing.Lock here in order to prevent IO bottlenecks?
            # read subset area from disk
            if bL == list(range(0, B)):
                tempArr = gdalnumeric.LoadFile(path, cS, rS, cE - cS + 1, rE - rS + 1)
                if tempArr is None:
                    raise Exception('Error reading file:  ' + gdal.GetLastErrorMsg())
                out_arr = np.swapaxes(np.swapaxes(tempArr, 0, 2), 0, 1) if B > 1 else tempArr
            else:
                ds = gdal.Open(path)
                if len(bL) == 1:
                    band = ds.GetRasterBand(bL[0] + 1)
                    out_arr= band.ReadAsArray(cS, rS, cE - cS + 1, rE - rS + 1)
                    band = None
                else:
                    out_arr = np.empty((rE - rS + 1, cE - cS + 1, len(bL)))
                    for i, bIdx in enumerate(bL):
                        band = ds.GetRasterBand(bIdx + 1)
                        out_arr[:, :, i] = band.ReadAsArray(cS, rS, cE - cS + 1, rE - rS + 1)
                        band = None

                ds = None

            if out_arr is None:
                raise Exception('Error reading file:  ' + gdal.GetLastErrorMsg())

            # only set self.arr if the whole cube has been read (in order to avoid sudden shape changes)
            if out_arr.shape==self.shape:
                self.arr = out_arr

            # write _arr_cache
            self._arr_cache = dict(pos=arr_pos, arr_cached=out_arr)

        return out_arr # TODO implement check of returned datatype (e.g. NoDataMask should always return np.bool
                       # TODO -> would be np.int8 if an int8 file is read from disk


    def save(self, out_path, fmt='ENVI', creationOptions=None):
        # type: (str, str, list) -> None
        """Write the raster data to disk.

        :param out_path:        <str> output path
        :param fmt:             <str> the output format / GDAL driver code to be used for output creation, e.g. 'ENVI'
                                Refer to http://www.gdal.org/formats_list.html to get a full list of supported formats.
        :param creationOptions: <list> GDAL creation options, e.g. ["QUALITY=80", "REVERSIBLE=YES", "WRITE_METADATA=YES"]
        :return:
        """

        if not self.q:
            print('Writing GeoArray of size %s to %s.' %(self.shape, out_path))
        assert self.ndim in [2,3], 'Only 2D- or 3D arrays are supported.'

        driver = gdal.GetDriverByName(fmt)
        if driver is None:
            raise Exception("'%s' is not a supported GDAL driver. Refer to www.gdal.org/formats_list.html for full "
                            "list of GDAL driver codes." % fmt)

        if not os.path.isdir(os.path.dirname(out_path)):
            os.makedirs(os.path.dirname(out_path))

        if self.is_inmem:
            ds = get_GDAL_ds_inmem(self.arr,self.geotransform, self.projection, self.nodata) # expects rows,columns,bands

            # set metadata
            if not self.metadata.empty:
                for bidx in range(self.bands):
                    band = ds.GetRasterBand(bidx+1)
                    meta2write = self.metadata[bidx].to_dict()
                    meta2write = dict((k, v) for k, v in meta2write.items() if not v is np.nan)
                    band.SetMetadata(meta2write)
                    band = None

            driver.CreateCopy(out_path, ds, options=creationOptions if creationOptions else [])

            #out_arr = self.arr if self.ndim == 2 else np.swapaxes(np.swapaxes(self.arr, 0, 2), 1, 2)  # rows, columns, bands => bands, rows, columns
            #gdalnumeric.SaveArray(out_arr, out_path, format=fmt, prototype=ds) # expects bands,rows,columns
            ds = None

        else:
            src_ds = gdal.Open(self.filePath)
855
856
857
            if not src_ds:
                raise Exception('Error reading file:  ' + gdal.GetLastErrorMsg())

858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
            gdal_Translate = get_gdal_func('Translate')
            gdal_Translate(out_path, src_ds, format=fmt, creationOptions=creationOptions)
            src_ds = None

        if not os.path.exists(out_path):
            raise Exception(gdal.GetLastErrorMsg())


    def dump(self, out_path):
        # type: (str) -> None
        """Serialize the whole object instance to disk using dill."""

        import dill
        with open(out_path,'wb') as outF:
            dill.dump(self,outF)


    def _get_plottable_image(self, xlim=None, ylim=None, band=None, boundsMap=None, boundsMapPrj=None, res_factor=None,
                             nodataVal=None, out_prj=None):
        # handle limits
        if boundsMap:
            boundsMapPrj = boundsMapPrj if boundsMapPrj else self.prj
            image2plot, gt, prj = self.get_mapPos(boundsMap, boundsMapPrj, band2get=band,
                                                  fillVal=nodataVal if nodataVal is not None else self.nodata)
        else:
883
884
            cS, cE = xlim if isinstance(xlim, (tuple, list)) else (0, self.columns)
            rS, rE = ylim if isinstance(ylim, (tuple, list)) else (0, self.rows)
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224

            image2plot = self[rS:rE, cS:cE, band] if band is not None else self[rS:rE, cS:cE]
            gt, prj    = self.geotransform, self.projection


        transOpt   = ['SRC_METHOD=NO_GEOTRANSFORM'] if tuple(gt) == (0, 1, 0, 0, 0, -1) else None
        xdim, ydim = None, None
        nodataVal  = nodataVal if nodataVal is not None else self.nodata

        if res_factor != 1. and image2plot.shape[0] * image2plot.shape[1] > 1e6:  # shape > 1000*1000
            # sample image down
            xdim, ydim = (self.columns * res_factor, self.rows * res_factor) if res_factor else \
                tuple((np.array([self.columns, self.rows]) / (np.array([self.columns, self.rows]).max() / 1000)))  # normalize
            xdim, ydim = int(xdim), int(ydim)

        if xdim or ydim or out_prj:
            from py_tools_ds.ptds.geo.raster.reproject import warp_ndarray
            image2plot, gt, prj = warp_ndarray(image2plot, self.geotransform, self.projection,
                                               out_XYdims=(xdim, ydim), in_nodata=nodataVal, out_nodata=nodataVal,
                                               transformerOptions=transOpt, out_prj=out_prj, q=True)
            if transOpt and 'NO_GEOTRANSFORM' in ','.join(transOpt):
                image2plot = np.flipud(image2plot)
                gt=list(gt)
                gt[3]=0

            if xdim or ydim:
                print('Note: array has been downsampled to %s x %s for faster visualization.' % (xdim, ydim))

        return image2plot, gt, prj


    def show(self, xlim=None, ylim=None, band=None, boundsMap=None, boundsMapPrj=None, figsize=None,
             interpolation='none', vmin=None, vmax=None, cmap=None, nodataVal=None, res_factor=None, interactive=False):
        """Plots the desired array position into a figure.

        :param xlim:            [start_column, end_column]
        :param ylim:            [start_row, end_row]
        :param band:            the band index of the band to be plotted (if None and interactive==True all bands are
                                shown, otherwise the first band is chosen)
        :param boundsMap:       xmin, ymin, xmax, ymax
        :param boundsMapPrj:
        :param figsize:
        :param interpolation:
        :param vmin:
        :param vmax:
        :param cmap:
        :param nodataVal:
        :param res_factor:
        :param interactive:     <bool> activates interactive plotting based on 'holoviews' library.
                                NOTE: this deactivates the magic '% matplotlib inline' in Jupyter Notebook
        :return:
        """

        band = (band if band is not None else 0) if not interactive else band

        # get image to plot
        nodataVal           = nodataVal if nodataVal is not None else self.nodata
        image2plot, gt, prj = self._get_plottable_image(xlim, ylim, band, boundsMap=boundsMap,
                                                        boundsMapPrj=boundsMapPrj, res_factor=res_factor,
                                                        nodataVal=nodataVal)

        # set color palette
        palette   = cmap if cmap else plt.cm.gray
        if nodataVal is not None and np.std(image2plot)!=0: # do not show nodata
            image2plot = np.ma.masked_equal(image2plot, nodataVal)
            vmin_auto, vmax_auto = np.percentile(image2plot.compressed(),2), np.percentile(image2plot.compressed(),98)
            palette.set_bad('aqua', 0)
        else:
            vmin_auto, vmax_auto = np.percentile(image2plot, 2), np.percentile(image2plot, 98)

        vmin = vmin if vmin is not None else vmin_auto
        vmax = vmax if vmax is not None else vmax_auto

        palette.set_over ('1')
        palette.set_under('0')

        if interactive and image2plot.ndim==3:
            import holoviews as hv
            from skimage.exposure import rescale_intensity
            hv.notebook_extension('matplotlib')

            cS, cE = xlim if isinstance(xlim, (tuple, list)) else (0, self.columns - 1)
            rS, rE = ylim if isinstance(ylim, (tuple, list)) else (0, self.rows - 1)

            image2plot   = np.array(rescale_intensity(image2plot, in_range=(vmin, vmax)))
            get_hv_image = lambda b: hv.Image(image2plot[:,:,b] if b is not None else image2plot,
                                              bounds=(cS, rS, cE, rE))(style={'cmap': 'gray'}, # FIXME ylabels have the wrong order
                                              plot={'fig_inches':4 if figsize is None else figsize, 'show_grid':True})

            #hvIm = hv.Image(image2plot)(style={'cmap': 'gray'}, figure_inches=figsize)
            hmap = hv.HoloMap([(band, get_hv_image(band)) for band in range(image2plot.shape[2])], kdims=['band'])

            return hmap

        else:
            if interactive:
                warnings.warn('Currently there is no interactive mode for single-band arrays. '
                              'Switching to standard matplotlib figure..') # TODO implement zoomable fig

            # show image
            plt.figure(figsize=figsize)
            rows, cols = image2plot.shape[:2]
            plt.imshow(image2plot, palette, interpolation=interpolation, extent=(0, cols, rows, 0),
                       vmin=vmin, vmax=vmax, ) # compressed excludes nodata values
            plt.show()


    def show_map(self, xlim=None, ylim=None, band=0, boundsMap=None, boundsMapPrj=None, ax=None, figsize=None,
                 interpolation='none', vmin=None, vmax=None, cmap=None, nodataVal=None, res_factor=None,
                 return_map=False, zoomable=False):
        """

        :param xlim:
        :param ylim:
        :param band:            band index (starting with 0)
        :param boundsMap:       xmin, ymin, xmax, ymax
        :param boundsMapPrj:
        :param ax:              allows to pass a matplotlib axis object where figure is plotted into
        :param figsize:
        :param interpolation:
        :param vmin:
        :param vmax:
        :param cmap:
        :param nodataVal:
        :param res_factor:
        :param return_map:
        :param zoomable:        <bool> enable or disable zooming via mpld3
        :return:
        """

        from mpl_toolkits.basemap import Basemap
        try:
            import mpld3
            if zoomable:
                mpld3.enable_notebook()
            else:
                mpld3.disable_notebook()
        except:
            if zoomable:
                zoomable = False
                warnings.warn('mpld3 is not available. Zooming disabled.')

        assert self.geotransform and tuple(self.geotransform) != (0,1,0,0,0,-1),\
            'A valid geotransform is needed for a map visualization. Got %s.' %self.geotransform
        assert self.projection,   'A projection is needed for a map visualization. Got %s.' %self.projection

        # get image to plot
        nodataVal           = nodataVal if nodataVal is not None else self.nodata
        image2plot, gt, prj = self._get_plottable_image(xlim, ylim, band, boundsMap=boundsMap,
                                                        boundsMapPrj=boundsMapPrj, res_factor=res_factor,
                                                        nodataVal=nodataVal, out_prj='epsg:4326')

        # calculate corner coordinates of plot
        #if boundsMap:
        #    boundsMapPrj = boundsMapPrj if boundsMapPrj else self.prj
        #    if not prj_equal(boundsMapPrj, 4326):
        #        boundsMap = reproject_shapelyGeometry(box(*boundsMap), boundsMapPrj, 4626).bounds
        #    xmin, ymin, xmax, ymax = boundsMap
        #    UL_XY, UR_XY, LR_XY, LL_XY = (xmin,ymax), (xmax, ymax), (xmax,ymin), (xmin, ymin)
        #else:
        UL_XY, UR_XY, LR_XY, LL_XY = [(YX[1],YX[0]) for YX in GeoArray(image2plot, gt, prj).box.boxMapYX]
        center_lon, center_lat     = (UL_XY[0]+UR_XY[0])/2., (UL_XY[1]+LL_XY[1])/2.

        # create map
        fig = plt.figure(figsize=figsize)
        plt.subplots_adjust(left=0.05, right=0.95, top=0.90, bottom=0.05, wspace=0.15, hspace=0.05)
        ax = ax if ax is not None else plt.subplot(111)

        m = Basemap(projection='tmerc', resolution=None,    lon_0=center_lon,   lat_0=center_lat,
                    urcrnrlon=UR_XY[0], urcrnrlat=UR_XY[1], llcrnrlon=LL_XY[0], llcrnrlat=LL_XY[1])

        # set color palette
        palette = cmap if cmap else plt.cm.gray
        if nodataVal is not None and np.std(image2plot)!=0: # do not show nodata
            image2plot = np.ma.masked_equal(image2plot, nodataVal)
            vmin_auto, vmax_auto = np.percentile(image2plot.compressed(), 2), np.percentile(image2plot.compressed(), 98)
            palette.set_bad('aqua', 0)
        else:
            vmin_auto, vmax_auto = np.percentile(image2plot, 2), np.percentile(image2plot, 98)
        vmin = vmin if vmin is not None else vmin_auto
        vmax = vmax if vmax is not None else vmax_auto
        palette.set_over ('1')
        palette.set_under('0')

        # add image to map (y-axis must be inverted for basemap)
        if zoomable:
            m.imshow(image2plot, palette, interpolation=interpolation, vmin=vmin, vmax=vmax)
        else:
            m.imshow(np.flipud(image2plot), palette, interpolation=interpolation, vmin=vmin, vmax=vmax)

        # add coordinate grid lines
        parallels = np.arange(-90, 90., 0.25) # TODO make this adjustable
        #parallels = np.arange(-90, 90., 0.1)
        m.drawparallels(parallels, labels=[1, 0, 0, 0], fontsize=12, linewidth=0.4)

        meridians = np.arange(-180., 180., 0.25)
        #meridians = np.arange(-180., 180., 0.1)
        m.drawmeridians(meridians, labels=[0, 0, 0, 1], fontsize=12, linewidth=0.4)

        if return_map:
            return fig,ax, m
        else:
            plt.show()


    def show_map_utm(self, xlim=None, ylim=None, band=0, figsize=None, interpolation='none', cmap=None,
                     nodataVal=None, vmin=None, vmax=None, res_factor=None, return_map=False):

        from mpl_toolkits.basemap import Basemap
        warnings.warn(UserWarning('This function is still under construction and may not work as expected!'))
        # TODO debug this function

        # get image to plot
        nodataVal           = nodataVal if nodataVal is not None else self.nodata
        image2plot, gt, prj = self._get_plottable_image(xlim, ylim, band, res_factor, nodataVal)

        # calculate corner coordinates of plot
        box2plot = GeoArray(image2plot, gt, prj).box
        UL_XY, UR_XY, LR_XY, LL_XY = [(YX[1], YX[0]) for YX in GeoArray(image2plot, gt, prj).box.boxMapYX]
        # Xarr, Yarr = self.box.get_coordArray_MapXY(prj=EPSG2WKT(4326))
        UL_XY, UR_XY, LR_XY, LL_XY = [transform_any_prj(self.projection,'epsg:4326',x,y)  for y,x in box2plot.boxMapYX]
        center_X, center_Y = (UL_XY[0] + UR_XY[0]) / 2., (UL_XY[1] + LL_XY[1]) / 2.
        center_lon, center_lat = transform_any_prj(prj,'epsg:4326', center_X, center_Y)
        print(center_lon, center_lat)

        # create map
        fig = plt.figure(figsize=figsize)
        plt.subplots_adjust(left=0.05, right=0.95, top=0.90, bottom=0.05, wspace=0.15, hspace=0.05)
        ax = plt.subplot(111)
        print(UL_XY, UR_XY, LR_XY, LL_XY)
#        m = Basemap(projection='tmerc', resolution=None, lon_0=center_lon, lat_0=center_lat,
#                    urcrnrx=UR_XY[0], urcrnry=UR_XY[1], llcrnrx=LL_XY[0], llcrnry=LL_XY[1])
        m = Basemap(projection='tmerc', resolution=None, lon_0=center_lon, lat_0=center_lat,
                    urcrnrlon=UR_XY[0], urcrnrlat=UR_XY[1], llcrnrlon=LL_XY[0], llcrnrlat=LL_XY[1],
                    k_0=0.9996, rsphere=(6378137.00, 6356752.314245179),suppress_ticks=False)
        # m.pcolormesh(Xarr, Yarr, self[:], cmap=plt.cm.jet)

        # set color palette
        palette = cmap if cmap else plt.cm.gray
        if nodataVal is not None: # do not show nodata
            image2plot = np.ma.masked_equal(image2plot, nodataVal)
            vmin_auto, vmax_auto = np.percentile(image2plot.compressed(), 2), np.percentile(image2plot.compressed(), 98)
            palette.set_bad('aqua', 0)
        else:
            vmin_auto, vmax_auto = np.percentile(image2plot, 2), np.percentile(image2plot, 98)
        vmin = vmin if vmin is not None else vmin_auto
        vmax = vmax if vmax is not None else vmax_auto
        palette.set_over('1')
        palette.set_under('0')

        # add image to map (y-axis must be inverted for basemap)
        m.imshow(np.flipud(image2plot), palette, interpolation=interpolation, vmin=vmin, vmax=vmax)

        # add coordinate grid lines
        parallels = np.arange(-90, 90., 0.25)
        m.drawparallels(parallels, labels=[1, 0, 0, 0], fontsize=12, linewidth=0.4)

        meridians = np.arange(-180., 180., 0.25)
        m.drawmeridians(meridians, labels=[0, 0, 0, 1], fontsize=12, linewidth=0.4)

        if return_map:
            return fig, ax, m
        else:
            plt.show()


    def show_footprint(self):
        """This method is intended to be called from Jupyter Notebook and shows a web map containing the calculated
        footprint of GeoArray.
        """

        try:
            import folium, geojson
        except ImportError:
            folium, geojson = None, None
        if not folium or not geojson:
            raise ImportError(
                "This method requires the libraries 'folium' and 'geojson'. They can be installed with "
                "the shell command 'pip install folium geojson'.")

        lonlatPoly = reproject_shapelyGeometry(self.footprint_poly, self.epsg, 4326)

        m   = folium.Map(location=tuple(np.array(lonlatPoly.centroid.coords.xy).flatten())[::-1])
        gjs = geojson.Feature(geometry=lonlatPoly, properties={})
        folium.GeoJson(gjs).add_to(m)
        return m


    def get_mapPos(self, mapBounds, mapBounds_prj, band2get=None, out_prj=None, arr_gt=None, arr_prj=None, fillVal=None,
                   rspAlg='near', progress=None, v=False): # TODO implement slice for indexing bands
        # type: (tuple, str, int, str, tuple, str, int, str, bool, bool) -> (np.ndarray, tuple, str)
        """Returns the array data of GeoArray at a given geographic position.

        NOTE: The given mapBounds are snapped to the pixel grid of GeoArray. If the given mapBounds include areas
        outside of the extent of GeoArray, these areas are filled with the fill value of GeoArray.

        :param mapBounds:       xmin, ymin, xmax, ymax
        :param mapBounds_prj:   WKT projection string corresponding to mapBounds
        :param band2get:        band index of the band to be returned (full array if not given)
        :param out_prj:         output projection as WKT string. If not given, the self.projection is used.
        :param arr_gt:          GDAL GeoTransform (taken from self if not given)
        :param arr_prj:         WKT projection string (taken from self if not given)
        :param fillVal:         nodata value
        :param rspAlg:          <str> Resampling method to use. Available methods are:
                                near, bilinear, cubic, cubicspline, lanczos, average, mode, max, min, med, q1, q2
        :param progress:        whether to show progress bars or not
        :param v:               verbose mode (not related to GeoArray.v; must be explicitly set)
        :return:
        """

        arr_gt   = arr_gt  if arr_gt  else self.geotransform
        arr_prj  = arr_prj if arr_prj else self.projection
        out_prj  = out_prj if out_prj else arr_prj
        fillVal  = fillVal if fillVal is not None else self.nodata
        progress = progress if progress is not None else self.progress

        if self.is_inmem and (not arr_gt or not arr_prj):
            raise ValueError('In case of in-mem arrays the respective geotransform and projection of the array '
                             'has to be passed.')

        if v:
            print('%s.get_mapPos() input parameters:')
            print('\tmapBounds', mapBounds, '<==>', self.box.boundsMap)
            print('\tEPSG', WKT2EPSG(mapBounds_prj), self.epsg)
            print('\tarr_gt', arr_gt, self.gt)
            print('\tarr_prj', WKT2EPSG(arr_prj), self.epsg)
            print('\tfillVal', fillVal, self.nodata, '\n')

        sub_arr, sub_gt, sub_prj = get_array_at_mapPos(self, arr_gt, arr_prj,
                                                       out_prj       = out_prj,
                                                       mapBounds     = mapBounds,
                                                       mapBounds_prj = mapBounds_prj,
                                                       fillVal       = fillVal,
                                                       rspAlg        = rspAlg,
                                                       out_gsd       = (self.xgsd,self.ygsd),
                                                       band2get      = band2get,
                                                       progress      = progress)
        return sub_arr, sub_gt, sub_prj


1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
    def get_subset(self, xslice=None, yslice=None, zslice=None, return_GeoArray=True):
        # type: (slice, slice, slice, bool) -> GeoArray
        """Returns a new instatnce of GeoArray representing a subset of the initial one wit respect to given array position.
        
        :param xslice:          a slice providing the X-position for the subset in the form slice(xstart, xend, xstep)
        :param yslice:          a slice providing the Y-position for the subset in the form slice(ystart, yend, ystep)
        :param zslice:          a slice providing the Z-position for the subset in the form slice(zstart, zend, zstep)
        :param return_GeoArray: whether to return an instance of GeoArray (default) or a tuple(np.ndarray, gt, prj)
        :return: 
        """

        sub_arr = self[xslice if xslice else slice(None),
                       yslice if yslice else slice(None),
                       zslice if zslice else slice(None)]
        sub_ulXY = imXY2mapXY((xslice.start, yslice.start), self.gt)
        sub_gt = (sub_ulXY[0], self.gt[1], self.gt[2], sub_ulXY[1], self.gt[4], self.gt[5])
        sub_gA = GeoArray(sub_arr, sub_gt, self.prj)

        return sub_gA if return_GeoArray else (sub_arr, sub_gt, self.prj)

        # import copy   # TODO implement that in order to include all previously set attribute values
        # sub_gA = copy.deepcopy(self)
        # print(type(sub_gA))
        # sub_gA.arr = self[xslice if xslice else slice(None),
        #                   yslice if yslice else slice(None),
        #                   zslice if zslice else slice(None)]
        # #sub_gA.deepcopy_array()
        # sub_ulXY = imXY2mapXY((xslice.start, yslice.start), self.gt)
        # sub_gA.gt = (sub_ulXY[0], self.gt[1], self.gt[2], sub_ulXY[1], self.gt[4], self.gt[5])
        #
        # return sub_gA if return_GeoArray else (sub_gA.arr, sub_gA.gt, sub_gA.prj)


1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
    def reproject_to_new_grid(self, prototype=None, tgt_prj=None, tgt_xygrid=None, rspAlg='cubic', CPUs=None):
        """Reproject all array-like attributes to a given target grid.

        :param prototype:   <GeoArray> an instance of GeoArray to be used as pixel grid reference
        :param tgt_prj:     <str> WKT string of the projection
        :param tgt_xygrid:  <list> target XY grid, e.g. [[xmin,xmax], [ymax, ymin]] for the UL corner
        :param rspAlg:      <str, int> GDAL compatible resampling algorithm code
        :param CPUs:        <int> number of CPUs to use (default: None -> use all available CPUs)
        :return:
        """

        assert (tgt_prj and tgt_xygrid) or prototype, "Provide either 'prototype' or 'tgt_prj' and 'tgt_xygrid'!"
        tgt_prj    = tgt_prj    if tgt_prj    else prototype.prj
        tgt_xygrid = tgt_xygrid if tgt_xygrid is not None else prototype.xygrid_specs
        assert tgt_xygrid[1][0]>tgt_xygrid[1][1]

        # set target GSD
        tgt_xgsd, tgt_ygsd = abs(tgt_xygrid[0][0]-tgt_xygrid[0][1]), abs(tgt_xygrid[1][0]-tgt_xygrid[1][1])

        # set target bounds
        tgt_bounds = reproject_shapelyGeometry(self.box.mapPoly, self.prj, tgt_prj).bounds

        gt = (tgt_xygrid[0][0], tgt_xgsd, 0, max(tgt_xygrid[1]), 0, -tgt_ygsd)
        xmin, ymin, xmax, ymax = snap_bounds_to_pixGrid(tgt_bounds, gt, roundAlg='on')

        from py_tools_ds.ptds.geo.raster.reproject import warp_ndarray
        self.arr, self.gt, self.prj = \
            warp_ndarray(self[:], self.gt, self.prj, tgt_prj,
                         out_gsd    = (tgt_xgsd, tgt_ygsd),
                         out_bounds = (xmin, ymin, xmax, ymax),
                         out_bounds_prj = tgt_prj,
                         rspAlg     = rspAlg,
                         in_nodata  = self.nodata,
                         CPUs       = CPUs,
                         progress   = self.progress,
                         q          = self.q)

        if hasattr(self, '_mask_nodata') and self._mask_nodata is not None:
            self.mask_nodata.reproject_to_new_grid(prototype  = prototype,
                                                   tgt_prj    = tgt_prj,
                                                   tgt_xygrid = tgt_xygrid,
                                                   rspAlg     = 'near',
                                                   CPUs       = CPUs)

        if hasattr(self, '_mask_baddata') and self._mask_baddata is not None:
            self.mask_baddata.reproject_to_new_grid(prototype  = prototype,
                                                    tgt_prj    = tgt_prj,
                                                    tgt_xygrid = tgt_xygrid,
                                                    rspAlg     = 'near',
                                                    CPUs       = CPUs)

        # update footprint polygon
        if self._footprint_poly:
            if not (self.footprint_poly.within(self.box.mapPoly) or self.footprint_poly.equals(self.box.mapPoly)):
                self.footprint_poly = self.footprint_poly.intersection(self.box.mapPoly)


    def read_pointData(self, mapXY_points, mapXY_points_prj=None, band=None):
        """Returns the array values for the given set of X/Y coordinates.
         NOTE: If GeoArray has been instanced with a file path, the function will read the dataset into memory.

        :param mapXY_points:        <np.ndarray, tuple> X/Y coordinates of the points of interest. If a numpy array is
                                    given, it must have the shape [Nx2]
        :param mapXY_points_prj:    <str, int> WKT string or EPSG code of the projection corresponding to the given
                                    coordinates.
        :param band:                <int> the band index of the band of interest. If None, the values of all bands are
                                    returned.
        :return:                    np.ndarray with shape [Nx2xbands]
        """

        mapXY = mapXY_points if isinstance(mapXY_points, np.ndarray) else np.array(mapXY_points).reshape(1,2)
        prj   = mapXY_points_prj if mapXY_points_prj else self.prj

        assert prj, 'A projection is needed for returning image DNs at specific map X/Y coordinates!'
        if not prj_equal(prj1=prj, prj2=self.prj):
            mapXY = transform_any_prj(prj, self.prj, mapXY[:,0], mapXY[:,1])

        imXY = mapXY2imXY(mapXY, self.geotransform)
        imYX = np.fliplr(np.array(imXY)).astype(np.int16)

        if imYX.size==2: # only one coordinate pair
            Y,X = imYX[0].tolist()
            return self[Y,X,band]
        else: # multiple coordinate pairs
            return self[imYX.T.tolist()+[band]]


    def to_mem(self):
        """Reads the whole dataset into memory and sets self.arr to the read data."""

        self.arr = self[:]
        return self


    def to_disk(self):
        """Sets self.arr back to None if GeoArray has been instanced with a file path
        and the whole dataset has been read."""

        if self.filePath and os.path.isfile(self.filePath):
            self._arr = None
        else:
            warnings.warn('GeoArray object cannot be turned into disk mode because this asserts that GeoArray.filePath '
                          'contains a valid file path. Got %s.' %self.filePath)
        return self


    def deepcopy_array(self):
        if self.is_inmem:
            temp     = np.empty_like(self.arr)
            temp[:]  = self.arr
            self.arr = temp  # deep copy: converts view to its own array in order to avoid wrong output


    def cache_array_subset(self, arr_pos):
        # type: (list) -> None
        """Sets the array cache of the GeoArray instance to the given array in order to speed up calculations
        afterwards.

        :param arr_pos: a list of array indices as passed to __getitem__
        """

        if not self.is_inmem:
            self[arr_pos] # runs __getitem__ and sets self._arr_cache
        else:
            pass # no array cache needed because array is in memory anyways


    def flush_cache(self):
        """Clear the array cache of the GeoArray instance."""

        self._arr_cache = None



def get_GeoArray_from_GDAL_ds(ds):

    # TODO implement as class method of GeoArray
    arr = gdalnumeric.DatasetReadAsArray(ds)
    if len(arr.shape) == 3:
        arr = np.swapaxes(np.swapaxes(arr, 0, 2), 0, 1)
    return GeoArray(arr, ds.GetGeoTransform(), ds.GetProjection())



class MultiGeoArray(object):
    def __init__(self, GeoArray_list):
        """

        :param GeoArray_list:   a list of GeoArray instances having a geographic overlap
        """

        self._arrs = None

        self.arrs  = GeoArray_list

        raise NotImplementedError('This class is not yet working.') # FIXME

    @property
    def arrs(self):
        return self._arrs


    @arrs.setter
    def arrs(self, GeoArray_list):
        for geoArr in GeoArray_list:
            assert isinstance(geoArr, GeoArray), "'arrs' can only be set to a list of GeoArray instances."

        self._arrs = GeoArray_list