gms_object.py 88.6 KB
Newer Older
1
2
3
4
5
# -*- coding: utf-8 -*-

import collections
import copy
import datetime
6
import functools
7
8
9
10
11
12
13
import glob
import json
import os
import re
import shutil
import sys
import warnings
14
import logging
15
from collections import OrderedDict
16
from itertools import chain
17
from typing import Iterable, List, Union, TYPE_CHECKING  # noqa F401  # flake8 issue
18
19
20

import numpy as np
import spectral
21
from spectral.io import envi
22
from numba import jit
23
from pandas import DataFrame, read_csv
24
from nested_dict import nested_dict
25

26
27
28
29
try:
    from osgeo import gdalnumeric
except ImportError:
    import gdalnumeric
30

31
from geoarray import GeoArray
32
from py_tools_ds.geo.coord_grid import is_coord_grid_equal
33
from py_tools_ds.geo.projection import EPSG2WKT
34
35
36
from py_tools_ds.geo.map_info import geotransform2mapinfo, mapinfo2geotransform
from py_tools_ds.geo.coord_calc import calc_FullDataset_corner_positions
from py_tools_ds.geo.coord_trafo import pixelToLatLon, pixelToMapYX
37
from sicor.options import get_options as get_ac_options
38

39
from ..misc.logging import GMS_logger as DatasetLogger
40
from ..model.mgrs_tile import MGRS_tile
41
from ..model.metadata import METADATA, get_dict_LayerOptTherm, metaDict_to_metaODict
42
43
44
from ..model.dataset import Dataset
from ..misc import path_generator as PG
from ..misc import database_tools as DB_T
45
from ..options.config import GMS_config as CFG
46
47
48
from ..algorithms import geoprocessing as GEOP
from ..io import input_reader as INP_R
from ..io import output_writer as OUT_W
49
50
from ..misc import helper_functions as HLP_F
from ..misc import definition_dicts as DEF_D
51
from ..misc.locks import MultiSlotLock
52

53
54
55
if TYPE_CHECKING:
    from ..algorithms.L1C_P import L1C_object  # noqa F401  # flake8 issue

56
__author__ = 'Daniel Scheffler'
57
58


59
class GMS_object(Dataset):
60
61
62
63
    # class attributes
    # NOTE: these attributes can be modified and seen by ALL GMS_object instances
    proc_status_all_GMSobjs = nested_dict()

64
65
66
67
    def __init__(self, pathImage=''):
        # get all attributes of base class "Dataset"
        super(GMS_object, self).__init__()

68
        # add private attributes
69
        self._dict_LayerOptTherm = None
70
71
        self._cloud_masking_algorithm = None
        self._meta_odict = None
72
        self._coreg_info = None
73

74
        self.job_ID = CFG.ID
75
        # FIXME not needed anymore?:
76
        # self.dataset_ID = int(DB_T.get_info_from_postgreSQLdb(CFG.conn_database, 'scenes', ['datasetid'],
77
78
79
        #                                {'id': self.scene_ID})[0][0]) if self.scene_ID !=-9999 else -9999
        self.scenes_proc_ID = None  # set by Output writer after creation/update of db record in table scenes_proc
        self.mgrs_tiles_proc_ID = None  # set by Output writer after creation/update of db rec in table mgrs_tiles_proc
80
        self.MGRS_info = None
81
82

        # set pathes
83
84
85
86
        self.path_cloud_class_obj = ''

        # handle initialization arguments
        if pathImage:
87
88
            # run the setter for 'arr' of the base class 'Dataset' which creates an Instance of GeoArray
            self.arr = pathImage
89
90
91
92

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

Daniel Scheffler's avatar
Bugfix    
Daniel Scheffler committed
93
        self.close_loggers()
94
        del self.pathGen  # path generator can only be used for the current processing level
95
96

        # delete arrays if their in-mem size is to big to be pickled
97
        # => (avoids MaybeEncodingError: Error sending result: '[<gms_preprocessing.algorithms.L2C_P.L2C_object
98
        #    object at 0x7fc44f6399e8>]'. Reason: 'error("'i' format requires -2147483648 <= number <= 2147483647",)')
99
        if self.proc_level == 'L2C' and CFG.inmem_serialization:
100
101
            # FIXME check by bandname
            if self.mask_nodata is not None and self.masks.bands > 1 and self.mask_clouds is not None:
102
                del self.masks
103

Daniel Scheffler's avatar
Bugfix    
Daniel Scheffler committed
104
105
        return self.__dict__

106
    def set_pathes(self):
107
108
109
110
111
112
        self.baseN = self.pathGen.get_baseN()
        self.path_procdata = self.pathGen.get_path_procdata()
        self.ExtractedFolder = self.pathGen.get_path_tempdir()
        self.path_logfile = self.pathGen.get_path_logfile()
        self.pathGen = PG.path_generator(self.__dict__)  # passes a logger in addition to previous attributes
        self.path_archive = self.pathGen.get_local_archive_path_baseN()
113

114
        if not CFG.inmem_serialization:
Daniel Scheffler's avatar
Daniel Scheffler committed
115
            self.path_InFilePreprocessor = os.path.join(self.ExtractedFolder, '%s%s_DN.bsq'
116
117
                                                        % (self.entity_ID,
                                                           ('_%s' % self.subsystem if self.subsystem else '')))
118
        else:  # keep data in memory
119
            self.path_InFilePreprocessor = None  # None: keeps all produced data in memory (numpy array attributes)
120
121
122
123
124
125

        self.path_MetaPreprocessor = self.path_archive

    def validate_pathes(self):
        if not os.path.isfile(self.path_archive) and not os.path.isdir(self.path_archive):
            self.logger.info("The %s dataset '%s' has not been processed earlier and no corresponding raw data archive"
126
                             "has been found at %s." % (self.sensor, self.entity_ID, self.path_archive))
127
            self.logger.info('Trying to download the dataset...')
128
            self.path_archive_valid = self._data_downloader(self.sensor, self.entity_ID)
129
130
131
        else:
            self.path_archive_valid = True

132
        if not CFG.inmem_serialization and self.ExtractedFolder and not os.path.isdir(self.ExtractedFolder):
133
134
135
136
137
138
139
            os.makedirs(self.ExtractedFolder)

        assert os.path.exists(self.path_archive), 'Invalid path to RAW data. File %s does not exist at %s.' \
                                                  % (os.path.basename(self.path_archive),
                                                     os.path.dirname(self.path_archive))
        assert isinstance(self.path_archive, str), 'Invalid path to RAW data. Got %s instead of string or unicode.' \
                                                   % type(self.path_archive)
140
        if not CFG.inmem_serialization and self.ExtractedFolder:
141
142
            assert os.path.exists(self.path_archive), \
                'Invalid path for temporary files. Directory %s does not exist.' % self.ExtractedFolder
143
144
145
146
147
148
149
150

    @property
    def logger(self):
        if self._loggers_disabled:
            return None
        if self._logger and self._logger.handlers[:]:
            return self._logger
        else:
151
            self._logger = DatasetLogger('log__' + self.baseN, fmt_suffix=self.scene_ID, path_logfile=self.path_logfile,
152
                                         log_level=CFG.log_level, append=True)
153
154
155
156
            return self._logger

    @logger.setter
    def logger(self, logger):
157
        assert isinstance(logger, logging.Logger) or logger in ['not set', None], \
158
            "GMS_obj.logger can not be set to %s." % logger
159
160

        # save prior logs
161
        # if logger is None and self._logger is not None:
162
        #    self.log += self.logger.captured_stream
163
164
        self._logger = logger

165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
    @property
    def proc_status(self):
        # type: () -> str
        """
        Get the processing status of the current GMS_object (subclass) instance for the current processing level.

        Possible values: 'initialized', 'running', 'finished', 'failed'
        """
        # NOTE: self.proc_status_all_GMSobjs is a class attribute (visible and modifyable from any other subsystem)
        return self.proc_status_all_GMSobjs[self.scene_ID][self.subsystem][self.proc_level]

    @proc_status.setter
    def proc_status(self, new_status):
        # type: (str) -> None
        self.proc_status_all_GMSobjs[self.scene_ID][self.subsystem][self.proc_level] = new_status

181
182
    @property
    def GMS_identifier(self):
183
184
        return collections.OrderedDict(zip(
            ['image_type', 'Satellite', 'Sensor', 'Subsystem', 'proc_level', 'dataset_ID', 'logger'],
185
186
            [self.image_type, self.satellite, self.sensor, self.subsystem, self.proc_level, self.dataset_ID,
             self.logger]))
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202

    @property
    def MetaObj(self):
        if self._meta_odict:
            # if there is already a meta_odict -> create a new MetaObj from it (ensures synchronization!)
            self._MetaObj = METADATA(self.GMS_identifier).from_odict(self._meta_odict)
            del self.meta_odict
        elif not self._MetaObj:
            # if there is no meta_odict and no MetaObj -> create MetaObj by reading metadata from disk
            pass  # reading from disk should use L1A_P.L1A_object.import_metadata -> so just return None

        return self._MetaObj

    @MetaObj.setter
    def MetaObj(self, MetaObj):
        assert isinstance(MetaObj, METADATA), "'MetaObj' can only be set to an instance of METADATA class. " \
203
                                              "Got %s." % type(MetaObj)
204
205
206
        self._MetaObj = MetaObj

        # update meta_odict
207
        del self.meta_odict  # it is recreated if getter is used the next time
208
209
210

    @MetaObj.deleter
    def MetaObj(self):
211
212
213
214
215
        if hasattr(self, '_MetaObj') and self._MetaObj and hasattr(self._MetaObj, 'logger') and \
                self._MetaObj.logger not in [None, 'not set']:
            self._MetaObj.logger.close()
            self._MetaObj.logger = None

216
217
218
219
220
        self._MetaObj = None

    @property
    def meta_odict(self):
        if self._MetaObj:
221
            # if there is already a MetaObj -> create new meta_odict from it (ensures synchronization!)
222
223
224
225
            self._meta_odict = self._MetaObj.to_odict()
            del self.MetaObj
        elif not self._meta_odict:
            # if there is no MetaObj and no meta_odict -> use MetaObj getter to read metadata from disk
226
            pass  # reading from disk should use L1A_P.L1A_object.import_metadata -> so just return None
227
228
229
230
231
232
            self._meta_odict = None

        return self._meta_odict

    @meta_odict.setter
    def meta_odict(self, odict):
233
234
        assert isinstance(odict, (collections.OrderedDict, dict)), "'meta_odict' can only be set to an instance of " \
                                                                   "collections.OrderedDict. Got %s." % type(odict)
235
236
237
        self._meta_odict = odict

        # update MetaObj
238
        del self.MetaObj  # it is recreated if getter is used the next time
239
240
241
242
243

    @meta_odict.deleter
    def meta_odict(self):
        self._meta_odict = None

244
245
246
247
248
    @property
    def dict_LayerOptTherm(self):
        if self._dict_LayerOptTherm:
            return self._dict_LayerOptTherm
        elif self.LayerBandsAssignment:
249
            self._dict_LayerOptTherm = get_dict_LayerOptTherm(self.identifier, self.LayerBandsAssignment)
250
251
252
253
254
255
256
            return self._dict_LayerOptTherm
        else:
            return None

    @property
    def georef(self):
        """Returns True if the current dataset can serve as spatial reference."""
Daniel Scheffler's avatar
Daniel Scheffler committed
257

258
259
260
261
        return True if self.image_type == 'RSD' and re.search('OLI', self.sensor, re.I) else False

    @property
    def coreg_needed(self):
262
        if self._coreg_needed is None:
263
            self._coreg_needed = not (self.dataset_ID == CFG.datasetid_spatial_ref)
264
        return self._coreg_needed
265
266
267
268
269

    @coreg_needed.setter
    def coreg_needed(self, value):
        self._coreg_needed = value

270
271
272
273
274
275
276
277
278
279
280
281
282
283
    @property
    def coreg_info(self):
        if not self._coreg_info:
            self._coreg_info = {
                'corrected_shifts_px': {'x': 0, 'y': 0},
                'corrected_shifts_map': {'x': 0, 'y': 0},
                'original map info': self.meta_odict['map info'],
                'updated map info': None,
                'reference scene ID': None,
                'reference entity ID': None,
                'reference geotransform': None,
                # reference projection must be the own projection in order to avoid overwriting with a wrong EPSG
                'reference projection': self.meta_odict['coordinate system string'],
                'reference extent': {'rows': None, 'cols': None},
284
285
                'reference grid': [list(CFG.spatial_ref_gridx),
                                   list(CFG.spatial_ref_gridy)],
286
287
288
289
290
291
292
293
294
                'success': False
            }

        return self._coreg_info

    @coreg_info.setter
    def coreg_info(self, val):
        self._coreg_info = val

295
296
297
298
    @property
    def resamp_needed(self):
        if self._resamp_needed is None:
            gt = mapinfo2geotransform(self.meta_odict['map info'])
299
300
            self._resamp_needed = not is_coord_grid_equal(gt, CFG.spatial_ref_gridx,
                                                          CFG.spatial_ref_gridy)
301
302
303
304
305
306
307
308
        return self._resamp_needed

    @resamp_needed.setter
    def resamp_needed(self, value):
        self._resamp_needed = value

    @property
    def masks(self):
309
        # if self.mask_nodata is not None and self.mask_clouds is not None and \
310
311
312
313
        #     self._masks is not None and self._masks.bands==1:

        #     self.build_combined_masks_array()

314
315
316
        return self._masks

    @masks.setter
317
    def masks(self, *geoArr_initArgs):
318
319
320
321
        """
        NOTE: This does not automatically update mask_nodata and mask_clouds BUT if mask_nodata and mask_clouds are
        None their getters will automatically synchronize!
        """
Daniel Scheffler's avatar
Daniel Scheffler committed
322

323
        if geoArr_initArgs[0] is not None:
324
            self._masks = GeoArray(*geoArr_initArgs)
325
            self._masks.nodata = 0
326
327
            self._masks.gt = self.arr.gt
            self._masks.prj = self.arr.prj
328
329
        else:
            del self.masks
330

331
332
333
334
    @masks.deleter
    def masks(self):
        self._masks = None

335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
    @property
    def mask_clouds_confidence(self):
        return self._mask_clouds_confidence

    @mask_clouds_confidence.setter
    def mask_clouds_confidence(self, *geoArr_initArgs):
        if geoArr_initArgs[0] is not None:
            cnfArr = GeoArray(*geoArr_initArgs)

            assert cnfArr.shape == self.arr.shape[:2], \
                "The 'mask_clouds_confidence' GeoArray can only be instanced with an array of the same dimensions " \
                "like GMS_obj.arr. Got %s." % str(cnfArr.shape)

            if cnfArr._nodata is None:
                cnfArr.nodata = DEF_D.get_outFillZeroSaturated(cnfArr.dtype)[0]
            cnfArr.gt = self.arr.gt
            cnfArr.prj = self.arr.prj
352
            cnfArr.bandnames = ['confidence']
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

            self._mask_clouds_confidence = cnfArr
        else:
            del self._mask_clouds_confidence

    @mask_clouds_confidence.deleter
    def mask_clouds_confidence(self):
        self._mask_clouds_confidence = None

    @property
    def ac_errors(self):
        """Returns an instance of GeoArray containing error information calculated by the atmospheric correction.

        :return:
        """

        return self._ac_errors  # FIXME should give a warning if None

    @ac_errors.setter
    def ac_errors(self, *geoArr_initArgs):
        if geoArr_initArgs[0] is not None:
            errArr = GeoArray(*geoArr_initArgs)

            if CFG.ac_bandwise_accuracy:
                assert errArr.shape == self.arr.shape, \
                    "The 'ac_errors' GeoArray can only be instanced with an array of the same dimensions like " \
                    "GMS_obj.arr. Got %s." % str(errArr.shape)
            else:
                assert errArr.shape[:2] == self.arr.shape[:2], \
                    "The 'ac_errors' GeoArray can only be instanced with an array of the same X/Y dimensions like " \
                    "GMS_obj.arr. Got %s." % str(errArr.shape)

            if errArr._nodata is None:
                errArr.nodata = DEF_D.get_outFillZeroSaturated(errArr.dtype)[0]
            errArr.gt = self.arr.gt
            errArr.prj = self.arr.prj
389
            errArr.bandnames = self.LBA2bandnames(self.LayerBandsAssignment) if errArr.ndim == 3 else ['median']
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

            self._ac_errors = errArr
        else:
            del self.ac_errors

    @ac_errors.deleter
    def ac_errors(self):
        self._ac_errors = None

    @property
    def spec_homo_errors(self):
        """Returns an instance of GeoArray containing error information calculated during spectral homogenization.

        :return:
        """

        return self._spec_homo_errors  # FIXME should give a warning if None

    @spec_homo_errors.setter
    def spec_homo_errors(self, *geoArr_initArgs):
        if geoArr_initArgs[0] is not None:
            errArr = GeoArray(*geoArr_initArgs)

            if CFG.spechomo_bandwise_accuracy:
                assert errArr.shape == self.arr.shape, \
                    "The 'spec_homo_errors' GeoArray can only be instanced with an array of the same dimensions like " \
                    "GMS_obj.arr. Got %s." % str(errArr.shape)
            else:
                assert errArr.shape[:2] == self.arr.shape[:2], \
                    "The 'spec_homo_errors' GeoArray can only be instanced with an array of the same X/Y dimensions " \
                    "like GMS_obj.arr. Got %s." % str(errArr.shape)

            if errArr._nodata is None:
                errArr.nodata = DEF_D.get_outFillZeroSaturated(errArr.dtype)[0]
            errArr.gt = self.arr.gt
            errArr.prj = self.arr.prj
426
            errArr.bandnames = self.LBA2bandnames(self.LayerBandsAssignment) if errArr.ndim == 3 else ['median']
427
428
429
430
431
432
433
434
435
436
437

            self._spec_homo_errors = errArr
        else:
            del self.spec_homo_errors

    @spec_homo_errors.deleter
    def spec_homo_errors(self):
        self._spec_homo_errors = None

    @property
    def accuracy_layers(self):
438
439
440
441
        if not self._accuracy_layers:
            if not self.proc_level.startswith('L2'):
                self.logger.warning('Attempt to get %s accuracy layers failed - they are a Level 2 feature only.'
                                    % self.proc_level)
442

443
444
445
446
            self.logger.info('Generating combined accuracy layers array..')
            try:
                from ..algorithms.L2C_P import AccuracyCube
                self._accuracy_layers = AccuracyCube(self)
447
448
449
450

            except ValueError as e:
                if str(e) == 'The given GMS_object contains no accuracy layers for combination.':
                    if CFG.ac_estimate_accuracy or CFG.spechomo_estimate_accuracy:
451
452
                        self.logger.warning('The given GMS_object contains no accuracy layers although computation '
                                            'of accurracy layers was enabled in job configuration.')
453
454
455
456
457
                    else:
                        pass  # self._accuracy_layers keeps None
                else:
                    raise

458
459
            except Exception as e:
                raise RuntimeError('Failed to generate AccuracyCube!', e)
460

461
        return self._accuracy_layers
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482

    @accuracy_layers.setter
    def accuracy_layers(self, geoArr_initArgs):
        if geoArr_initArgs[0] is not None:
            acc_lay = GeoArray(geoArr_initArgs)
            assert acc_lay.shape[:2] == self.arr.shape[:2],\
                "The 'accuracy_layers' GeoArray can only be instanced with an array of the same dimensions like " \
                "GMS_obj.arr. Got %s." % str(acc_lay.shape)

            if acc_lay._nodata is None:
                acc_lay.nodata = DEF_D.get_outFillZeroSaturated(acc_lay.dtype)[0]
            acc_lay.gt = self.arr.gt
            acc_lay.prj = self.arr.prj

            if not acc_lay.bandnames:
                raise ValueError

            self._accuracy_layers = acc_lay
        else:
            del self._accuracy_layers

483
484
485
486
    @accuracy_layers.deleter
    def accuracy_layers(self):
        self._accuracy_layers = None

487
488
489
490
491
492
493
494
495
496
497
    @property
    def accuracy_layers_meta(self):
        if self._accuracy_layers is not None:
            return {'map info': geotransform2mapinfo(self._accuracy_layers.gt, self._accuracy_layers.projection),
                    'coordinate system string': self._accuracy_layers.projection,
                    'bands': self._accuracy_layers.bands,
                    'band names': list(self._accuracy_layers.bandnames),
                    'data ignore value': self._accuracy_layers.nodata}
        else:
            return None

498
499
500
    @property
    def cloud_masking_algorithm(self):
        if not self._cloud_masking_algorithm:
501
            self._cloud_masking_algorithm = CFG.cloud_masking_algorithm[self.satellite]
502
503
        return self._cloud_masking_algorithm

504
505
    @property
    def ac_options(self):
506
        # type: () -> dict
507
        """
508
509
        Returns the options dictionary needed as input for atmospheric correction. If an empty dictionary is returned,
        atmospheric correction is not yet available for the current sensor and will later be skipped.
510
        """
Daniel Scheffler's avatar
Daniel Scheffler committed
511

512
        if not self._ac_options:
513
            path_ac_options = CFG.path_custom_sicor_options or PG.get_path_ac_options(self.GMS_identifier)
514

515
            if path_ac_options and os.path.exists(path_ac_options):
516
517
                # don't validate because options contain pathes that do not exist on another server:
                opt_dict = get_ac_options(path_ac_options, validation=False)
518

Daniel Scheffler's avatar
Daniel Scheffler committed
519
                # update some file paths depending on the current environment
520
521
                opt_dict['DEM']['fn'] = CFG.path_dem_proc_srtm_90m
                opt_dict['ECMWF']['path_db'] = CFG.path_ECMWF_db
522
523
524
                opt_dict['S2Image'][
                    'S2_MSI_granule_path'] = None  # only a placeholder -> will always be None for GMS usage
                opt_dict['output'] = []  # outputs are not needed for GMS -> so
525
                opt_dict['report']['report_path'] = os.path.join(self.pathGen.get_path_procdata(), '[TYPE]')
526
                if 'uncertainties' in opt_dict:
527
528
529
530
                    if CFG.ac_estimate_accuracy:
                        opt_dict['uncertainties']['snr_model'] = PG.get_path_snr_model(self.GMS_identifier)
                    else:
                        del opt_dict['uncertainties']  # SICOR will not compute uncertainties if that key is missing
531

532
533
534
535
536
537
                # apply custom configuration
                opt_dict["logger"]['level'] = CFG.log_level
                opt_dict["ram"]['upper_limit'] = CFG.ac_max_ram_gb
                opt_dict["ram"]['unit'] = 'GB'
                opt_dict["AC"]['fill_nonclear_areas'] = CFG.ac_fillnonclear_areas
                opt_dict["AC"]['clear_area_labels'] = CFG.ac_clear_area_labels
538
                # opt_dict['AC']['n_cores'] = CFG.CPUs if CFG.allow_subMultiprocessing else 1
539

540
                self._ac_options = opt_dict
541
542
543
            else:
                self.logger.warning('There is no options file available for atmospheric correction. '
                                    'Atmospheric correction must be skipped.')
544

545
546
        return self._ac_options

547
    def get_copied_dict_and_props(self, remove_privates=False):
548
        # type: (bool) -> dict
549
        """Returns a copy of the current object dictionary including the current values of all object properties."""
550
551
552

        # loggers must be closed
        self.close_GMS_loggers()
553
554
        # this disables automatic recreation of loggers (otherwise loggers are created by using getattr()):
        self._loggers_disabled = True
555
556
557
558
559

        out_dict = self.__dict__.copy()

        # add properties
        property_names = [p for p in dir(self.__class__) if isinstance(getattr(self.__class__, p), property)]
560
        [out_dict.update({propK: copy.copy(getattr(self, propK))}) for propK in property_names]
561
562
563
564
565
566
567
568
569

        # remove private attributes
        if remove_privates:
            out_dict = {k: v for k, v in out_dict.items() if not k.startswith('_')}

        self._loggers_disabled = False  # enables automatic recreation of loggers

        return out_dict

570
571
    def attributes2dict(self, remove_privates=False):
        # type: (bool) -> dict
572
        """Returns a copy of the current object dictionary including the current values of all object properties."""
573
574
575

        # loggers must be closed
        self.close_GMS_loggers()
576
577
        # this disables automatic recreation of loggers (otherwise loggers are created by using getattr()):
        self._loggers_disabled = True
578
579
580
581

        out_dict = self.__dict__.copy()

        # add some selected property values
582
583
        for i in ['GMS_identifier', 'LayerBandsAssignment', 'coreg_needed', 'coreg_info', 'resamp_needed',
                  'dict_LayerOptTherm', 'georef', 'meta_odict']:
584
            out_dict[i] = getattr(self, i)
585
586
587
588
589

        # remove private attributes
        if remove_privates:
            out_dict = {k: v for k, v in out_dict.items() if not k.startswith('_')}

590
        self._loggers_disabled = False  # enables automatic recreation of loggers
591
592
        return out_dict

593
    def _data_downloader(self, sensor, entity_ID):
594
595
596
597
        self.logger.info('Data downloader started.')
        success = False
        " > download source code for Landsat here < "
        if not success:
598
599
            self.logger.critical(
                "Download for %s dataset '%s' failed. No further processing possible." % (sensor, entity_ID))
600
            raise RuntimeError('Archive download failed.')
601
602
        return success

603
604
605
606
607
    def from_disk(self, tuple_GMS_subset):
        """Fills an already instanced GMS object with data from disk. Excludes array attributes in Python mode.

        :param tuple_GMS_subset:    <tuple> e.g. ('/path/gms_file.gms', ['cube', None])
        """
608

609
        path_GMS_file = tuple_GMS_subset[0]
610
        GMSfileDict = INP_R.GMSfile2dict(path_GMS_file)
611
612

        # copy all attributes from GMS file (private attributes are not touched since they are not included in GMS file)
613
        self.meta_odict = GMSfileDict['meta_odict']  # set that first in order to make some getters and setters work
614
615
        for key, value in GMSfileDict.items():
            if key in ['GMS_identifier', 'georef', 'dict_LayerOptTherm']:
616
                continue  # properties that should better be created on the fly
617
618
            try:
                setattr(self, key, value)
619
620
            except Exception:
                raise AttributeError("Can't set attribute %s." % key)
621

622
        self.acq_datetime = datetime.datetime.strptime(self.acq_datetime, '%Y-%m-%d %H:%M:%S.%f%z')
623
624
        self.arr_shape, self.arr_pos = tuple_GMS_subset[1]

625
626
627
        self.arr = self.pathGen.get_path_imagedata()
        # self.mask_nodata and self.mask_clouds are auto-synchronized via self.masks (see their getters):
        self.masks = self.pathGen.get_path_maskdata()
628

629
630
        return copy.copy(self)

631
    def from_sensor_subsystems(self, list_GMS_objs):
632
633
        # type: (List[GMS_object]) -> GMS_object
        # TODO convert to classmethod
634
635
636
637
638
639
        """Merge separate GMS objects belonging to the same scene-ID into ONE GMS object.

        :param list_GMS_objs:   <list> of GMS objects covering the same geographic area but representing different
                                sensor subsystems (e.g. 3 GMS_objects for Sentinel-2 10m/20m/60m bands)
        """

640
        # assertions
641
642
        assert len(list_GMS_objs) > 1, "'GMS_object.from_sensor_subsystems()' expects multiple input GMS objects. " \
                                       "Got %d." % len(list_GMS_objs)
643
        assert all([is_coord_grid_equal(list_GMS_objs[0].arr.gt, *obj.arr.xygrid_specs) for obj in list_GMS_objs[1:]]),\
644
645
646
            "The input GMS objects must have the same pixel grid. Received: %s" \
            % np.array([obj.arr.xygrid_specs for obj in list_GMS_objs])
        assert len(list(set([GMS_obj.proc_level for GMS_obj in list_GMS_objs]))) == 1, \
647
648
649
            "The input GMS objects for GMS_object.from_sensor_subsystems() must have the same processing level."
        subsystems = [GMS_obj.subsystem for GMS_obj in list_GMS_objs]
        assert len(subsystems) == len(list(set(subsystems))), \
650
            "The input 'list_GMS_objs' contains duplicates: %s" % subsystems
651

652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
        ##################
        # merge logfiles #
        ##################

        # read all logs into DataFrame, sort it by the first column
        [GMS_obj.close_GMS_loggers() for GMS_obj in list_GMS_objs]  # close the loggers of the input objects
        paths_inLogs = [GMS_obj.pathGen.get_path_logfile() for GMS_obj in list_GMS_objs]
        allLogs_df = DataFrame()
        for log in paths_inLogs:
            df = read_csv(log, sep='\n', delimiter=':   ', header=None,
                          engine='python')  # engine suppresses a pandas warning
            allLogs_df = allLogs_df.append(
                df)  # FIXME this will log e.g. atm. corr 3 times for S2A -> use captured streams instead?

        allLogs_df = allLogs_df.sort_values(0)

        # set common metadata, needed for logfile
        self.baseN = list_GMS_objs[0].pathGen.get_baseN(merged_subsystems=True)
        self.path_logfile = list_GMS_objs[0].pathGen.get_path_logfile(merged_subsystems=True)
        self.scene_ID = list_GMS_objs[0].scene_ID

        # write the merged logfile and flush previous logger
        np.savetxt(self.path_logfile, np.array(allLogs_df), delimiter=':   ', fmt="%s")
        self.close_GMS_loggers()

677
        # log
678
679
        self.logger.info('Merging the subsystems %s to a single GMS object...'
                         % ', '.join([GMS_obj.subsystem for GMS_obj in list_GMS_objs]))
680
681

        # find the common extent. NOTE: boundsMap is expected in the order [xmin,xmax,ymin,ymax]
682
683
        geoExtents = np.array([GMS_obj.arr.box.boundsMap for GMS_obj in list_GMS_objs])
        common_extent = (min(geoExtents[:, 0]), max(geoExtents[:, 1]), min(geoExtents[:, 2]), max(geoExtents[:, 3]))
684

685
686
687
688
        ##################
        # MERGE METADATA #
        ##################

689
690
691
        # copy all attributes from the first input GMS file (private attributes are not touched)
        for key, value in list_GMS_objs[0].__dict__.copy().items():
            if key in ['GMS_identifier', 'georef', 'dict_LayerOptTherm']:
692
                continue  # properties that should better be created on the fly
693
694
            elif key in ['baseN', 'path_logfile', 'scene_ID', 'subsystem']:
                continue  # either previously set with common values or not needed for merged GMS_object
695
696
            try:
                setattr(self, key, value)
697
698
            except Exception:
                raise AttributeError("Can't set attribute %s." % key)
699

700
        # update LayerBandsAssignment and get full list of output bandnames
701
        from .metadata import get_LayerBandsAssignment
702
703
        # use identifier of first input GMS object for getting LBA (respects current proc_level):
        gms_idf = list_GMS_objs[0].GMS_identifier
704
        self.LayerBandsAssignment = get_LayerBandsAssignment(gms_idf, return_fullLBA=True)
705
        bandnames = ['B%s' % i if len(i) == 2 else 'B0%s' % i for i in self.LayerBandsAssignment]
706
707
708

        # update layer-dependent metadata with respect to remaining input GMS objects
        self.meta_odict.update({
709
710
711
712
            'band names': [('Band %s' % i) for i in self.LayerBandsAssignment],
            'LayerBandsAssignment': self.LayerBandsAssignment,
            'Subsystem': '',
            'PhysUnit': self.meta_odict['PhysUnit'],  # TODO can contain val_optical / val_thermal
713
714
        })
        self.subsystem = ''
715
716
        del self.pathGen  # must be refreshed because subsystem is now ''
        self.close_GMS_loggers()  # must also be refreshed because it depends on pathGen
717

718
719
        for attrN in ['SolIrradiance', 'CWL', 'FWHM', 'Offsets', 'OffsetsRef', 'Gains', 'GainsRef',
                      'ThermalConstK1', 'ThermalConstK2', 'ViewingAngle_arrProv', 'IncidenceAngle_arrProv']:
720
721
722
723
724
725
726

            # combine values from separate subsystems to a single value
            attrDic_fullLBA = {}
            for GMS_obj in list_GMS_objs:
                attr_val = getattr(GMS_obj.MetaObj, attrN)
                if isinstance(attr_val, list):
                    attrDic_fullLBA.update(dict(zip(GMS_obj.LayerBandsAssignment, attr_val)))
727
                elif isinstance(attr_val, (dict, collections.OrderedDict)):
728
729
730
731
732
733
734
                    attrDic_fullLBA.update(attr_val)
                else:
                    raise ValueError(attrN)

            # update the attribute in self.MetaObj
            if attrDic_fullLBA:
                val2set = [attrDic_fullLBA[bN] for bN in self.LayerBandsAssignment] \
735
                    if isinstance(getattr(list_GMS_objs[0].MetaObj, attrN), list) else attrDic_fullLBA
736
737
                setattr(self.MetaObj, attrN, val2set)

738
739
740
        ####################
        # MERGE ARRAY DATA #
        ####################
741

742
        # overwrite array data with merged arrays, clipped to common_extent and reordered according to FullLayerBandsAss
743
744
745
        for attrname in ['arr', 'ac_errors', 'dem', 'mask_nodata', 'mask_clouds', 'mask_clouds_confidence', 'masks']:

            # get current attribute of each subsystem without running property getters
746
            all_arrays = [getattr(GMS_obj, '_%s' % attrname) for GMS_obj in list_GMS_objs]
747
748
749
750
751
752
753
754
755

            # get the same geographical extent for each input GMS object
            if len(set(tuple(ext) for ext in geoExtents.tolist())) > 1:
                # in case of different extents
                geoArrs_same_extent = []

                for geoArr in all_arrays:

                    if geoArr is not None:
756
                        # FIXME mask_clouds_confidence is no GeoArray until here
757
                        # FIXME -> has no nodata value -> calculation throughs warning
758
759
                        geoArr_same_extent = \
                            GeoArray(*geoArr.get_mapPos(
760
761
762
763
                                mapBounds=np.array(common_extent)[[0, 2, 1, 3]],  # pass (xmin, ymin, xmax, ymax)
                                mapBounds_prj=geoArr.prj),
                                     bandnames=list(geoArr.bandnames.keys()),
                                     nodata=geoArr.nodata)
764
765
                        geoArrs_same_extent.append(geoArr_same_extent)
                    else:
766
767
                        # e.g. in case of cloud mask that is only appended to the GMS object with the same
                        # spatial resolution)
768
769
770
771
772
773
                        geoArrs_same_extent.append(None)

            else:
                # skip get_mapPos() if all input GMS objects have the same extent
                geoArrs_same_extent = all_arrays

774
775
            # validate output GeoArrays #
            #############################
776

777
778
            if len([gA for gA in geoArrs_same_extent if gA is not None]) > 1:
                equal_bounds = all([geoArrs_same_extent[0].box.boundsMap == gA.box.boundsMap
779
                                    for gA in geoArrs_same_extent[1:]])
780
781
                equal_epsg = all([geoArrs_same_extent[0].epsg == gA.epsg for gA in geoArrs_same_extent[1:]])
                equal_xydims = all([geoArrs_same_extent[0].shape[:2] == gA.shape[:2] for gA in geoArrs_same_extent[1:]])
782
783
784
785
786
                if not all([equal_bounds, equal_epsg, equal_xydims]):
                    raise RuntimeError('Something went wrong during getting the same geographical extent for all the '
                                       'input GMS objects. The extents, projections or pixel dimensions of the '
                                       'calculated input GMS objects are not equal.')

787
788
            # set output arrays #
            #####################
789

790
791
            # handle those arrays where bands have to be reordered according to FullLayerBandsAssignment
            if attrname in ['arr', 'ac_errors'] and list(set(geoArrs_same_extent)) != [None]:
792
793
                # check that each desired band name for the current attribute is provided by one of the input
                # GMS objects
794
795
                available_bandNs = list(chain.from_iterable([list(gA.bandnames) for gA in geoArrs_same_extent]))
                for bN in bandnames:
796
                    if bN not in available_bandNs:
797
                        raise ValueError("The given input GMS objects (subsystems) do not provide a bandname '%s' for "
798
799
                                         "the attribute '%s'. Available band names amongst all input GMS objects are: "
                                         "%s" % (bN, attrname, str(available_bandNs)))
800
801

                # merge arrays
802
803
                def get_band(bandN):
                    return [gA[bandN] for gA in geoArrs_same_extent if gA and bandN in gA.bandnames][0]
804
805
                full_geoArr = GeoArray(np.dstack((get_band(bandN) for bandN in bandnames)),
                                       geoArrs_same_extent[0].gt, geoArrs_same_extent[0].prj,
806
807
                                       bandnames=bandnames,
                                       nodata=geoArrs_same_extent[0].nodata)
808
809
                setattr(self, attrname, full_geoArr)

810
            # handle the remaining arrays
811
812
813
            else:
                # masks, dem, mask_nodata, mask_clouds, mask_clouds_confidence
                if attrname == 'dem':
814
815
                    # use the DEM of the first input object
                    # (if the grid is the same, the DEMs should be the same anyway)
816
                    self.dem = geoArrs_same_extent[0]
817

818
819
820
                elif attrname == 'mask_nodata':
                    # must not be merged -> self.arr is already merged, so just recalculate it (np.all)
                    self.mask_nodata = self.calc_mask_nodata(overwrite=True)
821

822
823
824
                elif attrname == 'mask_clouds':
                    # possibly only present in ONE subsystem (set by atm. Corr.)
                    mask_clouds = [msk for msk in geoArrs_same_extent if msk is not None]
825
826
                    if len(mask_clouds) > 1:
                        raise ValueError('Expected mask clouds in only one subsystem. Got %s.' % len(mask_clouds))
827
                    self.mask_clouds = mask_clouds[0] if mask_clouds else None
828

829
830
831
                elif attrname == 'mask_clouds_confidence':
                    # possibly only present in ONE subsystem (set by atm. Corr.)
                    mask_clouds_conf = [msk for msk in geoArrs_same_extent if msk is not None]
832
833
834
                    if len(mask_clouds_conf) > 1:
                        raise ValueError(
                            'Expected mask_clouds_conf in only one subsystem. Got %s.' % len(mask_clouds_conf))
835
                    self.mask_clouds_confidence = mask_clouds_conf[0] if mask_clouds_conf else None
836

837
                elif attrname == 'masks':
838
839
840
841
842
843
844
845
                    # self.mask_nodata and self.mask_clouds will already be set here -> so just recreate it from there
                    self.masks = None

        # recreate self.masks
        self.build_combined_masks_array()

        # update array-dependent metadata
        self.meta_odict.update({
846
847
            'samples': self.arr.cols, 'lines': self.arr.rows, 'bands': self.arr.bands,
            'map info': geotransform2mapinfo(self.arr.gt, self.arr.prj), 'coordinate system string': self.arr.prj, })
848
849

        # set shape of full array
850
        self.shape_fullArr = self.arr.shape
851
852
853

        return copy.copy(self)

854
855
856
857
858
859
    def from_tiles(self, list_GMS_tiles):
        # type: (list) -> self
        """Merge separate GMS objects with different spatial coverage but belonging to the same scene-ID to ONE GMS object.

        :param list_GMS_tiles: <list> of GMS objects that have been created by cut_GMS_obj_into_blocks()
        """
Daniel Scheffler's avatar
Daniel Scheffler committed
860

861
862
863
864
865
        if 'IMapUnorderedIterator' in str(type(list_GMS_tiles)):
            list_GMS_tiles = list(list_GMS_tiles)

        # copy all attributes except of array attributes
        tile1 = list_GMS_tiles[0]
866
867
        [setattr(self, i, getattr(tile1, i)) for i in tile1.__dict__
         if not callable(getattr(tile1, i)) and not isinstance(getattr(tile1, i), (np.ndarray, GeoArray))]
868
869

        # MERGE ARRAY-ATTRIBUTES
870
        list_arraynames = [i for i in tile1.__dict__ if not callable(getattr(tile1, i)) and
871
                           isinstance(getattr(tile1, i), (np.ndarray, GeoArray))]
872
873
        list_arraynames = ['_arr'] + [i for i in list_arraynames if
                                      i != '_arr']  # list must start with _arr, otherwise setters will not work
874
875
876
877

        for arrname in list_arraynames:
            samplearray = getattr(tile1, arrname)
            assert isinstance(samplearray, (np.ndarray, GeoArray)), \
878
                'Received a %s object for attribute %s. Expected a numpy array or an instance of GeoArray.' \
879
                % (type(samplearray), arrname)
880
881
            is_3d = samplearray.ndim == 3
            bands = (samplearray.shape[2],) if is_3d else ()  # dynamic -> works for arr, cld_arr,...
882
883
884
885
            target_shape = tuple(self.shape_fullArr[:2]) + bands
            target_dtype = samplearray.dtype
            merged_array = self._numba_array_merger(list_GMS_tiles, arrname, target_shape, target_dtype)

886
887
            setattr(self, arrname if not arrname.startswith('_') else arrname[1:],
                    merged_array)  # use setters if possible
888
889
890
891
            # NOTE: this asserts that each attribute starting with '_' has also a property with a setter!

        # UPDATE ARRAY-DEPENDENT ATTRIBUTES
        self.arr_shape = 'cube'
892
        self.arr_pos = None
893
894
895

        # update MetaObj attributes
        self.meta_odict.update({
896
897
            'samples': self.arr.cols, 'lines': self.arr.rows, 'bands': self.arr.bands,
            'map info': geotransform2mapinfo(self.arr.gt, self.arr.prj), 'coordinate system string': self.arr.prj, })
898
899
900
901

        # calculate data_corners_imXY (mask_nodata is always an array here because get_mapPos always returns an array)
        corners_imYX = calc_FullDataset_corner_positions(
            self.mask_nodata, assert_four_corners=False, algorithm='shapely')
902
        self.trueDataCornerPos = [(YX[1], YX[0]) for YX in corners_imYX]  # [UL, UR, LL, LR]
903
904
905
906
907
908

        # calculate trueDataCornerLonLat
        data_corners_LatLon = pixelToLatLon(self.trueDataCornerPos, geotransform=self.arr.gt, projection=self.arr.prj)
        self.trueDataCornerLonLat = [(YX[1], YX[0]) for YX in data_corners_LatLon]

        # calculate trueDataCornerUTM
909
910
        data_corners_utmYX = pixelToMapYX(self.trueDataCornerPos, geotransform=self.arr.gt,
                                          projection=self.arr.prj)  # FIXME asserts gt in UTM coordinates
911
912
913
914
        self.trueDataCornerUTM = [(YX[1], YX[0]) for YX in data_corners_utmYX]

        return copy.copy(self)

915
916
917
    @staticmethod
    @jit
    def _numba_array_merger(list_GMS_tiles, arrname2merge, target_shape, target_dtype):
Daniel Scheffler's avatar
Daniel Scheffler committed
918
        # type: (list, str, tuple, np.dtype) -> np.ndarray
919
920
921
922
923
924
925
926
927
        """
        private function, e.g. called by merge_GMS_tiles_to_GMS_obj() in order to fasten array merging

        :param list_GMS_tiles:
        :param arrname2merge:
        :param target_shape:
        :param target_dtype:
        :return:
        """
Daniel Scheffler's avatar
Daniel Scheffler committed
928

929
930
931
932
933
934
935
        out_arr = np.empty(target_shape, dtype=target_dtype)
        for idx, tile in enumerate(list_GMS_tiles):
            rowStart, rowEnd = tile.arr_pos[0]
            colStart, colEnd = tile.arr_pos[1]
            out_arr[rowStart:rowEnd + 1, colStart:colEnd + 1] = getattr(tile, arrname2merge)
        return out_arr

Daniel Scheffler's avatar
Daniel Scheffler committed
936
    def log_for_fullArr_or_firstTile(self, log_msg, subset=None):
937
938
939
940
941
942
943
        """Send a message to the logger only if full array or the first tile is currently processed.
        This function can be called when processing any tile but log message will only be sent from first tile.

        :param log_msg:  the log message to be logged
        :param subset:   subset argument as sent to e.g. DN2TOARadRefTemp that indicates which tile is to be processed.
                         Not needed if self.arr_pos is not None.
        """
Daniel Scheffler's avatar
Daniel Scheffler committed
944

945
946
947
948
        if subset is None and \
            (self.arr_shape == 'cube' or self.arr_pos is None or [self.arr_pos[0][0], self.arr_pos[1][0]] == [0, 0]) or\
                subset == ['cube', None] or (subset and [subset[1][0][0], subset[1][1][0]] == [0, 0]) or \
                hasattr(self, 'logAtThisTile') and getattr(self, 'logAtThisTile'):  # cube or 1st tile
Daniel Scheffler's avatar
Daniel Scheffler committed
949
            self.logger.info(log_msg)
950
951
952
953
        else:
            pass

    def apply_nodata_mask_to_ObjAttr(self, attrname, out_nodata_val=None):
954
        # type: (str,int) -> None
955
        """Applies self.mask_nodata to the specified array attribute by setting all values where mask_nodata is 0 to the
956
957
958
959
960
961
962
        given nodata value.

        :param attrname:         The attribute to apply the nodata mask to. Must be an array attribute or
                                 a string path to a previously saved ENVI-file.
        :param out_nodata_val:   set the values of the given attribute to this value.
        """

963
        assert hasattr(self, attrname)
964

965
        if getattr(self, attrname) is not None:
966

967
968
969
            if isinstance(getattr(self, attrname), str):
                update_spec_vals = True if attrname == 'arr' else False
                self.apply_nodata_mask_to_saved_ENVIfile(getattr(self, attrname), out_nodata_val, update_spec_vals)
970
            else:
971
                assert isinstance(getattr(self, attrname), (np.ndarray, GeoArray)), \
972
                    'L1A_obj.%s must be a numpy array or an instance of GeoArray. Got type %s.' \
973
974
                    % (attrname, type(getattr(self, attrname)))
                assert hasattr(self, 'mask_nodata') and self.mask_nodata is not None
975

976
                self.log_for_fullArr_or_firstTile('Applying nodata mask to L1A_object.%s...' % attrname)
977

978
                nodata_val = out_nodata_val if out_nodata_val else \
979
                    DEF_D.get_outFillZeroSaturated(getattr(self, attrname).dtype)[0]
980
                getattr(self, attrname)[self.mask_nodata.astype(np.int8) == 0] = nodata_val
981

982
983
                if attrname == 'arr':
                    self.MetaObj.spec_vals['fill'] = nodata_val
984
985
986
987
988
989

    def build_combined_masks_array(self):
        # type: () -> dict
        """Generates self.masks attribute (unsigned integer 8bit) from by concatenating all masks included in GMS obj.
        The corresponding metadata is assigned to L1A_obj.masks_meta. Empty mask attributes are skipped."""

990
        arrays2combine = [aN for aN in ['mask_nodata', 'mask_clouds']
991
                          if hasattr(self, aN) and isinstance(getattr(self, aN), (GeoArray, np.ndarray))]
992
993
        if arrays2combine:
            self.log_for_fullArr_or_firstTile('Combining masks...')
994
995

            def get_data(arrName): return getattr(self, arrName).astype(np.uint8)[:, :, None]
996
997

            for aN in arrays2combine:
998
                if False in np.equal(getattr(self, aN), getattr(self, aN).astype(np.uint8)):
999
1000
1001
                    warnings.warn('Some pixel values of attribute %s changed during data type '
                                  'conversion within build_combined_masks_array().')

1002
            # set self.masks
1003
1004
1005
            self.masks = get_data(arrays2combine[0]) if len(arrays2combine) == 1 else \
                np.concatenate([get_data(aN) for aN in arrays2combine], axis=2)
            self.masks.bandnames = arrays2combine  # set band names of GeoArray (allows later indexing by band name)
1006

1007
            # set self.masks_meta
1008
            nodataVal = DEF_D.get_outFillZeroSaturated(self.masks.dtype)[0]
1009
            self.masks_meta = {'map info': self.MetaObj.map_info, 'coordinate system string': self.MetaObj.projection,
1010
1011
                               'bands': len(arrays2combine), 'band names': arrays2combine,
                               'data ignore value': nodataVal}
1012
1013

            return {'desc': 'masks', 'row_start': 0, 'row_end': self.shape_fullArr[0],
1014
                    'col_start': 0, 'col_end': self.shape_fullArr[1], 'data': self.masks}  # usually not needed
1015
1016

    def apply_nodata_mask_to_saved_ENVIfile(self, path_saved_ENVIhdr, custom_nodata_val=None, update_spec_vals=False):
1017
        # type: (str,int,bool) -> None
1018
1019
        """Applies self.mask_nodata to a saved ENVI file with the same X/Y dimensions like self.mask_nodata by setting all
         values where mask_nodata is 0 to the given nodata value.
1020
1021
1022
1023
1024
1025
1026
1027

        :param path_saved_ENVIhdr:  <str> The path of the ENVI file to apply the nodata mask to.
        :param custom_nodata_val:   <int> set the values of the given attribute to this value.
        :param update_spec_vals:    <bool> whether to update self.MetaObj.spec_vals['fill']
        """

        self.log_for_fullArr_or_firstTile('Applying nodata mask to saved ENVI file...')
        assert os.path.isfile(path_saved_ENVIhdr)
1028
1029
1030
        assert hasattr(self, 'mask_nodata') and self.mask_nodata is not None
        if not path_saved_ENVIhdr.endswith('.hdr') and os.path.isfile(os.path.splitext(path_saved_ENVIhdr)[0] + '.hdr'):
            path_saved_ENVIhdr = os.path.splitext(path_saved_ENVIhdr)[0] + '.hdr'
1031
        if custom_nodata_val is None:
1032
            dtype_IDL = int(INP_R.read_ENVIhdr_to_dict(path_saved_ENVIhdr)['data type'])
1033
            nodata_val = DEF_D.get_outFillZeroSaturated(DEF_D.dtype_lib_IDL_Python[dtype_IDL])[0]