source: anuga_core/source/anuga/geospatial_data/geospatial_data.py @ 6438

Last change on this file since 6438 was 6438, checked in by ole, 15 years ago

Fixed bug in ensure_numeric where input was always copied irrespective of type.

Then fixed snowball bugs by copying input to ensure_numeric where appropriate.

Thanks Ross

File size: 70.5 KB
Line 
1"""Class Geospatial_data - Manipulation of locations on the planet and
2associated attributes.
3
4"""
5
6from sys import maxint
7from os import access, F_OK, R_OK,remove
8from types import DictType
9from warnings import warn
10from string import lower
11from RandomArray import randint, seed, get_seed
12from copy import deepcopy
13from Scientific.IO.NetCDF import NetCDFFile
14
15import Numeric as num
16
17from anuga.coordinate_transforms.lat_long_UTM_conversion import UTMtoLL
18from anuga.utilities.numerical_tools import ensure_numeric
19from anuga.coordinate_transforms.geo_reference import Geo_reference, \
20     TitleError, DEFAULT_ZONE, ensure_geo_reference, write_NetCDF_georeference
21from anuga.coordinate_transforms.redfearn import convert_from_latlon_to_utm
22from anuga.utilities.anuga_exceptions import ANUGAError
23from anuga.config import points_file_block_line_size as MAX_READ_LINES
24from anuga.config import netcdf_mode_r, netcdf_mode_w, netcdf_mode_a
25
26DEFAULT_ATTRIBUTE = 'elevation'
27
28
29##
30# @brief ??
31class Geospatial_data:
32
33    ##
34    # @brief
35    # @param data_points Mx2 iterable of tuples or array of x,y coordinates.
36    # @param attributes Associated values for each data point.
37    # @param geo_reference ??
38    # @param default_attribute_name ??
39    # @param file_name
40    # @param latitudes ??
41    # @param longitudes ??
42    # @param points_are_lats_longs True if points are lat/long, not UTM.
43    # @param max_read_lines Size of block to read, if blocking.
44    # @param load_file_now True if blocking but we eant to read file now.
45    # @param verbose True if this class instance is verbose.
46    def __init__(self,
47                 data_points=None, # this can also be a points file name
48                 attributes=None,
49                 geo_reference=None,
50                 default_attribute_name=None,
51                 file_name=None,
52                 latitudes=None,
53                 longitudes=None,
54                 points_are_lats_longs=False,
55                 max_read_lines=None,
56                 load_file_now=True,
57                 verbose=False):
58        """
59        Create instance from data points and associated attributes
60
61        data_points: x,y coordinates in meters. Type must be either a
62        sequence of 2-tuples or an Mx2 Numeric array of floats.  A file name
63        with extension .txt, .cvs or .pts can also be passed in here.
64
65        attributes: Associated values for each data point. The type
66        must be either a list or an array of length M or a dictionary
67        of lists (or arrays) of length M. In the latter case the keys
68        in the dictionary represent the attribute names, in the former
69        the attribute will get the default name "elevation".
70
71        geo_reference: Object representing the origin of the data
72        points. It contains UTM zone, easting and northing and data
73        points are assumed to be relative to this origin.
74        If geo_reference is None, the default geo ref object is used.
75
76        default_attribute_name: Name of default attribute to be used with
77        get_attribute_values. The idea is that the dataset can be
78        equipped with information about which attribute to return.
79        If None, the default is the "first"
80
81        latitudes, longitudes: Vectors of latitudes and longitudes,
82        used to specify location instead of points.
83
84        points_are_lats_longs: Set this as true if the points are actually
85        lats and longs, not UTM
86
87        max_read_lines: The number of rows read into memory when using
88        blocking to read a file.
89
90        load_file_now:  If true the file is automatically loaded
91        into the geospatial instance. Used when blocking.
92
93        file_name: Name of input netCDF file or .txt file. netCDF file must
94        have dimensions "points" etc.
95        .txt file is a comma seperated file with x, y and attribute
96        data.
97
98        The first line has the titles of the columns.  The first two
99        column titles are checked to see if they start with lat or
100        long (not case sensitive).  If so the data is assumed to be
101        latitude and longitude, in decimal format and converted to
102        UTM.  Otherwise the first two columns are assumed to be the x
103        and y, and the title names acually used are ignored.
104
105
106        The format for a .txt file is:
107            1st line:     [column names]
108            other lines:  x y [attributes]
109
110            for example:
111            x, y, elevation, friction
112            0.6, 0.7, 4.9, 0.3
113            1.9, 2.8, 5, 0.3
114            2.7, 2.4, 5.2, 0.3
115
116        The first two columns have to be x, y or lat, long
117        coordinates.
118
119
120        The format for a Points dictionary is:
121          ['pointlist'] a 2 column array describing points. 1st column x,
122          2nd column y.
123          ['attributelist'], a dictionary of 1D arrays, representing
124          attribute values at the point.  The dictionary key is the attribute
125          header.
126          ['geo_reference'] a Geo_refernece object. Use if the point
127          information is relative. This is optional.
128            eg
129            dic['pointlist'] = [[1.0,2.0],[3.0,5.0]]
130            dic['attributelist']['elevation'] = [[7.0,5.0]
131
132        verbose:
133
134        """
135
136        if isinstance(data_points, basestring):
137            # assume data point is really a file name
138            file_name = data_points
139
140        self.set_verbose(verbose)
141        self.geo_reference = None #create the attribute
142        self.file_name = file_name
143
144        if max_read_lines is None:
145            self.max_read_lines = MAX_READ_LINES
146        else:
147            self.max_read_lines = max_read_lines
148
149        if file_name is None:
150            if latitudes is not None \
151               or longitudes is not None \
152               or points_are_lats_longs:
153                data_points, geo_reference = \
154                    _set_using_lat_long(latitudes=latitudes,
155                                        longitudes=longitudes,
156                                        geo_reference=geo_reference,
157                                        data_points=data_points,
158                                        points_are_lats_longs=\
159                                            points_are_lats_longs)
160            self.check_data_points(data_points)
161            self.set_attributes(attributes)
162            self.set_geo_reference(geo_reference)
163            self.set_default_attribute_name(default_attribute_name)
164        elif load_file_now is True:
165            # watch for case where file name and points,
166            # attributes etc are provided!!
167            # if file name then all provided info will be removed!
168
169            if verbose is True:
170                if file_name is not None:
171                    print 'Loading Geospatial data from file: %s' % file_name
172
173            self.import_points_file(file_name, verbose=verbose)
174
175            self.check_data_points(self.data_points)
176            self.set_attributes(self.attributes)
177            self.set_geo_reference(self.geo_reference)
178            self.set_default_attribute_name(default_attribute_name)
179
180        if verbose is True:
181            if file_name is not None:
182                print 'Geospatial data created from file: %s' % file_name
183                if load_file_now is False:
184                    print 'Data will be loaded blockwise on demand'
185
186                    if file_name.endswith('csv') or file_name.endswith('txt'):
187                        pass
188                        # This message was misleading.
189                        # FIXME (Ole): Are we blocking here or not?
190                        #print 'ASCII formats are not that great for '
191                        #print 'blockwise reading. Consider storing this'
192                        #print 'data as a pts NetCDF format'
193
194    ##
195    # @brief Return length of the points set.
196    def __len__(self):
197        return len(self.data_points)
198
199    ##
200    # @brief Return a string representation of the points set.
201    def __repr__(self):
202        return str(self.get_data_points(absolute=True))
203
204    ##
205    # @brief Save points data in instance.
206    # @param data_points Points data to store in instance and check.
207    # @note Throws ValueError exception if no data.
208    def check_data_points(self, data_points):
209        """Checks data points
210        """
211
212        if data_points is None:
213            self.data_points = None
214            msg = 'There is no data or file provided!'
215            raise ValueError, msg
216
217        else:
218            self.data_points = ensure_numeric(data_points)
219            if not (0,) == self.data_points.shape:
220                assert len(self.data_points.shape) == 2
221                assert self.data_points.shape[1] == 2
222
223    ##
224    # @brief Check and assign attributes data.
225    # @param attributes Dictionary or scalar to save as .attributes.
226    # @note Throws exception if unable to convert dict keys to numeric.
227    def set_attributes(self, attributes):
228        """Check and assign attributes dictionary
229        """
230
231        if attributes is None:
232            self.attributes = None
233            return
234
235        if not isinstance(attributes, DictType):
236            #Convert single attribute into dictionary
237            attributes = {DEFAULT_ATTRIBUTE: attributes}
238
239        #Check input attributes
240        for key in attributes.keys():
241            try:
242                attributes[key] = ensure_numeric(attributes[key])
243            except:
244                msg = 'Attribute %s could not be converted' %key
245                msg += 'to a numeric vector'
246                raise msg
247
248        self.attributes = attributes
249
250    #def set_geo_reference(self, geo_reference):
251    #    # FIXME (Ole): Backwards compatibility - deprecate
252    #    self.setgeo_reference(geo_reference)
253
254    ##
255    # @brief Set the georeference of geospatial data.
256    # @param geo_reference The georeference data    to set.
257    # @note Will raise exception if param not instance of Geo_reference.
258    def set_geo_reference(self, geo_reference):
259        """Set the georeference of geospatial data.
260
261        It can also be used to change the georeference and will ensure that
262        the absolute coordinate values are unchanged.
263        """
264
265        from anuga.coordinate_transforms.geo_reference import Geo_reference
266
267        if geo_reference is None:
268            # Use default - points are in absolute coordinates
269            geo_reference = Geo_reference()
270
271        # Allow for tuple (zone, xllcorner, yllcorner)
272        geo_reference = ensure_geo_reference(geo_reference)
273
274        if not isinstance(geo_reference, Geo_reference):
275            # FIXME (Ole): This exception will be raised even
276            # if geo_reference is None. Is that the intent Duncan?
277            msg = 'Argument geo_reference must be a valid Geo_reference '
278            msg += 'object or None.'
279            raise msg
280
281        # If a geo_reference already exists, change the point data according to
282        # the new geo reference
283        if  self.geo_reference is not None:
284            self.data_points = self.get_data_points(geo_reference=geo_reference)
285
286        self.geo_reference = geo_reference
287
288    ##
289    # @brief Set default attribute name.
290    # @param default_attribute_name The default to save.
291    def set_default_attribute_name(self, default_attribute_name):
292        self.default_attribute_name = default_attribute_name
293
294    ##
295    # @brief Set the instance verbose flag.
296    # @param verbose The value to save.
297    # @note Will raise exception if param is not True or False.
298    def set_verbose(self, verbose=False):
299        if verbose in [False, True]:
300            self.verbose = verbose
301        else:
302            msg = 'Illegal value: %s' % str(verbose)
303            raise Exception, msg
304
305    ##
306    # @brief Clip geospatial data by a given polygon.
307    # @param polygon The polygon to clip with.
308    # @param closed True if points on clip boundary are not included in result.
309    # @param verbose True if this function is verbose.
310    def clip(self, polygon, closed=True, verbose=False):
311        """Clip geospatial data by a polygon
312
313        Input
314          polygon - Either a list of points, an Nx2 array or
315                    a Geospatial data object.
316          closed - (optional) determine whether points on boundary should be
317          regarded as belonging to the polygon (closed = True)
318          or not (closed = False). Default is True.
319
320        Output
321          New geospatial data object representing points inside
322          specified polygon.
323
324
325        Note - this method is non-destructive and leaves the data in 'self'
326        unchanged
327        """
328
329        from anuga.utilities.polygon import inside_polygon
330
331        if isinstance(polygon, Geospatial_data):
332            # Polygon is an object - extract points
333            polygon = polygon.get_data_points()
334
335        points = self.get_data_points()
336        inside_indices = inside_polygon(points, polygon, closed, verbose)
337
338        clipped_G = self.get_sample(inside_indices)
339
340        return clipped_G
341
342    ##
343    # @brief Clip points data by polygon, return points outside polygon.
344    # @param polygon The polygon to clip with.
345    # @param closed True if points on clip boundary are not included in result.
346    # @param verbose True if this function is verbose.
347    def clip_outside(self, polygon, closed=True, verbose=False):
348        """Clip geospatial date by a polygon, keeping data OUTSIDE of polygon
349
350        Input
351          polygon - Either a list of points, an Nx2 array or
352                    a Geospatial data object.
353          closed - (optional) determine whether points on boundary should be
354          regarded as belonging to the polygon (closed = True)
355          or not (closed = False). Default is True.
356
357        Output
358          Geospatial data object representing point OUTSIDE specified polygon
359        """
360
361        from anuga.utilities.polygon import outside_polygon
362
363        if isinstance(polygon, Geospatial_data):
364            # Polygon is an object - extract points
365            polygon = polygon.get_data_points()
366
367        points = self.get_data_points()
368        outside_indices = outside_polygon(points, polygon, closed,verbose)
369
370        clipped_G = self.get_sample(outside_indices)
371
372        return clipped_G
373
374    ##
375    # @brief Get instance geo_reference data.
376    def get_geo_reference(self):
377        return self.geo_reference
378
379    ##
380    # @brief Get coordinates for all data points as an Nx2 array.
381    # @param absolute If True, return UTM, else relative to xll/yll corners.
382    # @param geo_reference If supplied, points are relative to it.
383    # @param as_lat_long If True, return points as lat/lon.
384    # @param isSouthHemisphere If True, return lat/lon points in S.Hemi.
385    # @return A set of data points, in appropriate form.
386    def get_data_points(self, absolute=True, geo_reference=None,
387                        as_lat_long=False, isSouthHemisphere=True):
388        """Get coordinates for all data points as an Nx2 array
389
390        If absolute is False returned coordinates are relative to the
391        internal georeference's xll and yll corners, otherwise
392        absolute UTM coordinates are returned.
393
394        If a geo_reference is passed the points are returned relative
395        to that geo_reference.
396
397        isSH (isSouthHemisphere) is only used when getting data
398        points "as_lat_long" is True and if FALSE will return lats and
399        longs valid for the Northern Hemisphere.
400
401        Default: absolute is True.
402        """
403
404        if as_lat_long is True:
405            msg = "Points need a zone to be converted into lats and longs"
406            assert self.geo_reference is not None, msg
407            zone = self.geo_reference.get_zone()
408            assert self.geo_reference.get_zone() is not DEFAULT_ZONE, msg
409            lats_longs = []
410            for point in self.get_data_points(True):
411                # UTMtoLL(northing, easting, zone,
412                lat_calced, long_calced = UTMtoLL(point[1], point[0],
413                                                  zone, isSouthHemisphere)
414                lats_longs.append((lat_calced, long_calced)) # to hash
415            return lats_longs
416
417        if absolute is True and geo_reference is None:
418            return self.geo_reference.get_absolute(self.data_points)
419        elif geo_reference is not None:
420            return geo_reference.change_points_geo_ref(self.data_points,
421                                                       self.geo_reference)
422        else:
423            # If absolute is False
424            return self.data_points
425
426    ##
427    # @brief Get value for attribute name.
428    # @param attribute_name Name to get value for.
429    # @note If name passed is None, return default attribute value.
430    def get_attributes(self, attribute_name=None):
431        """Return values for one named attribute.
432
433        If attribute_name is None, default_attribute_name is used
434        """
435
436        if attribute_name is None:
437            if self.default_attribute_name is not None:
438                attribute_name = self.default_attribute_name
439            else:
440                attribute_name = self.attributes.keys()[0]
441                # above line takes the first one from keys
442
443        if self.verbose is True:
444            print 'Using attribute %s' %attribute_name
445            print 'Available attributes: %s' %(self.attributes.keys())
446
447        msg = 'Attribute name %s does not exist in data set' % attribute_name
448        assert self.attributes.has_key(attribute_name), msg
449
450        return self.attributes[attribute_name]
451
452    ##
453    # @brief Get all instance attributes.
454    # @return The instance attribute dictionary, or None if no attributes.
455    def get_all_attributes(self):
456        """Return values for all attributes.
457        The return value is either None or a dictionary (possibly empty).
458        """
459
460        return self.attributes
461
462    ##
463    # @brief Override __add__() to allow addition of geospatial objects.
464    # @param self This object.
465    # @param other The second object.
466    # @return The new geospatial object.
467    def __add__(self, other):
468        """Returns the addition of 2 geospatical objects,
469        objects are concatencated to the end of each other
470
471        NOTE: doesn't add if objects contain different
472        attributes
473
474        Always return absolute points!
475        This also means, that if you add None to the object,
476        it will be turned into absolute coordinates
477
478        other can be None in which case nothing is added to self.
479        """
480
481        # find objects zone and checks if the same
482        geo_ref1 = self.get_geo_reference()
483        zone1 = geo_ref1.get_zone()
484
485        if other is not None:
486            geo_ref2 = other.get_geo_reference()
487            zone2 = geo_ref2.get_zone()
488            geo_ref1.reconcile_zones(geo_ref2)
489            new_points = num.concatenate((self.get_data_points(absolute=True),
490                                          other.get_data_points(absolute=True)),
491                                         axis = 0)
492
493            # Concatenate attributes if any
494            if self.attributes is None:
495                if other.attributes is not None:
496                    msg = 'Geospatial data must have the same '
497                    msg += 'attributes to allow addition.'
498                    raise Exception, msg
499
500                new_attributes = None
501            else:
502                new_attributes = {}
503                for x in self.attributes.keys():
504                    if other.attributes.has_key(x):
505                        attrib1 = self.attributes[x]
506                        attrib2 = other.attributes[x]
507                        new_attributes[x] = num.concatenate((attrib1, attrib2))
508                    else:
509                        msg = 'Geospatial data must have the same \n'
510                        msg += 'attributes to allow addition.'
511                        raise Exception, msg
512        else:
513            #other is None:
514            new_points = self.get_data_points(absolute=True)
515            new_attributes = self.attributes
516
517        # Instantiate new data object and return absolute coordinates
518        new_geo_ref = Geo_reference(geo_ref1.get_zone(), 0.0, 0.0)
519        return Geospatial_data(new_points, new_attributes, new_geo_ref)
520
521    ##
522    # @brief Override the addition case where LHS isn't geospatial object.
523    # @param self This object.
524    # @param other The second object.
525    # @return The new geospatial object.
526    def __radd__(self, other):
527        """Handle cases like None + Geospatial_data(...)
528        """
529
530        return self + other
531
532    ###
533    #  IMPORT/EXPORT POINTS FILES
534    ###
535
536    ##
537    # @brief Import a .txt, .csv or .pts points data file.
538    # @param file_name
539    # @param delimiter
540    # @param verbose True if this function is to be verbose.
541    # @note Will throw IOError or SyntaxError if there is a problem.
542    def import_points_file(self, file_name, delimiter=None, verbose=False):
543        """ load an .txt, .csv or .pts file
544        Note: will throw an IOError/SyntaxError if it can't load the file.
545        Catch these!
546
547        Post condition: self.attributes dictionary has been set
548        """
549
550        if access(file_name, F_OK) == 0 :
551            msg = 'File %s does not exist or is not accessible' % file_name
552            raise IOError, msg
553
554        attributes = {}
555        if file_name[-4:]== ".pts":
556            try:
557                data_points, attributes, geo_reference = \
558                             _read_pts_file(file_name, verbose)
559            except IOError, e:
560                msg = 'Could not open file %s ' % file_name
561                raise IOError, msg
562        elif file_name[-4:]== ".txt" or file_name[-4:]== ".csv":
563            try:
564                data_points, attributes, geo_reference = \
565                             _read_csv_file(file_name, verbose)
566            except IOError, e:
567                # This should only be if a file is not found
568                msg = 'Could not open file %s. ' % file_name
569                msg += 'Check the file location.'
570                raise IOError, msg
571            except SyntaxError, e:
572                # This should only be if there is a format error
573                msg = 'Could not open file %s. \n' %file_name
574                msg += Error_message['IOError']
575                raise SyntaxError, msg
576        else:
577            msg = 'Extension %s is unknown' %file_name[-4:]
578            raise IOError, msg
579
580        self.data_points = data_points
581        self.attributes = attributes
582        self.geo_reference = geo_reference
583
584    ##
585    # @brief Write points data to a file (.csv or .pts).
586    # @param file_name Path to file to write.
587    # @param absolute ??
588    # @param as_lat_long ??
589    # @param isSouthHemisphere ??
590    def export_points_file(self, file_name, absolute=True,
591                           as_lat_long=False, isSouthHemisphere=True):
592
593        """
594        write a points file, file_name, as a text (.csv) or binary (.pts) file
595        file_name is the file name, including the extension
596        The point_dict is defined at the top of this file.
597
598        If absolute is True data the xll and yll are added to the points value
599        and the xll and yll of the geo_reference are set to 0.
600
601        If absolute is False data points at returned as relative to the xll
602        and yll and geo_reference remains uneffected
603
604        isSouthHemisphere: is only used when getting data
605        points "as_lat_long" is True and if FALSE will return lats and
606        longs valid for the Northern Hemisphere.
607        """
608
609        if (file_name[-4:] == ".pts"):
610            if absolute is True:
611                geo_ref = deepcopy(self.geo_reference)
612                geo_ref.xllcorner = 0
613                geo_ref.yllcorner = 0
614                _write_pts_file(file_name,
615                                self.get_data_points(absolute),
616                                self.get_all_attributes(),
617                                geo_ref)
618            else:
619                _write_pts_file(file_name,
620                                self.get_data_points(absolute),
621                                self.get_all_attributes(),
622                                self.get_geo_reference())
623
624        elif file_name[-4:] == ".txt" or file_name[-4:] == ".csv":
625            msg = "ERROR: trying to write a .txt file with relative data."
626            assert absolute, msg
627            _write_csv_file(file_name,
628                            self.get_data_points(absolute=True,
629                                                 as_lat_long=as_lat_long,
630                                  isSouthHemisphere=isSouthHemisphere),
631                            self.get_all_attributes(),
632                            as_lat_long=as_lat_long)
633
634        elif file_name[-4:] == ".urs" :
635            msg = "ERROR: Can not write a .urs file as a relative file."
636            assert absolute, msg
637            _write_urs_file(file_name,
638                            self.get_data_points(as_lat_long=True,
639                                  isSouthHemisphere=isSouthHemisphere))
640
641        else:
642            msg = 'Unknown file type %s ' %file_name
643            raise IOError, msg
644
645    ##
646    # @brief Get a subset of data that is referred to by 'indices'.
647    # @param indices A list of indices to select data subset with.
648    # @return A geospatial object containing data subset.
649    def get_sample(self, indices):
650        """ Returns a object which is a subset of the original
651        and the data points and attributes in this new object refer to
652        the indices provided
653
654        Input
655            indices- a list of integers that represent the new object
656        Output
657            New geospatial data object representing points specified by
658            the indices
659        """
660
661        #FIXME: add the geo_reference to this
662        points = self.get_data_points()
663        sampled_points = num.take(points, indices)
664
665        attributes = self.get_all_attributes()
666
667        sampled_attributes = {}
668        if attributes is not None:
669            for key, att in attributes.items():
670                sampled_attributes[key] = num.take(att, indices)
671
672        return Geospatial_data(sampled_points, sampled_attributes)
673
674    ##
675    # @brief Split one geospatial object into two.
676    # @param factor Relative size to make first result object.
677    # @param seed_num Random 'seed' - used only for unit test.
678    # @param verbose True if this function is to be verbose.
679    # @note Points in each result object are selected randomly.
680    def split(self, factor=0.5, seed_num=None, verbose=False):
681        """Returns two
682        geospatial_data object, first is the size of the 'factor'
683        smaller the original and the second is the remainder. The two
684        new object are disjoin set of each other.
685
686        Points of the two new object have selected RANDOMLY.
687
688        This method create two lists of indices which are passed into
689        get_sample.  The lists are created using random numbers, and
690        they are unique sets eg.  total_list(1,2,3,4,5,6,7,8,9)
691        random_list(1,3,6,7,9) and remainder_list(0,2,4,5,8)
692
693        Input -  the factor which to split the object, if 0.1 then 10% of the
694                 together object will be returned
695
696        Output - two geospatial_data objects that are disjoint sets of the
697                 original
698        """
699
700        i=0
701        self_size = len(self)
702        random_list = []
703        remainder_list = []
704        new_size = round(factor*self_size)
705
706        # Find unique random numbers
707        if verbose: print "make unique random number list and get indices"
708
709        total=num.array(range(self_size), num.Int)     #array default#
710        total_list = total.tolist()
711
712        if verbose: print "total list len", len(total_list)
713
714        # There will be repeated random numbers however will not be a
715        # problem as they are being 'pop'ed out of array so if there
716        # are two numbers the same they will pop different indicies,
717        # still basically random
718        ## create list of non-unquie random numbers
719        if verbose: print "create random numbers list %s long" %new_size
720
721        # Set seed if provided, mainly important for unit test!
722        # plus recalcule seed when no seed provided.
723        if seed_num != None:
724            seed(seed_num, seed_num)
725        else:
726            seed()
727
728        if verbose: print "seed:", get_seed()
729
730        random_num = randint(0, self_size-1, (int(new_size),))
731        random_num = random_num.tolist()
732
733        #need to sort and reverse so the pop() works correctly
734        random_num.sort()
735        random_num.reverse()
736
737        if verbose: print "make random number list and get indices"
738
739        j=0
740        k=1
741        remainder_list = total_list[:]
742
743        # pops array index (random_num) from remainder_list
744        # (which starts as the
745        # total_list and appends to random_list
746        random_num_len = len(random_num)
747        for i in random_num:
748            random_list.append(remainder_list.pop(i))
749            j += 1
750            #prints progress
751            if verbose and round(random_num_len/10*k) == j:
752                print '(%s/%s)' % (j, random_num_len)
753                k += 1
754
755        # FIXME: move to tests, it might take a long time
756        # then create an array of random lenght between 500 and 1000,
757        # and use a random factor between 0 and 1
758        # setup for assertion
759        test_total = random_list[:]
760        test_total.extend(remainder_list)
761        test_total.sort()
762        msg = 'The two random lists made from the original list when added ' \
763              'together DO NOT equal the original list'
764        assert total_list == test_total, msg
765
766        # Get new samples
767        if verbose: print "get values of indices for random list"
768        G1 = self.get_sample(random_list)
769        if verbose: print "get values of indices for opposite of random list"
770        G2 = self.get_sample(remainder_list)
771
772        return G1, G2
773
774    ##
775    # @brief Allow iteration over this object.
776    def __iter__(self):
777        """Read in the header, number_of_points and save the
778        file pointer position
779        """
780        from Scientific.IO.NetCDF import NetCDFFile
781        #FIXME - what to do if the file isn't there
782
783        # FIXME (Ole): Shouldn't this go into the constructor?
784        # This method acts like the constructor when blocking.
785        # ... and shouldn't it be called block_size?
786        #
787        if self.max_read_lines is None:
788            self.max_read_lines = MAX_READ_LINES
789
790        if self.file_name[-4:] == ".pts":
791            # See if the file is there.  Throw a QUIET IO error if it isn't
792            fd = open(self.file_name,'r')
793            fd.close()
794
795            # Throws prints to screen if file not present
796            self.fid = NetCDFFile(self.file_name, netcdf_mode_r)
797
798            (self.blocking_georef,
799             self.blocking_keys,
800             self.number_of_points) = _read_pts_file_header(self.fid,
801                                                            self.verbose)
802            self.start_row = 0
803            self.last_row = self.number_of_points
804            self.show_verbose = 0
805            self.verbose_block_size = (self.last_row + 10)/10
806            self.block_number = 0
807            self.number_of_blocks = self.number_of_points/self.max_read_lines
808            # This computes the number of full blocks. The last block may be
809            # smaller and won't be ircluded in this estimate.
810
811            if self.verbose is True:
812                print 'Reading %d points (in ~%d blocks) from file %s. ' \
813                      % (self.number_of_points,
814                         self.number_of_blocks,
815                         self.file_name),
816                print 'Each block consists of %d data points' \
817                      % self.max_read_lines
818
819        else:
820            # Assume the file is a csv file
821            file_pointer = open(self.file_name)
822            self.header, self.file_pointer = _read_csv_file_header(file_pointer)
823            self.blocking_georef = None # Used for reconciling zones
824
825        return self
826
827    ##
828    # @brief Read another block into the instance.
829    def next(self):
830        """read a block, instanciate a new geospatial and return it"""
831
832        if self.file_name[-4:] == ".pts":
833            if self.start_row == self.last_row:
834                # Read the end of the file last iteration
835                # Remove blocking attributes
836                self.fid.close()
837                del self.max_read_lines
838                del self.blocking_georef
839                del self.last_row
840                del self.start_row
841                del self.blocking_keys
842                del self.fid
843                raise StopIteration
844            fin_row = self.start_row + self.max_read_lines
845            if fin_row > self.last_row:
846                fin_row = self.last_row
847
848            if self.verbose is True:
849                if self.show_verbose >= self.start_row \
850                   and self.show_verbose < fin_row:
851                    print 'Reading block %d (points %d to %d) out of %d'\
852                          %(self.block_number,
853                            self.start_row,
854                            fin_row,
855                            self.number_of_blocks)
856
857                    self.show_verbose += max(self.max_read_lines,
858                                             self.verbose_block_size)
859
860
861            # Read next block
862            pointlist, att_dict, = _read_pts_file_blocking(self.fid,
863                                                           self.start_row,
864                                                           fin_row,
865                                                           self.blocking_keys)
866
867            geo = Geospatial_data(pointlist, att_dict, self.blocking_georef)
868            self.start_row = fin_row
869
870            self.block_number += 1
871
872        else:
873            # Assume the file is a csv file
874            try:
875                (pointlist,
876                 att_dict,
877                 geo_ref,
878                 self.file_pointer) = \
879                        _read_csv_file_blocking(self.file_pointer,
880                                                self.header[:],
881                                                max_read_lines=\
882                                                    self.max_read_lines,
883                                                verbose=self.verbose)
884
885                # Check that the zones haven't changed.
886                if geo_ref is not None:
887                    geo_ref.reconcile_zones(self.blocking_georef)
888                    self.blocking_georef = geo_ref
889                elif self.blocking_georef is not None:
890                    msg = 'Geo reference given, then not given.'
891                    msg += ' This should not happen.'
892                    raise ValueError, msg
893                geo = Geospatial_data(pointlist, att_dict, geo_ref)
894            except StopIteration:
895                self.file_pointer.close()
896                del self.header
897                del self.file_pointer
898                raise StopIteration
899            except ANUGAError:
900                self.file_pointer.close()
901                del self.header
902                del self.file_pointer
903                raise
904            except SyntaxError:
905                self.file_pointer.close()
906                del self.header
907                del self.file_pointer
908                # This should only be if there is a format error
909                msg = 'Could not open file %s. \n' % self.file_name
910                msg += Error_message['IOError']
911                raise SyntaxError, msg
912        return geo
913
914
915##################### Error messages ###########
916Error_message = {}
917Em = Error_message
918Em['IOError'] = "NOTE: The format for a comma separated .txt/.csv file is:\n"
919Em['IOError'] += "        1st line:     [column names]\n"
920Em['IOError'] += "        other lines:  [x value], [y value], [attributes]\n"
921Em['IOError'] += "\n"
922Em['IOError'] += "           for example:\n"
923Em['IOError'] += "           x, y, elevation, friction\n"
924Em['IOError'] += "           0.6, 0.7, 4.9, 0.3\n"
925Em['IOError'] += "           1.9, 2.8, 5, 0.3\n"
926Em['IOError'] += "           2.7, 2.4, 5.2, 0.3\n"
927Em['IOError'] += "\n"
928Em['IOError'] += "The first two columns are assumed to be x, y coordinates.\n"
929Em['IOError'] += "The attribute values must be numeric.\n"
930
931##
932# @brief ??
933# @param latitudes ??
934# @param longitudes ??
935# @param geo_reference ??
936# @param data_points ??
937# @param points_are_lats_longs ??
938def _set_using_lat_long(latitudes,
939                        longitudes,
940                        geo_reference,
941                        data_points,
942                        points_are_lats_longs):
943    """
944    if the points has lat long info, assume it is in (lat, long) order.
945    """
946
947    if geo_reference is not None:
948        msg = "A georeference is specified yet latitude and longitude " \
949              "are also specified!"
950        raise ValueError, msg
951
952    if data_points is not None and not points_are_lats_longs:
953        msg = "Data points are specified yet latitude and longitude are " \
954              "also specified."
955        raise ValueError, msg
956
957    if points_are_lats_longs:
958        if data_points is None:
959            msg = "Data points are not specified."
960            raise ValueError, msg
961        lats_longs = ensure_numeric(data_points)
962        latitudes = num.ravel(lats_longs[:,0:1])
963        longitudes = num.ravel(lats_longs[:,1:])
964
965    if latitudes is None and longitudes is None:
966        msg = "Latitudes and Longitudes are not specified."
967        raise ValueError, msg
968
969    if latitudes is None:
970        msg = "Longitudes are specified yet latitudes aren't."
971        raise ValueError, msg
972
973    if longitudes is None:
974        msg = "Latitudes are specified yet longitudes aren't."
975        raise ValueError, msg
976
977    data_points, zone  = convert_from_latlon_to_utm(latitudes=latitudes,
978                                                    longitudes=longitudes)
979    return data_points, Geo_reference(zone=zone)
980
981
982##
983# @brief Read a .pts data file.
984# @param file_name Path to file to read.
985# @param verbose True if this function is to be verbose.
986# @return (pointlist, attributes, geo_reference)
987def _read_pts_file(file_name, verbose=False):
988    """Read .pts NetCDF file
989
990    Return a dic of array of points, and dic of array of attribute
991    eg
992    dic['points'] = [[1.0,2.0],[3.0,5.0]]
993    dic['attributelist']['elevation'] = [[7.0,5.0]]
994    """
995
996    from Scientific.IO.NetCDF import NetCDFFile
997
998    if verbose: print 'Reading ', file_name
999
1000    # See if the file is there.  Throw a QUIET IO error if it isn't
1001    fd = open(file_name,'r')
1002    fd.close()
1003
1004    # Throws prints to screen if file not present
1005    fid = NetCDFFile(file_name, netcdf_mode_r)
1006
1007    pointlist = num.array(fid.variables['points'])
1008    keys = fid.variables.keys()
1009
1010    if verbose: print 'Got %d variables: %s' % (len(keys), keys)
1011
1012    try:
1013        keys.remove('points')
1014    except IOError, e:
1015        fid.close()
1016        msg = "Expected keyword 'points' but could not find it"
1017        raise IOError, msg
1018
1019    attributes = {}
1020    for key in keys:
1021        if verbose: print "reading attribute '%s'" % key
1022
1023        attributes[key] = num.array(fid.variables[key])
1024
1025    try:
1026        geo_reference = Geo_reference(NetCDFObject=fid)
1027    except AttributeError, e:
1028        geo_reference = None
1029
1030    fid.close()
1031
1032    return pointlist, attributes, geo_reference
1033
1034
1035##
1036# @brief Read a .csv data file.
1037# @param file_name Path to the .csv file to read.
1038# @param verbose True if this function is to be verbose.
1039def _read_csv_file(file_name, verbose=False):
1040    """Read .csv file
1041
1042    Return a dic of array of points, and dic of array of attribute
1043    eg
1044    dic['points'] = [[1.0,2.0],[3.0,5.0]]
1045    dic['attributelist']['elevation'] = [[7.0,5.0]]
1046    """
1047
1048    file_pointer = open(file_name)
1049    header, file_pointer = _read_csv_file_header(file_pointer)
1050    try:
1051        (pointlist,
1052         att_dict,
1053         geo_ref,
1054         file_pointer) = _read_csv_file_blocking(file_pointer,
1055                                                 header,
1056                                                 max_read_lines=1e30)
1057                                    #If the file is bigger that this, block..
1058                                    # FIXME (Ole) What's up here?
1059    except ANUGAError:
1060        file_pointer.close()
1061        raise
1062
1063    file_pointer.close()
1064
1065    return pointlist, att_dict, geo_ref
1066
1067
1068##
1069# @brief Read a .csv file header.
1070# @param file_pointer Open descriptor of the file to read.
1071# @param delimiter Header line delimiter string, split on this string.
1072# @param verbose True if this function is to be verbose.
1073# @return A tuple of (<cleaned header string>, <input file_pointer>)
1074
1075CSV_DELIMITER = ','
1076
1077def _read_csv_file_header(file_pointer,
1078                          delimiter=CSV_DELIMITER,
1079                          verbose=False):
1080    """Read the header of a .csv file
1081    Return a list of the header names
1082    """
1083
1084    line = file_pointer.readline()
1085    header = clean_line(line, delimiter)
1086
1087    return header, file_pointer
1088
1089##
1090# @brief Read a .csv file, with blocking.
1091# @param file_pointer Open descriptor of the file to read.
1092# @param header List of already read .csv header fields.
1093# @param delimiter Delimiter string header was split on.
1094# @param max_read_lines The max number of lines to read before blocking.
1095# @param verbose True if this function is to be verbose.
1096# @note Will throw IndexError, SyntaxError exceptions.
1097def _read_csv_file_blocking(file_pointer, header,
1098                            delimiter=CSV_DELIMITER,
1099                            max_read_lines=MAX_READ_LINES,
1100                            verbose=False):
1101    """Read the body of a .csv file.
1102    header: The list header of the csv file, with the x and y labels.
1103    """
1104
1105    points = []
1106    pointattributes = []
1107    att_dict = {}
1108
1109    # This is to remove the x and y headers.
1110    header = header[:]
1111    try:
1112        x_header = header.pop(0)
1113        y_header = header.pop(0)
1114    except IndexError:
1115        # if there are not two columns this will occur.
1116        # eg if it is a space seperated file
1117        raise SyntaxError
1118
1119    read_lines = 0
1120    while read_lines < max_read_lines:
1121        line = file_pointer.readline()
1122        numbers = clean_line(line, delimiter)
1123        if len(numbers) <= 1:
1124            break
1125        if line[0] == '#':
1126            continue
1127
1128        read_lines += 1
1129
1130        try:
1131            x = float(numbers[0])
1132            y = float(numbers[1])
1133            points.append([x,y])
1134            numbers.pop(0)
1135            numbers.pop(0)
1136            if len(header) != len(numbers):
1137                file_pointer.close()
1138                msg = "File load error. " \
1139                      "There might be a problem with the file header."
1140                raise SyntaxError, msg
1141            for i,n in enumerate(numbers):
1142                n.strip()
1143                if n != '\n' and n != '':
1144                    #attributes.append(float(n))
1145                    att_dict.setdefault(header[i],[]).append(float(n))
1146        except ValueError:
1147            raise SyntaxError
1148
1149    if points == []:
1150        raise StopIteration
1151
1152    pointlist = num.array(points, num.Float)
1153    for key in att_dict.keys():
1154        att_dict[key] = num.array(att_dict[key], num.Float)
1155
1156    # Do stuff here so the info is in lat's and longs
1157    geo_ref = None
1158    x_header = lower(x_header[:3])
1159    y_header = lower(y_header[:3])
1160    if (x_header == 'lon' or  x_header == 'lat') \
1161       and (y_header == 'lon' or  y_header == 'lat'):
1162        if x_header == 'lon':
1163            longitudes = num.ravel(pointlist[:,0:1])
1164            latitudes = num.ravel(pointlist[:,1:])
1165        else:
1166            latitudes = num.ravel(pointlist[:,0:1])
1167            longitudes = num.ravel(pointlist[:,1:])
1168
1169        pointlist, geo_ref = _set_using_lat_long(latitudes,
1170                                                 longitudes,
1171                                                 geo_reference=None,
1172                                                 data_points=None,
1173                                                 points_are_lats_longs=False)
1174
1175    return pointlist, att_dict, geo_ref, file_pointer
1176
1177
1178##
1179# @brief Read a .pts file header.
1180# @param fid Handle to the open .pts file.
1181# @param verbose True if the function is to be verbose.
1182# @return (geo_reference, keys, fid.dimensions['number_of_points'])
1183# @note Will throw IOError and AttributeError exceptions.
1184def _read_pts_file_header(fid, verbose=False):
1185    """Read the geo_reference and number_of_points from a .pts file
1186    """
1187
1188    keys = fid.variables.keys()
1189    try:
1190        keys.remove('points')
1191    except IOError, e:
1192        fid.close()
1193        msg = "Expected keyword 'points' but could not find it."
1194        raise IOError, msg
1195
1196    if verbose: print 'Got %d variables: %s' % (len(keys), keys)
1197
1198    try:
1199        geo_reference = Geo_reference(NetCDFObject=fid)
1200    except AttributeError, e:
1201        geo_reference = None
1202
1203    return geo_reference, keys, fid.dimensions['number_of_points']
1204
1205
1206##
1207# @brief Read the body of a .csf file, with blocking.
1208# @param fid Handle to already open file.
1209# @param start_row Start row index of points to return.
1210# @param fin_row End row index of points to return.
1211# @param keys Iterable of keys to return.
1212# @return Tuple of (pointlist, attributes).
1213def _read_pts_file_blocking(fid, start_row, fin_row, keys):
1214    """Read the body of a .csv file.
1215    """
1216
1217    pointlist = num.array(fid.variables['points'][start_row:fin_row])
1218
1219    attributes = {}
1220    for key in keys:
1221        attributes[key] = num.array(fid.variables[key][start_row:fin_row])
1222
1223    return pointlist, attributes
1224
1225
1226##
1227# @brief Write a .pts data file.
1228# @param file_name Path to the file to write.
1229# @param write_data_points Data points to write.
1230# @param write_attributes Attributes to write.
1231# @param write_geo_reference Georef to write.
1232def _write_pts_file(file_name,
1233                    write_data_points,
1234                    write_attributes=None,
1235                    write_geo_reference=None):
1236    """Write .pts NetCDF file
1237
1238    NOTE: Below might not be valid ask Duncan : NB 5/2006
1239
1240    WARNING: This function mangles the point_atts data structure
1241    #F??ME: (DSG)This format has issues.
1242    # There can't be an attribute called points
1243    # consider format change
1244    # method changed by NB not sure if above statement is correct
1245
1246    should create new test for this
1247    legal_keys = ['pointlist', 'attributelist', 'geo_reference']
1248    for key in point_atts.keys():
1249        msg = 'Key %s is illegal. Valid keys are %s' %(key, legal_keys)
1250        assert key in legal_keys, msg
1251    """
1252
1253    from Scientific.IO.NetCDF import NetCDFFile
1254
1255    # NetCDF file definition
1256    outfile = NetCDFFile(file_name, netcdf_mode_w)
1257
1258    # Create new file
1259    outfile.institution = 'Geoscience Australia'
1260    outfile.description = 'NetCDF format for compact and portable storage ' \
1261                          'of spatial point data'
1262
1263    # Dimension definitions
1264    shape = write_data_points.shape[0]
1265    outfile.createDimension('number_of_points', shape)
1266    outfile.createDimension('number_of_dimensions', 2) #This is 2d data
1267
1268    # Variable definition
1269    outfile.createVariable('points', num.Float, ('number_of_points',
1270                                                 'number_of_dimensions'))
1271
1272    #create variables
1273    outfile.variables['points'][:] = write_data_points #.astype(Float32)
1274
1275    if write_attributes is not None:
1276        for key in write_attributes.keys():
1277            outfile.createVariable(key, num.Float, ('number_of_points',))
1278            outfile.variables[key][:] = write_attributes[key] #.astype(Float32)
1279
1280    if write_geo_reference is not None:
1281        write_NetCDF_georeference(write_geo_reference, outfile)
1282
1283    outfile.close()
1284
1285
1286##
1287# @brief Write a .csv data file.
1288# @param file_name Path to the file to write.
1289# @param write_data_points Data points to write.
1290# @param write_attributes Attributes to write.
1291# @param as_lat_long True if points are lat/lon, else x/y.
1292# @param delimiter The CSV delimiter to use.
1293def _write_csv_file(file_name,
1294                    write_data_points,
1295                    write_attributes=None,
1296                    as_lat_long=False,
1297                    delimiter=','):
1298    """Write a .csv file.
1299    """
1300
1301    points = write_data_points
1302    pointattributes = write_attributes
1303
1304    fd = open(file_name, 'w')
1305
1306    if as_lat_long:
1307        titlelist = "latitude" + delimiter + "longitude"  + delimiter
1308    else:
1309        titlelist = "x" + delimiter + "y"  + delimiter
1310
1311    if pointattributes is not None:
1312        for title in pointattributes.keys():
1313            titlelist = titlelist + title + delimiter
1314        titlelist = titlelist[0:-len(delimiter)] # remove the last delimiter
1315
1316    fd.write(titlelist + "\n")
1317
1318    # <x/lat> <y/long> [attributes]
1319    for i, vert in enumerate( points):
1320        if pointattributes is not None:
1321            attlist = ","
1322            for att in pointattributes.keys():
1323                attlist = attlist + str(pointattributes[att][i]) + delimiter
1324            attlist = attlist[0:-len(delimiter)] # remove the last delimiter
1325            attlist.strip()
1326        else:
1327            attlist = ''
1328
1329        fd.write(str(vert[0]) + delimiter + str(vert[1]) + attlist + "\n")
1330
1331    fd.close()
1332
1333
1334##
1335# @brief Write a URS file.
1336# @param file_name The path of the file to write.
1337# @param points
1338# @param delimiter
1339def _write_urs_file(file_name, points, delimiter=' '):
1340    """Write a URS format file.
1341    export a file, file_name, with the urs format
1342    the data points are in lats and longs
1343    """
1344
1345    fd = open(file_name, 'w')
1346
1347    # first line is # points
1348    fd.write(str(len(points)) + "\n")
1349
1350    # <lat> <long> <id#>
1351    for i, vert in enumerate( points):
1352        fd.write(str(round(vert[0],7)) + delimiter +
1353                 str(round(vert[1],7)) + delimiter + str(i) + "\n")
1354
1355    fd.close()
1356
1357
1358##
1359# @brief ??
1360# @param point_atts ??
1361# @return ??
1362def _point_atts2array(point_atts):
1363    point_atts['pointlist'] = num.array(point_atts['pointlist'], num.Float)
1364
1365    for key in point_atts['attributelist'].keys():
1366        point_atts['attributelist'][key] = \
1367                num.array(point_atts['attributelist'][key], num.Float)
1368
1369    return point_atts
1370
1371
1372##
1373# @brief Convert geospatial object to a points dictionary.
1374# @param geospatial_data The geospatial object to convert.
1375# @return A points dictionary.
1376def geospatial_data2points_dictionary(geospatial_data):
1377    """Convert geospatial data to points_dictionary
1378    """
1379
1380    points_dictionary = {}
1381    points_dictionary['pointlist'] = geospatial_data.data_points
1382
1383    points_dictionary['attributelist'] = {}
1384
1385    for attribute_name in geospatial_data.attributes.keys():
1386        val = geospatial_data.attributes[attribute_name]
1387        points_dictionary['attributelist'][attribute_name] = val
1388
1389    points_dictionary['geo_reference'] = geospatial_data.geo_reference
1390
1391    return points_dictionary
1392
1393
1394##
1395# @brief Convert a points dictionary to a geospatial object.
1396# @param points_dictionary A points dictionary to convert.
1397def points_dictionary2geospatial_data(points_dictionary):
1398    """Convert points_dictionary to geospatial data object
1399    """
1400
1401    msg = 'Points dictionary must have key pointlist'
1402    assert points_dictionary.has_key('pointlist'), msg
1403
1404    msg = 'Points dictionary must have key attributelist'
1405    assert points_dictionary.has_key('attributelist'), msg
1406
1407    if points_dictionary.has_key('geo_reference'):
1408        geo = points_dictionary['geo_reference']
1409    else:
1410        geo = None
1411
1412    return Geospatial_data(points_dictionary['pointlist'],
1413                           points_dictionary['attributelist'],
1414                           geo_reference = geo)
1415
1416
1417##
1418# @brief Split a string into 'clean' fields.
1419# @param line The string to process.
1420# @param delimiter The delimiter string to split 'line' with.
1421# @return A list of 'cleaned' field strings.
1422# @note Any fields that were initially zero length will be removed.
1423def clean_line(line,delimiter):
1424    """Remove whitespace
1425    """
1426
1427    line = line.strip()                 # probably unnecessary RW
1428    numbers = line.split(delimiter)
1429
1430    i = len(numbers) - 1
1431    while i >= 0:
1432        if numbers[i] == '':
1433            numbers.pop(i)
1434        else:
1435            numbers[i] = numbers[i].strip()
1436        i += -1
1437
1438    return numbers
1439
1440
1441##
1442# @brief Ensure that points are in absolute coordinates.
1443# @param points A list or array of points to check, or geospatial object.
1444# @param geo_reference If supplied,
1445# @return ??
1446def ensure_absolute(points, geo_reference=None):
1447    """Ensure that points are in absolute coordinates.
1448
1449    This function inputs several formats and
1450    outputs one format. - a numeric array of absolute points.
1451
1452    Input formats are;
1453      points: List or numeric array of coordinate pairs [xi, eta] of
1454              points or geospatial object or points file name
1455
1456    mesh_origin: A geo_reference object or 3-tuples consisting of
1457                 UTM zone, easting and northing.
1458                 If specified vertex coordinates are assumed to be
1459                 relative to their respective origins.
1460    """
1461
1462    import copy
1463    # Input check
1464    if isinstance(points, basestring):
1465        #It's a string - assume it is a point file
1466        points = Geospatial_data(file_name=points)
1467
1468    if isinstance(points, Geospatial_data):
1469        points = points.get_data_points(absolute=True)
1470        msg = 'Use a Geospatial_data object or a mesh origin, not both.'
1471        assert geo_reference == None, msg
1472    else:
1473        points = ensure_numeric(copy.copy(points), num.Float)
1474
1475    # Sort of geo_reference and convert points
1476    if geo_reference is None:
1477        geo = None #Geo_reference()
1478    else:
1479        if isinstance(geo_reference, Geo_reference):
1480            geo = geo_reference
1481        else:
1482            geo = Geo_reference(geo_reference[0],
1483                                geo_reference[1],
1484                                geo_reference[2])
1485        points = geo.get_absolute(points)
1486
1487    return points
1488
1489
1490##
1491# @brief
1492# @param points
1493# @param geo_reference
1494# @return A geospatial object.
1495def ensure_geospatial(points, geo_reference=None):
1496    """
1497    This function inputs several formats and
1498    outputs one format. - a geospatial_data instance.
1499
1500    Inputed formats are;
1501    points:      List or numeric array of coordinate pairs [xi, eta] of
1502                 points or geospatial object
1503
1504    mesh_origin: A geo_reference object or 3-tuples consisting of
1505                 UTM zone, easting and northing.
1506                 If specified vertex coordinates are assumed to be
1507                 relative to their respective origins.
1508    """
1509
1510    import copy
1511    # Input check
1512    if isinstance(points, Geospatial_data):
1513        msg = "Use a Geospatial_data object or a mesh origin, not both."
1514        assert geo_reference is None, msg
1515        return points
1516    else:
1517        # List or numeric array of absolute points
1518        points = ensure_numeric(points, num.Float)
1519
1520    # Sort out geo reference
1521    if geo_reference is None:
1522        geo = None
1523    else:
1524        if isinstance(geo_reference, Geo_reference):
1525            geo = geo_reference
1526        else:
1527            geo = Geo_reference(geo_reference[0],
1528                                geo_reference[1],
1529                                geo_reference[2])
1530
1531    # Create Geospatial_data object with appropriate geo reference and return
1532    points = Geospatial_data(data_points=points, geo_reference=geo)
1533
1534    return points
1535
1536
1537##
1538# @brief
1539# @param data_file
1540# @param alpha_list
1541# @param mesh_file
1542# @param boundary_poly
1543# @param mesh_resolution
1544# @param north_boundary
1545# @param south_boundary
1546# @param east_boundary
1547# @param west_boundary
1548# @param plot_name
1549# @param split_factor
1550# @param seed_num
1551# @param cache
1552# @param verbose
1553def find_optimal_smoothing_parameter(data_file,
1554                                     alpha_list=None,
1555                                     mesh_file=None,
1556                                     boundary_poly=None,
1557                                     mesh_resolution=100000,
1558                                     north_boundary=None,
1559                                     south_boundary=None,
1560                                     east_boundary=None,
1561                                     west_boundary=None,
1562                                     plot_name='all_alphas',
1563                                     split_factor=0.1,
1564                                     seed_num=None,
1565                                     cache=False,
1566                                     verbose=False):
1567    """
1568    Removes a small random sample of points from 'data_file'. Then creates
1569    models with different alpha values from 'alpha_list' and cross validates
1570    the predicted value to the previously removed point data. Returns the
1571    alpha value which has the smallest covariance.
1572
1573    data_file: must not contain points outside the boundaries defined
1574               and it either a pts, txt or csv file.
1575
1576    alpha_list: the alpha values to test in a single list
1577
1578    mesh_file: name of the created mesh file or if passed in will read it.
1579               NOTE, if there is a mesh file mesh_resolution,
1580               north_boundary, south... etc will be ignored.
1581
1582    mesh_resolution: the maximum area size for a triangle
1583
1584    north_boundary... west_boundary: the value of the boundary
1585
1586    plot_name: the name for the plot contain the results
1587
1588    seed_num: the seed to the random number generator
1589
1590    USAGE:
1591        value, alpha = find_optimal_smoothing_parameter(data_file=fileName,
1592                                             alpha_list=[0.0001, 0.01, 1],
1593                                             mesh_file=None,
1594                                             mesh_resolution=3,
1595                                             north_boundary=5,
1596                                             south_boundary=-5,
1597                                             east_boundary=5,
1598                                             west_boundary=-5,
1599                                             plot_name='all_alphas',
1600                                             seed_num=100000,
1601                                             verbose=False)
1602
1603    OUTPUT: returns the minumum normalised covalance calculate AND the
1604           alpha that created it. PLUS writes a plot of the results
1605
1606    NOTE: code will not work if the data_file extent is greater than the
1607    boundary_polygon or any of the boundaries, eg north_boundary...west_boundary
1608    """
1609
1610    from anuga.shallow_water import Domain
1611    from anuga.geospatial_data.geospatial_data import Geospatial_data
1612    from anuga.pmesh.mesh_interface import create_mesh_from_regions
1613    from anuga.utilities.numerical_tools import cov
1614    from anuga.utilities.polygon import is_inside_polygon
1615    from anuga.fit_interpolate.benchmark_least_squares import mem_usage
1616
1617
1618    attribute_smoothed='elevation'
1619
1620    if mesh_file is None:
1621        if verbose: print "building mesh"
1622        mesh_file = 'temp.msh'
1623
1624        if north_boundary is None or south_boundary is None \
1625           or east_boundary is None or west_boundary is None:
1626            no_boundary = True
1627        else:
1628            no_boundary = False
1629
1630        if no_boundary is True:
1631            msg = 'All boundaries must be defined'
1632            raise msg
1633
1634        poly_topo = [[east_boundary,south_boundary],
1635                     [east_boundary,north_boundary],
1636                     [west_boundary,north_boundary],
1637                     [west_boundary,south_boundary]]
1638
1639        create_mesh_from_regions(poly_topo,
1640                                 boundary_tags={'back': [2],
1641                                                'side': [1,3],
1642                                                'ocean': [0]},
1643                                 maximum_triangle_area=mesh_resolution,
1644                                 filename=mesh_file,
1645                                 use_cache=cache,
1646                                 verbose=verbose)
1647
1648    else: # if mesh file provided
1649        #test mesh file exists?
1650        if verbose: "reading from file: %s" % mesh_file
1651        if access(mesh_file,F_OK) == 0:
1652            msg = "file %s doesn't exist!" % mesh_file
1653            raise IOError, msg
1654
1655    #split topo data
1656    if verbose: print 'Reading elevation file: %s' % data_file
1657    G = Geospatial_data(file_name = data_file)
1658    if verbose: print 'Start split'
1659    G_small, G_other = G.split(split_factor, seed_num, verbose=verbose)
1660    if verbose: print 'Finish split'
1661    points = G_small.get_data_points()
1662
1663    if verbose: print "Number of points in sample to compare: ", len(points)
1664
1665    if alpha_list == None:
1666        alphas = [0.001,0.01,100]
1667        #alphas = [0.000001, 0.00001, 0.0001, 0.001, 0.01,\
1668         #         0.1, 1.0, 10.0, 100.0,1000.0,10000.0]
1669    else:
1670        alphas = alpha_list
1671
1672    #creates array with columns 1 and 2 are x, y. column 3 is elevation
1673    #4 onwards is the elevation_predicted using the alpha, which will
1674    #be compared later against the real removed data
1675    data = num.array([], typecode=num.Float)
1676
1677    data=num.resize(data, (len(points), 3+len(alphas)))
1678
1679    #gets relative point from sample
1680    data[:,0] = points[:,0]
1681    data[:,1] = points[:,1]
1682    elevation_sample = G_small.get_attributes(attribute_name=attribute_smoothed)
1683    data[:,2] = elevation_sample
1684
1685    normal_cov=num.array(num.zeros([len(alphas), 2]), typecode=num.Float)
1686
1687    if verbose: print 'Setup computational domains with different alphas'
1688
1689    for i, alpha in enumerate(alphas):
1690        #add G_other data to domains with different alphas
1691        if verbose:
1692            print '\n Calculating domain and mesh for Alpha = ', alpha, '\n'
1693        domain = Domain(mesh_file, use_cache=cache, verbose=verbose)
1694        if verbose: print domain.statistics()
1695        domain.set_quantity(attribute_smoothed,
1696                            geospatial_data=G_other,
1697                            use_cache=cache,
1698                            verbose=verbose,
1699                            alpha=alpha)
1700
1701        # Convert points to geospatial data for use with get_values below
1702        points_geo = Geospatial_data(points, domain.geo_reference)
1703
1704        #returns the predicted elevation of the points that were "split" out
1705        #of the original data set for one particular alpha
1706        if verbose: print 'Get predicted elevation for location to be compared'
1707        elevation_predicted = \
1708                domain.quantities[attribute_smoothed].\
1709                    get_values(interpolation_points=points_geo)
1710
1711        #add predicted elevation to array that starts with x, y, z...
1712        data[:,i+3] = elevation_predicted
1713
1714        sample_cov = cov(elevation_sample)
1715        ele_cov = cov(elevation_sample - elevation_predicted)
1716        normal_cov[i,:] = [alpha,ele_cov / sample_cov]
1717
1718        if verbose:
1719            print 'Covariance for alpha ', normal_cov[i][0], '= ', \
1720                      normal_cov[i][1]
1721            print '-------------------------------------------- \n'
1722
1723    normal_cov0 = normal_cov[:,0]
1724    normal_cov_new = num.take(normal_cov, num.argsort(normal_cov0))
1725
1726    if plot_name is not None:
1727        from pylab import savefig, semilogx, loglog
1728
1729        semilogx(normal_cov_new[:,0], normal_cov_new[:,1])
1730        loglog(normal_cov_new[:,0], normal_cov_new[:,1])
1731        savefig(plot_name, dpi=300)
1732
1733    if mesh_file == 'temp.msh':
1734        remove(mesh_file)
1735
1736    if verbose:
1737        print 'Final results:'
1738        for i, alpha in enumerate(alphas):
1739            print 'covariance for alpha %s = %s ' \
1740                  % (normal_cov[i][0], normal_cov[i][1])
1741        print '\n Optimal alpha is: %s ' \
1742              % normal_cov_new[(num.argmin(normal_cov_new, axis=0))[1], 0]
1743
1744    # covariance and optimal alpha
1745    return (min(normal_cov_new[:,1]),
1746            normal_cov_new[(num.argmin(normal_cov_new,axis=0))[1],0])
1747
1748
1749##
1750# @brief
1751# @param data_file
1752# @param alpha_list
1753# @param mesh_file
1754# @param boundary_poly
1755# @param mesh_resolution
1756# @param north_boundary
1757# @param south_boundary
1758# @param east_boundary
1759# @param west_boundary
1760# @param plot_name
1761# @param split_factor
1762# @param seed_num
1763# @param cache
1764# @param verbose
1765def old_find_optimal_smoothing_parameter(data_file,
1766                                         alpha_list=None,
1767                                         mesh_file=None,
1768                                         boundary_poly=None,
1769                                         mesh_resolution=100000,
1770                                         north_boundary=None,
1771                                         south_boundary=None,
1772                                         east_boundary=None,
1773                                         west_boundary=None,
1774                                         plot_name='all_alphas',
1775                                         split_factor=0.1,
1776                                         seed_num=None,
1777                                         cache=False,
1778                                         verbose=False):
1779    """
1780    data_file: must not contain points outside the boundaries defined
1781               and it either a pts, txt or csv file.
1782
1783    alpha_list: the alpha values to test in a single list
1784
1785    mesh_file: name of the created mesh file or if passed in will read it.
1786               NOTE, if there is a mesh file mesh_resolution,
1787               north_boundary, south... etc will be ignored.
1788
1789    mesh_resolution: the maximum area size for a triangle
1790
1791    north_boundary... west_boundary: the value of the boundary
1792
1793    plot_name: the name for the plot contain the results
1794
1795    seed_num: the seed to the random number generator
1796
1797    USAGE:
1798        value, alpha = find_optimal_smoothing_parameter(data_file=fileName,
1799                                             alpha_list=[0.0001, 0.01, 1],
1800                                             mesh_file=None,
1801                                             mesh_resolution=3,
1802                                             north_boundary=5,
1803                                             south_boundary=-5,
1804                                             east_boundary=5,
1805                                             west_boundary=-5,
1806                                             plot_name='all_alphas',
1807                                             seed_num=100000,
1808                                             verbose=False)
1809
1810    OUTPUT: returns the minumum normalised covalance calculate AND the
1811            alpha that created it. PLUS writes a plot of the results
1812
1813    NOTE: code will not work if the data_file extend is greater than the
1814          boundary_polygon or the north_boundary...west_boundary
1815    """
1816
1817    from anuga.shallow_water import Domain
1818    from anuga.geospatial_data.geospatial_data import Geospatial_data
1819    from anuga.pmesh.mesh_interface import create_mesh_from_regions
1820    from anuga.utilities.numerical_tools import cov
1821    from anuga.utilities.polygon import is_inside_polygon
1822    from anuga.fit_interpolate.benchmark_least_squares import mem_usage
1823
1824
1825    attribute_smoothed = 'elevation'
1826
1827    if mesh_file is None:
1828        mesh_file = 'temp.msh'
1829
1830        if north_boundary is None or south_boundary is None \
1831           or east_boundary is None or west_boundary is None:
1832            no_boundary = True
1833        else:
1834            no_boundary = False
1835
1836        if no_boundary is True:
1837            msg = 'All boundaries must be defined'
1838            raise msg
1839
1840        poly_topo = [[east_boundary,south_boundary],
1841                     [east_boundary,north_boundary],
1842                     [west_boundary,north_boundary],
1843                     [west_boundary,south_boundary]]
1844
1845        create_mesh_from_regions(poly_topo,
1846                                 boundary_tags={'back': [2],
1847                                                'side': [1,3],
1848                                                'ocean': [0]},
1849                                 maximum_triangle_area=mesh_resolution,
1850                                 filename=mesh_file,
1851                                 use_cache=cache,
1852                                 verbose=verbose)
1853
1854    else: # if mesh file provided
1855        #test mesh file exists?
1856        if access(mesh_file,F_OK) == 0:
1857            msg = "file %s doesn't exist!" % mesh_file
1858            raise IOError, msg
1859
1860    #split topo data
1861    G = Geospatial_data(file_name=data_file)
1862    if verbose: print 'start split'
1863    G_small, G_other = G.split(split_factor, seed_num, verbose=verbose)
1864    if verbose: print 'finish split'
1865    points = G_small.get_data_points()
1866
1867    #FIXME: Remove points outside boundary polygon
1868#    print 'new point',len(points)
1869#
1870#    new_points=[]
1871#    new_points=array([],typecode=Float)
1872#    new_points=resize(new_points,(len(points),2))
1873#    print "BOUNDARY", boundary_poly
1874#    for i,point in enumerate(points):
1875#        if is_inside_polygon(point,boundary_poly, verbose=True):
1876#            new_points[i] = point
1877#            print"WOW",i,new_points[i]
1878#    points = new_points
1879
1880    if verbose: print "Number of points in sample to compare: ", len(points)
1881
1882    if alpha_list == None:
1883        alphas = [0.001,0.01,100]
1884        #alphas = [0.000001, 0.00001, 0.0001, 0.001, 0.01,\
1885         #         0.1, 1.0, 10.0, 100.0,1000.0,10000.0]
1886    else:
1887        alphas = alpha_list
1888
1889    domains = {}
1890
1891    if verbose: print 'Setup computational domains with different alphas'
1892
1893    for alpha in alphas:
1894        #add G_other data to domains with different alphas
1895        if verbose:
1896            print '\n Calculating domain and mesh for Alpha = ', alpha, '\n'
1897        domain = Domain(mesh_file, use_cache=cache, verbose=verbose)
1898        if verbose: print domain.statistics()
1899        domain.set_quantity(attribute_smoothed,
1900                            geospatial_data=G_other,
1901                            use_cache=cache,
1902                            verbose=verbose,
1903                            alpha=alpha)
1904        domains[alpha] = domain
1905
1906    #creates array with columns 1 and 2 are x, y. column 3 is elevation
1907    #4 onwards is the elevation_predicted using the alpha, which will
1908    #be compared later against the real removed data
1909    data = num.array([], typecode=num.Float)
1910
1911    data = num.resize(data, (len(points), 3+len(alphas)))
1912
1913    #gets relative point from sample
1914    data[:,0] = points[:,0]
1915    data[:,1] = points[:,1]
1916    elevation_sample = G_small.get_attributes(attribute_name=attribute_smoothed)
1917    data[:,2] = elevation_sample
1918
1919    normal_cov = num.array(num.zeros([len(alphas), 2]), typecode=num.Float)
1920
1921    if verbose:
1922        print 'Determine difference between predicted results and actual data'
1923    for i, alpha in enumerate(domains):
1924        if verbose: print'Alpha =', alpha
1925
1926        points_geo = domains[alpha].geo_reference.change_points_geo_ref(points)
1927        #returns the predicted elevation of the points that were "split" out
1928        #of the original data set for one particular alpha
1929        elevation_predicted = \
1930                domains[alpha].quantities[attribute_smoothed].\
1931                        get_values(interpolation_points=points_geo)
1932
1933        #add predicted elevation to array that starts with x, y, z...
1934        data[:,i+3] = elevation_predicted
1935
1936        sample_cov = cov(elevation_sample)
1937        ele_cov = cov(elevation_sample - elevation_predicted)
1938        normal_cov[i,:] = [alpha,ele_cov / sample_cov]
1939        print 'memory usage during compare', mem_usage()
1940        if verbose: print 'cov', normal_cov[i][0], '= ', normal_cov[i][1]
1941
1942    normal_cov0 = normal_cov[:,0]
1943    normal_cov_new = num.take(normal_cov, num.argsort(normal_cov0))
1944
1945    if plot_name is not None:
1946        from pylab import savefig,semilogx,loglog
1947
1948        semilogx(normal_cov_new[:,0], normal_cov_new[:,1])
1949        loglog(normal_cov_new[:,0], normal_cov_new[:,1])
1950        savefig(plot_name, dpi=300)
1951    if mesh_file == 'temp.msh':
1952        remove(mesh_file)
1953
1954    return (min(normal_cov_new[:,1]),
1955            normal_cov_new[(num.argmin(normal_cov_new, axis=0))[1],0])
1956
1957
1958if __name__ == "__main__":
1959    pass
1960
Note: See TracBrowser for help on using the repository browser.