Context Navigation

source: branches/numpy/anuga/geospatial_data/geospatial_data.py @ 6428

Last change on this file since 6428 was 6428, checked in by rwilson, 15 years ago
numpy changes
File size: 69.4 KB

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

Note: See TracBrowser for help on using the repository browser.

Download in other formats: