5 |
# This program is free software under the GPL (>=v2) |
# This program is free software under the GPL (>=v2) |
6 |
# Read the file COPYING coming with Thuban for details. |
# Read the file COPYING coming with Thuban for details. |
7 |
|
|
8 |
|
""" |
9 |
|
Functions to generate Classifications |
10 |
|
""" |
11 |
|
|
12 |
|
__version__ = "$Revision$" |
13 |
|
# $Source$ |
14 |
|
# $Id$ |
15 |
|
|
16 |
import operator |
import operator |
17 |
|
|
18 |
from color import Color |
from color import Color |
20 |
from classification import Classification, ClassGroupSingleton, \ |
from classification import Classification, ClassGroupSingleton, \ |
21 |
ClassGroupRange, ClassGroupProperties |
ClassGroupRange, ClassGroupProperties |
22 |
|
|
23 |
class ClassGenerator: |
def generate_singletons(_list, numGroups, ramp): |
24 |
|
"""Generate a new classification consisting solely of singletons. |
25 |
|
|
26 |
def GenSingletonsFromList(self, list, numGroups, ramp): |
The resulting classification will consist of at most 'numGroups' |
27 |
"""Generate a new classification consisting solely of singletons. |
groups whose group properties ramp between 'prop1' and 'prop2'. There |
28 |
|
could be fewer groups if '_list' contains fewer that 'numGroups' items. |
29 |
|
|
30 |
The resulting classification will consist of at most 'numGroups' |
_list -- any object that implements the iterator interface |
|
groups whose group properties ramp between 'prop1' and 'prop2'. There |
|
|
could be fewer groups if 'list' contains fewer that 'numGroups' items. |
|
31 |
|
|
32 |
list -- any object that implements the iterator interface |
numGroups -- how many groups to generate. This can not be |
33 |
|
determined while the classification is being |
34 |
|
generated because the stepping values must |
35 |
|
be precalculated to ramp between prop1 and prop2. |
36 |
|
|
37 |
numGroups -- how many groups to generate. This can not be |
ramp -- an object which implements the CustomRamp interface |
38 |
determined while the classification is being |
""" |
|
generated because the stepping values must |
|
|
be precalculated to ramp between prop1 and prop2. |
|
39 |
|
|
40 |
prop1 -- initial group property values |
clazz = Classification() |
41 |
|
if numGroups == 0: return clazz |
42 |
|
|
43 |
prop2 -- final group property values |
ramp.SetNumGroups(numGroups) |
|
""" |
|
44 |
|
|
45 |
clazz = Classification() |
for value, prop in zip(_list, ramp): |
46 |
if numGroups == 0: return clazz |
clazz.AppendGroup(ClassGroupSingleton(value, prop)) |
47 |
|
|
48 |
ramp.SetNumGroups(numGroups) |
return clazz |
49 |
|
|
50 |
for value, prop in zip(list, ramp): |
def generate_uniform_distribution(min, max, numGroups, ramp, intStep = False): |
51 |
clazz.AppendGroup(ClassGroupSingleton(value, prop)) |
"""Generate a classification with numGroups range groups |
52 |
|
each with the same interval. |
53 |
|
|
54 |
return clazz |
intStep -- force the calculated stepping to an integer. |
55 |
|
Useful if the values are integers but the |
56 |
|
number of groups specified doesn't evenly |
57 |
|
divide (max - min). |
58 |
|
""" |
59 |
|
|
60 |
def GenSingletons(self, min, max, numGroups, ramp): |
clazz = Classification() |
61 |
|
if numGroups == 0: return clazz |
62 |
|
|
63 |
clazz = Classification() |
ramp.SetNumGroups(numGroups) |
64 |
|
|
65 |
#step = int((max - min) / float(numGroups)) |
cur_min = min |
66 |
|
|
67 |
if numGroups > 0: |
i = 1 |
68 |
|
end = "[" |
69 |
|
for prop in ramp: |
70 |
|
|
71 |
|
if intStep: |
72 |
|
cur_max = min + int(round((i * (max - min + 1)) / float(numGroups))) |
73 |
|
else: |
74 |
|
cur_max = min + (i * (max - min)) / float(numGroups) |
75 |
|
|
76 |
|
if i == numGroups: |
77 |
|
cur_max = max |
78 |
|
end = "]" |
79 |
|
|
80 |
|
if cur_min == cur_max: |
81 |
|
range = Range(("[", cur_min, cur_max, "]")) |
82 |
|
else: |
83 |
|
range = Range(("[", cur_min, cur_max, end)) |
84 |
|
|
85 |
|
clazz.AppendGroup(ClassGroupRange(range, None, prop)) |
86 |
|
|
87 |
|
cur_min = cur_max |
88 |
|
i += 1 |
89 |
|
|
90 |
|
return clazz |
91 |
|
|
92 |
|
|
93 |
|
def generate_quantiles(_list, percents, ramp, _range): |
94 |
|
"""Generates a Classification which has groups of ranges that |
95 |
|
represent quantiles of _list at the percentages given in percents. |
96 |
|
Only the values that fall within _range are considered. |
97 |
|
|
98 |
|
Returns a tuple (adjusted, Classification) where adjusted is |
99 |
|
True if the Classification does not exactly represent the given |
100 |
|
range, or if the Classification is empty. |
101 |
|
|
102 |
|
_list -- a sort list of values |
103 |
|
|
104 |
step = int((max - min + 1) / float(numGroups)) |
percents -- a sorted list of floats in the range 0.0-1.0 which |
105 |
cur_value = min |
represent the upper bound of each quantile. the |
106 |
|
union of all percentiles should be the entire |
107 |
|
range from 0.0-1.0 |
108 |
|
|
109 |
|
ramp -- an object which implements the CustomRamp interface |
110 |
|
|
111 |
|
_range -- a Range object |
112 |
|
|
113 |
|
Raises a Value Error if 'percents' has fewer than two items, or |
114 |
|
does not cover the entire range. |
115 |
|
""" |
116 |
|
|
117 |
|
clazz = Classification() |
118 |
|
quantiles = calculate_quantiles(_list, percents, _range) |
119 |
|
adjusted = True |
120 |
|
|
121 |
|
if quantiles is not None: |
122 |
|
|
123 |
|
numGroups = len(quantiles[3]) |
124 |
|
|
125 |
|
if numGroups != 0: |
126 |
|
|
127 |
|
adjusted = quantiles[0] |
128 |
|
|
129 |
ramp.SetNumGroups(numGroups) |
ramp.SetNumGroups(numGroups) |
130 |
|
|
131 |
for prop in ramp: |
start, min, endMax, right = _range.GetRange() |
|
clazz.AppendGroup(ClassGroupSingleton(cur_value), prop) |
|
|
cur_value += step |
|
132 |
|
|
133 |
return clazz |
oldp = 0 |
134 |
|
i = 1 |
135 |
|
end = "]" |
136 |
|
|
137 |
def GenUnifromDistribution(self, min, max, numGroups, |
for (q, p), prop in zip(quantiles[3], ramp): |
138 |
ramp, intStep = False): |
if i == numGroups: |
139 |
"""Generate a classification with numGroups range groups |
max = endMax |
140 |
each with the same interval. |
end = right |
141 |
|
else: |
142 |
|
max = _list[q] |
143 |
|
|
144 |
intStep -- force the calculated stepping to an integer. |
group = ClassGroupRange(Range((start, min, max, end)), |
145 |
Useful if the values are integers but the |
None, prop) |
146 |
number of groups specified doesn't evenly |
|
147 |
divide (max - min). |
group.SetLabel("%s%% - %s%%" % (round(oldp*100, 2), |
148 |
""" |
round(p*100, 2))) |
149 |
|
oldp = p |
150 |
|
start = "]" |
151 |
|
min = max |
152 |
|
clazz.AppendGroup(group) |
153 |
|
i += 1 |
154 |
|
|
155 |
clazz = Classification() |
return (adjusted, clazz) |
|
if numGroups == 0: return clazz |
|
156 |
|
|
157 |
ramp.SetNumGroups(numGroups) |
def GenQuantiles0(_list, percents, ramp, _range): |
158 |
|
"""Same as GenQuantiles, but the first class won't be added to |
159 |
|
the classification. |
160 |
|
|
161 |
step = (max - min) / float(numGroups) |
Returns a tuple (adjusted, Classification, upper_class0) where |
162 |
|
upper_class0 is the highest value inside the first class. |
163 |
|
|
164 |
if intStep: |
_list -- a sort list of values |
165 |
step = int(step) |
|
166 |
|
percents -- a sorted list of floats in the range 0.0-1.0 which |
167 |
|
represent the upper bound of each quantile. the |
168 |
|
union of all percentiles should be the entire |
169 |
|
range from 0.0-1.0 |
170 |
|
|
171 |
|
ramp -- an object which implements the CustomRamp interface |
172 |
|
|
173 |
|
_range -- a Range object |
174 |
|
|
175 |
|
Raises a Value Error if 'percents' has fewer than two items, or |
176 |
|
does not cover the entire range. |
177 |
|
""" |
178 |
|
|
179 |
|
clazz = Classification() |
180 |
|
quantiles = calculate_quantiles(_list, percents, _range) |
181 |
|
adjusted = True |
182 |
|
|
183 |
cur_min = min |
if quantiles is not None: |
|
cur_max = cur_min + step |
|
184 |
|
|
185 |
i = 0 |
numGroups = len(quantiles[3]) - 1 |
|
end = "[" |
|
|
for prop in ramp: |
|
186 |
|
|
187 |
if i == (numGroups - 1): |
if numGroups > 0: |
188 |
cur_max = max |
adjusted = quantiles[0] |
|
end = "]" |
|
189 |
|
|
190 |
|
ramp.SetNumGroups(numGroups) |
191 |
|
|
192 |
# this check guards against rounding issues |
start, min, endMax, right = _range.GetRange() |
|
if cur_min != cur_max: |
|
|
range = Range("[" + str(float(cur_min)) + ";" + |
|
|
str(float(cur_max)) + end) |
|
|
clazz.AppendGroup(ClassGroupRange(range, None, prop)) |
|
193 |
|
|
194 |
cur_min = cur_max |
class0 = quantiles[3][0] |
195 |
cur_max += step |
min = _list[class0[0]] |
196 |
i += 1 |
oldp = class0[1] |
197 |
|
i = 1 |
198 |
|
end = "]" |
199 |
|
|
200 |
|
for (q, p), prop in zip(quantiles[3][1:], ramp): |
201 |
|
if i == numGroups: |
202 |
|
max = endMax |
203 |
|
end = right |
204 |
|
else: |
205 |
|
max = _list[q] |
206 |
|
|
207 |
|
group = ClassGroupRange(Range((start, min, max, end)), |
208 |
|
None, prop) |
209 |
|
|
210 |
|
group.SetLabel("%s%% - %s%%" % (round(oldp*100, 2), |
211 |
|
round(p*100, 2))) |
212 |
|
oldp = p |
213 |
|
start = "]" |
214 |
|
min = max |
215 |
|
clazz.AppendGroup(group) |
216 |
|
i += 1 |
217 |
|
|
218 |
|
return (adjusted, clazz, _list[class0[0]]) |
219 |
|
|
220 |
|
|
221 |
|
def calculate_quantiles(_list, percents, _range): |
222 |
|
"""Calculate quantiles for the given _list of percents from the |
223 |
|
sorted list of values that are in range. |
224 |
|
|
225 |
|
This may not actually generate len(percents) quantiles if |
226 |
|
many of the values that fall on quantile borders are the same. |
227 |
|
|
228 |
|
Returns a tuple of the form: |
229 |
|
(adjusted, minIndex, maxIndex, [quantile_list]) |
230 |
|
|
231 |
|
where adjusted is True if the the quantile percentages differ from |
232 |
|
those supplied, minIndex is the index into _list where the |
233 |
|
minimum value used is located, maxIndex is the index into _list |
234 |
|
where the maximum value used is located, and quantile_list is a |
235 |
|
list of tuples of the form: (list_index, quantile_percentage) |
236 |
|
|
237 |
|
Returns None, if no quantiles could be generated based on the |
238 |
|
given range or input list. |
239 |
|
|
240 |
|
_list -- a sort list of values |
241 |
|
|
242 |
|
percents -- a sorted list of floats in the range 0.0-1.0 which |
243 |
|
represent the upper bound of each quantile. the |
244 |
|
union of all percentiles should be the entire |
245 |
|
range from 0.0-1.0 |
246 |
|
|
247 |
|
_range -- a Range object |
248 |
|
|
249 |
|
Raises a Value Error if 'percents' has fewer than two items, or |
250 |
|
does not cover the entire range. |
251 |
|
""" |
252 |
|
|
253 |
|
quantiles = [] |
254 |
|
adjusted = False |
255 |
|
|
256 |
|
if len(percents) <= 1: |
257 |
|
raise ValueError("percents parameter must have more than one item") |
258 |
|
|
259 |
|
if percents[len(percents) - 1] != 1.0: |
260 |
|
raise ValueError("percents does not cover the entire range") |
261 |
|
|
262 |
|
# |
263 |
|
# find what part of the _list range covers |
264 |
|
# |
265 |
|
minIndex = -1 |
266 |
|
maxIndex = -2 |
267 |
|
for i in xrange(0, len(_list), 1): |
268 |
|
if operator.contains(_range, _list[i]): |
269 |
|
minIndex = i |
270 |
|
break |
271 |
|
|
272 |
|
for i in xrange(len(_list)-1, -1, -1): |
273 |
|
if operator.contains(_range, _list[i]): |
274 |
|
maxIndex = i |
275 |
|
break |
276 |
|
|
277 |
return clazz |
numValues = maxIndex - minIndex + 1 |
278 |
|
|
279 |
|
if numValues > 0: |
280 |
|
|
281 |
def GenQuantiles(self, list, percents, ramp, _range): |
# |
282 |
clazz = Classification() |
# build a list of unique indices into list of where each |
283 |
quantiles = self.CalculateQuantiles(list, percents, _range) |
# quantile *should* be. set adjusted if the resulting |
284 |
numGroups = len(quantiles[1]) |
# indices are different |
285 |
if numGroups == 0: return clazz |
# |
286 |
|
quantiles = {} |
287 |
|
for p in percents: |
288 |
|
index = min(minIndex + int(p*numValues)-1, maxIndex) |
289 |
|
|
290 |
|
adjusted = adjusted \ |
291 |
|
or quantiles.has_key(index) \ |
292 |
|
or ((index - minIndex + 1) / float(numValues)) != p |
293 |
|
|
294 |
ramp.SetNumGroups(numGroups) |
quantiles[index] = 0 |
295 |
|
|
296 |
left, min, max, right = _range.GetRange() |
quantiles = quantiles.keys() |
297 |
|
quantiles.sort() |
298 |
|
|
299 |
start = "[" |
# |
300 |
oldp = 0 |
# the current quantile index must be strictly greater than |
301 |
for (q, p), prop in zip(quantiles[1], ramp): |
# the lowerBound |
302 |
max = list[q] |
# |
303 |
group = ClassGroupRange(Range(start + str(min) + ";" + |
lowerBound = minIndex - 1 |
|
str(max) + "]"), |
|
|
None, prop) |
|
304 |
|
|
305 |
group.SetLabel("%s%% - %s%%" % (round(oldp*100, 2), |
for qindex in xrange(len(quantiles)): |
306 |
round(p*100, 2))) |
if lowerBound >= maxIndex: |
307 |
oldp = p |
# discard higher quantiles |
308 |
start = "]" |
quantiles = quantiles[:qindex] |
309 |
min = max |
break |
|
clazz.AppendGroup(group) |
|
310 |
|
|
311 |
return (quantiles[0], clazz) |
# lowerBound + 1 is always a valid index |
312 |
|
|
313 |
def CalculateQuantiles(self, list, percents, _range): |
# |
314 |
"""Calculate quantiles for the given list of percents from the |
# bump up the current quantile index to be a usable index |
315 |
sorted list of values that are in range. |
# if it currently falls below the lowerBound |
316 |
|
# |
317 |
percents is a sorted list of floats in the range 0.0-1.0 |
if quantiles[qindex] <= lowerBound: |
318 |
|
quantiles[qindex] = lowerBound + 1 |
319 |
|
|
320 |
This may not actually generate numGroups quantiles if |
listIndex = quantiles[qindex] |
321 |
many of the values that fall on quantile borders are the same. |
value = _list[listIndex] |
322 |
|
|
323 |
Returns a tuple of the form: (adjusted, [quantile_list]) |
# |
324 |
|
# look for similar values around the quantile index |
325 |
|
# |
326 |
|
lindex = listIndex - 1 |
327 |
|
while lindex > lowerBound and value == _list[lindex]: |
328 |
|
lindex -= 1 |
329 |
|
lcount = (listIndex - 1) - lindex |
330 |
|
|
331 |
|
rindex = listIndex + 1 |
332 |
|
while rindex < maxIndex + 1 and value == _list[rindex]: |
333 |
|
rindex += 1 |
334 |
|
rcount = (listIndex + 1) - rindex |
335 |
|
|
|
where adjusted is true if the the quantile percentages differ from |
|
|
those supplied, and quantile_list is a list of tuples of the form: |
|
|
(list_index, quantile_percentage) |
|
|
""" |
|
|
|
|
|
quantiles = [] |
|
|
|
|
|
adjusted = False |
|
|
if len(percents) != 0: |
|
|
|
|
336 |
# |
# |
337 |
# find what part of the list range covers |
# adjust the current quantile index based on how many |
338 |
|
# numbers in the _list are the same as the current value |
339 |
# |
# |
340 |
minIndex = -1 |
newIndex = listIndex |
341 |
maxIndex = -2 |
if lcount == rcount: |
342 |
for i in xrange(0, len(list), 1): |
if lcount != 0: |
|
if operator.contains(_range, list[i]): |
|
|
minIndex = i |
|
|
break |
|
|
|
|
|
for i in xrange(len(list)-1, -1, -1): |
|
|
if operator.contains(_range, list[i]): |
|
|
maxIndex = i |
|
|
break; |
|
|
|
|
|
numValues = maxIndex - minIndex + 1 |
|
|
if minIndex <= maxIndex: |
|
|
|
|
|
# |
|
|
# build a list of unique indices into list of where each |
|
|
# quantile *should* be. set adjusted if the resulting |
|
|
# indices are different |
|
|
# |
|
|
quantiles = {} |
|
|
for p in percents: |
|
|
index = min(minIndex + int(p*numValues)-1, maxIndex) |
|
|
|
|
|
adjusted = adjusted \ |
|
|
or quantiles.has_key(index) \ |
|
|
or ((index - minIndex + 1) / float(numValues)) != p |
|
|
|
|
|
quantiles[index] = 0 |
|
|
|
|
|
quantiles = quantiles.keys() |
|
|
quantiles.sort() |
|
|
|
|
|
# |
|
|
# the current quantile index must be strictly greater than |
|
|
# the lowerBound |
|
|
# |
|
|
lowerBound = minIndex - 1 |
|
|
|
|
|
for qindex in range(len(quantiles)): |
|
|
if lowerBound >= maxIndex: |
|
|
# discard higher quantiles |
|
|
quantiles = quantiles[:qindex] |
|
|
break |
|
|
|
|
|
# lowerBound + 1 is always a valid index |
|
|
|
|
|
# |
|
|
# bump up the current quantile index to be a usable index |
|
|
# if it currently falls below the lowerBound |
|
343 |
# |
# |
344 |
if quantiles[qindex] <= lowerBound: |
# there are an equal number of numbers to the left |
345 |
quantiles[qindex] = min(lowerBound + 1, maxIndex) |
# and right, try going to the left first unless |
346 |
|
# doing so creates an empty quantile. |
|
listIndex = quantiles[qindex] |
|
|
value = list[quantiles[qindex]] |
|
|
|
|
|
# |
|
|
# look for similar values around the quantile index |
|
|
# |
|
|
lindex = listIndex - 1 |
|
|
lcount = 0 |
|
|
while lindex > lowerBound: |
|
|
if value != list[lindex]: break |
|
|
lcount += 1 |
|
|
lindex -= 1 |
|
|
|
|
|
rindex = listIndex + 1 |
|
|
rcount = 0 |
|
|
while rindex < maxIndex + 1: |
|
|
if value != list[rindex]: break |
|
|
rcount += 1 |
|
|
rindex += 1 |
|
|
|
|
347 |
# |
# |
348 |
# adjust the current quantile index based on how many |
if lindex != lowerBound: |
349 |
# numbers in the list are the same as the current value |
newIndex = lindex |
350 |
# |
else: |
|
newIndex = listIndex |
|
|
if lcount == rcount: |
|
|
if lcount != 0: |
|
|
# |
|
|
# there are an equal number of numbers to the left |
|
|
# and right, try going to the left first unless |
|
|
# doing so creates an empty quantile. |
|
|
# |
|
|
if lindex != lowerBound: |
|
|
newIndex = lindex |
|
|
else: |
|
|
newIndex = rindex - 1 |
|
|
|
|
|
elif lcount < rcount: |
|
|
# there are fewer items to the left, so |
|
|
# try going to the left first unless |
|
|
# doing so creates an empty quantile. |
|
|
if lindex != lowerBound: |
|
|
newIndex = lindex |
|
|
else: |
|
|
newIndex = rindex - 1 |
|
|
|
|
|
elif rcount < lcount: |
|
|
# there are fewer items to the right, so go to the right |
|
351 |
newIndex = rindex - 1 |
newIndex = rindex - 1 |
352 |
|
|
353 |
quantiles[qindex] = newIndex |
elif lcount < rcount: |
354 |
lowerBound = quantiles[qindex] |
# there are fewer items to the left, so |
355 |
|
# try going to the left first unless |
356 |
# |
# doing so creates an empty quantile. |
357 |
# since quantiles is only set if the code is at least a little |
if lindex != lowerBound: |
358 |
# successful, an empty list will be generated in the case that |
newIndex = lindex |
359 |
# we fail to get to the real body of the algorithm |
else: |
360 |
# |
newIndex = rindex - 1 |
361 |
return (adjusted, |
|
362 |
[(q, (q - minIndex+1) / float(numValues)) for q in quantiles]) |
elif rcount < lcount: |
363 |
|
# there are fewer items to the right, so go to the right |
364 |
|
newIndex = rindex - 1 |
365 |
|
|
366 |
|
adjusted = adjusted or newIndex != listIndex |
367 |
|
|
368 |
|
quantiles[qindex] = newIndex |
369 |
|
lowerBound = quantiles[qindex] |
370 |
|
|
371 |
|
if len(quantiles) == 0: |
372 |
|
return None |
373 |
|
else: |
374 |
|
return (adjusted, minIndex, maxIndex, |
375 |
|
[(q, (q - minIndex+1) / float(numValues)) \ |
376 |
|
for q in quantiles]) |
377 |
|
|
378 |
CLR = 0 |
CLR = 0 |
379 |
STEP = 1 |
STEP = 1 |
479 |
color = [color1.red * 255, |
color = [color1.red * 255, |
480 |
color1.green * 255, |
color1.green * 255, |
481 |
color1.blue * 255] |
color1.blue * 255] |
482 |
step = ((color2.red * 255 - color1.red * 255) / numGroups, |
step = ((color2.red * 255 - color1.red * 255) / numGroups, |
483 |
(color2.green * 255 - color1.green * 255) / numGroups, |
(color2.green * 255 - color1.green * 255) / numGroups, |
484 |
(color2.blue * 255 - color1.blue * 255) / numGroups) |
(color2.blue * 255 - color1.blue * 255) / numGroups) |
485 |
|
|
486 |
|
|
487 |
return (color, step) |
return (color, step) |
561 |
prop.SetFill(Color(clr[0], clr[1], clr[2])) |
prop.SetFill(Color(clr[0], clr[1], clr[2])) |
562 |
|
|
563 |
return prop |
return prop |
|
|
|
|
#class Colors16Ramp: |
|
|
# |
|
|
#def __iter__(self): |
|
|
#return self |
|
|
# |
|
|
#def GetRamp(self): |
|
|
#return self |
|
|
# |
|
|
#def SetNumGroups(self, num): |
|
|
#if num < 0: |
|
|
#return False |
|
|
# |
|
|
#self.index = 0 |
|
|
# |
|
|
#return True |
|