aboutsummaryrefslogtreecommitdiffstats
path: root/csv_tool
diff options
context:
space:
mode:
Diffstat (limited to 'csv_tool')
-rwxr-xr-xcsv_tool214
1 files changed, 0 insertions, 214 deletions
diff --git a/csv_tool b/csv_tool
deleted file mode 100755
index 17ad949..0000000
--- a/csv_tool
+++ /dev/null
@@ -1,214 +0,0 @@
1#!/usr/bin/env python
2
3"""
4Do stuff with csv files.
5"""
6
7import optparse
8import defapp
9
10import csv
11import operator
12import os.path
13from collections import defaultdict as defdict
14from itertools import izip
15
16o = optparse.make_option
17
18opts = [
19
20 o('-c', '--column', action='append', dest='col', type='int',
21 help='The column(s) on which to operate.'),
22
23 o(None, '--write-to-file', action='store_true', dest='write_to_file',
24 help='Write the output of operation xyz on file abc.csv to xyz_abc.csv.'),
25
26 o('-u', '--upper-bound', action='store_true', dest='upper_bound',
27 help="The linear regression is modified to represent an upper bound."),
28
29# o(None, '--degree', action='store', type='float', dest='thruthiness',
30# help='Not quite absolut truth.'),
31 ]
32
33defaults = {
34 'col' : [],
35 'write_to_file' : False,
36 'upper_bound' : False,
37 }
38
39def pair_iter(it):
40 it = iter(it)
41 while True:
42 yield (it.next(), it.next())
43
44def make_vector_op(op):
45 def vector_op(a, b, defvalue=0):
46 if len(a) > len(b):
47 shorter = b
48 longer = a
49 else:
50 shorter = a
51 longer = b
52 c = list(longer)
53 for i in xrange(len(shorter)):
54 c[i] = op(longer[i], shorter[i])
55 for i in xrange(len(shorter), len(longer)):
56 c[i] = op(longer[i], defvalue)
57 return c
58 return vector_op
59
60def make_scalar_op(op):
61 def scalar_op(scalar, a):
62 return [op(x, scalar) for x in a]
63 return scalar_op
64
65row_add = make_vector_op(operator.add)
66row_min = make_vector_op(min)
67row_max = make_vector_op(max)
68
69def row_reduce(row_op, fixup=lambda key, rows, res: res):
70 def _reduce(order, by_key):
71 for key in order:
72 if key in by_key:
73 rows = by_key[key]
74 res = reduce(row_op, rows)
75 del by_key[key]
76 yield fixup(key, rows, res)
77 return _reduce
78
79
80row_mul = make_scalar_op(operator.mul)
81row_div = make_scalar_op(operator.div)
82
83def transpose(rows):
84 rows = list(rows)
85 if rows:
86 r = len(rows)
87 c = max([len(x) for x in rows])
88 def at(x, y):
89 try:
90 return rows[x][y]
91 except IndexError:
92 return 0
93 for i in xrange(c):
94 yield [at(j, i) for j in xrange(r) ]
95
96def reorder_columns(rows, xchg_pairs):
97 for r in rows:
98 for (x,y) in xchg_pairs:
99 r[x], r[y] = r[y], r[x]
100 yield r
101
102def select_columns(rows, cols):
103 for r in rows:
104 yield [r[x] for x in cols]
105
106def numpy_lstsq(x, y):
107 from numpy import ones, array
108 from numpy.linalg import lstsq
109 A = ones((len(y), 2), dtype=float)
110 A[:,0] = array(x)
111 b = array(y)
112 return lstsq(A, b)[0]
113
114def max_delta(c0, c1, x, y):
115 return max([float(y) - (c0 + c1 * float(x)) for (x,y) in izip(x, y)])
116
117def least_squares(rows, xy_pairs, upper_bound):
118 cols = []
119 for (x, y) in xy_pairs:
120 cols += [x, y]
121 rows = select_columns(rows, cols)
122 cols = transpose(rows)
123 for ((x, y), (xval, yval)) in izip(xy_pairs, pair_iter(cols)):
124 c1, c0 = numpy_lstsq(xval, yval)
125 if upper_bound:
126 c0 += max_delta(c0, c1, xval, yval)
127 yield [x + 1, y + 1, c0, c1]
128
129def select_by_key(rows, col, cast=None):
130 by_key = defdict(list)
131 order = []
132 for r in rows:
133 key = r[col]
134 if cast:
135 by_key[key] += [[cast(x) for x in r]]
136 else:
137 by_key[key] += [r]
138 order += [key]
139 return (order, by_key)
140
141class CsvApp(defapp.App):
142 def __init__(self):
143 defapp.App.__init__(self, opts, defaults)
144 # fixup human-friendly offsets
145 if not self.options.col:
146 self.options.col = [1]
147 self.options.col = [x - 1 for x in self.options.col]
148 self.options.col_pairs = list(pair_iter(self.options.col))
149
150 def transform(self, make_iterator, ordered=True):
151 """
152 Read a file, pass the rows in the file to an iterator factory, and
153 write out the output of the iterator. The iterator performs the
154 desired transformation.
155 """
156 files = list(self.args)
157 del files[0]
158 for fn in files:
159 try:
160 # read in content
161 rows = csv.reader(open(fn, 'r'))
162 # set up transformation
163 if ordered:
164 (order, by_key) = select_by_key(rows, self.options.col[0],
165 float)
166 rows = make_iterator(order, by_key)
167 else:
168 rows = make_iterator(rows)
169 # write out
170 outfile = self.outfile()
171 if self.options.write_to_file:
172 (dir, file) = os.path.split(fn)
173 fn = os.path.join(dir, self.args[0] + '_' + file)
174 outfile = open(fn, 'w')
175 csv.writer(outfile).writerows(rows)
176 if self.options.write_to_file:
177 outfile.close()
178 except IOError, ex:
179 self.err("%s:%s" % (fn, str(ex)))
180 except IndexError, ex:
181 self.err("%s: Sorry, index out of range." % fn)
182
183 def do_avg(self, _):
184 def fixup_avg(key, rows, res):
185 res = row_div(len(rows), res)
186 res[self.options.col[0]] = key
187 return res
188 self.transform(row_reduce(row_add, fixup_avg))
189
190 def do_max(self, _):
191 self.transform(row_reduce(row_max))
192
193 def do_min(self, _):
194 self.transform(row_reduce(row_min))
195
196 def do_transpose(self, _):
197 self.transform(transpose, ordered=False)
198
199 def do_reorder(self, _):
200 self.transform(lambda rows: reorder_columns(
201 rows, self.options.col_pairs), ordered=False)
202
203 def do_select(self, _):
204 self.transform(lambda rows: select_columns(
205 rows, self.options.col), ordered=False)
206
207 def do_lstsqrs(self, _):
208 self.transform(
209 lambda rows: least_squares(rows, self.options.col_pairs,
210 self.options.upper_bound),
211 ordered=False)
212
213if __name__ == '__main__':
214 CsvApp().launch()
generated by cgit v1.2.2 (git 2.25.0) at 2025-05-18 17:10:38 -0400