line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Encode::Float; |
2
|
1
|
|
|
1
|
|
20707
|
use strict; |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
38
|
|
3
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
BEGIN |
5
|
|
|
|
|
|
|
{ |
6
|
1
|
|
|
1
|
|
5
|
use Exporter (); |
|
1
|
|
|
|
|
2
|
|
|
1
|
|
|
|
|
22
|
|
7
|
1
|
|
|
1
|
|
4
|
use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS); |
|
1
|
|
|
|
|
5
|
|
|
1
|
|
|
|
|
143
|
|
8
|
1
|
|
|
1
|
|
2
|
$VERSION = '0.11'; |
9
|
1
|
|
|
|
|
15
|
@ISA = qw(Exporter); |
10
|
1
|
|
|
|
|
2
|
@EXPORT = qw(); |
11
|
1
|
|
|
|
|
2
|
@EXPORT_OK = qw(); |
12
|
1
|
|
|
|
|
691
|
%EXPORT_TAGS = (); |
13
|
|
|
|
|
|
|
} |
14
|
|
|
|
|
|
|
|
15
|
|
|
|
|
|
|
#01234567890123456789012345678901234567890123 |
16
|
|
|
|
|
|
|
#Encode/decode float as a string for sorting. |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
=head1 NAME |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
C - Encode/decode float as a string for sorting. |
21
|
|
|
|
|
|
|
|
22
|
|
|
|
|
|
|
=head1 SYNOPSIS |
23
|
|
|
|
|
|
|
|
24
|
|
|
|
|
|
|
use Encode::Float; |
25
|
|
|
|
|
|
|
my $encoder = Encode::Float->new(); |
26
|
|
|
|
|
|
|
my @list; |
27
|
|
|
|
|
|
|
for (my $i = 0 ; $i < 10 ; $i++) |
28
|
|
|
|
|
|
|
{ |
29
|
|
|
|
|
|
|
my $float = (.5 - rand) * 10**int(10 - 20 * rand); |
30
|
|
|
|
|
|
|
$float = 0 if $i == 0; |
31
|
|
|
|
|
|
|
my $encoded = $encoder->encode($float); |
32
|
|
|
|
|
|
|
my $decoded = $encoder->decode($encoded); |
33
|
|
|
|
|
|
|
my $error = $encoder->getRelativeDifference($float, $decoded); |
34
|
|
|
|
|
|
|
push @list, [ $encoded, $float, $decoded, $error ]; |
35
|
|
|
|
|
|
|
} |
36
|
|
|
|
|
|
|
@list = sort { $a->[0] cmp $b->[0] } @list; |
37
|
|
|
|
|
|
|
foreach (@list) |
38
|
|
|
|
|
|
|
{ |
39
|
|
|
|
|
|
|
print join(',', @$_) . "\n"; |
40
|
|
|
|
|
|
|
} |
41
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
=head1 DESCRIPTION |
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
C encodes and decodes floating point numbers |
45
|
|
|
|
|
|
|
as fixed length positive decimal integers that preserve their order (less |
46
|
|
|
|
|
|
|
rounding errors), that is, sorting the encoded integers also sorts the |
47
|
|
|
|
|
|
|
floating point numbers. |
48
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
=head1 CONSTRUCTOR |
50
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
=head2 C |
52
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
The method C creates an instance of the C |
54
|
|
|
|
|
|
|
class with the following parameter: |
55
|
|
|
|
|
|
|
|
56
|
|
|
|
|
|
|
=over |
57
|
|
|
|
|
|
|
|
58
|
|
|
|
|
|
|
=item C |
59
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
digitsOfAccuracy => 16 |
61
|
|
|
|
|
|
|
|
62
|
|
|
|
|
|
|
C is an optional parameter that sets the number of |
63
|
|
|
|
|
|
|
decimal digits to preserve in the floating point number; the default is 16. |
64
|
|
|
|
|
|
|
|
65
|
|
|
|
|
|
|
=back |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
=cut |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
sub new |
70
|
|
|
|
|
|
|
{ |
71
|
1
|
|
|
1
|
1
|
11
|
my ($Class, %Parameters) = @_; |
72
|
1
|
|
33
|
|
|
7
|
my $Self = bless({}, ref($Class) || $Class); |
73
|
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
# set the number of digits used to represent a float. |
75
|
1
|
|
|
|
|
6
|
$Self->{digitsOfAccuracy} = 16; |
76
|
1
|
50
|
|
|
|
4
|
$Self->{digitsOfAccuracy} = int abs $Parameters{digitsOfAccuracy} if exists $Parameters{digitsOfAccuracy}; |
77
|
1
|
50
|
|
|
|
3
|
$Self->{digitsOfAccuracy} = 1 if $Self->{digitsOfAccuracy} < 1; |
78
|
|
|
|
|
|
|
|
79
|
|
|
|
|
|
|
# get the maximum integer value. |
80
|
1
|
|
|
|
|
4
|
my $mantissaMaxStr = '9' x $Self->{digitsOfAccuracy}; |
81
|
1
|
|
|
|
|
3
|
my $mantissaMax = $mantissaMaxStr + 0; |
82
|
1
|
50
|
|
|
|
4
|
if ($mantissaMax ne $mantissaMaxStr) |
83
|
|
|
|
|
|
|
{ |
84
|
0
|
|
|
|
|
0
|
die "digitsOfAccuracy = $Self->{digitsOfAccuracy} is too large.\n"; |
85
|
|
|
|
|
|
|
} |
86
|
1
|
|
|
|
|
3
|
$Self->{mantissaMax} = $mantissaMax; |
87
|
1
|
|
|
|
|
3
|
$Self->{floatFormat} = '%+' . '.' . ($Self->{digitsOfAccuracy} - 1) . 'E'; |
88
|
1
|
|
|
|
|
3
|
$Self->{exponentSize} = 3; |
89
|
1
|
|
|
|
|
3
|
$Self->{exponentFormat} = '%+0' . ($Self->{exponentSize} + 1) . 'd'; |
90
|
1
|
|
|
|
|
3
|
$Self->{exponentMax} = '9' x $Self->{exponentSize}; |
91
|
1
|
|
|
|
|
3
|
$Self->{mantissaFormat} = '%0' . $Self->{digitsOfAccuracy} . 'd'; |
92
|
1
|
|
|
|
|
3
|
return $Self; |
93
|
|
|
|
|
|
|
} |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=head1 METHODS |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
=head2 C |
98
|
|
|
|
|
|
|
|
99
|
|
|
|
|
|
|
The method C takes a floating point number as its only parameter |
100
|
|
|
|
|
|
|
and returns its integer encoding. |
101
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
=cut |
103
|
|
|
|
|
|
|
|
104
|
|
|
|
|
|
|
sub encode |
105
|
|
|
|
|
|
|
{ |
106
|
256230
|
|
|
256230
|
1
|
3738079
|
my ($Self, $Float) = @_; |
107
|
|
|
|
|
|
|
|
108
|
|
|
|
|
|
|
# convert the float to a string. |
109
|
256230
|
|
|
|
|
1281217
|
my $string = sprintf($Self->{floatFormat}, $Float); |
110
|
|
|
|
|
|
|
|
111
|
|
|
|
|
|
|
# get the mantissa of the float. |
112
|
256230
|
|
|
|
|
576507
|
my $mantissa = substr($string, 1, 1) . substr($string, 3, $Self->{digitsOfAccuracy} - 1); |
113
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
# get the exponent of the float, with its sign. |
115
|
256230
|
|
|
|
|
873198
|
my $exponent = sprintf($Self->{exponentFormat}, substr($string, $Self->{digitsOfAccuracy} + 3)); |
116
|
|
|
|
|
|
|
|
117
|
|
|
|
|
|
|
# encode the sign of the float and the exponent to a single leading digit. |
118
|
256230
|
|
|
|
|
378238
|
my $lead; |
119
|
256230
|
100
|
|
|
|
485319
|
if ($Float < 0) |
120
|
|
|
|
|
|
|
{ |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
# the float is negative, so take the complement of the mantissa. |
123
|
127676
|
|
|
|
|
367783
|
$mantissa = sprintf($Self->{mantissaFormat}, $Self->{mantissaMax} - $mantissa); |
124
|
127676
|
100
|
|
|
|
251379
|
if (substr($exponent, 0, 1) eq '-') |
125
|
|
|
|
|
|
|
{ |
126
|
63982
|
|
|
|
|
88935
|
$lead = 2; |
127
|
|
|
|
|
|
|
} |
128
|
|
|
|
|
|
|
else |
129
|
|
|
|
|
|
|
{ |
130
|
63694
|
|
|
|
|
78170
|
$lead = 1; |
131
|
|
|
|
|
|
|
|
132
|
|
|
|
|
|
|
# negative float but positive exponent, so take the complement of the exponent. |
133
|
63694
|
|
|
|
|
158390
|
$exponent = sprintf($Self->{exponentFormat}, $Self->{exponentMax} - $exponent); |
134
|
|
|
|
|
|
|
} |
135
|
|
|
|
|
|
|
} |
136
|
|
|
|
|
|
|
else |
137
|
|
|
|
|
|
|
{ |
138
|
128554
|
100
|
|
|
|
239065
|
if (substr($exponent, 0, 1) eq '-') |
139
|
|
|
|
|
|
|
{ |
140
|
64206
|
|
|
|
|
77047
|
$lead = 3; |
141
|
|
|
|
|
|
|
|
142
|
|
|
|
|
|
|
# positive float but negative exponent, so take the complement of the exponent. |
143
|
64206
|
|
|
|
|
182866
|
$exponent = sprintf($Self->{exponentFormat}, $Self->{exponentMax} + $exponent); |
144
|
|
|
|
|
|
|
} |
145
|
|
|
|
|
|
|
else |
146
|
|
|
|
|
|
|
{ |
147
|
64348
|
|
|
|
|
90887
|
$lead = 4; |
148
|
|
|
|
|
|
|
} |
149
|
|
|
|
|
|
|
} |
150
|
|
|
|
|
|
|
|
151
|
|
|
|
|
|
|
# zero is a special case. |
152
|
256230
|
100
|
|
|
|
527255
|
$lead = 3 if $Float == 0; |
153
|
|
|
|
|
|
|
|
154
|
|
|
|
|
|
|
# encode the float as a long integer that preserves sort order. |
155
|
256230
|
|
|
|
|
912788
|
return $lead . substr($exponent, 1) . $mantissa; |
156
|
|
|
|
|
|
|
} |
157
|
|
|
|
|
|
|
|
158
|
|
|
|
|
|
|
=head2 C |
159
|
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
The method C takes an encoded floating point number (a positive |
161
|
|
|
|
|
|
|
integer) and returns its floating point number. |
162
|
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
=cut |
164
|
|
|
|
|
|
|
|
165
|
|
|
|
|
|
|
sub decode |
166
|
|
|
|
|
|
|
{ |
167
|
128115
|
|
|
128115
|
1
|
179934
|
my ($Self, $EncodedFloat) = @_; |
168
|
|
|
|
|
|
|
|
169
|
|
|
|
|
|
|
# holds the sign of the float. |
170
|
128115
|
|
|
|
|
131890
|
my $sign; |
171
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
# get the leading digit that encodes the sign of the float and exponent. |
173
|
128115
|
|
|
|
|
183635
|
my $lead = substr($EncodedFloat, 0, 1); |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
# get the exponent of the float. |
176
|
128115
|
|
|
|
|
203585
|
my $exponent = substr($EncodedFloat, 1, $Self->{exponentSize}); |
177
|
|
|
|
|
|
|
|
178
|
|
|
|
|
|
|
# get the mantissa of the float. |
179
|
128115
|
|
|
|
|
225544
|
my $mantissa = substr($EncodedFloat, $Self->{exponentSize} + 1); |
180
|
|
|
|
|
|
|
|
181
|
|
|
|
|
|
|
# adjust the exponent and sign via the leading digit. |
182
|
128115
|
100
|
|
|
|
345856
|
if ($lead == 1) |
|
|
100
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
183
|
|
|
|
|
|
|
{ |
184
|
31847
|
|
|
|
|
37327
|
$sign = -1; |
185
|
31847
|
|
|
|
|
48349
|
$mantissa = $Self->{mantissaMax} - $mantissa; |
186
|
31847
|
|
|
|
|
57702
|
$exponent = $Self->{exponentMax} - $exponent; |
187
|
|
|
|
|
|
|
} |
188
|
|
|
|
|
|
|
elsif ($lead == 2) |
189
|
|
|
|
|
|
|
{ |
190
|
31991
|
|
|
|
|
42043
|
$sign = -1; |
191
|
31991
|
|
|
|
|
48921
|
$mantissa = $Self->{mantissaMax} - $mantissa; |
192
|
31991
|
|
|
|
|
46975
|
$exponent = -$exponent; |
193
|
|
|
|
|
|
|
} |
194
|
|
|
|
|
|
|
elsif ($lead == 3) |
195
|
|
|
|
|
|
|
{ |
196
|
32108
|
|
|
|
|
37504
|
$sign = 1; |
197
|
32108
|
|
|
|
|
52580
|
$exponent -= $Self->{exponentMax}; |
198
|
|
|
|
|
|
|
} |
199
|
|
|
|
|
|
|
else |
200
|
|
|
|
|
|
|
{ |
201
|
32169
|
|
|
|
|
41992
|
$sign = 1; |
202
|
|
|
|
|
|
|
} |
203
|
|
|
|
|
|
|
|
204
|
|
|
|
|
|
|
# return the float. |
205
|
128115
|
|
|
|
|
186245
|
$mantissa =~ s/^0+//; |
206
|
128115
|
100
|
|
|
|
243777
|
return 0 unless length $mantissa; |
207
|
128110
|
|
|
|
|
232644
|
my $decimal = substr($mantissa, 0, 1) . '.' . substr($mantissa, 1); |
208
|
128110
|
|
|
|
|
1391076
|
return $sign * $decimal * 10**$exponent; |
209
|
|
|
|
|
|
|
} |
210
|
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
=head2 C |
212
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
The method C computes the relative |
214
|
|
|
|
|
|
|
difference between the floating point numbers C and C, which |
215
|
|
|
|
|
|
|
is C or zero if both |
216
|
|
|
|
|
|
|
numbers are zero. |
217
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
=cut |
219
|
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
sub getRelativeDifference |
221
|
|
|
|
|
|
|
{ |
222
|
0
|
|
|
0
|
1
|
|
my ($Self, $FloatA, $FloatB) = @_; |
223
|
0
|
|
|
|
|
|
my $absMax = abs $FloatA; |
224
|
0
|
|
|
|
|
|
my $absFloatB = abs $FloatB; |
225
|
0
|
0
|
|
|
|
|
$absMax = $absFloatB if $absFloatB > $absMax; |
226
|
0
|
0
|
|
|
|
|
return 0 unless $absMax; |
227
|
0
|
|
|
|
|
|
return abs($FloatA - $FloatB) / $absMax; |
228
|
|
|
|
|
|
|
} |
229
|
|
|
|
|
|
|
|
230
|
|
|
|
|
|
|
=head1 INSTALLATION |
231
|
|
|
|
|
|
|
|
232
|
|
|
|
|
|
|
Use L to install the module and all its prerequisites: |
233
|
|
|
|
|
|
|
|
234
|
|
|
|
|
|
|
perl -MCPAN -e shell |
235
|
|
|
|
|
|
|
>install Encode::Float |
236
|
|
|
|
|
|
|
|
237
|
|
|
|
|
|
|
=head1 BUGS |
238
|
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
Please email bugs reports or feature requests to C, or through |
240
|
|
|
|
|
|
|
the web interface at L. The author |
241
|
|
|
|
|
|
|
will be notified and you can be automatically notified of progress on the bug fix or feature request. |
242
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
=head1 AUTHOR |
244
|
|
|
|
|
|
|
|
245
|
|
|
|
|
|
|
Jeff Kubina |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
=head1 COPYRIGHT |
248
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
Copyright (c) 2013 Jeff Kubina. All rights reserved. |
250
|
|
|
|
|
|
|
This program is free software; you can redistribute |
251
|
|
|
|
|
|
|
it and/or modify it under the same terms as Perl itself. |
252
|
|
|
|
|
|
|
|
253
|
|
|
|
|
|
|
The full text of the license can be found in the |
254
|
|
|
|
|
|
|
LICENSE file included with this module. |
255
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
=head1 KEYWORDS |
257
|
|
|
|
|
|
|
|
258
|
|
|
|
|
|
|
decoding, double, encoding, float |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
=head1 SEE ALSO |
261
|
|
|
|
|
|
|
|
262
|
|
|
|
|
|
|
L, L |
263
|
|
|
|
|
|
|
|
264
|
|
|
|
|
|
|
=cut |
265
|
|
|
|
|
|
|
|
266
|
|
|
|
|
|
|
1; |
267
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
# The preceding line will help the module return a true value |