mirror of
https://github.com/jkriege2/JKQtPlotter.git
synced 2024-11-16 02:25:50 +08:00
60a4656dd8
NEW: Added new JKQTPCALabelType element JKQTPCALTscientific, so axis label ticks can be displayed as numbers in scientific notation like 1.2E-34 auto-generate JKQTPCALabelType documentation
64 lines
2.1 KiB
C++
64 lines
2.1 KiB
C++
/*
|
|
Copyright (c) 2008-2022 Jan W. Krieger (<jan@jkrieger.de>)
|
|
|
|
|
|
This software is free software: you can redistribute it and/or modify
|
|
it under the terms of the GNU Lesser General Public License (LGPL) as published by
|
|
the Free Software Foundation, either version 2.1 of the License, or
|
|
(at your option) any later version.
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
GNU Lesser General Public License (LGPL) for more details.
|
|
|
|
You should have received a copy of the GNU Lesser General Public License (LGPL)
|
|
along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
*/
|
|
|
|
|
|
|
|
#include "jkqtcommon/jkqtpmathtools.h"
|
|
#include <cmath>
|
|
|
|
|
|
|
|
uint64_t jkqtp_gcd(uint64_t a, uint64_t b)
|
|
{
|
|
if (a == 0)
|
|
return b;
|
|
else if (b == 0)
|
|
return a;
|
|
|
|
if (a < b)
|
|
return jkqtp_gcd(a, b % a);
|
|
else
|
|
return jkqtp_gcd(b, a % b);
|
|
}
|
|
|
|
void jkqtp_estimateFraction(double input, int& sign, uint64_t &intpart, uint64_t &num, uint64_t &denom, unsigned int precision)
|
|
{
|
|
// see https://stackoverflow.com/a/28662740
|
|
// https://stackoverflow.com/questions/4819075/how-to-output-fraction-instead-of-decimal-number
|
|
// https://stackoverflow.com/a/64828741
|
|
// https://stackoverflow.com/questions/26643695/converting-a-floating-point-decimal-value-to-a-fraction
|
|
sign=+1;
|
|
if (input<0.0) {
|
|
sign=-1;
|
|
input=fabs(input);
|
|
}
|
|
// now input is guaranteed positive and sign stores the sign +1 or -1
|
|
const double integral = std::floor(input);
|
|
const double frac = input - integral;
|
|
|
|
const double prec = pow(10,precision); // This is the accuracy.
|
|
|
|
uint64_t gcd_ = jkqtp_gcd(static_cast<uint64_t>(round(frac * prec)), static_cast<uint64_t>(prec));
|
|
|
|
denom = static_cast<uint64_t>(prec / gcd_);
|
|
num = static_cast<uint64_t>(round(frac * prec) / gcd_);
|
|
|
|
intpart=static_cast<uint64_t>(integral);
|
|
|
|
}
|