1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
|
/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4; fill-column: 100 -*- */
/*
* This file is part of the LibreOffice project.
*
* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/.
*/
#pragma once
#include <o3tl/safeint.hxx>
#include <sal/macros.h>
#include <sal/types.h>
#include <array>
#include <cassert>
#include <numeric>
#include <utility>
#include <type_traits>
namespace o3tl
{
// Length units
enum class Length
{
mm100 = 0, // 1/100th mm
mm10, // 1/10 mm, corresponds to MapUnit::Map10thMM
mm, // millimeter
cm, // centimeter
m, // meter
km, // kilometer
emu, // English Metric Unit: 1/360000 cm, 1/914400 in
twip, // "Twentieth of a point" aka "dxa": 1/20 pt
pt, // Point: 1/72 in
pc, // Pica: 1/6 in, corresponds to FieldUnit::PICA and MeasureUnit::PICA
in1000, // 1/1000 in, corresponds to MapUnit::Map1000thInch
in100, // 1/100 in, corresponds to MapUnit::Map100thInch
in10, // 1/10 in, corresponds to MapUnit::Map10thInch
in, // inch
ft, // foot
mi, // mile
master, // PPT Master Unit: 1/576 in
px, // "pixel" unit: 15 twip (96 ppi), corresponds to MeasureUnit::PIXEL
ch, // "char" unit: 210 twip (14 px), corresponds to FieldUnit::CHAR
line, // "line" unit: 312 twip, corresponds to FieldUnit::LINE
count, // <== add new units above this last entry
invalid = -1
};
// If other categories of units would be needed (like time), a separate scoped enum
// should be created, respective conversion array prepared in detail namespace, and
// respective md(NewUnit, NewUnit) overload introduced, which would allow using
// o3tl::convert(), o3tl::convertSaturate() and o3tl::getConversionMulDiv() with the
// new category in a type-safe way, without mixing unrelated units.
namespace detail
{
// Common utilities
// A special function to avoid compiler warning comparing signed and unsigned values
template <typename I> constexpr bool isBetween(I n, sal_Int64 min, sal_Int64 max)
{
assert(max > 0 && min < 0);
if constexpr (std::is_signed_v<I>)
return n >= min && n <= max;
else
return n <= sal_uInt64(max);
}
// Ensure correct rounding for both positive and negative integers
template <typename I, std::enable_if_t<std::is_integral_v<I>, int> = 0>
constexpr sal_Int64 MulDiv(I n, sal_Int64 m, sal_Int64 d)
{
assert(m > 0 && d > 0);
assert(isBetween(n, (SAL_MIN_INT64 + d / 2) / m, (SAL_MAX_INT64 - d / 2) / m));
return (n >= 0 ? (n * m + d / 2) : (n * m - d / 2)) / d;
}
template <typename F, std::enable_if_t<std::is_floating_point_v<F>, int> = 0>
constexpr double MulDiv(F f, sal_Int64 m, sal_Int64 d)
{
assert(m > 0 && d > 0);
return f * (double(m) / d);
}
template <typename I, std::enable_if_t<std::is_integral_v<I>, int> = 0>
constexpr sal_Int64 MulDiv(I n, sal_Int64 m, sal_Int64 d, bool& bOverflow, sal_Int64 nDefault)
{
if (!isBetween(n, (SAL_MIN_INT64 + d / 2) / m, (SAL_MAX_INT64 - d / 2) / m))
{
bOverflow = true;
return nDefault;
}
bOverflow = false;
return MulDiv(n, m, d);
}
template <typename I, std::enable_if_t<std::is_integral_v<I>, int> = 0>
constexpr sal_Int64 MulDivSaturate(I n, sal_Int64 m, sal_Int64 d)
{
if (!isBetween(n, (SAL_MIN_INT64 + d / 2) / m, (SAL_MAX_INT64 - d / 2) / m))
{
if (m > d && !isBetween(n, SAL_MIN_INT64 / m * d + d / 2, SAL_MAX_INT64 / m * d - d / 2))
return n > 0 ? SAL_MAX_INT64 : SAL_MIN_INT64; // saturate
return (n >= 0 ? n + d / 2 : n - d / 2) / d * m; // divide before multiplication
}
return MulDiv(n, m, d);
}
template <class M, class N> constexpr std::common_type_t<M, N> asserting_gcd(M m, N n)
{
auto ret = std::gcd(m, n);
assert(ret != 0);
return ret;
}
// Packs integral multiplier and divisor for conversion from one unit to another
struct m_and_d
{
sal_Int64 m; // multiplier
sal_Int64 d; // divisor
constexpr m_and_d(sal_Int64 _m, sal_Int64 _d)
: m(_m / asserting_gcd(_m, _d)) // make sure to use smallest quotients here because
, d(_d / asserting_gcd(_m, _d)) // they will be multiplied when building final table
{
assert(_m > 0 && _d > 0);
}
};
// Resulting static array N x N of all quotients to convert between all units. The
// quotients are minimal to allow largest range of converted numbers without overflow.
// Maybe o3tl::enumarray could be used here, but it's not constexpr yet.
template <int N> constexpr auto prepareMDArray(const m_and_d (&mdBase)[N])
{
std::array<std::array<sal_Int64, N>, N> a{};
for (int i = 0; i < N; ++i)
{
a[i][i] = 1;
for (int j = 0; j < i; ++j)
{
assert(mdBase[i].m < SAL_MAX_INT64 / mdBase[j].d);
assert(mdBase[i].d < SAL_MAX_INT64 / mdBase[j].m);
const sal_Int64 m = mdBase[i].m * mdBase[j].d, d = mdBase[i].d * mdBase[j].m;
const sal_Int64 g = asserting_gcd(m, d);
a[i][j] = m / g;
a[j][i] = d / g;
}
}
return a;
}
// A generic template used for fundamental arithmetic types
template <typename U> constexpr sal_Int64 md(U i, U /*j*/) { return i; }
// Length units implementation
// Array of conversion quotients for mm, used to build final conversion table. Entries
// are { multiplier, divider } to convert respective unit *to* mm. Order of elements
// corresponds to order in o3tl::Length enum (Length::count and Length::invalid omitted).
constexpr m_and_d mdBaseLen[] = {
{ 1, 100 }, // mm100 => mm
{ 1, 10 }, // mm10 => mm
{ 1, 1 }, // mm => mm
{ 10, 1 }, // cm => mm
{ 1000, 1 }, // m => mm
{ 1000000, 1 }, // km => mm
{ 1, 36000 }, // emu => mm
{ 254, 10 * 1440 }, // twip => mm
{ 254, 10 * 72 }, // pt => mm
{ 254, 10 * 6 }, // pc => mm
{ 254, 10000 }, // in1000 => mm
{ 254, 1000 }, // in100 => mm
{ 254, 100 }, // in10 => mm
{ 254, 10 }, // in => mm
{ 254 * 12, 10 }, // ft => mm
{ 254 * 12 * 5280, 10 }, // mi => mm
{ 254, 10 * 576 }, // master => mm
{ 254 * 15, 10 * 1440 }, // px => mm
{ 254 * 210, 10 * 1440 }, // ch => mm
{ 254 * 312, 10 * 1440 }, // line => mm
};
static_assert(std::size(mdBaseLen) == static_cast<int>(Length::count),
"mdBaseL must have an entry for each unit in o3tl::Length");
// The resulting multipliers and divisors array
constexpr auto aLengthMDArray = prepareMDArray(mdBaseLen);
// an overload taking Length
constexpr sal_Int64 md(Length i, Length j)
{
const int nI = static_cast<int>(i), nJ = static_cast<int>(j);
assert(nI >= 0 && o3tl::make_unsigned(nI) < aLengthMDArray.size());
assert(nJ >= 0 && o3tl::make_unsigned(nJ) < aLengthMDArray.size());
return aLengthMDArray[nI][nJ];
}
// here might go overloads of md() taking other units ...
}
// Unchecked conversion. Takes a number value, multiplier and divisor
template <typename N> constexpr auto convert(N n, sal_Int64 mul, sal_Int64 div)
{
return detail::MulDiv(n, mul, div);
}
// Unchecked conversion. Takes a number value and units defined in this header
template <typename N, typename U> constexpr auto convert(N n, U from, U to)
{
return convert(n, detail::md(from, to), detail::md(to, from));
}
// Convert to twips - for convenience as we do this a lot
template <typename N> constexpr auto toTwips(N number, Length from)
{
return convert(number, from, Length::twip);
}
// Returns nDefault if intermediate multiplication overflows sal_Int64 (only for integral types).
// On return, bOverflow indicates if overflow happened. nDefault is returned when overflow occurs.
template <typename N, typename U>
constexpr auto convert(N n, U from, U to, bool& bOverflow, sal_Int64 nDefault = 0)
{
return detail::MulDiv(n, detail::md(from, to), detail::md(to, from), bOverflow, nDefault);
}
// Conversion with saturation (only for integral types). For too large input returns SAL_MAX_INT64.
// When intermediate multiplication would overflow, but the end result is in sal_Int64 range, the
// precision is decreased because of inversion of multiplication and division.
template <typename N, typename U> constexpr auto convertSaturate(N n, U from, U to)
{
return detail::MulDivSaturate(n, detail::md(from, to), detail::md(to, from));
}
// Conversion with saturation (only for integral types), optimized for return types smaller than
// sal_Int64. In this case, it's easier to clamp input values to known bounds, than to do some
// preprocessing to handle too large input values, just to clamp the result anyway. Use it like:
//
// sal_Int32 n = convertNarrowing<sal_Int32, o3tl::Length::mm100, o3tl::Length::emu>(m);
template <typename Out, auto from, auto to, typename N,
std::enable_if_t<
std::is_integral_v<N> && std::is_integral_v<Out> && sizeof(Out) < sizeof(sal_Int64),
int> = 0>
constexpr Out convertNarrowing(N n)
{
constexpr sal_Int64 nMin = convertSaturate(std::numeric_limits<Out>::min(), to, from);
constexpr sal_Int64 nMax = convertSaturate(std::numeric_limits<Out>::max(), to, from);
if (static_cast<sal_Int64>(n) > nMax)
return std::numeric_limits<Out>::max();
if (static_cast<sal_Int64>(n) < nMin)
return std::numeric_limits<Out>::min();
return convert(n, from, to);
}
// Return a pair { multiplier, divisor } for a given conversion
template <typename U> constexpr std::pair<sal_Int64, sal_Int64> getConversionMulDiv(U from, U to)
{
return { detail::md(from, to), detail::md(to, from) };
}
}
/* vim:set shiftwidth=4 softtabstop=4 expandtab cinoptions=b1,g0,N-s cinkeys+=0=break: */
|