authorgravatar for andrew@ziglang.orgAndrew Kelley <andrew@ziglang.org> 2018-07-11 11:56:14-04:00
committergravatar for noreply@github.comGitHub <noreply@github.com> 2018-07-11 11:56:14-04:00
log3aaf814b9df6eecb6be025c4c73b8e9c46a112ba
tree2f5fe1ebbdf54f3e2d1847444f04c4450e99f4be
parentc6c49389ebd1503de38d9bb6ff6d9f6fba94d63b
parent9b054e73f6b9f946a404a7f0c8b6a90d83e4b928
signature Signed by PGP key 4AEE18F83AFDEB23

Merge pull request #1216 from ziglang/sort-improvements

Add generic comparator generator functions for sorting

2 files changed, 87 insertions(+), 240 deletions(-)

std/macho.zig+1-1
......@@ -42,7 +42,7 @@ pub const Symbol = struct {
4242 name: []const u8,
4343 address: u64,
4444
45 fn addressLessThan(lhs: *const Symbol, rhs: *const Symbol) bool {
45 fn addressLessThan(lhs: Symbol, rhs: Symbol) bool {
4646 return lhs.address < rhs.address;
4747 }
4848};
std/sort.zig+86-239
......@@ -5,7 +5,7 @@ const math = std.math;
55const builtin = @import("builtin");
66
77/// Stable in-place sort. O(n) best case, O(pow(n, 2)) worst case. O(1) memory (no allocator required).
8pub fn insertionSort(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *const T) bool) void {
8pub fn insertionSort(comptime T: type, items: []T, lessThan: fn (lhs: T, rhs: T) bool) void {
99 {
1010 var i: usize = 1;
1111 while (i < items.len) : (i += 1) {
......@@ -30,7 +30,7 @@ const Range = struct {
3030 };
3131 }
3232
33 fn length(self: *const Range) usize {
33 fn length(self: Range) usize {
3434 return self.end - self.start;
3535 }
3636};
......@@ -108,7 +108,7 @@ const Pull = struct {
108108
109109/// Stable in-place sort. O(n) best case, O(n*log(n)) worst case and average case. O(1) memory (no allocator required).
110110/// Currently implemented as block sort.
111pub fn sort(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *const T) bool) void {
111pub fn sort(comptime T: type, items: []T, lessThan: fn (lhs: T, rhs: T) bool) void {
112112 // Implementation ported from https://github.com/BonzaiThePenguin/WikiSort/blob/master/WikiSort.c
113113 var cache: [512]T = undefined;
114114
......@@ -131,16 +131,7 @@ pub fn sort(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *con
131131 // http://pages.ripco.net/~jgamble/nw.html
132132 var iterator = Iterator.init(items.len, 4);
133133 while (!iterator.finished()) {
134 var order = []u8{
135 0,
136 1,
137 2,
138 3,
139 4,
140 5,
141 6,
142 7,
143 };
134 var order = []u8{ 0, 1, 2, 3, 4, 5, 6, 7 };
144135 const range = iterator.nextRange();
145136
146137 const sliced_items = items[range.start..];
......@@ -741,7 +732,7 @@ pub fn sort(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *con
741732}
742733
743734// merge operation without a buffer
744fn mergeInPlace(comptime T: type, items: []T, A_arg: *const Range, B_arg: *const Range, lessThan: fn (*const T, *const T) bool) void {
735fn mergeInPlace(comptime T: type, items: []T, A_arg: Range, B_arg: Range, lessThan: fn (T, T) bool) void {
745736 if (A_arg.length() == 0 or B_arg.length() == 0) return;
746737
747738 // this just repeatedly binary searches into B and rotates A into position.
......@@ -762,8 +753,8 @@ fn mergeInPlace(comptime T: type, items: []T, A_arg: *const Range, B_arg: *const
762753 // again, this is NOT a general-purpose solution – it only works well in this case!
763754 // kind of like how the O(n^2) insertion sort is used in some places
764755
765 var A = A_arg.*;
766 var B = B_arg.*;
756 var A = A_arg;
757 var B = B_arg;
767758
768759 while (true) {
769760 // find the first place in B where the first item in A needs to be inserted
......@@ -783,7 +774,7 @@ fn mergeInPlace(comptime T: type, items: []T, A_arg: *const Range, B_arg: *const
783774}
784775
785776// merge operation using an internal buffer
786fn mergeInternal(comptime T: type, items: []T, A: *const Range, B: *const Range, lessThan: fn (*const T, *const T) bool, buffer: *const Range) void {
777fn mergeInternal(comptime T: type, items: []T, A: Range, B: Range, lessThan: fn (T, T) bool, buffer: Range) void {
787778 // whenever we find a value to add to the final array, swap it with the value that's already in that spot
788779 // when this algorithm is finished, 'buffer' will contain its original contents, but in a different order
789780 var A_count: usize = 0;
......@@ -819,7 +810,7 @@ fn blockSwap(comptime T: type, items: []T, start1: usize, start2: usize, block_s
819810
820811// combine a linear search with a binary search to reduce the number of comparisons in situations
821812// where have some idea as to how many unique values there are and where the next value might be
822fn findFirstForward(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool, unique: usize) usize {
813fn findFirstForward(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool, unique: usize) usize {
823814 if (range.length() == 0) return range.start;
824815 const skip = math.max(range.length() / unique, usize(1));
825816
......@@ -833,7 +824,7 @@ fn findFirstForward(comptime T: type, items: []T, value: *const T, range: *const
833824 return binaryFirst(T, items, value, Range.init(index - skip, index), lessThan);
834825}
835826
836fn findFirstBackward(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool, unique: usize) usize {
827fn findFirstBackward(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool, unique: usize) usize {
837828 if (range.length() == 0) return range.start;
838829 const skip = math.max(range.length() / unique, usize(1));
839830
......@@ -847,7 +838,7 @@ fn findFirstBackward(comptime T: type, items: []T, value: *const T, range: *cons
847838 return binaryFirst(T, items, value, Range.init(index, index + skip), lessThan);
848839}
849840
850fn findLastForward(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool, unique: usize) usize {
841fn findLastForward(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool, unique: usize) usize {
851842 if (range.length() == 0) return range.start;
852843 const skip = math.max(range.length() / unique, usize(1));
853844
......@@ -861,7 +852,7 @@ fn findLastForward(comptime T: type, items: []T, value: *const T, range: *const
861852 return binaryLast(T, items, value, Range.init(index - skip, index), lessThan);
862853}
863854
864fn findLastBackward(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool, unique: usize) usize {
855fn findLastBackward(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool, unique: usize) usize {
865856 if (range.length() == 0) return range.start;
866857 const skip = math.max(range.length() / unique, usize(1));
867858
......@@ -875,7 +866,7 @@ fn findLastBackward(comptime T: type, items: []T, value: *const T, range: *const
875866 return binaryLast(T, items, value, Range.init(index, index + skip), lessThan);
876867}
877868
878fn binaryFirst(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool) usize {
869fn binaryFirst(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool) usize {
879870 var start = range.start;
880871 var end = range.end - 1;
881872 if (range.start >= range.end) return range.end;
......@@ -893,7 +884,7 @@ fn binaryFirst(comptime T: type, items: []T, value: *const T, range: *const Rang
893884 return start;
894885}
895886
896fn binaryLast(comptime T: type, items: []T, value: *const T, range: *const Range, lessThan: fn (*const T, *const T) bool) usize {
887fn binaryLast(comptime T: type, items: []T, value: T, range: Range, lessThan: fn (T, T) bool) usize {
897888 var start = range.start;
898889 var end = range.end - 1;
899890 if (range.start >= range.end) return range.end;
......@@ -911,7 +902,7 @@ fn binaryLast(comptime T: type, items: []T, value: *const T, range: *const Range
911902 return start;
912903}
913904
914fn mergeInto(comptime T: type, from: []T, A: *const Range, B: *const Range, lessThan: fn (*const T, *const T) bool, into: []T) void {
905fn mergeInto(comptime T: type, from: []T, A: Range, B: Range, lessThan: fn (T, T) bool, into: []T) void {
915906 var A_index: usize = A.start;
916907 var B_index: usize = B.start;
917908 const A_last = A.end;
......@@ -941,7 +932,7 @@ fn mergeInto(comptime T: type, from: []T, A: *const Range, B: *const Range, less
941932 }
942933}
943934
944fn mergeExternal(comptime T: type, items: []T, A: *const Range, B: *const Range, lessThan: fn (*const T, *const T) bool, cache: []T) void {
935fn mergeExternal(comptime T: type, items: []T, A: Range, B: Range, lessThan: fn (T, T) bool, cache: []T) void {
945936 // A fits into the cache, so use that instead of the internal buffer
946937 var A_index: usize = 0;
947938 var B_index: usize = B.start;
......@@ -969,27 +960,32 @@ fn mergeExternal(comptime T: type, items: []T, A: *const Range, B: *const Range,
969960 mem.copy(T, items[insert_index..], cache[A_index..A_last]);
970961}
971962
972fn swap(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *const T) bool, order: *[8]u8, x: usize, y: usize) void {
963fn swap(comptime T: type, items: []T, lessThan: fn (lhs: T, rhs: T) bool, order: *[8]u8, x: usize, y: usize) void {
973964 if (lessThan(items[y], items[x]) or ((order.*)[x] > (order.*)[y] and !lessThan(items[x], items[y]))) {
974965 mem.swap(T, &items[x], &items[y]);
975966 mem.swap(u8, &(order.*)[x], &(order.*)[y]);
976967 }
977968}
978969
979fn i32asc(lhs: *const i32, rhs: *const i32) bool {
980 return lhs.* < rhs.*;
981}
970// Use these to generate a comparator function for a given type. e.g. `sort(u8, slice, asc(u8))`.
971pub fn asc(comptime T: type) fn (T, T) bool {
972 const impl = struct {
973 fn inner(a: T, b: T) bool {
974 return a < b;
975 }
976 };
982977
983fn i32desc(lhs: *const i32, rhs: *const i32) bool {
984 return rhs.* < lhs.*;
978 return impl.inner;
985979}
986980
987fn u8asc(lhs: *const u8, rhs: *const u8) bool {
988 return lhs.* < rhs.*;
989}
981pub fn desc(comptime T: type) fn (T, T) bool {
982 const impl = struct {
983 fn inner(a: T, b: T) bool {
984 return a > b;
985 }
986 };
990987
991fn u8desc(lhs: *const u8, rhs: *const u8) bool {
992 return rhs.* < lhs.*;
988 return impl.inner;
993989}
994990
995991test "stable sort" {
......@@ -998,119 +994,38 @@ test "stable sort" {
998994}
999995fn testStableSort() void {
1000996 var expected = []IdAndValue{
1001 IdAndValue{
1002 .id = 0,
1003 .value = 0,
1004 },
1005 IdAndValue{
1006 .id = 1,
1007 .value = 0,
1008 },
1009 IdAndValue{
1010 .id = 2,
1011 .value = 0,
1012 },
1013 IdAndValue{
1014 .id = 0,
1015 .value = 1,
1016 },
1017 IdAndValue{
1018 .id = 1,
1019 .value = 1,
1020 },
1021 IdAndValue{
1022 .id = 2,
1023 .value = 1,
1024 },
1025 IdAndValue{
1026 .id = 0,
1027 .value = 2,
1028 },
1029 IdAndValue{
1030 .id = 1,
1031 .value = 2,
1032 },
1033 IdAndValue{
1034 .id = 2,
1035 .value = 2,
1036 },
997 IdAndValue{ .id = 0, .value = 0 },
998 IdAndValue{ .id = 1, .value = 0 },
999 IdAndValue{ .id = 2, .value = 0 },
1000 IdAndValue{ .id = 0, .value = 1 },
1001 IdAndValue{ .id = 1, .value = 1 },
1002 IdAndValue{ .id = 2, .value = 1 },
1003 IdAndValue{ .id = 0, .value = 2 },
1004 IdAndValue{ .id = 1, .value = 2 },
1005 IdAndValue{ .id = 2, .value = 2 },
10371006 };
10381007 var cases = [][9]IdAndValue{
10391008 []IdAndValue{
1040 IdAndValue{
1041 .id = 0,
1042 .value = 0,
1043 },
1044 IdAndValue{
1045 .id = 0,
1046 .value = 1,
1047 },
1048 IdAndValue{
1049 .id = 0,
1050 .value = 2,
1051 },
1052 IdAndValue{
1053 .id = 1,
1054 .value = 0,
1055 },
1056 IdAndValue{
1057 .id = 1,
1058 .value = 1,
1059 },
1060 IdAndValue{
1061 .id = 1,
1062 .value = 2,
1063 },
1064 IdAndValue{
1065 .id = 2,
1066 .value = 0,
1067 },
1068 IdAndValue{
1069 .id = 2,
1070 .value = 1,
1071 },
1072 IdAndValue{
1073 .id = 2,
1074 .value = 2,
1075 },
1009 IdAndValue{ .id = 0, .value = 0 },
1010 IdAndValue{ .id = 0, .value = 1 },
1011 IdAndValue{ .id = 0, .value = 2 },
1012 IdAndValue{ .id = 1, .value = 0 },
1013 IdAndValue{ .id = 1, .value = 1 },
1014 IdAndValue{ .id = 1, .value = 2 },
1015 IdAndValue{ .id = 2, .value = 0 },
1016 IdAndValue{ .id = 2, .value = 1 },
1017 IdAndValue{ .id = 2, .value = 2 },
10761018 },
10771019 []IdAndValue{
1078 IdAndValue{
1079 .id = 0,
1080 .value = 2,
1081 },
1082 IdAndValue{
1083 .id = 0,
1084 .value = 1,
1085 },
1086 IdAndValue{
1087 .id = 0,
1088 .value = 0,
1089 },
1090 IdAndValue{
1091 .id = 1,
1092 .value = 2,
1093 },
1094 IdAndValue{
1095 .id = 1,
1096 .value = 1,
1097 },
1098 IdAndValue{
1099 .id = 1,
1100 .value = 0,
1101 },
1102 IdAndValue{
1103 .id = 2,
1104 .value = 2,
1105 },
1106 IdAndValue{
1107 .id = 2,
1108 .value = 1,
1109 },
1110 IdAndValue{
1111 .id = 2,
1112 .value = 0,
1113 },
1020 IdAndValue{ .id = 0, .value = 2 },
1021 IdAndValue{ .id = 0, .value = 1 },
1022 IdAndValue{ .id = 0, .value = 0 },
1023 IdAndValue{ .id = 1, .value = 2 },
1024 IdAndValue{ .id = 1, .value = 1 },
1025 IdAndValue{ .id = 1, .value = 0 },
1026 IdAndValue{ .id = 2, .value = 2 },
1027 IdAndValue{ .id = 2, .value = 1 },
1028 IdAndValue{ .id = 2, .value = 0 },
11141029 },
11151030 };
11161031 for (cases) |*case| {
......@@ -1125,8 +1040,8 @@ const IdAndValue = struct {
11251040 id: usize,
11261041 value: i32,
11271042};
1128fn cmpByValue(a: *const IdAndValue, b: *const IdAndValue) bool {
1129 return i32asc(a.value, b.value);
1043fn cmpByValue(a: IdAndValue, b: IdAndValue) bool {
1044 return asc(i32)(a.value, b.value);
11301045}
11311046
11321047test "std.sort" {
......@@ -1161,7 +1076,7 @@ test "std.sort" {
11611076 var buf: [8]u8 = undefined;
11621077 const slice = buf[0..case[0].len];
11631078 mem.copy(u8, slice, case[0]);
1164 sort(u8, slice, u8asc);
1079 sort(u8, slice, asc(u8));
11651080 assert(mem.eql(u8, slice, case[1]));
11661081 }
11671082
......@@ -1175,48 +1090,20 @@ test "std.sort" {
11751090 []i32{1},
11761091 },
11771092 [][]const i32{
1178 []i32{
1179 0,
1180 1,
1181 },
1182 []i32{
1183 0,
1184 1,
1185 },
1093 []i32{ 0, 1 },
1094 []i32{ 0, 1 },
11861095 },
11871096 [][]const i32{
1188 []i32{
1189 1,
1190 0,
1191 },
1192 []i32{
1193 0,
1194 1,
1195 },
1097 []i32{ 1, 0 },
1098 []i32{ 0, 1 },
11961099 },
11971100 [][]const i32{
1198 []i32{
1199 1,
1200 -1,
1201 0,
1202 },
1203 []i32{
1204 -1,
1205 0,
1206 1,
1207 },
1101 []i32{ 1, -1, 0 },
1102 []i32{ -1, 0, 1 },
12081103 },
12091104 [][]const i32{
1210 []i32{
1211 2,
1212 1,
1213 3,
1214 },
1215 []i32{
1216 1,
1217 2,
1218 3,
1219 },
1105 []i32{ 2, 1, 3 },
1106 []i32{ 1, 2, 3 },
12201107 },
12211108 };
12221109
......@@ -1224,7 +1111,7 @@ test "std.sort" {
12241111 var buf: [8]i32 = undefined;
12251112 const slice = buf[0..case[0].len];
12261113 mem.copy(i32, slice, case[0]);
1227 sort(i32, slice, i32asc);
1114 sort(i32, slice, asc(i32));
12281115 assert(mem.eql(i32, slice, case[1]));
12291116 }
12301117}
......@@ -1240,48 +1127,20 @@ test "std.sort descending" {
12401127 []i32{1},
12411128 },
12421129 [][]const i32{
1243 []i32{
1244 0,
1245 1,
1246 },
1247 []i32{
1248 1,
1249 0,
1250 },
1130 []i32{ 0, 1 },
1131 []i32{ 1, 0 },
12511132 },
12521133 [][]const i32{
1253 []i32{
1254 1,
1255 0,
1256 },
1257 []i32{
1258 1,
1259 0,
1260 },
1134 []i32{ 1, 0 },
1135 []i32{ 1, 0 },
12611136 },
12621137 [][]const i32{
1263 []i32{
1264 1,
1265 -1,
1266 0,
1267 },
1268 []i32{
1269 1,
1270 0,
1271 -1,
1272 },
1138 []i32{ 1, -1, 0 },
1139 []i32{ 1, 0, -1 },
12731140 },
12741141 [][]const i32{
1275 []i32{
1276 2,
1277 1,
1278 3,
1279 },
1280 []i32{
1281 3,
1282 2,
1283 1,
1284 },
1142 []i32{ 2, 1, 3 },
1143 []i32{ 3, 2, 1 },
12851144 },
12861145 };
12871146
......@@ -1289,28 +1148,16 @@ test "std.sort descending" {
12891148 var buf: [8]i32 = undefined;
12901149 const slice = buf[0..case[0].len];
12911150 mem.copy(i32, slice, case[0]);
1292 sort(i32, slice, i32desc);
1151 sort(i32, slice, desc(i32));
12931152 assert(mem.eql(i32, slice, case[1]));
12941153 }
12951154}
12961155
12971156test "another sort case" {
1298 var arr = []i32{
1299 5,
1300 3,
1301 1,
1302 2,
1303 4,
1304 };
1305 sort(i32, arr[0..], i32asc);
1306
1307 assert(mem.eql(i32, arr, []i32{
1308 1,
1309 2,
1310 3,
1311 4,
1312 5,
1313 }));
1157 var arr = []i32{ 5, 3, 1, 2, 4 };
1158 sort(i32, arr[0..], asc(i32));
1159
1160 assert(mem.eql(i32, arr, []i32{ 1, 2, 3, 4, 5 }));
13141161}
13151162
13161163test "sort fuzz testing" {
......@@ -1345,7 +1192,7 @@ fn fuzzTest(rng: *std.rand.Random) void {
13451192 }
13461193}
13471194
1348pub fn min(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *const T) bool) T {
1195pub fn min(comptime T: type, items: []T, lessThan: fn (lhs: T, rhs: T) bool) T {
13491196 var i: usize = 0;
13501197 var smallest = items[0];
13511198 for (items[1..]) |item| {
......@@ -1356,7 +1203,7 @@ pub fn min(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *cons
13561203 return smallest;
13571204}
13581205
1359pub fn max(comptime T: type, items: []T, lessThan: fn (lhs: *const T, rhs: *const T) bool) T {
1206pub fn max(comptime T: type, items: []T, lessThan: fn (lhs: T, rhs: T) bool) T {
13601207 var i: usize = 0;
13611208 var biggest = items[0];
13621209 for (items[1..]) |item| {