This module implements some common generic algorithms on openArrays.
Basic usage
Example:
import std/algorithm type People = tuple year: int name: string var a: seq[People] a.add((2000, "John")) a.add((2005, "Marie")) a.add((2010, "Jane")) # Sorting with default system.cmp a.sort() assert a == @[(year: 2000, name: "John"), (year: 2005, name: "Marie"), (year: 2010, name: "Jane")] proc myCmp(x, y: People): int = cmp(x.name, y.name) # Sorting with custom proc a.sort(myCmp) assert a == @[(year: 2010, name: "Jane"), (year: 2000, name: "John"), (year: 2005, name: "Marie")]
See also
- sequtils module for working with the built-in seq type
- tables module for sorting tables
Procs
-
Flips the sign of x if order == Descending. If order == Ascending then x is returned.
x is supposed to be the result of a comparator, i.e.
< 0 for less than,
== 0 for equal,
> 0 for greater than.Example:
Source Editassert -123 * Descending == 123 assert 123 * Descending == -123 assert -123 * Ascending == -123 assert 123 * Ascending == 123
proc binarySearch[T, K](a: openArray[T]; key: K; cmp: proc (x: T; y: K): int {.closure.}): int {. effectsOf: cmp.}
-
Binary search for key in a. Return the index of key or -1 if not found. Assumes that a is sorted according to cmp.
cmp is the comparator function to use, the expected return values are the same as those of system.cmp.
Example:
Source Editassert binarySearch(["a", "b", "c", "d"], "d", system.cmp[string]) == 3 assert binarySearch(["a", "b", "c", "d"], "c", system.cmp[string]) == 2
proc binarySearch[T](a: openArray[T]; key: T): int
-
Binary search for key in a. Return the index of key or -1 if not found. Assumes that a is sorted.
Example:
Source Editassert binarySearch([0, 1, 2, 3, 4], 4) == 4 assert binarySearch([0, 1, 2, 3, 4], 2) == 2
-
Checks to see whether a is already sorted in order using cmp for the comparison. The parameters are identical to sort. Requires O(n) time.
See also:
Example:
Source Editlet a = [2, 3, 1, 5, 4] b = [1, 2, 3, 4, 5] c = [5, 4, 3, 2, 1] d = ["adam", "brian", "cat", "dande"] e = ["adam", "dande", "brian", "cat"] assert isSorted(a) == false assert isSorted(b) == true assert isSorted(c) == false assert isSorted(c, Descending) == true assert isSorted(d) == true assert isSorted(e) == false
-
Shortcut version of isSorted that uses system.cmp[T] as the comparison function.
See also:
Example:
Source Editlet a = [2, 3, 1, 5, 4] b = [1, 2, 3, 4, 5] c = [5, 4, 3, 2, 1] d = ["adam", "brian", "cat", "dande"] e = ["adam", "dande", "brian", "cat"] assert isSorted(a) == false assert isSorted(b) == true assert isSorted(c) == false assert isSorted(c, Descending) == true assert isSorted(d) == true assert isSorted(e) == false
proc lowerBound[T, K](a: openArray[T]; key: K; cmp: proc (x: T; k: K): int {.closure.}): int {. effectsOf: cmp.}
-
Returns the index of the first element in a that is not less than (i.e. greater or equal to) key, or last if no such element is found. In other words if you have a sorted sequence and you call insert(thing, elm, lowerBound(thing, elm)) the sequence will still be sorted. Assumes that a is sorted according to cmp.
If an invalid range is passed, it raises IndexDefect.
This version uses cmp to compare the elements. The expected return values are the same as those of system.cmp.
See also:
- upperBound proc sorted by cmp in the specified order
- upperBound proc
Example:
Source Editvar arr = @[1, 2, 3, 5, 6, 7, 8, 9] assert arr.lowerBound(3, system.cmp[int]) == 2 assert arr.lowerBound(4, system.cmp[int]) == 3 assert arr.lowerBound(5, system.cmp[int]) == 3 arr.insert(4, arr.lowerBound(4, system.cmp[int])) assert arr == [1, 2, 3, 4, 5, 6, 7, 8, 9]
proc lowerBound[T](a: openArray[T]; key: T): int
-
Returns the index of the first element in a that is not less than (i.e. greater or equal to) key, or last if no such element is found. In other words if you have a sorted sequence and you call insert(thing, elm, lowerBound(thing, elm)) the sequence will still be sorted. Assumes that a is sorted.
This version uses the default comparison function cmp.
See also:
- upperBound proc sorted by cmp in the specified order
- upperBound proc
-
Shortcut version of merge that uses system.cmp[T] as the comparison function.
See also:
Example:
Source Editlet x = [5, 10, 15, 20, 25] let y = [50, 40, 30, 20, 10].sorted var merged: seq[int] merged.merge(x, y) assert merged.isSorted assert merged == @[5, 10, 10, 15, 20, 20, 25, 30, 40, 50]
-
Merges two sorted openArray. x and y are assumed to be sorted. If you do not wish to provide your own cmp, you may use system.cmp or instead call the overloaded version of merge, which uses system.cmp.Note: The original data of result is not cleared, new data is appended to result.
See also:
Example:
Source Editlet x = @[1, 3, 6] let y = @[2, 3, 4] block: var merged = @[7] # new data is appended to merged sequence merged.merge(x, y, system.cmp[int]) assert merged == @[7, 1, 2, 3, 3, 4, 6] block: var merged = @[7] # if you only want new data, clear merged sequence first merged.setLen(0) merged.merge(x, y, system.cmp[int]) assert merged.isSorted assert merged == @[1, 2, 3, 3, 4, 6] import std/sugar var res: seq[(int, int)] res.merge([(1, 1)], [(1, 2)], (a, b) => a[0] - b[0]) assert res == @[(1, 1), (1, 2)] assert seq[int].default.dup(merge([1, 3], [2, 4])) == @[1, 2, 3, 4]
proc nextPermutation[T](x: var openArray[T]): bool {.discardable.}
-
Calculates the next lexicographic permutation, directly modifying x. The result is whether a permutation happened, otherwise we have reached the last-ordered permutation.
If you start with an unsorted array/seq, the repeated permutations will not give you all permutations but stop with the last.
See also:
Example:
Source Editvar v = @[0, 1, 2, 3] assert v.nextPermutation() == true assert v == @[0, 1, 3, 2] assert v.nextPermutation() == true assert v == @[0, 2, 1, 3] assert v.prevPermutation() == true assert v == @[0, 1, 3, 2] v = @[3, 2, 1, 0] assert v.nextPermutation() == false assert v == @[3, 2, 1, 0]
proc prevPermutation[T](x: var openArray[T]): bool {.discardable.}
-
Calculates the previous lexicographic permutation, directly modifying x. The result is whether a permutation happened, otherwise we have reached the first-ordered permutation.
See also:
Example:
Source Editvar v = @[0, 1, 2, 3] assert v.prevPermutation() == false assert v == @[0, 1, 2, 3] assert v.nextPermutation() == true assert v == @[0, 1, 3, 2] assert v.prevPermutation() == true assert v == @[0, 1, 2, 3]
-
Produces the Cartesian product of the array. Every element of the result is a combination of one element from each seq in x, with the ith element coming from x[i].Warning: complexity may explode.
Example:
Source Editassert product(@[@[1], @[2]]) == @[@[1, 2]] assert product(@[@["A", "K"], @["Q"]]) == @[@["K", "Q"], @["A", "Q"]]
-
Reverses the contents of the container a.
See also:
- reversed proc reverse a slice and returns a seq[T]
- reversed proc reverse and returns a seq[T]
Example:
Source Editvar a = [1, 2, 3, 4, 5, 6] a.reverse() assert a == [6, 5, 4, 3, 2, 1] a.reverse() assert a == [1, 2, 3, 4, 5, 6]
-
Reverses the slice a[first..last].
If an invalid range is passed, it raises IndexDefect.
See also:
- reversed proc reverse a slice and returns a seq[T]
- reversed proc reverse and returns a seq[T]
Example:
Source Editvar a = [1, 2, 3, 4, 5, 6] a.reverse(1, 3) assert a == [1, 4, 3, 2, 5, 6] a.reverse(1, 3) assert a == [1, 2, 3, 4, 5, 6] doAssertRaises(IndexDefect, a.reverse(1, 7))
proc rotatedLeft[T](arg: openArray[T]; dist: int): seq[T]
-
Same as rotateLeft, just with the difference that it does not modify the argument. It creates a new seq instead.
See also:
- rotateLeft proc for the in-place version of this proc
- rotatedLeft proc for a version which rotates a range
Example:
Source Editvar a = @[1, 2, 3, 4, 5] a = rotatedLeft(a, 2) assert a == @[3, 4, 5, 1, 2] a = rotatedLeft(a, 4) assert a == @[2, 3, 4, 5, 1] a = rotatedLeft(a, -6) assert a == @[1, 2, 3, 4, 5]
-
Same as rotateLeft, just with the difference that it does not modify the argument. It creates a new seq instead.
Elements outside of slice will be left unchanged. If an invalid range (HSlice) is passed, it raises IndexDefect.
- slice
- The indices of the element range that should be rotated.
- dist
- The distance in amount of elements that the data should be rotated. Can be negative, can be any number.
See also:
- rotateLeft proc for the in-place version of this proc
- rotatedLeft proc for a version which rotates the whole container
Example:
Source Editvar a = @[1, 2, 3, 4, 5] a = rotatedLeft(a, 1 .. 4, 3) assert a == @[1, 5, 2, 3, 4] a = rotatedLeft(a, 1 .. 3, 2) assert a == @[1, 3, 5, 2, 4] a = rotatedLeft(a, 1 .. 3, -2) assert a == @[1, 5, 2, 3, 4]
proc rotateLeft[T](arg: var openArray[T]; dist: int): int {.discardable.}
-
Same as rotateLeft, but with default arguments for slice, so that this procedure operates on the entire arg, and not just on a part of it.
See also:
- rotateLeft proc for a version which rotates a range
- rotatedLeft proc for a version which returns a seq[T]
Example:
Source Editvar a = [1, 2, 3, 4, 5] a.rotateLeft(2) assert a == [3, 4, 5, 1, 2] a.rotateLeft(4) assert a == [2, 3, 4, 5, 1] a.rotateLeft(-6) assert a == [1, 2, 3, 4, 5]
-
Performs a left rotation on a range of elements. If you want to rotate right, use a negative dist. Specifically, rotateLeft rotates the elements at slice by dist positions.
The element at index slice.a + dist will be at index slice.a.
The element at index slice.b will be at slice.a + dist - 1.
The element at index slice.a will be at slice.b + 1 - dist.
The element at index slice.a + dist - 1 will be at slice.b.Elements outside of slice will be left unchanged. The time complexity is linear to slice.b - slice.a + 1. If an invalid range (HSlice) is passed, it raises IndexDefect.
- slice
- The indices of the element range that should be rotated.
- dist
- The distance in amount of elements that the data should be rotated. Can be negative, can be any number.
See also:
- rotateLeft proc for a version which rotates the whole container
- rotatedLeft proc for a version which returns a seq[T]
Example:
Source Editvar a = [0, 1, 2, 3, 4, 5] a.rotateLeft(1 .. 4, 3) assert a == [0, 4, 1, 2, 3, 5] a.rotateLeft(1 .. 4, 3) assert a == [0, 3, 4, 1, 2, 5] a.rotateLeft(1 .. 4, -3) assert a == [0, 4, 1, 2, 3, 5] doAssertRaises(IndexDefect, a.rotateLeft(1 .. 7, 2))
-
Default Nim sort (an implementation of merge sort). The sorting is guaranteed to be stable (that is, equal elements stay in the same order) and the worst case is guaranteed to be O(n log n). Sorts by cmp in the specified order.
The current implementation uses an iterative mergesort to achieve this. It uses a temporary sequence of length a.len div 2. If you do not wish to provide your own cmp, you may use system.cmp or instead call the overloaded version of sort, which uses system.cmp.
sort(myIntArray, system.cmp[int]) # do not use cmp[string] here as we want to use the specialized # overload: sort(myStrArray, system.cmp)
You can inline adhoc comparison procs with the do notation. Example:
people.sort do (x, y: Person) -> int: result = cmp(x.surname, y.surname) if result == 0: result = cmp(x.name, y.name)
See also:
- sort proc
- sorted proc sorted by cmp in the specified order
- sorted proc
- sortedByIt template
Example:
Source Editvar d = ["boo", "fo", "barr", "qux"] proc myCmp(x, y: string): int = if x.len() > y.len() or x.len() == y.len(): 1 else: -1 sort(d, myCmp) assert d == ["fo", "qux", "boo", "barr"]
-
Shortcut version of sort that uses system.cmp[T] as the comparison function.
See also:
- sort func
- sorted proc sorted by cmp in the specified order
- sorted proc
- sortedByIt template
-
Returns a sorted by cmp in the specified order.
See also:
Example:
Source Editlet a = [2, 3, 1, 5, 4] b = sorted(a, system.cmp[int]) c = sorted(a, system.cmp[int], Descending) d = sorted(["adam", "dande", "brian", "cat"], system.cmp[string]) assert b == @[1, 2, 3, 4, 5] assert c == @[5, 4, 3, 2, 1] assert d == @["adam", "brian", "cat", "dande"]
-
Shortcut version of sorted that uses system.cmp[T] as the comparison function.
See also:
Example:
Source Editlet a = [2, 3, 1, 5, 4] b = sorted(a) c = sorted(a, Descending) d = sorted(["adam", "dande", "brian", "cat"]) assert b == @[1, 2, 3, 4, 5] assert c == @[5, 4, 3, 2, 1] assert d == @["adam", "brian", "cat", "dande"]
proc upperBound[T, K](a: openArray[T]; key: K; cmp: proc (x: T; k: K): int {.closure.}): int {. effectsOf: cmp.}
-
Returns the index of the first element in a that is greater than key, or last if no such element is found. In other words if you have a sorted sequence and you call insert(thing, elm, upperBound(thing, elm)) the sequence will still be sorted. Assumes that a is sorted according to cmp.
If an invalid range is passed, it raises IndexDefect.
This version uses cmp to compare the elements. The expected return values are the same as those of system.cmp.
See also:
- lowerBound proc sorted by cmp in the specified order
- lowerBound proc
Example:
Source Editvar arr = @[1, 2, 3, 5, 6, 7, 8, 9] assert arr.upperBound(2, system.cmp[int]) == 2 assert arr.upperBound(3, system.cmp[int]) == 3 assert arr.upperBound(4, system.cmp[int]) == 3 arr.insert(4, arr.upperBound(3, system.cmp[int])) assert arr == [1, 2, 3, 4, 5, 6, 7, 8, 9]
proc upperBound[T](a: openArray[T]; key: T): int
-
Returns the index of the first element in a that is greater than key, or last if no such element is found. In other words if you have a sorted sequence and you call insert(thing, elm, upperBound(thing, elm)) the sequence will still be sorted. Assumes that a is sorted.
This version uses the default comparison function cmp.
See also:
- lowerBound proc sorted by cmp in the specified order
- lowerBound proc
Templates
template sortedByIt(seq1, op: untyped): untyped
-
Convenience template around the sorted proc to reduce typing.
The template injects the it variable which you can use directly in an expression.
Because the underlying cmp() is defined for tuples you can also do a nested sort.
See also:
- sort func
- sort proc
- sorted proc sorted by cmp in the specified order
- sorted proc
Example:
Source Edittype Person = tuple[name: string, age: int] var p1: Person = (name: "p1", age: 60) p2: Person = (name: "p2", age: 20) p3: Person = (name: "p3", age: 30) p4: Person = (name: "p4", age: 30) people = @[p1, p2, p4, p3] assert people.sortedByIt(it.name) == @[(name: "p1", age: 60), (name: "p2", age: 20), (name: "p3", age: 30), (name: "p4", age: 30)] # Nested sort assert people.sortedByIt((it.age, it.name)) == @[(name: "p2", age: 20), (name: "p3", age: 30), (name: "p4", age: 30), (name: "p1", age: 60)]