Although this module has seq in its name, it implements operations not only for seq type, but for three built-in container types under the openArray umbrella:
- sequences
- strings
- array
The system module defines several common functions, such as:
- newSeq[T] for creating new sequences of type T
- @ for converting arrays and strings to sequences
- add for adding new elements to strings and sequences
- & for string and seq concatenation
- in (alias for contains) and notin for checking if an item is in a container
This module builds upon that, providing additional functionality in form of procs, iterators and templates inspired by functional programming languages.
For functional style programming you have different options at your disposal:
- sugar.collect macro
- pass anonymous proc
- import sugar module and use => macro
- use ...It templates (mapIt, filterIt, etc.)
The chaining of functions is possible thanks to the method call syntax.
import sequtils, sugar # Creating a sequence from 1 to 10, multiplying each member by 2, # keeping only the members which are not divisible by 6. let foo = toSeq(1..10).map(x => x*2).filter(x => x mod 6 != 0) bar = toSeq(1..10).mapIt(it*2).filterIt(it mod 6 != 0) baz = collect(newSeq): for i in 1..10: let j = 2*i if j mod 6 != 0: j doAssert foo == bar doAssert foo == baz echo foo # @[2, 4, 8, 10, 14, 16, 20] echo foo.any(x => x > 17) # true echo bar.allIt(it < 20) # false echo foo.foldl(a + b) # 74; sum of all members
import sequtils from strutils import join let vowels = @"aeiou" # creates a sequence @['a', 'e', 'i', 'o', 'u'] foo = "sequtils is an awesome module" echo foo.filterIt(it notin vowels).join # "sqtls s n wsm mdl"
See also:
- strutils module for common string functions
- sugar module for syntactic sugar macros
- algorithm module for common generic algorithms
- json module for a structure which allows heterogeneous members
Procs
proc concat[T](seqs: varargs[seq[T]]): seq[T]
-
Takes several sequences' items and returns them inside a new sequence. All sequences must be of the same type.
See also:
- distribute proc for a reverse operation
Example:
let s1 = @[1, 2, 3] s2 = @[4, 5] s3 = @[6, 7] total = concat(s1, s2, s3) assert total == @[1, 2, 3, 4, 5, 6, 7]
Source Edit proc count[T](s: openArray[T]; x: T): int
-
Returns the number of occurrences of the item x in the container s.
Example:
let a = @[1, 2, 2, 3, 2, 4, 2] b = "abracadabra" assert count(a, 2) == 4 assert count(a, 99) == 0 assert count(b, 'r') == 2
Source Edit proc cycle[T](s: openArray[T]; n: Natural): seq[T]
-
Returns a new sequence with the items of the container s repeated n times. n must be a non-negative number (zero or more).
Example:
let s = @[1, 2, 3] total = s.cycle(3) assert total == @[1, 2, 3, 1, 2, 3, 1, 2, 3]
Source Edit proc repeat[T](x: T; n: Natural): seq[T]
-
Returns a new sequence with the item x repeated n times. n must be a non-negative number (zero or more).
Example:
let total = repeat(5, 3) assert total == @[5, 5, 5]
Source Edit proc deduplicate[T](s: openArray[T]; isSorted: bool = false): seq[T]
-
Returns a new sequence without duplicates.
Setting the optional argument isSorted to true (default: false) uses a faster algorithm for deduplication.
Example:
let dup1 = @[1, 1, 3, 4, 2, 2, 8, 1, 4] dup2 = @["a", "a", "c", "d", "d"] unique1 = deduplicate(dup1) unique2 = deduplicate(dup2, isSorted = true) assert unique1 == @[1, 3, 4, 2, 8] assert unique2 == @["a", "c", "d"]
Source Edit proc minIndex[T](s: openArray[T]): int
-
Returns the index of the minimum value of s. T needs to have a < operator.
Example:
let a = @[1, 2, 3, 4] b = @[6, 5, 4, 3] c = [2, -7, 8, -5] d = "ziggy" assert minIndex(a) == 0 assert minIndex(b) == 3 assert minIndex(c) == 1 assert minIndex(d) == 2
Source Edit proc maxIndex[T](s: openArray[T]): int
-
Returns the index of the maximum value of s. T needs to have a < operator.
Example:
let a = @[1, 2, 3, 4] b = @[6, 5, 4, 3] c = [2, -7, 8, -5] d = "ziggy" assert maxIndex(a) == 3 assert maxIndex(b) == 0 assert maxIndex(c) == 2 assert maxIndex(d) == 0
Source Edit proc zip[S, T](s1: openArray[S]; s2: openArray[T]): seq[(S, T)]
-
Returns a new sequence with a combination of the two input containers.
The input containers can be of different types. If one container is shorter, the remaining items in the longer container are discarded.
Note: For Nim 1.0.x and older version, zip returned a seq of named tuple with fields a and b. For Nim versions 1.1.x and newer, zip returns a seq of unnamed tuples.
Example:
let short = @[1, 2, 3] long = @[6, 5, 4, 3, 2, 1] words = @["one", "two", "three"] letters = "abcd" zip1 = zip(short, long) zip2 = zip(short, words) assert zip1 == @[(1, 6), (2, 5), (3, 4)] assert zip2 == @[(1, "one"), (2, "two"), (3, "three")] assert zip1[2][0] == 3 assert zip2[1][1] == "two" when (NimMajor, NimMinor) <= (1, 0): let zip3 = zip(long, letters) assert zip3 == @[(a: 6, b: 'a'), (5, 'b'), (4, 'c'), (3, 'd')] assert zip3[0].b == 'a' else: let zip3: seq[tuple[num: int, letter: char]] = zip(long, letters) assert zip3 == @[(6, 'a'), (5, 'b'), (4, 'c'), (3, 'd')] assert zip3[0].letter == 'a'
Source Edit proc unzip[S, T](s: openArray[(S, T)]): (seq[S], seq[T])
-
Returns a tuple of two sequences split out from a sequence of 2-field tuples.
Example:
let zipped = @[(1, 'a'), (2, 'b'), (3, 'c')] unzipped1 = @[1, 2, 3] unzipped2 = @['a', 'b', 'c'] assert zipped.unzip() == (unzipped1, unzipped2) assert zip(unzipped1, unzipped2).unzip() == (unzipped1, unzipped2)
Source Edit proc distribute[T](s: seq[T]; num: Positive; spread = true): seq[seq[T]]
-
Splits and distributes a sequence s into num sub-sequences.
Returns a sequence of num sequences. For some input values this is the inverse of the concat proc. The input sequence s can be empty, which will produce num empty sequences.
If spread is false and the length of s is not a multiple of num, the proc will max out the first sub-sequence with 1 + len(s) div num entries, leaving the remainder of elements to the last sequence.
On the other hand, if spread is true, the proc will distribute evenly the remainder of the division across all sequences, which makes the result more suited to multithreading where you are passing equal sized work units to a thread pool and want to maximize core usage.
Example:
let numbers = @[1, 2, 3, 4, 5, 6, 7] assert numbers.distribute(3) == @[@[1, 2, 3], @[4, 5], @[6, 7]] assert numbers.distribute(3, false) == @[@[1, 2, 3], @[4, 5, 6], @[7]] assert numbers.distribute(6)[0] == @[1, 2] assert numbers.distribute(6)[1] == @[3]
Source Edit proc map[T, S](s: openArray[T]; op: proc (x: T): S {...}{.closure.}): seq[S] {...}{.inline.}
-
Returns a new sequence with the results of op proc applied to every item in the container s.
Since the input is not modified you can use it to transform the type of the elements in the input container.
Instead of using map and filter, consider using the collect macro from the sugar module.
See also:
- sugar.collect macro
- mapIt template
- apply proc for the in-place version
Example:
let a = @[1, 2, 3, 4] b = map(a, proc(x: int): string = $x) assert b == @["1", "2", "3", "4"]
Source Edit proc apply[T](s: var openArray[T]; op: proc (x: var T) {...}{.closure.}) {...}{.inline.}
-
Applies op to every item in s modifying it directly.
Note that container s must be declared as a var and it is required for your input and output types to be the same, since s is modified in-place. The parameter function takes a var T type parameter.
See also:
Example:
var a = @["1", "2", "3", "4"] apply(a, proc(x: var string) = x &= "42") assert a == @["142", "242", "342", "442"]
Source Edit proc apply[T](s: var openArray[T]; op: proc (x: T): T {...}{.closure.}) {...}{.inline.}
-
Applies op to every item in s modifying it directly.
Note that container s must be declared as a var and it is required for your input and output types to be the same, since s is modified in-place. The parameter function takes and returns a T type variable.
See also:
Example:
var a = @["1", "2", "3", "4"] apply(a, proc(x: string): string = x & "42") assert a == @["142", "242", "342", "442"]
Source Edit proc apply[T](s: openArray[T]; op: proc (x: T) {...}{.closure.}) {...}{.inline.}
-
Same as apply but for proc that do not return and do not mutate s directly.
Example:
apply([0, 1, 2, 3, 4], proc(item: int) = echo item)
Source Edit proc filter[T](s: openArray[T]; pred: proc (x: T): bool {...}{.closure.}): seq[T] {...}{. inline.}
-
Returns a new sequence with all the items of s that fulfilled the predicate pred (function that returns a bool).
Instead of using map and filter, consider using the collect macro from the sugar module.
See also:
- sugar.collect macro
- filterIt template
- filter iterator
- keepIf proc for the in-place version
Example:
let colors = @["red", "yellow", "black"] f1 = filter(colors, proc(x: string): bool = x.len < 6) f2 = filter(colors, proc(x: string): bool = x.contains('y')) assert f1 == @["red", "black"] assert f2 == @["yellow"]
Source Edit proc keepIf[T](s: var seq[T]; pred: proc (x: T): bool {...}{.closure.}) {...}{.inline.}
-
Keeps the items in the passed sequence s if they fulfilled the predicate pred (function that returns a bool).
Note that s must be declared as a var.
Similar to the filter proc, but modifies the sequence directly.
See also:
Example:
var floats = @[13.0, 12.5, 5.8, 2.0, 6.1, 9.9, 10.1] keepIf(floats, proc(x: float): bool = x > 10) assert floats == @[13.0, 12.5, 10.1]
Source Edit proc delete[T](s: var seq[T]; first, last: Natural)
-
Deletes in the items of a sequence s at positions first..last (including both ends of a range). This modifies s itself, it does not return a copy.
Example:
let outcome = @[1, 1, 1, 1, 1, 1, 1, 1] var dest = @[1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1] dest.delete(3, 8) assert outcome == dest
Source Edit proc insert[T](dest: var seq[T]; src: openArray[T]; pos = 0)
-
Inserts items from src into dest at position pos. This modifies dest itself, it does not return a copy.
Notice that src and dest must be of the same type.
Example:
var dest = @[1, 1, 1, 1, 1, 1, 1, 1] let src = @[2, 2, 2, 2, 2, 2] outcome = @[1, 1, 1, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1] dest.insert(src, 3) assert dest == outcome
Source Edit proc all[T](s: openArray[T]; pred: proc (x: T): bool {...}{.closure.}): bool
-
Iterates through a container and checks if every item fulfills the predicate.
See also:
Example:
let numbers = @[1, 4, 5, 8, 9, 7, 4] assert all(numbers, proc (x: int): bool = return x < 10) == true assert all(numbers, proc (x: int): bool = return x < 9) == false
Source Edit proc any[T](s: openArray[T]; pred: proc (x: T): bool {...}{.closure.}): bool
-
Iterates through a container and checks if some item fulfills the predicate.
See also:
Example:
let numbers = @[1, 4, 5, 8, 9, 7, 4] assert any(numbers, proc (x: int): bool = return x > 8) == true assert any(numbers, proc (x: int): bool = return x > 9) == false
Source Edit
Iterators
iterator filter[T](s: openArray[T]; pred: proc (x: T): bool {...}{.closure.}): T
-
Iterates through a container s and yields every item that fulfills the predicate pred (function that returns a bool).
Instead of using map and filter, consider using the collect macro from the sugar module.
See also:
Example:
let numbers = @[1, 4, 5, 8, 9, 7, 4] var evens = newSeq[int]() for n in filter(numbers, proc (x: int): bool = x mod 2 == 0): evens.add(n) assert evens == @[4, 8, 4]
Source Edit iterator items[T](xs: iterator (): T): T
- iterates over each element yielded by a closure iterator. This may not seem particularly useful on its own, but this allows closure iterators to be used by the the mapIt, filterIt, allIt, anyIt, etc. templates. Source Edit
Macros
macro mapLiterals(constructor, op: untyped; nested = true): untyped
-
Applies op to each of the atomic literals like 3 or "abc" in the specified constructor AST. This can be used to map every array element to some target type:
Example:
let x = mapLiterals([0.1, 1.2, 2.3, 3.4], int) doAssert x is array[4, int]
Short notation for:
let x = [int(0.1), int(1.2), int(2.3), int(3.4)]
If nested is true (which is the default), the literals are replaced everywhere in the constructor AST, otherwise only the first level is considered:
let a = mapLiterals((1.2, (2.3, 3.4), 4.8), int) let b = mapLiterals((1.2, (2.3, 3.4), 4.8), int, nested=false) assert a == (1, (2, 3), 4) assert b == (1, (2.3, 3.4), 4) let c = mapLiterals((1, (2, 3), 4, (5, 6)), `$`) let d = mapLiterals((1, (2, 3), 4, (5, 6)), `$`, nested=false) assert c == ("1", ("2", "3"), "4", ("5", "6")) assert d == ("1", (2, 3), "4", (5, 6))
There are no constraints for the constructor AST, it works for nested tuples of arrays of sets etc.
Source Edit
Templates
template filterIt(s, pred: untyped): untyped
-
Returns a new sequence with all the items of s that fulfilled the predicate pred.
Unlike the filter proc and filter iterator, the predicate needs to be an expression using the it variable for testing, like: filterIt("abcxyz", it == 'x').
Instead of using mapIt and filterIt, consider using the collect macro from the sugar module.
See also:
Example:
let temperatures = @[-272.15, -2.0, 24.5, 44.31, 99.9, -113.44] acceptable = temperatures.filterIt(it < 50 and it > -10) notAcceptable = temperatures.filterIt(it > 50 or it < -10) assert acceptable == @[-2.0, 24.5, 44.31] assert notAcceptable == @[-272.15, 99.9, -113.44]
Source Edit template keepItIf(varSeq: seq; pred: untyped)
-
Keeps the items in the passed sequence (must be declared as a var) if they fulfilled the predicate.
Unlike the keepIf proc, the predicate needs to be an expression using the it variable for testing, like: keepItIf("abcxyz", it == 'x').
See also:
Example:
var candidates = @["foo", "bar", "baz", "foobar"] candidates.keepItIf(it.len == 3 and it[0] == 'b') assert candidates == @["bar", "baz"]
Source Edit template countIt(s, pred: untyped): int
-
Returns a count of all the items that fulfilled the predicate.
The predicate needs to be an expression using the it variable for testing, like: countIt(@[1, 2, 3], it > 2).
Example:
let numbers = @[-3, -2, -1, 0, 1, 2, 3, 4, 5, 6] iterator iota(n: int): int = for i in 0..<n: yield i assert numbers.countIt(it < 0) == 3 assert countIt(iota(10), it < 2) == 2
Source Edit template allIt(s, pred: untyped): bool
-
Iterates through a container and checks if every item fulfills the predicate.
Unlike the all proc, the predicate needs to be an expression using the it variable for testing, like: allIt("abba", it == 'a').
See also:
Example:
let numbers = @[1, 4, 5, 8, 9, 7, 4] assert numbers.allIt(it < 10) == true assert numbers.allIt(it < 9) == false
Source Edit template anyIt(s, pred: untyped): bool
-
Iterates through a container and checks if some item fulfills the predicate.
Unlike the any proc, the predicate needs to be an expression using the it variable for testing, like: anyIt("abba", it == 'a').
See also:
Example:
let numbers = @[1, 4, 5, 8, 9, 7, 4] assert numbers.anyIt(it > 8) == true assert numbers.anyIt(it > 9) == false
Source Edit template toSeq(iter: untyped): untyped
-
Transforms any iterable (anything that can be iterated over, e.g. with a for-loop) into a sequence.
Example:
let myRange = 1..5 mySet: set[int8] = {5'i8, 3, 1} assert typeof(myRange) is HSlice[system.int, system.int] assert typeof(mySet) is set[int8] let mySeq1 = toSeq(myRange) mySeq2 = toSeq(mySet) assert mySeq1 == @[1, 2, 3, 4, 5] assert mySeq2 == @[1'i8, 3, 5]
Source Edit template foldl(sequence, operation: untyped): untyped
-
Template to fold a sequence from left to right, returning the accumulation.
The sequence is required to have at least a single element. Debug versions of your program will assert in this situation but release versions will happily go ahead. If the sequence has a single element it will be returned without applying operation.
The operation parameter should be an expression which uses the variables a and b for each step of the fold. Since this is a left fold, for non associative binary operations like subtraction think that the sequence of numbers 1, 2 and 3 will be parenthesized as (((1) - 2) - 3).
See also:
- foldl template with a starting parameter
- foldr template
Example:
let numbers = @[5, 9, 11] addition = foldl(numbers, a + b) subtraction = foldl(numbers, a - b) multiplication = foldl(numbers, a * b) words = @["nim", "is", "cool"] concatenation = foldl(words, a & b) procs = @["proc", "Is", "Also", "Fine"] proc foo(acc, cur: string): string = result = acc & cur assert addition == 25, "Addition is (((5)+9)+11)" assert subtraction == -15, "Subtraction is (((5)-9)-11)" assert multiplication == 495, "Multiplication is (((5)*9)*11)" assert concatenation == "nimiscool" assert foldl(procs, foo(a, b)) == "procIsAlsoFine"
Source Edit template foldl(sequence, operation, first): untyped
-
Template to fold a sequence from left to right, returning the accumulation.
This version of foldl gets a starting parameter. This makes it possible to accumulate the sequence into a different type than the sequence elements.
The operation parameter should be an expression which uses the variables a and b for each step of the fold. The first parameter is the start value (the first a) and therefor defines the type of the result.
See also:
Example:
let numbers = @[0, 8, 1, 5] digits = foldl(numbers, a & (chr(b + ord('0'))), "") assert digits == "0815"
Source Edit template foldr(sequence, operation: untyped): untyped
-
Template to fold a sequence from right to left, returning the accumulation.
The sequence is required to have at least a single element. Debug versions of your program will assert in this situation but release versions will happily go ahead. If the sequence has a single element it will be returned without applying operation.
The operation parameter should be an expression which uses the variables a and b for each step of the fold. Since this is a right fold, for non associative binary operations like subtraction think that the sequence of numbers 1, 2 and 3 will be parenthesized as (1 - (2 - (3))).
See also:
- foldl template
- foldl template with a starting parameter
Example:
let numbers = @[5, 9, 11] addition = foldr(numbers, a + b) subtraction = foldr(numbers, a - b) multiplication = foldr(numbers, a * b) words = @["nim", "is", "cool"] concatenation = foldr(words, a & b) assert addition == 25, "Addition is (5+(9+(11)))" assert subtraction == 7, "Subtraction is (5-(9-(11)))" assert multiplication == 495, "Multiplication is (5*(9*(11)))" assert concatenation == "nimiscool"
Source Edit template mapIt(s: typed; op: untyped): untyped
-
Returns a new sequence with the results of op proc applied to every item in the container s.
Since the input is not modified you can use it to transform the type of the elements in the input container.
The template injects the it variable which you can use directly in an expression.
Instead of using mapIt and filterIt, consider using the collect macro from the sugar module.
See also:
- sugar.collect macro
- map proc
- applyIt template for the in-place version
Example:
let nums = @[1, 2, 3, 4] strings = nums.mapIt($(4 * it)) assert strings == @["4", "8", "12", "16"]
Source Edit template applyIt(varSeq, op: untyped)
-
Convenience template around the mutable apply proc to reduce typing.
The template injects the it variable which you can use directly in an expression. The expression has to return the same type as the sequence you are mutating.
See also:
Example:
var nums = @[1, 2, 3, 4] nums.applyIt(it * 3) assert nums[0] + nums[3] == 15
Source Edit template newSeqWith(len: int; init: untyped): untyped
-
Creates a new sequence of length len, calling init to initialize each value of the sequence.
Useful for creating "2D" sequences - sequences containing other sequences or to populate fields of the created sequence.
Example:
## Creates a sequence containing 5 bool sequences, each of length of 3. var seq2D = newSeqWith(5, newSeq[bool](3)) assert seq2D.len == 5 assert seq2D[0].len == 3 assert seq2D[4][2] == false ## Creates a sequence of 20 random numbers from 1 to 10 import random var seqRand = newSeqWith(20, rand(10))
Source Edit