Search Algorithms

In this lesson, we cover the 8 most useful searching algorithms within the standard library: find, find_if, find_if_not, find_first_of, adjacent_find, search_n, search, and find_end.

All the algorithms in this section are available within the <algorithm> header:

1#include <algorithm>

We use these algorithms to traverse through our containers to find objects, or sequences of objects, in different ways.

std::ranges::find()

The std::ranges::find() algorithm is a basic, bread-and-butter search. We pass it a container and an object we’re looking for in that container. The algorithm returns an iterator to the first element that it finds:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 5};
7
8  auto Result{std::ranges::find(Numbers, 3)};
9
10  std::cout << "Found a " << *Result
11            << " in position "
12            << std::distance(Numbers.begin(),
13                             Result);
14}

1Found a 3 in position 2

If our object is not found, the algorithm will return an end iterator, which is equivalent to what is returned from calling end() on our input range.

Therefore, if we’re not sure if our range includes what we’re looking for, we should perform that equality check before dereferencing the returned iterator:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 5};
7
8  auto Result{std::ranges::find(Numbers, 6)};
9
10  if (Result == Numbers.end()) {
11    std::cout << "We did not find a 6";
12  } else {
13    std::cout << "We found a 6 - safe to "
14                 "dereference";
15  }
16}

1We did not find a 6

Sentinel and Iterator Based Algorithms

As with any range-based algorithm, we can provide our range as an iterator-sentinel pair:

1#include <algorithm>
2#include <vector>
3
4int main() {
5  std::vector Numbers{1, 2, 3, 4, 5};
6
7  std::ranges::find(
8    Numbers.begin(), Numbers.end(), 3);
9}

std::ranges::find(), and and every other algorithm we cover in this lesson, has an older variation that works with iterators. These are accessible by excluding the ranges namespace qualifier when calling the functions - for example, we'd replace std::ranges::find() with std::find():

1#include <algorithm>
2#include <vector>
3
4int main() {
5  std::vector Numbers{1, 2, 3, 4, 5};
6
7  std::find(Numbers.begin(), Numbers.end(), 3);
8}

Search Algorithms with Custom Types

For search algorithms to determine if they found what they’re looking for, they use the equality operator == by default.

As such, for a type to be compatible with the search algorithms, it usually needs to implement that operator.

Below, we show an example where a user-defined type is being used:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5struct MyStruct {
6  int Value;
7  bool operator==(const MyStruct& Other) const {
8    return Value == Other.Value;
9  }
10};
11
12int main() {
13  std::vector Numbers{MyStruct{1}, MyStruct{2},
14                      MyStruct{3}, MyStruct{4},
15                      MyStruct{5}};
16
17  auto Result{
18      std::ranges::find(Numbers, MyStruct{2})};
19
20  std::cout << "Found struct with value "
21            << Result->Value;
22}

1Found struct with value: 2

Projection Functions

Like with most range-based algorithms, we can execute searches on projections of our inputs.

Projection Functions

Learn how to use projection functions to apply range-based algorithms on derived data

Below, we project our objects to their absolute value, allowing us to search for 3, and find either 3 or -3:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{-1, 2, -3, 4, -5};
7
8  auto Projector{[](int a) {
9    return std::abs(a);
10  }};
11
12  auto Result{
13    std::ranges::find(Numbers, 3, Projector)};  
14
15  std::cout << "Found a " << *Result;
16}

1Found a -3

std::ranges::find_if()

The find_if() algorithm works similarly to find(). However, instead of performing an equality check using == on each object to determine if it matches an argument, the algorithm passes each object into a predicate function that we provide.

If that predicate returns true, that object is considered to have satisfied the search criteria, and the find_if() algorithm returns an iterator to it.

Below, we search our range for an even number:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 5};
7
8  auto isEven{[](int x) { return x % 2 == 0; }};
9
10  auto Result{
11      std::ranges::find_if(Numbers, isEven)};
12
13  std::cout << "Found a " << *Result
14            << " in position "
15            << std::distance(Numbers.begin(),
16                             Result);
17}

1Found a 2 in position 1

Similar to find(), find_if() will return an end iterator if it doesn’t find anything.

std::ranges::find_if_not()

The find_if_not() algorithm is similar to find_if(), except it will find the first element for which the predicate function returns false.

Below, we find the first odd number, by finding the first number for which our isEven() function returns false:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 5, 3};
7
8  auto isEven{[](int x) { return x % 2 == 0; }};
9
10  auto Result{
11      std::ranges::find_if_not(Numbers, isEven)};
12
13  std::cout << "Found a " << *Result
14            << " in position "
15            << std::distance(Numbers.begin(),
16                             Result);
17}

1Found a 1 in position 0

Similar to the other algorithms, find_if_not() will return an end iterator if it doesn’t find anything.

std::ranges::find_first_of()

The find_first_of() algorithm is useful when we want to find any of a range of possibilities. In the below example, we search our collection for any of 0, 5, or 1. As soon as it finds any of those values in the input range, it will return an iterator to it.

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 5};
7  std::vector SearchTerms{0, 5, 1};
8
9  auto Result{std::ranges::find_first_of(
10      Numbers, SearchTerms)};
11
12  std::cout << "Found a " << *Result
13            << " in position "
14            << std::distance(Numbers.begin(),
15                             Result);
16}

1Found a 1 in position 0

Similar to the other algorithms, find_first_of() will return an end iterator for the input range if it doesn’t find anything.

std::ranges::adjacent_find()

The adjacent_find() algorithm will search our container for the first instance where at least two adjacent objects are equal to each other. It will return an iterator to the first object in the sequence.

In the following example, our input range has two adjacent 4s, which the algorithm finds successfully:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 4, 5};
7
8  auto Result{
9      std::ranges::adjacent_find(Numbers)};
10
11  std::cout << "Found two adjacent " << *Result
12            << "s starting at position "
13            << std::distance(Numbers.begin(),
14                             Result);
15}

1Found two adjacent 4s starting at position 3

adjacent_find() will return an end iterator if it doesn’t find anything.

std::ranges::search_n()

The search_n() algorithm looks for a sequence of elements that match an equality check. As such, it takes 3 inputs:

• The range to search
• The number of sequential elements to search for
• The object to search for

The following example passes Numbers, 2, and 4, therefore the algorithm searches Numbers for at least 2 sequential 4s. It finds it and returns that sequence as a subrange.

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 4, 5};
7
8  auto Result{
9      std::ranges::search_n(Numbers, 2, 4)};
10
11  std::cout << "Found two adjacent 4s starting "
12               "at position "
13            << std::distance(Numbers.begin(),
14                             Result.begin());
15}

1Found two adjacent 4s starting at position 3

A subrange is an example of a C++ view. We introduced these concepts earlier in the course:

Creating Views using std::ranges::subrange

This lesson introduces std::ranges::subrange, allowing us to create non-owning ranges that view some underlying container

Standard Library Views

Learn how to create and use views in C++ using examples from std::views

If the search_n() algorithm does not find what we asked, the subrange it returns will be an empty range, with its begin() iterator being equal to its end() iterator. Both iterators will also be equal to the end() iterator of the input range.

Therefore, we can check if our search did not find anything by checking for these conditions. Below, we show both options:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 4, 5};
7
8  auto Result{std::ranges::search_n(Numbers, 3, 4)};
9
10  if (Result.begin() == Result.end()) {
11    std::cout << "Not found\n";
12  }
13
14  if (Result.begin() == Numbers.end()) {
15    std::cout << "Not found\n";
16  }
17}

1Not found
2Not found

Depending on the specific type of range our view is based on, we may have access to the empty() method, which gives us a friendlier way to check if no results were found:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 4, 4, 5};
7
8  auto Result{std::ranges::search_n(Numbers, 3, 4)};
9
10  if (Result.empty()) {
11    std::cout << "Not found";
12  }
13}

1Not found

std::ranges::search()

The std::ranges::search() algorithm searches a range for a specific subrange. In the following example, we search the range 1, 2, 3, 1, 2, 3 for the subrange 1, 2, 3.

The search() algorithm finds the first matching subrange, which begins at position 0 in this case:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 1, 2, 3};
7  std::vector Subsequence{1, 2, 3};
8
9  auto Result{
10      std::ranges::search(Numbers, Subsequence)};
11
12  if (Result.empty()) {
13    std::cout << "Subsequence not found";
14  } else {
15    std::cout << "Found Subsequence: ";
16    for (int& i : Result) {
17      std::cout << i << ", ";
18    }
19    std::cout
20        << "\nSubsequence starts at position "
21        << std::distance(Numbers.begin(),
22                         Result.begin());
23  }
24}

1Found Subsequence: 1, 2, 3,
2Subsequence starts at position 0

The main use case for searching a range for a subrange comes when working with strings. Standard library strings are also ranges, so are compatible with search() and the other algorithms we’ve shown here.

Below, we use search() to find a string within another string, thereby determining a user’s email provider:

1#include <algorithm>
2#include <iostream>
3#include <string>
4
5int main() {
6  std::string EmailAddress{"new-user@gmail.com"};
7  std::string SearchString{"@gmail."};
8
9  auto Result{std::ranges::search(EmailAddress,
10                                  SearchString)};
11
12  if (!Result.empty()) {
13    std::cout << "User is using Gmail";
14  }
15}

1User is using Gmail

Many string types may support this functionality natively, giving a cleaner API. As of C++23, this includes std::string, which provides the contains() method:

1#include <iostream>
2#include <string>
3
4int main() {
5  std::string EmailAddress{"new-user@gmail.com"};
6  std::string SearchString{"@gmail."};
7
8  auto Result{
9    EmailAddress.contains(SearchString)};  
10
11  if (Result) {
12    std::cout << "User is using Gmail";
13  }
14}

1User is using Gmail

Custom Searchers and Boyer-Moore

There are many different ways to search through a collection.

The std::search() algorithm accepts an optional additional argument, where we can provide a specific implementation. The standard library also bundles a few options we can use directly:

• std::default_searcher (used if we don’t specify a searcher)
• std::boyer_moore_searcher
• std::boyer_moore_horspool_searcher

All of these algorithms are available by including <functional>

Note, the ability to specify a custom searcher is currently restricted to std::search(). It is not available to other algorithms, including std::ranges::search.

The following is an example of how to use the Boyer-Moore searcher:

1#include <algorithm>
2#include <functional>
3#include <iostream>
4#include <string>
5
6int main() {
7  std::string EmailAddress{"new-user@gmail.com"};
8  std::string SearchString{"@gmail."};
9
10  std::boyer_moore_searcher Searcher{
11      SearchString.begin(), SearchString.end()};
12
13  auto Result{std::search(EmailAddress.begin(),
14                          EmailAddress.end(),
15                          Searcher)};
16
17  if (Result != EmailAddress.end()) {
18    std::cout << "User is using Gmail";
19  }
20}

1User is using Gmail

When it comes to choosing which algorithm to use, there generally is no correct answer. The performance characteristics of different algorithms vary on situational factors, such as:

• the length of the substring we’re looking for (sometimes called the needle)
• the length of text we’re searching for the needle in (sometimes called the haystack)
• whether either of those strings is known at compile time

For example, the Boyer-Moore searchers tend to be faster than the default searcher when working with large inputs, like searching through the text of a book. But, they are generally slower when the inputs are smaller strings, like names and email addresses.

In general, if your code is running in a performance-critical context, the best option is just to try out different options with arguments that are representative of your data, and see what works best.

std::ranges::find_end()

Finally, the find_end() algorithm works in much the same way as search(), except it will search in the opposite direction. That means that find_end() will return the last instance of the subsequence we’re searching for.

Here, we find the last instance of 1, 2, 3 within our 1, 2, 3, 1, 2, 3 range:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3, 1, 2, 3};
7  std::vector Subsequence{1, 2, 3};
8
9  auto Result{std::ranges::find_end(Numbers,
10                                    Subsequence)};
11
12  if (Result.empty()) {
13    std::cout << "Subsequence not found";
14  } else {
15    std::cout << "Found Subsequence: ";
16    for (int& i : Result) {
17      std::cout << i << ", ";
18    }
19    std::cout
20        << "\nSubsequence starts at position "
21        << std::distance(Numbers.begin(),
22                         Result.begin());
23  }
24}

1Found Subsequence: 1, 2, 3,
2Subsequence starts at position 3

Even though find_end() works in reverse, when the subsequence is not found, its behavior is equivalent to search(), returning an empty subrange that begins and ends at the end iterator of our input range:

1#include <algorithm>
2#include <iostream>
3#include <vector>
4
5int main() {
6  std::vector Numbers{1, 2, 3};
7  std::vector Subsequence{4, 5, 6};
8
9  auto Result{std::ranges::find_end(
10    Numbers, Subsequence)};
11
12  
13  if (Result.empty()) {
14    std::cout << "Subsequence not found\n";
15  }
16
17  if (Result.end() == Numbers.end()) {
18    std::cout << "Subsequence not found\n";
19  }
20
21  if (Result.begin() == Result.end()) {
22    std::cout << "Subsequence not found\n";
23  }
24}

1Subsequence not found
2Subsequence not found
3Subsequence not found

Summary

In this lesson, we explored the standard library’s search algorithms, from simple searches to working with custom types and comparison functions.

Main Points Covered

• Introduced the eight main search algorithms: find(), find_if(), find_if_not(), find_first_of(), adjacent_find(), search_n(), search(), and find_end().
• Demonstrated the use of std::ranges::find() for basic search operations and handling cases where the search item is not found.
• Explained the difference between range-based algorithms and their iterator-sentinel pair counterparts.
• Covered the requirements for custom types to work with search algorithms, specifically the need for an equality operator ==.
• Illustrated the use of projection functions to transform elements before comparison.
• Showcased std::ranges::find_if() and std::ranges::find_if_not() for conditional searches.
• Demonstrated how to use std::ranges::find_first_of() to find any of a set of possible values.
• Explained std::ranges::adjacent_find() for finding the first pair of adjacent equal elements.
• Showed how std::ranges::search_n() searches for a sequence of repeated elements.
• Described std::ranges::search() for finding subranges and std::ranges::find_end() for finding the last instance of a subsequence.
• Discussed the customization of search operations with custom searchers, including the Boyer-Moore algorithm.

Next Lesson

Binary Search in C++

An introduction to the advantages of binary search, and how to use it with the C++ standard library algorithms binary_search(), lower_bound(), upper_bound(), and equal_range()

ContinueView Course