Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
roth_stephan_clean_c20_sustainable_software_development_patt.pdf
Скачиваний:
29
Добавлен:
27.03.2023
Размер:
7.26 Mб
Скачать

Chapter 7 Functional Programming

int main() {

const bool result1 = containsEvenValue(10, 7, 11, 9, 33, 14); const bool result2 = containsEvenValue(17, 7, 11, 9, 33, 29);

std::cout << std::boolalpha;

std::cout << "result1 is " << result1 << "\n"; std::cout << "result2 is " << result2 << std::endl; return 0;

}

The output of this program is as follows:

result1 is true result2 is false

Pipelining with Range Adaptors (C++20)

People who like to work with UNIX or Linux operating systems emphasize, among other things, a particularly convenient and efficient way to perform tasks on these operating systems: using shell programming. Basically, a UNIX/Linux shell is a command-line-­based human-machine interface. There are a number of text-based shells available for UNIX/ Linux OS, such as Bash (an acronym for Bourne Again Shell), the Korn Shell, and the Z Shell.

One of the reasons that you can perform complex tasks very elegantly and efficiently with the help of these shells is the possibility of pipelining. Some people say the concept of pipelining, introduced in 1972, has been one of the most important UNIX innovations in history, perhaps apart from regular expressions. Basically, a pipeline is a message-­passing pattern and describes a chain of processing elements. In a shell, a pipeline is a set of processes, usually small programs, chained together by their standard streams, so that the output text of each process (stdout) is passed directly as input (stdin) to the next one.

As an example, suppose we have a text file named customers.txt, in which hundreds of customer names are listed line-by-line, preceded by a date, as follows (an excerpt):

2020-11-05,Stephan Roth

2020-11-22,John Doe

2020-10-15,Mark Powers [...]

329

Chapter 7 Functional Programming

On the Bash command line, we now execute the following command sequence:

$ cat customers.txt | sort -r > customers2.txt

What will happen here? Well, first the cat command (short for “concatenate”) lists the contents of the text file customers.txt on stdout. This output stream is redirected with the help of the pipe operator (a vertical bar: |) and is used as the input stream for the program sort. The sort command is used to sort lines in a text file. In our case, we have specified by the command-line option -r that the sort order should be reversed.

Instead of descending sorting, ascending sorting is used. The output stream of the sort command is redirected again, because we don’t want to see its output on the screen. Instead, it should be written to a new text file. The greater than symbol, >, tells the shell to redirect the sort’s output to the customers2.txt file.

In Chapter 5 I briefly introduced the new C++20 Ranges library. However, I neglected a few of the new features there: range adaptors and chaining using the pipe operator! Just as you can chain commands together on a UNIX shell using pipes, this is also possible with C++20 Range adaptors.

You may remember the following small code example (excerpt) from Listing 5-26? It’s shown in again in Listing 7-30.

Listing 7-30.  The Code Snippet from Listing 5-26

#include <iostream>

#include <ranges> #include <vector>

std::vector<int> integers = { 2, 5, 8, 22, 45, 67, 99 };

auto view = std::views::reverse(integers); // does not change 'integers'

Recall that a view is lazy evaluated, i.e., whatever transformation it applies, the view performs it at the moment someone requests an element. And as a range adaptor, it does not modify the underlying range, in our case the vector named integers.

Due to their property as adaptors, views can be easily chained. We now extend this example by adding more views and some lambda expressions; see Listing 7-31.

330

Chapter 7 Functional Programming

Listing 7-31.  Chaining Range Adaptors

#include <algorithm> // required for std::ranges::for_each #include <iostream>

#include <ranges> #include <vector>

int main() {

std::vector<int> integers = { 2, 5, 8, 22, 45, 67, 99 }; auto isOdd = [] (const int value) { return value % 2 != 0; }; auto square = [] (const int value) { return value * value; };

auto printOnStdOut = [] (const int value) { std::cout << value << '\n';

};

auto view = std::views::transform(std::views::reverse(std::views::filter( integers, isOdd)),

square);

std::ranges::for_each(view, printOnStdOut); return 0;

}

The output of the program is as follows:

9801

4489

2025

25

Well, from a clean code developer’s perspective, this code sample still has one unsightly flaw: the nested Range adaptors for creating the view. This is just one line of code, but it is not easy to comprehend on first sight what’s happening here. Just think about that in a real software application; nested Range adaptors could become much more complex than in this relatively simple example.

331

Chapter 7 Functional Programming

This is where the new C++20 pipe operator comes into play. It is “syntactic sugar”1 and can be used for easier function chaining. The line of code in which the view view is created can also be written as follows:

auto view = integers | std::views::filter(isOdd) | std::views::reverse | std::views::transform(square);

That’s pretty convenient, isn’t it? This looks very similar to building pipelines in a UNIX shell, as we saw it at the beginning of this section. You can simply read the

expression from left to right and easily understand how the view is composed. Thanks to the power of views, the C++20 Ranges Library enables developers to write code in an even more functional programming-style.

Before we come to the end of this chapter on functional programming, I previously announced that I would revisit and improve the code example in Listing  7-15 once again. With all the new functional programming concepts we have now learned, we can refactor this example and make it much more compact and elegant; see Listing 7-32.

Listing 7-32.    The Refactored Code Example

#include <concepts> #include <iostream>

#include <ranges>

template <typename T>

concept Streamable = requires (std::ostream& os, const T& value) { { os << value };

};

int main() {

auto toSquare = [] (const auto value) { return value * value; }; auto isAnEvenNumber = [] <std::integral T> (const T value) {

return (value % 2) == 0;

};

auto print = [] <Streamable T> (const T& printable) { std::cout << printable << '\n';

};

1The term “syntactic sugar” describes syntax within a programming language that is designed to make parts of the code easier to read or to express.

332