Range-based for loop with auto specifier combined with static_cast - c++

Imaged that I have an std::vector of std::string, and I want to convert those std::string to std::string_view in range-based for loop:
auto v = std::vector<std::string>{"abc", "def", "ghi"};
for (std::string_view sv : v) {
// do something with string_view
}
The above code is perfectly valid, but I want to remain the auto specifier to do that, how to do static_cast in one line range-based for loop like this? It seems like C++20 ranges can do this in a concise way, can someone give an example?
for (auto sv : v | static_cast<std::string_view>) {
// do something with std::string_view
}

Not that this is a good idea as written, but this might be a useful example of a more general transform concept (and an evil lambda trick):
for(auto sv : v |
views::transform([](std::string_view x) {return x;})) …

You could, as #Elijay comments, simply create a new vector of string_views:
for (auto sv : vector<string_view>(v.begin(), v.end()))
but this kind of defeats the whole purpose of using string_view in the first place: to avoid copying.
It also, as commented below, kind of defeats the whole purpose of auto: to avoid unnecessarily restating type information. Here the type is being introduced for the first time so must be explicitly stated. Why not right up front in plain view?
Less verbose and using string_view as intended is where you started:
for (string_view sv : v)

You can do something like this:
#include <string>
#include <vector>
#include <iostream>
#include <string_view>
int main() {
auto v = std::vector<std::string>{"abc", "def", "ghi"};
for (auto sv : std::vector<std::string_view>(v.begin(), v.end())) {
// use sv ...
}
}
However note that creating a whole new vector is not at all recommended. It allocates memory again and causes a lot of unnecessary overhead. Besides, you have to spell the type somewhere anyway, so auto doesn't have any advantages at all here. The Right ThingTM to do is specifying the type name explicitly and not using auto.

Iterators is a good customization point, unfortunately it requires quite some boilerplate:
#include <vector>
#include <iostream>
#include <string_view>
template <typename T>
struct container_view {
const T& container;
container_view(const T& t) : container(t) {}
struct iterator{
typename T::const_iterator base;
iterator(const typename T::const_iterator& it) : base(it) {}
bool operator!=(const iterator& other) { return base != other.base; }
iterator& operator++() { ++base; return *this; }
std::string_view operator*() { return {*base}; }
// ^--- string_view
};
iterator begin() { return {container.begin()};}
iterator end() { return {container.end()};}
};
int main (){
auto v = std::vector<std::string>{"abc", "def", "ghi"};
// v--- auto
for (auto sv : container_view(v)) {
std::cout << sv << '\n';
}
}

Arguably more/less succinct. Opinions will vary.
#include <vector>
#include <string_view>
#include <string>
#include <iostream>
#include <range/v3/view/transform.hpp>
int main()
{
auto v = std::vector<std::string>{"abc", "def", "ghi"};
using namespace ranges;
auto as_string_view = views::transform([](auto&& x) { return std::string_view(x); });
for (auto sv : v | as_string_view) {
std::cout << sv << '\n';
}
}

Related

How to write an infinite sequence compatible with std::ranges?

I would like to write a struct which will generate an infinite sequence of fibonacci numbers in a way compatible with std::ranges and range adaptors.
So if I wanted the first 5 even fibonacci numbers, I would write something like this:
#include <iostream>
#include <ranges>
using namespace std;
int main() {
auto is_even = [](auto a) { return a % 2 == 0; };
for (auto v :
something | ranges::views::filter(is_even) | ranges::views::take(5))
std::cout << v << std::endl;
return 0;
}
What "something" needs to be ?
It seems to me that it has to be a class with a forward iterator, but I can't find any example.
Edit: As it was pointed out by 康桓瑋 in the comments, there has been a better, cleaner solution presented at Cppcon, link, by Tristan Brindle.
I think this could serve as a quick reference to make custom iterator-based generators.
By your requirements, something must be a std::ranges::view, meaning it must be a moveable std::ranges::range deriving from std::ranges::view_interface<something>.
We can tackle all three with the following:
#include <ranges>
template<typename T>
class fib : public std::ranges::view_interface<fib<T>>{
public:
struct iterator;
auto begin() const { return iterator{}; }
auto end() const { return std::unreachable_sentinel; }
};
Notice std::unreachable_sentinel which makes creating sequences without end really simple.
We still have to define the iterator which does the actual work. In your case we want fib to be "the source" of the values, so our iterator should actually be std::input_iterator. There's some boiler plate code needed for that but it's basically just a type which can be incremented and dereferenced to yield its current value.
Something like this will do:
#include <iterator>
template<typename T>
struct fib<T>::iterator {
using iterator_category = std::input_iterator_tag;
using value_type = T;
using difference_type = std::ptrdiff_t;
using pointer = T*;
using reference = T;
constexpr iterator() noexcept = default;
iterator& operator++() {
auto old_next = next;
next = next + current;
current = old_next;
return *this;
}
iterator operator++(int) {
iterator current{*this};
++(*this);
return current;
}
value_type operator*() const {
return current;
}
bool operator==(const iterator& other) const { return current == other.current && next==other.next; }
private:
T current= {};
T next = T{} + 1; // Could perhaps be fancier.
};
The increment operator does the computation itself, it's the simple iterating algorithm.
That's it, here's a working example:
#include <cstdint>
#include <iostream>
int main() {
auto is_even = [](auto a) { return a % 2 == 0; };
for (auto v :
fib<std::uint64_t>{} | std::ranges::views::filter(is_even) | std::ranges::views::take(10))
std::cout << v << std::endl;
return 0;
}
which outputs:
0
2
8
34
144
610
2584
10946
46368
196418
Of course you won't get very far even with std::uint64_t. But T can be anything numeric enough.
One can easily generalize the iterator to hold a stateful functor, likely passed from the range itself, call it during each increment and store the yielded value for dereferencing later. This would be very crude, but simple, way how to at least simulate "yield-based" generators.

`c++17` ranged for loop over std::vectors with value and index [duplicate]

Is there an equivalent to the range-based enumerate loop from python in C++?
I would imagine something like this.
enumerateLoop (auto counter, auto el, container) {
charges.at(counter) = el[0];
aa.at(counter) = el[1];
}
Can this be done with templates or macros?
I'm aware that I can just use an old school for-loop and iterate until I reach container.size(). But I'm interested how this would be solved using templates or macros.
EDIT
I played a bit with boost iterators after the hint in the comments. I got another working solution using C++14.
template <typename... T>
auto zip(const T &... containers) -> boost::iterator_range<boost::zip_iterator<
decltype(boost::make_tuple(std::begin(containers)...))>> {
auto zip_begin =
boost::make_zip_iterator(boost::make_tuple(std::begin(containers)...));
auto zip_end =
boost::make_zip_iterator(boost::make_tuple(std::end(containers)...));
return boost::make_iterator_range(zip_begin, zip_end);
}
template <typename T>
auto enumerate(const T &container) {
return zip(boost::counting_range(0, static_cast<int>(container.size())),
container);
}
https://gist.github.com/kain88-de/fef962dc1c15437457a8
Enumeration of multiple variables has been an idiom since C. The only complication is that you can't declare both variables in the initializer of the for loop.
int index;
for (auto p = container.begin(), index = 0; p != container.end(); ++p, ++index)
I don't think it gets any simpler (or more powerful) than that.
There is a pre C++11 solution in boost to this: boost.range.indexed.
Unfortunately it doesn't work with C++11 range based for-loops, only old style verbose loops. However with C++17 it should be become (almost) as easy as in python using structured bindings
Then it should be possible implement something that works like this:
for (auto& [n,x] : enumerate(vec)) x = n;
So, a bit of waiting still ;)
I wrote something for this a while back.
Essentially, you need to wrap an iterator and give it pair semantics.
AFAIK, there's nothing like this built into the language. And I don't think boost has it either. You pretty much have to roll your own.
// Wraps a forward-iterator to produce {value, index} pairs, similar to
// python's enumerate()
template <typename Iterator>
struct EnumerateIterator {
private:
Iterator current;
Iterator last;
size_t index;
bool atEnd;
public:
typedef decltype(*std::declval<Iterator>()) IteratorValue;
typedef pair<IteratorValue const&, size_t> value_type;
EnumerateIterator()
: index(0), atEnd(true) {}
EnumerateIterator(Iterator begin, Iterator end)
: current(begin), last(end), index(0) {
atEnd = current == last;
}
EnumerateIterator begin() const {
return *this;
}
EnumerateIterator end() const {
return EnumerateIterator();
}
EnumerateIterator operator++() {
if (!atEnd) {
++current;
++index;
atEnd = current == last;
}
return *this;
}
value_type operator*() const {
return {*current, index};
}
bool operator==(EnumerateIterator const& rhs) const {
return
(atEnd && rhs.atEnd) ||
(!atEnd && !rhs.atEnd && current == rhs.current && last == rhs.last);
}
bool operator!=(EnumerateIterator const& rhs) const {
return !(*this == rhs);
}
explicit operator bool() const {
return !atEnd;
}
};
template<typename Iterable>
EnumerateIterator<decltype(std::declval<Iterable>().begin())> enumerateIterator(Iterable& list) {
return EnumerateIterator<decltype(std::declval<Iterable>().begin())>(list.begin(), list.end());
}
template<typename ResultContainer, typename Iterable>
ResultContainer enumerateConstruct(Iterable&& list) {
ResultContainer res;
for (auto el : enumerateIterator(list))
res.push_back(move(el));
return res;
}
C++17 and structured bindings makes this look OK - certainly better than some ugly mutable lambda with a local [i = 0](Element&) mutable or whatever I've done before admitting that probably not everything should be shoehorned into for_each() et al. - and than other solutions that require a counter with scope outside the for loop.
for (auto [it, end, i] = std::tuple{container.cbegin(), container.cend(), 0};
it != end; ++it, ++i)
{
// something that needs both `it` and `i`ndex
}
You could make this generic, if you use this pattern often enough:
template <typename Container>
auto
its_and_idx(Container&& container)
{
using std::begin, std::end;
return std::tuple{begin(container), end(container), 0};
}
// ...
for (auto [it, end, i] = its_and_idx(foo); it != end; ++it, ++i)
{
// something
}
C++ Standard proposal P2164 proposes to add views::enumerate, which would provide a view of a range giving both reference-to-element and index-of-element to a user iterating it.
We propose a view enumerate whose value type is a struct with 2 members index and value representing respectively the position and value of the elements in the adapted range.
[ . . .]
This feature exists in some form in Python, Rust, Go (backed into the language), and in many C++ libraries: ranges-v3, folly, boost::ranges (indexed).
The existence of this feature or lack thereof is the subject of recurring stackoverflow questions.
Hey, look! We're famous.
You can also more elegantly use the auto ranges available since C++11:
int i = 0;
for (auto& el : container){
charges.at(counter) = el[0];
aa.at(counter) = el[1];
++i;
}
You still have to count the i up by hand, though.
Here's a macro-based solution that probably beats most others on simplicity, compile time, and code generation quality:
#include <iostream>
#define fori(i, ...) if(size_t i = -1) for(__VA_ARGS__) if(i++, true)
int main() {
fori(i, auto const & x : {"hello", "world", "!"}) {
std::cout << i << " " << x << std::endl;
}
}
Result:
$ g++ -o enumerate enumerate.cpp -std=c++11 && ./enumerate
0 hello
1 world
2 !
Tobias Widlund wrote a nice MIT licensed Python style header only enumerate (C++17 though):
GitHub
Blog Post
Really nice to use:
std::vector<int> my_vector {1,3,3,7};
for(auto [i, my_element] : en::enumerate(my_vector))
{
// do stuff
}
Boost::Range supports this as of 1.56.
#include <boost/range/adaptor/indexed.hpp>
#include <boost/assign.hpp>
#include <iterator>
#include <iostream>
#include <vector>
int main(int argc, const char* argv[])
{
using namespace boost::assign;
using namespace boost::adaptors;
std::vector<int> input;
input += 10,20,30,40,50,60,70,80,90;
// for (const auto& element : index(input, 0)) // function version
for (const auto& element : input | indexed(0))
{
std::cout << "Element = " << element.value()
<< " Index = " << element.index()
<< std::endl;
}
return 0;
}

How do I create a range from a begin and end iterator?

I have an object with functions for getting the begin and end iterators:
const_iterator err_begin() const
const_iterator err_end() const
Because they are not named begin and end, I cannot pass my object directly to functions in range-v3.
Is there a simple wrapper I can use to make this object work with the range-v3 library?
For example:
auto hasErrors = !empty(something(x.err_begin(), x.err_end()));
Sounds like you're looking for iterator_range:
auto hasErrors = !empty(ranges::make_iterator_range(x.err_begin(), x.err_end()));
You clarified that the class in question is part of a library that you cannot change. Fine. Create a facade class:
class FacadeClass {
const RealClassWithErrBeginEnd &r;
public:
FacadeClass(const RealClassWithErrBeginEnd &r) : r(r) {}
auto begin() const { return r.err_begin(); }
auto end() const { return r.err_end(); }
};
This should be good enough to fool most code that expects a container. In the worst case, you might need to provide additional typedefs in the facade, i.e. value_type, etc...
boost::make_iterator_range will do the right thing. Now add a little ADL and we find that one free function solves all our problems:
#include <vector>
#include <iostream>
#include <string>
#include <boost/range.hpp>
// simulate the library class
struct X
{
auto err_begin() const { return errors.begin(); }
auto err_end() const { return errors.end(); }
std::vector<std::string> errors;
};
// provide a generator to build an iterator range
auto errors(const X& x)
{
return boost::make_iterator_range(x.err_begin(), x.err_end());
}
// do things with the iterator_range
int main()
{
X x;
for (const auto& err : errors(x))
{
std::cout << err << std::endl;
}
std::cout << empty(errors(x)) << std::endl;
}

boost variant copy semantics

I was wondering what the copy semantics of boost variants are. I've checked the source code and it's a bit baffling to me so I was wondering, in the example code, if my getVal(name) function makes a copy of the underlying vector when it's returned? If so, should I change it to be a reference (&) returned instead?
using Val = boost::variant<std::vector<int>, std::vector<std::string>>;
Val getVal(std::string& name) {
return map[name];// where map is std::map<std::string, Val>
}
Yes, your getVal returns a copy of the whole vectors (including copies of all the element strings, e.g.).
Yes, returning a reference instead solves this.
Note you can also have a variant that stores a reference. In this case, returning it by "value" still has the same semantics as returning the reference:
using Ref = variant<std::vector<int>&, std::vector<std::string>&>;
Ref getVal(std::string& name) {
return map[name]; // where map is std::map<std::string, Val>
}
Full sample with the necessary mechanics to convert from Ref to Val (and vice versa):
Live On Coliru
#include <boost/variant.hpp>
#include <map>
#include <vector>
#include <string>
using Val = boost::variant<std::vector<int>, std::vector<std::string>>;
using Ref = boost::variant<std::vector<int>&, std::vector<std::string>& >;
std::map<std::string, Val> map {
{ "first", std::vector<int> { 1,2,3,4 } },
{ "2nd", std::vector<std::string> { "five", "six", "seven", "eight" } }
};
namespace { // detail
template <typename T>
struct implicit_convert : boost::static_visitor<T> {
template <typename U> T operator()(U&& u) const { return std::forward<U>(u); }
};
}
Ref getVal(std::string& name) {
return boost::apply_visitor(implicit_convert<Ref>(), map[name]);
}
#include <iostream>
int main() {
for (auto i : boost::get<std::vector<int> >(map["first"])) std::cout << i << " ";
for (auto i : boost::get<std::vector<std::string> >(map["2nd"])) std::cout << i << " ";
}
Output:
1 2 3 4 five six seven eight
Without any vectors being copied

boost::any_range and operator []

Consider the following code:
#include <boost/range.hpp>
#include <boost/range/any_range.hpp>
#include <boost/range/join.hpp>
#include <iostream>
#include <algorithm>
#include <string>
#include <vector>
#include <list>
struct TestData {
TestData() : m_strMem01("test"), m_intMem02(42), m_boolMem03(true) {}
std::string m_strMem01;
int m_intMem02;
bool m_boolMem03;
};
struct IntComp {
bool operator()(const TestData &s, int i) { return s.m_intMem02 < i; }
bool operator()(int i, const TestData &s) { return i < s.m_intMem02; }
bool operator()(const TestData &i, const TestData &s) {
return i.m_intMem02 < s.m_intMem02;
}
};
struct StrComp {
bool operator()(const TestData &s, const std::string &str) {
return s.m_strMem01 < str;
}
bool operator()(const std::string &str, const TestData &s) {
return str < s.m_strMem01;
}
bool operator()(const TestData &i, const TestData &s) {
return i.m_strMem01 < s.m_strMem01;
}
};
typedef boost::any_range<TestData, boost::forward_traversal_tag,
const TestData &, std::ptrdiff_t> TestRange;
std::vector<TestData> vecData(10);
std::list<TestData> listData(20);
TestRange foo() {
TestRange retVal;
auto tmp1 = std::equal_range(vecData.cbegin(), vecData.cend(), 42, IntComp());
retVal = boost::join(retVal, tmp1);
auto tmp2 =
std::equal_range(listData.cbegin(), listData.cend(), "test", StrComp());
retVal = boost::join(retVal, tmp2);
return retVal;
}
int main(int argc, char *argv[]) {
auto res = foo();
for (auto a : res) {
std::cout << a.m_strMem01 << std::endl;
}
//std::cout << res[4].m_intMem02 << std::endl;
}
If you uncomment the last line the code fails since distance_to not implemented for any_forward_iterator_interface. I'm not sure what exactly I'm missing here, like implementing operator[] or distance_to but for what? My own version traversal tag? And why it doesn't work in the first place?
Coliru version
I would say the answer depends on your performance needs and your laziness when it comes to implementing a new iterator abstraction. The core reason for your [] operator not working is the fact that std::list<...> does not provide a random access traversal iterator. If you would have chosen a container that provides such an iterator. You any_range<...> could have taken the random_access_traversal_tag and everything would be fine.
I think it's fair to say that it is not such a big deal to implement a random access iterator on top of a list by simply encapsulating the current index and count forward and backward within the list whenever a specific position is meant to be accessed, but it's clearly against the nature of the list performance-wise.
Is there a good reason to hold one of the collection in a list ?
Is there a good reason to access the resulting any_range by random ?
Is it worth the effort to provide a inefficient random access interface for std::list ?
Of course any_iterator (which underlies the any_range implementation) doesn't gratuitously emulate RandomAccess iterators for any odd iterator you pass.
If you want that, just make an iterator adaptor that does this (making it very slow to random access elements in a list - so don't do this).