开发者

std::transform how to not return (and not throw), just skip?

开发者 https://www.devze.com 2023-04-11 22:00 出处:网络
I want to display folder contents displaying (not having any folder system) So Having a std::set<string_file_names> with std::strings in it and some given path to some dir we want to search for

I want to display folder contents displaying (not having any folder system) So Having a std::set<string_file_names> with std::strings in it and some given path to some dir we want to search for folder contents as in normal fs. so having set of:

    set<string> demo_set;
demo_set.insert("file1");
demo_set.insert("file2");
demo_set.insert("folder/file1");
demo_set.insert("folder/file2");
demo_set.insert("folder/folder/file1");
demo_set.insert("folder/folder/file2");
demo_set.insert("bin/obj/Debug/CloudServerPrototype/ra.write.1.tlog");
demo_set.insert("bin/obj/Debug/CloudServerPrototype/rc.write.1.tlog");
demo_set.insert("bin/obj/Debug/vc100.idb");
demo_set.insert("bin/obj/Debug/vc100.pdb");

and search string "bin/obj/Debug/" we want to get 3 Items - folder, 2 files:

CloudServerPrototype/
vc100.idb
vc100.pdb

But we get also an empty line. How to not get it and how to throw an error in case of non items found?

Entire code:

#include <iostream>
#include <algorithm>
#include <set>
#include <string>
#include <iterator>

using namespace std;

struct get_pertinent_part
{
    const std::string given_string;

    get_pertinent_part(const std::string& s)
        :given_string(s)
    {
    }

    std::string operator()(const std::string& s开发者_高级运维)
    {
        std::string::size_type first = 0;

        if (s.find(given_string) == 0)
        {
            first = given_string.length();
        }
        else
        {
            return "";
        }

        std::string::size_type count = std::string::npos;
        std::string::size_type pos = s.find_last_of("/");
        if (pos != std::string::npos && pos > first)
        {
            count = pos + 1 - first;
        }

        return s.substr(first, count);
    }
};

void directory_listning_without_directories_demo()
{
    set<string> output;
    set<string> demo_set;

    demo_set.insert("file1");
    demo_set.insert("file2");
    demo_set.insert("folder/file1");
    demo_set.insert("folder/file2");
    demo_set.insert("folder/folder/file1");
    demo_set.insert("folder/folder/file2");
    demo_set.insert("bin/obj/Debug/CloudServerPrototype/ra.write.1.tlog");
    demo_set.insert("bin/obj/Debug/CloudServerPrototype/rc.write.1.tlog");
    demo_set.insert("bin/obj/Debug/vc100.idb");
    demo_set.insert("bin/obj/Debug/vc100.pdb");


    std::transform(demo_set.begin(),
        demo_set.end(),
        std::inserter(output, output.end()),
        get_pertinent_part("bin/obj/Debug/"));

    std::copy(output.begin(),
        output.end(),
        std::ostream_iterator<std::string>(std::cout, "\n"));
}

int main()
{
    directory_listning_without_directories_demo();
    cin.get();
    return 0;
}

code sample based on this grate answer.


As there is no transform_if, a proper way to do this would be to to first copy_if the paths that have a non-zero-length get_pertinent_part result into another container, and then run the transform on that new container.

Alternatively, you could write something like transform_if that compares the result of your transform function with a predicate. Here's my untested shot at it:

template <class InIt, class OutIt, class UnaryFunction, class Predicate>
void transform_if_value(InIt first, InIt last, OutIt out, UnaryFunction fn, Predicate pred)
{
    for ( ; first != last; ++ first)
    {
        auto val = fn(*first);
        if (pred(val))
            *out++ = val;
    }
}

Then you could use that like

transform_if_value(
    demo_set.begin(),
    demo_set.end(),
    std::inserter(output, output.begin()),
    get_pertinent_part("bin/obj/Debug/"),
    [](const std::string& s) {return !s.empty();});

For even cooler syntax, check out boost's Range adaptors:

boost::copy(
    demo_set | boost::adaptors::transformed(get_pertinent_part("bin/obj/Debug/"))
             | boost::adaptors::filtered([](const std::string& s) {return !s.empty();}),
    std::inserter(output, output.begin()));


You can use std::accumulate (not from <algorithm> but from <numeric>) to, well, accumulate the matches into your output.

std::set<string> const&
operator()(std::set<string> const& out, std::string const& candidate) const
{
    if(/* same logic to check if candidate matches */) {
        out.insert(/* insert processed/transformed path */);
    }
    return out;
}

Call as:

std::accumulate(
    demo_set.begin(), demo_set.end()
    , std::ref(output)
    , get_pertinent_path("bin/obj/Debug/") );

Note the use of std::ref (from <functional>). You don't want the output set to be moved around (plus the result of the call to std::accumulate is ignored so no changes will be visible). Alternatively, if you don't want a dependency on std::ref, you can use pointers.

edit: heh, revisiting this idea, this is arguably no better than using std::for_each and passing a reference to the ouput to the constructor of the functor. YMMV.

0

精彩评论

暂无评论...
验证码 换一张
取 消

关注公众号