C++ 将目录中的所有文件与::文件系统进行比较

Question

I'm trying to compare all of the files in a directory.

The directory is constantly getting new files added to it from a different thread. Though the directory_iterator documentation says that the change may not be reflected, so I have solved this with an outer loop which tries again when the inner loop is done:

void mergeFolder(int aFolderNum)
{
bool aPreviousMergesFinished = false;       
while (!aPreviousMergesFinished)
    {
        filesystem::directory_iterator aEnd;
        for (filesystem::directory_iterator aIterator1("data/temp/MERGEME/" + std::to_string(aFolderNum)), 
                aIterator2 = ++aIterator1;
            aIterator2 != aEnd; ++aIterator1, ++aIterator2)
        {
            std::cout << "file 1: " << aIterator1->path() << std::endl;
            std::cout << "file 2: " << aIterator2->path() << std::endl;
        }
        std::cout << "Finished inner loop" << std::endl;

        aPreviousMergesFinished =
            std::filesystem::is_empty("data/temp/MERGEME/" + std::to_string(aFolderNum));
    }
}

However, the output is as follows:

...

file 1: "data/temp/MERGEME/1\1_999"

file 2: "data/temp/MERGEME/1\1_999"

file 1: "data/temp/MERGEME/1\1_9991"

file 2: "data/temp/MERGEME/1\1_9991"

file 1: "data/temp/MERGEME/1\1_9993"

file 2: "data/temp/MERGEME/1\1_9993"

file 1: "data/temp/MERGEME/1\1_9995"

file 2: "data/temp/MERGEME/1\1_9995"

file 1: "data/temp/MERGEME/1\1_9997"

file 2: "data/temp/MERGEME/1\1_9997"

file 1: "data/temp/MERGEME/1\1_9999"

file 2: "data/temp/MERGEME/1\1_9999"

Finished inner loop

As you can see, each file is being printed twice in the inner loop, but I'm almost certain I've done the inner loop correctly. What has gone wrong here?

Answer 1

directory_iterator is only an input iterator , not a forward iterator . This means that it guarantees correct behavior only on one single pass.

As soon as you increment the directory_iterator any copies referring to the previous state are invalidated. Using them causes undefined behavior.

What you are trying here is not possible. Even if what you are trying was allowed, you would only iterate adjacent pairs, not all pairs.

Instead, read in all the file names from one iterator pass into a container and iterate over that container with two loops.