Question

I have a std::vector<std::string> of all the files in a directory:

// fileList
folder/file1
folder/file2
file3
file4.ext

and a std::set<std::string> of filenames and the same for all used folder prefixes:

// set1
file2
file4.ext

// set2
folder

I need to generate the full (relative) paths to the ALL files in set1, but see no way of doing that without iterating over set2 set1.size() times, multiplied by fileList.size()

UPDATE: some clarification:

Expected output for above example:

folder/file2
file4.ext

Proposed (inefficient?) solution, maybe too verbose and with stupid implementation:

// pseudo-code!
vector<string> allpossibleFullPaths( set1.size()*set2.size() );
vector<string> output;
foreach( prefix_in_set2 )
    foreach( filename_in_set1 )
        allpossibleFullpaths.push_back( set2[i] + "/" set1[i] )

foreach( filename_in_fileList )
    files.push_back( find( fileList[i] in allpossibleFullPaths ) );

(fast pseudocode-ish)
This seems very innefficient, is there a better way to make these matches?

Thanks!

PS: better still would be a way to keep track of doubles, so that I can warn the user about that.

Answer 1

One area which you are not clear about is this:

• Given set1 & set2 as described above, what if fileList had “file4.ext” and “folder\file4.ext”. Would you want both? Or is the list of file in set1 guaranteed to be unique?

Assuming that you’d want both, pseudo-code:

 foreach(pathname in fileList)
    separate pathname into path & filename.
    if path is not empty, but not in set2, skip to next pathname.
    if filename is in set1, output pathname.

Since a set lookup should be O(1), total complexity is O(2 * fileList.Length)

If the filenames in set1 are unique, you can count the number of pathnames output, and exit early when set1.Length is reached.

It may seem counter-intuitive to step through the longest collection, but it also has the slowest lookup, so operations on fileList have to be minimized.

UPDATE: Here’s the full working C++ code (includes & usings elided)

void ListFiles()
{
    vector<string> fileList;
    fileList.push_back("folder/file1");
    fileList.push_back("folder/file2");
    fileList.push_back("file3");
    fileList.push_back("file4.ext");

    set<string> set1;
    set1.insert("file2");
    set1.insert("file4.ext");

    set<string> set2;
    set2.insert("folder");

    for(vector<string>::iterator iter = fileList.begin();
        iter != fileList.end();
        ++iter)
    {
        string pathname = *iter;
        string filename;
        string path;
        size_t pos = pathname.find('/');
        if (pos == string::npos || pos == 0)
            filename = pathname;
        else
        {
            path = pathname.substr(0, pos);
            if (set2.find(path) == set2.end())
                continue;
            filename = pathname.substr(pos+1);
        }
        if (set1.find(filename) != set1.end())
            cout << pathname << endl;
    }

}