I have a class with an initialiser list in the constructor where one of the fields I'm initialising is a std::filesystem::path but it doesn't seem to be initialising to the expected value.
MyClass::MyClass(
unsigned int deviceSerial,
const std::string& processName
) :
deviceSerial(deviceSerial),
processName(processName),
configFilePath(GetBasePath() / std::to_string(deviceSerial) / ("#" + processName + ".json"))
{
/* Parameter checks */
}
Using the debugger I can see that GetBasePath() is returning exactly what I expect (returns std::filesystem::path with correct path) but the / operator doesn't seem to be having an effect. Once inside the body of the constructor I can see that configFilePath is setup to the result of GetBasePath() without the extra info appended.
I'm using MSVS-2019, I have the C++ language standard set to C++17 and in debug mode I have all optimisations disabled.
I have also tested the following in the body of the class and I still see path as simply the result of GetBasePath() and the extra items are not being appended.
{
auto path = GetBasePath(); // path = "C:/Users/Me/Desktop/Devices"
path /= std::to_string(deviceSerial); // path = "C:/Users/Me/Desktop/Devices"
path /= ("#" + processName + ".json"); // path = "C:/Users/Me/Desktop/Devices"
}
On a slight side note I also tried the above test with += instead of /= and I still see the same results.
Edit
As requested, below is a minimal complete and verifiable example.
#include <Windows.h>
#include <cstdio>
#include <filesystem>
#include <memory>
#include <string>
std::string ExpandPath(const std::string &str) {
auto reqBufferLen = ExpandEnvironmentStrings(str.c_str(), nullptr, 0);
if (reqBufferLen == 0) {
throw std::system_error((int)GetLastError(), std::system_category(),
"ExpandEnvironmentStrings() failed.");
}
auto buffer = std::make_unique<char[]>(reqBufferLen);
auto setBufferLen =
ExpandEnvironmentStrings(str.c_str(), buffer.get(), reqBufferLen);
if (setBufferLen != reqBufferLen - 1) {
throw std::system_error((int)GetLastError(), std::system_category(),
"ExpandEnvironmentStrings() failed.");
}
return std::string{buffer.get(), setBufferLen};
}
int main() {
unsigned int serial = 12345;
std::string procName = "Bake";
std::filesystem::path p(ExpandPath("%USERPROFILE%\\Desktop\\Devices"));
std::printf("Path = %s\n", p.string().c_str());
// p = C:\Users\Me\Desktop\Devices
p /= std::to_string(serial);
std::printf("Path = %s\n", p.string().c_str());
// p = C:\Users\Me\Desktop\Devices
p /= "#" + procName + ".json";
std::printf("Path = %s\n", p.string().c_str());
// p = C:\Users\Me\Desktop\Devices
std::getchar();
}
I've also used this example and tested with `p.append()` and got the same result.
I'd like to give thanks to #rustyx and #Frank for their suggestions, following this advice has led me to discover a bug in the way I create the initial string that gets passed to the path constructor (Also #M.M who found the exact bug while I was typing this answer)
I created a function (that is in use in my class) std::string ExpandPath(const std::string& path) which uses the Windows API to expand any environment variables in a path and return a string. This string is generated from a char* and a count, that count includes the null byte so when creating an string using the constructor variant std::string(char* cstr, size_t len) this includes the null byte in the string itself.
Because I was using the debugger to interrogate the variables it reads C-style strings and stops at the null byte. In my original example I also use printf() as I just happen to prefer this function for output, but again this stops printing at the null byte. If I change the output to use std::cout I can see that the output has the expected path but with an extra space their (the null byte being printed as a space). Using std::cout I see that my paths result as the following with each append:
Path = C:\Users\Me\Desktop\Devices
Path = C:\Users\Me\Desktop\Devices \12345
Path = C:\Users\Me\Desktop\Devices \12345\#Bake.json
Summary:
Bug in my ExpandPath() where
return std::string{buffer.get(), setBufferLen};
Should be
return std::string{buffer.get(), setBufferLen - 1};
In MATLAB there's a nice function called fileparts that takes a full file path and parses it into path, filename (without extension), and extension as in the following example from the documentation:
file = 'H:\user4\matlab\classpath.txt';
[pathstr, name, ext] = fileparts(file)
>> pathstr = H:\user4\matlab
>> name = classpath
>> ext = .txt
So I was wondering if there's an equivalent function in any standard C++ or C libraries that I could use? Or would I have to implement this myself? I realize it's fairly simple, but I figured if there's already something pre-made that would be preferable.
Thanks.
The boost library has a file system component "basic_path" that allows you use iterators to discover each component in the filename. Such a component would be OS specific, and I believe you need to compile boost separately for Windows, Linux etc.
I just wrote this simple function. It behaves similar as Matlab's fileparts and works independent of platform.
struct FileParts
{
string path;
string name;
string ext;
};
FileParts fileparts(string filename)
{
int idx0 = filename.rfind("/");
int idx1 = filename.rfind(".");
FileParts fp;
fp.path = filename.substr(0,idx0+1);
fp.name = filename.substr(idx0+1,idx1-idx0-1);
fp.ext = filename.substr(idx1);
return fp;
}
A platform-independent way with C++11/14.
#include <experimental/filesystem>
namespace fs = std::experimental::filesystem;
void fileparts(string full, string& fpath, string& fname, string& fext)
{
auto source = fs::path(full);
fpath = source.parent_path().string();
fname = source.stem().string();
fext = source.extension().string();
}
...
string fpath, fname, fext;
fileparts(full_file_path,fpath,fname,fext);
Some possible solutions, depending on your OS:
Visual C++ _splitpath function
Win32 Shell Path Handling Functions such as PathFindExtension, PathFindFileName, PathStripPath, PathRemoveExtension, PathRemoveFileSpec
Ekalic's text-only approach is useful, but it didn't check for errors. Here's one that does, and also works with both / and \
struct FileParts
{
std::string path; //!< containing folder, if provided, including trailing slash
std::string name; //!< base file name, without extension
std::string ext; //!< extension, including '.'
};
//! Using only text manipulation, splits a full path into component file parts
FileParts fileparts(const std::string &fullpath)
{
using namespace std;
size_t idxSlash = fullpath.rfind("/");
if (idxSlash == string::npos) {
idxSlash = fullpath.rfind("\\");
}
size_t idxDot = fullpath.rfind(".");
FileParts fp;
if (idxSlash != string::npos && idxDot != string::npos) {
fp.path = fullpath.substr(0, idxSlash + 1);
fp.name = fullpath.substr(idxSlash + 1, idxDot - idxSlash - 1);
fp.ext = fullpath.substr(idxDot);
} else if (idxSlash == string::npos && idxDot == string::npos) {
fp.name = fullpath;
} else if (/* only */ idxSlash == string::npos) {
fp.name = fullpath.substr(0, idxDot);
fp.ext = fullpath.substr(idxDot);
} else { // only idxDot == string::npos
fp.path = fullpath.substr(0, idxSlash + 1);
fp.name = fullpath.substr(idxSlash + 1);
}
return fp;
}
I need to save all ".xml" file names in a directory to a vector. To make a long story short, I cannot use the dirent API. It seems as if C++ does not have any concept of "directories".
Once I have the filenames in a vector, I can iterate through and "fopen" these files.
Is there an easy way to get these filenames at runtime?
Easy way is to use Boost.Filesystem library.
namespace fs = boost::filesystem;
// ...
std::string path_to_xml = CUSTOM_DIR_PATH;
std::vector<string> xml_files;
fs::directory_iterator dir_iter( static_cast<fs::path>(path_to_xml) ), dir_end;
for (; dir_iter != dir_end; ++dir_iter ) {
if ( boost::iends_with( boost::to_lower_copy( dir_iter->filename() ), ".xml" ) )
xml_files.push_back( dir_iter->filename() );
}
I suggest having a look at boost::filesystem if it should be portable and bringing boost in isn't too heavy.
If you don't like boost, try Poco. It has a DirectoryIterator. http://pocoproject.org/
Something like this (Note, Format is a sprintf:ish funciton you can replace)
bool MakeFileList(const wchar_t* pDirectory,vector<wstring> *pFileList)
{
wstring sTemp = Format(L"%s\\*.%s",pDirectory,L"xml");
_wfinddata_t first_file;
long hFile = _wfindfirst(sTemp.c_str(),&first_file);
if(hFile != -1)
{
wstring sFile = first_file.name;
wstring sPath = Format(L"%s%s",pDirectory,sFile.c_str());
pFileList->push_back(sPath);
while(_wfindnext(hFile,&first_file) != -1)
{
wstring sFile = first_file.name;
wstring sPath = Format(L"%s%s",pDirectory,sFile.c_str());
pFileList->push_back(sPath);
}
_findclose(hFile);
}else
return false;
return true;
}
Given a string "filename.conf", how to I verify the extension part?
I need a cross platform solution.
Is this too simple of a solution?
#include <iostream>
#include <string>
int main()
{
std::string fn = "filename.conf";
if(fn.substr(fn.find_last_of(".") + 1) == "conf") {
std::cout << "Yes..." << std::endl;
} else {
std::cout << "No..." << std::endl;
}
}
The best way is to not write any code that does it but call existing methods. In windows, the PathFindExtension method is probably the simplest.
So why would you not write your own?
Well, take the strrchr example, what happens when you use that method on the following string "c:\program files\AppleGate.Net\readme"? Is ".Net\readme" the extension? It is easy to write something that works for a few example cases, but can be much harder to write something that works for all cases.
With C++17 and its std::filesystem::path::extension (the library is the successor to boost::filesystem) you would make your statement more expressive than using e.g. std::string.
#include <iostream>
#include <filesystem> // C++17
namespace fs = std::filesystem;
int main()
{
fs::path filePath = "my/path/to/myFile.conf";
if (filePath.extension() == ".conf") // Heed the dot.
{
std::cout << filePath.stem() << " is a valid type."; // Output: "myFile is a valid type."
}
else
{
std::cout << filePath.filename() << " is an invalid type."; // Output: e.g. "myFile.cfg is an invalid type"
}
}
See also std::filesystem::path::stem, std::filesystem::path::filename.
You have to make sure you take care of file names with more then one dot.
example: c:\.directoryname\file.name.with.too.many.dots.ext would not be handled correctly by strchr or find.
My favorite would be the boost filesystem library that have an extension(path) function
Assuming you have access to STL:
std::string filename("filename.conf");
std::string::size_type idx;
idx = filename.rfind('.');
if(idx != std::string::npos)
{
std::string extension = filename.substr(idx+1);
}
else
{
// No extension found
}
Edit: This is a cross platform solution since you didn't mention the platform. If you're specifically on Windows, you'll want to leverage the Windows specific functions mentioned by others in the thread.
Someone else mentioned boost but I just wanted to add the actual code to do this:
#include <boost/filesystem.hpp>
using std::string;
string texture = foo->GetTextureFilename();
string file_extension = boost::filesystem::extension(texture);
cout << "attempting load texture named " << texture
<< " whose extensions seems to be "
<< file_extension << endl;
// Use JPEG or PNG loader function, or report invalid extension
actually the STL can do this without much code, I advise you learn a bit about the STL because it lets you do some fancy things, anyways this is what I use.
std::string GetFileExtension(const std::string& FileName)
{
if(FileName.find_last_of(".") != std::string::npos)
return FileName.substr(FileName.find_last_of(".")+1);
return "";
}
this solution will always return the extension even on strings like "this.a.b.c.d.e.s.mp3" if it cannot find the extension it will return "".
Actually, the easiest way is
char* ext;
ext = strrchr(filename,'.')
One thing to remember: if '.' doesn't exist in filename, ext will be NULL.
I've stumbled onto this question today myself, even though I already had a working code I figured out that it wouldn't work in some cases.
While some people already suggested using some external libraries, I prefer to write my own code for learning purposes.
Some answers included the method I was using in the first place (looking for the last "."), but I remembered that on linux hidden files/folders start with ".".
So if file file is hidden and has no extension, the whole file name would be taken for extension.
To avoid that I wrote this piece of code:
bool getFileExtension(const char * dir_separator, const std::string & file, std::string & ext)
{
std::size_t ext_pos = file.rfind(".");
std::size_t dir_pos = file.rfind(dir_separator);
if(ext_pos>dir_pos+1)
{
ext.append(file.begin()+ext_pos,file.end());
return true;
}
return false;
}
I haven't tested this fully, but I think that it should work.
I'd go with boost::filesystem::extension (std::filesystem::path::extension with C++17) but if you cannot use Boost and you just have to verify the extension, a simple solution is:
bool ends_with(const std::string &filename, const std::string &ext)
{
return ext.length() <= filename.length() &&
std::equal(ext.rbegin(), ext.rend(), filename.rbegin());
}
if (ends_with(filename, ".conf"))
{ /* ... */ }
Using std::string's find/rfind solves THIS problem, but if you work a lot with paths then you should look at boost::filesystem::path since it will make your code much cleaner than fiddling with raw string indexes/iterators.
I suggest boost since it's a high quality, well tested, (open source and commercially) free and fully portable library.
For char array-type strings you can use this:
#include <ctype.h>
#include <string.h>
int main()
{
char filename[] = "apples.bmp";
char extension[] = ".jpeg";
if(compare_extension(filename, extension) == true)
{
// .....
} else {
// .....
}
return 0;
}
bool compare_extension(char *filename, char *extension)
{
/* Sanity checks */
if(filename == NULL || extension == NULL)
return false;
if(strlen(filename) == 0 || strlen(extension) == 0)
return false;
if(strchr(filename, '.') == NULL || strchr(extension, '.') == NULL)
return false;
/* Iterate backwards through respective strings and compare each char one at a time */
for(int i = 0; i < strlen(filename); i++)
{
if(tolower(filename[strlen(filename) - i - 1]) == tolower(extension[strlen(extension) - i - 1]))
{
if(i == strlen(extension) - 1)
return true;
} else
break;
}
return false;
}
Can handle file paths in addition to filenames. Works with both C and C++. And cross-platform.
If you use Qt library, you can give a try to QFileInfo's suffix()
Good answers but I see most of them has some problems:
First of all I think a good answer should work for complete file names which have their path headings, also it should work for linux or windows or as mentioned it should be cross platform. For most of answers; file names with no extension but a path with a folder name including dot, the function will fail to return the correct extension: examples of some test cases could be as follow:
const char filename1 = {"C:\\init.d\\doc"}; // => No extention
const char filename2 = {"..\\doc"}; //relative path name => No extention
const char filename3 = {""}; //emputy file name => No extention
const char filename4 = {"testing"}; //only single name => No extention
const char filename5 = {"tested/k.doc"}; // normal file name => doc
const char filename6 = {".."}; // parent folder => No extention
const char filename7 = {"/"}; // linux root => No extention
const char filename8 = {"/bin/test.d.config/lx.wize.str"}; // ordinary path! => str
"brian newman" suggestion will fail for filename1 and filename4.
and most of other answers which are based on reverse find will fail for filename1.
I suggest including the following method in your source:
which is function returning index of first character of extension or the length of given string if not found.
size_t find_ext_idx(const char* fileName)
{
size_t len = strlen(fileName);
size_t idx = len-1;
for(size_t i = 0; *(fileName+i); i++) {
if (*(fileName+i) == '.') {
idx = i;
} else if (*(fileName + i) == '/' || *(fileName + i) == '\\') {
idx = len - 1;
}
}
return idx+1;
}
you could use the above code in your c++ application like below:
std::string get_file_ext(const char* fileName)
{
return std::string(fileName).substr(find_ext_idx(fileName));
}
The last point in some cases the a folder is given to file name as argument and includes a dot in the folder name the function will return folder's dot trailing so better first to user check that the given name is a filename and not folder name.
This is a solution I came up with. Then, I noticed that it is similar to what #serengeor posted.
It works with std::string and find_last_of, but the basic idea will also work if modified to use char arrays and strrchr.
It handles hidden files, and extra dots representing the current directory. It is platform independent.
string PathGetExtension( string const & path )
{
string ext;
// Find the last dot, if any.
size_t dotIdx = path.find_last_of( "." );
if ( dotIdx != string::npos )
{
// Find the last directory separator, if any.
size_t dirSepIdx = path.find_last_of( "/\\" );
// If the dot is at the beginning of the file name, do not treat it as a file extension.
// e.g., a hidden file: ".alpha".
// This test also incidentally avoids a dot that is really a current directory indicator.
// e.g.: "alpha/./bravo"
if ( dotIdx > dirSepIdx + 1 )
{
ext = path.substr( dotIdx );
}
}
return ext;
}
Unit test:
int TestPathGetExtension( void )
{
int errCount = 0;
string tests[][2] =
{
{ "/alpha/bravo.txt", ".txt" },
{ "/alpha/.bravo", "" },
{ ".alpha", "" },
{ "./alpha.txt", ".txt" },
{ "alpha/./bravo", "" },
{ "alpha/./bravo.txt", ".txt" },
{ "./alpha", "" },
{ "c:\\alpha\\bravo.net\\charlie.txt", ".txt" },
};
int n = sizeof( tests ) / sizeof( tests[0] );
for ( int i = 0; i < n; ++i )
{
string ext = PathGetExtension( tests[i][0] );
if ( ext != tests[i][1] )
{
++errCount;
}
}
return errCount;
}
A NET/CLI version using System::String
System::String^ GetFileExtension(System::String^ FileName)
{
int Ext=FileName->LastIndexOf('.');
if( Ext != -1 )
return FileName->Substring(Ext+1);
return "";
}
_splitpath, _wsplitpath, _splitpath_s, _wsplitpath_w
This is Windows (Platform SDK) only
You can use strrchr() to find last occurence of .(dot) and get .(dot) based extensions files.
Check the below code for example.
#include<stdio.h>
void GetFileExtension(const char* file_name) {
int ext = '.';
const char* extension = NULL;
extension = strrchr(file_name, ext);
if(extension == NULL){
printf("Invalid extension encountered\n");
return;
}
printf("File extension is %s\n", extension);
}
int main()
{
const char* file_name = "c:\\.directoryname\\file.name.with.too.many.dots.ext";
GetFileExtension(file_name);
return 0;
}
Here's a function that takes a path/filename as a string and returns the extension as a string. It is all standard c++, and should work cross-platform for most platforms.
Unlike several other answers here, it handles the odd cases that windows' PathFindExtension handles, based on PathFindExtensions's documentation.
wstring get_file_extension( wstring filename )
{
size_t last_dot_offset = filename.rfind(L'.');
// This assumes your directory separators are either \ or /
size_t last_dirsep_offset = max( filename.rfind(L'\\'), filename.rfind(L'/') );
// no dot = no extension
if( last_dot_offset == wstring::npos )
return L"";
// directory separator after last dot = extension of directory, not file.
// for example, given C:\temp.old\file_that_has_no_extension we should return "" not "old"
if( (last_dirsep_offset != wstring::npos) && (last_dirsep_offset > last_dot_offset) )
return L"";
return filename.substr( last_dot_offset + 1 );
}
I use these two functions to get the extension and filename without extension:
std::string fileExtension(std::string file){
std::size_t found = file.find_last_of(".");
return file.substr(found+1);
}
std::string fileNameWithoutExtension(std::string file){
std::size_t found = file.find_last_of(".");
return file.substr(0,found);
}
And these regex approaches for certain extra requirements:
std::string fileExtension(std::string file){
std::regex re(".*[^\\.]+\\.([^\\.]+$)");
std::smatch result;
if(std::regex_match(file,result,re))return result[1];
else return "";
}
std::string fileNameWithoutExtension(std::string file){
std::regex re("(.*[^\\.]+)\\.[^\\.]+$");
std::smatch result;
if(std::regex_match(file,result,re))return result[1];
else return file;
}
Extra requirements that are met by the regex method:
If filename is like .config or something like this, extension will be an empty string and filename without extension will be .config.
If filename doesn't have any extension, extention will be an empty string, filename without extension will be the filename unchanged.
EDIT:
The extra requirements can also be met by the following:
std::string fileExtension(const std::string& file){
std::string::size_type pos=file.find_last_of('.');
if(pos!=std::string::npos&&pos!=0)return file.substr(pos+1);
else return "";
}
std::string fileNameWithoutExtension(const std::string& file){
std::string::size_type pos=file.find_last_of('.');
if(pos!=std::string::npos&&pos!=0)return file.substr(0,pos);
else return file;
}
Note:
Pass only the filenames (not path) in the above functions.
Try to use strstr
char* lastSlash;
lastSlash = strstr(filename, ".");
Or you can use this:
char *ExtractFileExt(char *FileName)
{
std::string s = FileName;
int Len = s.length();
while(TRUE)
{
if(FileName[Len] != '.')
Len--;
else
{
char *Ext = new char[s.length()-Len+1];
for(int a=0; a<s.length()-Len; a++)
Ext[a] = FileName[s.length()-(s.length()-Len)+a];
Ext[s.length()-Len] = '\0';
return Ext;
}
}
}
This code is cross-platform
So, using std::filesystem is the best answer, but if for whatever reason you don't have C++17 features available, this will work even if the input string includes directories:
string getextn (const string &fn) {
int sep = fn.find_last_of(".\\/");
return (sep >= 0 && fn[sep] == '.') ? fn.substr(sep) : "";
}
I'm adding this because the rest of the answers here are either strangely complicated or fail if the path to the file contains a dot and the file doesn't. I think the fact that find_last_of can look for multiple characters is often overlooked.
It works with both / and \ path separators. It fails if the extension itself contains a slash but that's usually too rare to matter. It doesn't do any filtering for filenames that start with a dot and contain no other dots -- if this matters to you then this is the least unreasonable answer here.
Example input / output:
/ => ''
./ => ''
./pathname/ => ''
./path.name/ => ''
pathname/ => ''
path.name/ => ''
c:\path.name\ => ''
/. => '.'
./. => '.'
./pathname/. => '.'
./path.name/. => '.'
pathname/. => '.'
path.name/. => '.'
c:\path.name\. => '.'
/.git_ignore => '.git_ignore'
./.git_ignore => '.git_ignore'
./pathname/.git_ignore => '.git_ignore'
./path.name/.git_ignore => '.git_ignore'
pathname/.git_ignore => '.git_ignore'
path.name/.git_ignore => '.git_ignore'
c:\path.name\.git_ignore => '.git_ignore'
/filename => ''
./filename => ''
./pathname/filename => ''
./path.name/filename => ''
pathname/filename => ''
path.name/filename => ''
c:\path.name\filename => ''
/filename. => '.'
./filename. => '.'
./pathname/filename. => '.'
./path.name/filename. => '.'
pathname/filename. => '.'
path.name/filename. => '.'
c:\path.name\filename. => '.'
/filename.tar => '.tar'
./filename.tar => '.tar'
./pathname/filename.tar => '.tar'
./path.name/filename.tar => '.tar'
pathname/filename.tar => '.tar'
path.name/filename.tar => '.tar'
c:\path.name\filename.tar => '.tar'
/filename.tar.gz => '.gz'
./filename.tar.gz => '.gz'
./pathname/filename.tar.gz => '.gz'
./path.name/filename.tar.gz => '.gz'
pathname/filename.tar.gz => '.gz'
path.name/filename.tar.gz => '.gz'
c:\path.name\filename.tar.gz => '.gz'
If you happen to use Poco libraries you can do:
#include <Poco/Path.h>
...
std::string fileExt = Poco::Path("/home/user/myFile.abc").getExtension(); // == "abc"
If you consider the extension as the last dot and the possible characters after it, but only if they don't contain the directory separator character, the following function returns the extension starting index, or -1 if no extension found. When you have that you can do what ever you want, like strip the extension, change it, check it etc.
long get_extension_index(string path, char dir_separator = '/') {
// Look from the end for the first '.',
// but give up if finding a dir separator char first
for(long i = path.length() - 1; i >= 0; --i) {
if(path[i] == '.') {
return i;
}
if(path[i] == dir_separator) {
return -1;
}
}
return -1;
}
I used PathFindExtension() function to know whether it is a valid tif file or not.
#include <Shlwapi.h>
bool A2iAWrapperUtility::isValidImageFile(string imageFile)
{
char * pStrExtension = ::PathFindExtension(imageFile.c_str());
if (pStrExtension != NULL && strcmp(pStrExtension, ".tif") == 0)
{
return true;
}
return false;
}