os— Miscellaneous operating system interfaces

Source code:Lib/os.py

---

This module provides a portable way of using operating system dependent functionality. If you just want to read or write a file seeopen(),if you want to manipulate paths, see theos.pathmodule, and if you want to read all the lines in all the files on the command line see thefileinput module. For creating temporary files and directories see thetempfile module, and for high-level file and directory handling see theshutil module.

Notes on the availability of these functions:

• The design of all built-in operating system dependent modules of Python is such that as long as the same functionality is available, it uses the same interface; for example, the functionos.stat(path)returns stat information aboutpathin the same format (which happens to have originated with the POSIX interface).

• Extensions peculiar to a particular operating system are also available through theosmodule, but using them is of course a threat to portability.

• All functions accepting path or file names accept both bytes and string objects, and result in an object of the same type, if a path or file name is returned.

• On VxWorks, os.popen, os.fork, os.execv and os.spawn*p* are not supported.

• On WebAssembly platforms, Android and iOS, large parts of theosmodule are not available or behave differently. APIs related to processes (e.g. fork(),execve()) and resources (e.g.nice()) are not available. Others likegetuid()andgetpid()are emulated or stubs. WebAssembly platforms also lack support for signals (e.g. kill(),wait()).

Note

All functions in this module raiseOSError(or subclasses thereof) in the case of invalid or inaccessible file names and paths, or other arguments that have the correct type, but are not accepted by the operating system.

exceptionos.error

An alias for the built-inOSErrorexception.

os.name

The name of the operating system dependent module imported. The following names have currently been registered:'posix','nt', 'java'.

See also

sys.platformhas a finer granularity.os.uname()gives system-dependent version information.

Theplatformmodule provides detailed checks for the system’s identity.

File Names, Command Line Arguments, and Environment Variables

In Python, file names, command line arguments, and environment variables are represented using the string type. On some systems, decoding these strings to and from bytes is necessary before passing them to the operating system. Python uses thefilesystem encoding and error handlerto perform this conversion (seesys.getfilesystemencoding()).

Thefilesystem encoding and error handlerare configured at Python startup by thePyConfig_Read()function: see filesystem_encodingand filesystem_errorsmembers ofPyConfig.

Changed in version 3.1:On some systems, conversion using the file system encoding may fail. In this case, Python uses thesurrogateescape encoding error handler,which means that undecodable bytes are replaced by a Unicode character U+DCxxon decoding, and these are again translated to the original byte on encoding.

Thefile system encodingmust guarantee to successfully decode all bytes below 128. If the file system encoding fails to provide this guarantee, API functions can raise UnicodeError.

See also thelocale encoding.

Python UTF-8 Mode

Added in version 3.7:SeePEP 540for more details.

The Python UTF-8 Mode ignores thelocale encodingand forces the usage of the UTF-8 encoding:

• Use UTF-8 as thefilesystem encoding.

• sys.getfilesystemencoding()returns'utf-8'.

• locale.getpreferredencoding()returns'utf-8'(thedo_setlocale argument has no effect).

• sys.stdin,sys.stdout,andsys.stderrall use UTF-8 as their text encoding, with thesurrogateescape error handlerbeing enabled forsys.stdin andsys.stdout(sys.stderrcontinues to use backslashreplaceas it does in the default locale-aware mode)

• On Unix,os.device_encoding()returns'utf-8'rather than the device encoding.

Note that the standard stream settings in UTF-8 mode can be overridden by PYTHONIOENCODING(just as they can be in the default locale-aware mode).

As a consequence of the changes in those lower level APIs, other higher level APIs also exhibit different default behaviours:

• Command line arguments, environment variables and filenames are decoded to text using the UTF-8 encoding.

• os.fsdecode()andos.fsencode()use the UTF-8 encoding.

• open(),io.open(),andcodecs.open()use the UTF-8 encoding by default. However, they still use the strict error handler by default so that attempting to open a binary file in text mode is likely to raise an exception rather than producing nonsense data.

ThePython UTF-8 Modeis enabled if the LC_CTYPE locale is CorPOSIXat Python startup (see thePyConfig_Read() function).

It can be enabled or disabled using the-Xutf8command line option and thePYTHONUTF8environment variable.

If thePYTHONUTF8environment variable is not set at all, then the interpreter defaults to using the current locale settings,unlessthe current locale is identified as a legacy ASCII-based locale (as described for PYTHONCOERCECLOCALE), and locale coercion is either disabled or fails. In such legacy locales, the interpreter will default to enabling UTF-8 mode unless explicitly instructed not to do so.

The Python UTF-8 Mode can only be enabled at the Python startup. Its value can be read fromsys.flags.utf8_mode.

See also theUTF-8 mode on Windows and thefilesystem encoding and error handler.

See also

PEP 686

Python 3.15 will makePython UTF-8 Modedefault.

Process Parameters

These functions and data items provide information and operate on the current process and user.

os.ctermid()

Return the filename corresponding to the controlling terminal of the process.

Availability:Unix, not WASI.

os.environ

Amappingobject where keys and values are strings that represent the process environment. For example,environ['HOME']is the pathname of your home directory (on some platforms), and is equivalent to getenv( "HOME" )in C.

This mapping is captured the first time theosmodule is imported, typically during Python startup as part of processingsite.py.Changes to the environment made after this time are not reflected inos.environ, except for changes made by modifyingos.environdirectly.

This mapping may be used to modify the environment as well as query the environment.putenv()will be called automatically when the mapping is modified.

On Unix, keys and values usesys.getfilesystemencoding()and 'surrogateescape'error handler. Useenvironbif you would like to use a different encoding.

On Windows, the keys are converted to uppercase. This also applies when getting, setting, or deleting an item. For example, environ['monty']='python'maps the key'MONTY'to the value 'python'.

Note

Callingputenv()directly does not changeos.environ,so it’s better to modifyos.environ.

Note

On some platforms, including FreeBSD and macOS, settingenvironmay cause memory leaks. Refer to the system documentation for putenv().

You can delete items in this mapping to unset environment variables. unsetenv()will be called automatically when an item is deleted from os.environ,and when one of thepop()orclear()methods is called.

Changed in version 3.9:Updated to supportPEP 584’s merge (|) and update (|=) operators.

os.environb

Bytes version ofenviron:amappingobject where both keys and values arebytesobjects representing the process environment. environandenvironbare synchronized (modifying environbupdatesenviron,and vice versa).

environbis only available ifsupports_bytes_environis True.

Added in version 3.2.

Changed in version 3.9:Updated to supportPEP 584’s merge (|) and update (|=) operators.

os.chdir(path)

os.fchdir(fd)

os.getcwd()

These functions are described inFiles and Directories.

os.fsencode(filename)

Encodepath-likefilenameto the filesystem encoding and error handler;returnbytes unchanged.

fsdecode()is the reverse function.

Added in version 3.2.

Changed in version 3.6:Support added to accept objects implementing theos.PathLike interface.

os.fsdecode(filename)

Decode thepath-likefilenamefrom the filesystem encoding and error handler;returnstr unchanged.

fsencode()is the reverse function.

Added in version 3.2.

Changed in version 3.6:Support added to accept objects implementing theos.PathLike interface.

os.fspath(path)

Return the file system representation of the path.

Ifstrorbytesis passed in, it is returned unchanged. Otherwise__fspath__()is called and its value is returned as long as it is astrorbytesobject. In all other cases,TypeErroris raised.

Added in version 3.6.

classos.PathLike

Anabstract base classfor objects representing a file system path, e.g.pathlib.PurePath.

Added in version 3.6.

abstractmethod__fspath__()

Return the file system path representation of the object.

The method should only return astrorbytesobject, with the preference being forstr.

os.getenv(key,default=None)

Return the value of the environment variablekeyas a string if it exists, or defaultif it doesn’t.keyis a string. Note that sincegetenv()usesos.environ,the mapping ofgetenv()is similarly also captured on import, and the function may not reflect future environment changes.

On Unix, keys and values are decoded withsys.getfilesystemencoding() and'surrogateescape'error handler. Useos.getenvb()if you would like to use a different encoding.

Availability:Unix, Windows.

os.getenvb(key,default=None)

Return the value of the environment variablekeyas bytes if it exists, or defaultif it doesn’t.keymust be bytes. Note that sincegetenvb()usesos.environb,the mapping ofgetenvb()is similarly also captured on import, and the function may not reflect future environment changes.

getenvb()is only available ifsupports_bytes_environ isTrue.

Availability:Unix.

Added in version 3.2.

os.get_exec_path(env=None)

Returns the list of directories that will be searched for a named executable, similar to a shell, when launching a process. env,when specified, should be an environment variable dictionary to lookup the PATH in. By default, whenenvisNone,environis used.

Added in version 3.2.

os.getegid()

Return the effective group id of the current process. This corresponds to the “set id” bit on the file being executed in the current process.

Availability:Unix, not WASI.

os.geteuid()

Return the current process’s effective user id.

Availability:Unix, not WASI.

os.getgid()

Return the real group id of the current process.

Availability:Unix.

The function is a stub on WASI, seeWebAssembly platformsfor more information.

os.getgrouplist(user,group,/)

Return list of group ids thatuserbelongs to. Ifgroupis not in the list, it is included; typically,groupis specified as the group ID field from the password record foruser,because that group ID will otherwise be potentially omitted.

Availability:Unix, not WASI.

Added in version 3.3.

os.getgroups()

Return list of supplemental group ids associated with the current process.

Availability:Unix, not WASI.

Note

On macOS,getgroups()behavior differs somewhat from other Unix platforms. If the Python interpreter was built with a deployment target of10.5or earlier,getgroups()returns the list of effective group ids associated with the current user process; this list is limited to a system-defined number of entries, typically 16, and may be modified by calls tosetgroups()if suitably privileged. If built with a deployment target greater than10.5, getgroups()returns the current group access list for the user associated with the effective user id of the process; the group access list may change over the lifetime of the process, it is not affected by calls tosetgroups(),and its length is not limited to 16. The deployment target value,MACOSX_DEPLOYMENT_TARGET,can be obtained withsysconfig.get_config_var().

os.getlogin()

Return the name of the user logged in on the controlling terminal of the process. For most purposes, it is more useful to use getpass.getuser()since the latter checks the environment variables LOGNAMEorUSERNAMEto find out who the user is, and falls back topwd.getpwuid(os.getuid())[0]to get the login name of the current real user id.

Availability:Unix, Windows, not WASI.

os.getpgid(pid)

Return the process group id of the process with process idpid.Ifpidis 0, the process group id of the current process is returned.

Availability:Unix, not WASI.

os.getpgrp()

Return the id of the current process group.

Availability:Unix, not WASI.

os.getpid()

Return the current process id.

The function is a stub on WASI, seeWebAssembly platformsfor more information.

os.getppid()

Return the parent’s process id. When the parent process has exited, on Unix the id returned is the one of the init process (1), on Windows it is still the same id, which may be already reused by another process.

Availability:Unix, Windows, not WASI.

Changed in version 3.2:Added support for Windows.

os.getpriority(which,who)

Get program scheduling priority. The valuewhichis one of PRIO_PROCESS,PRIO_PGRP,orPRIO_USER,andwho is interpreted relative towhich(a process identifier for PRIO_PROCESS,process group identifier forPRIO_PGRP,and a user ID forPRIO_USER). A zero value forwhodenotes (respectively) the calling process, the process group of the calling process, or the real user ID of the calling process.

Availability:Unix, not WASI.

Added in version 3.3.

os.PRIO_PROCESS

os.PRIO_PGRP

os.PRIO_USER

Parameters for thegetpriority()andsetpriority()functions.

Availability:Unix, not WASI.

Added in version 3.3.

os.PRIO_DARWIN_THREAD

os.PRIO_DARWIN_PROCESS

os.PRIO_DARWIN_BG

os.PRIO_DARWIN_NONUI

Parameters for thegetpriority()andsetpriority()functions.

Availability:macOS

Added in version 3.12.

os.getresuid()

Return a tuple (ruid, euid, suid) denoting the current process’s real, effective, and saved user ids.

Availability:Unix, not WASI.

Added in version 3.2.

os.getresgid()

Return a tuple (rgid, egid, sgid) denoting the current process’s real, effective, and saved group ids.

Availability:Unix, not WASI.

Added in version 3.2.

os.getuid()

Return the current process’s real user id.

Availability:Unix.

The function is a stub on WASI, seeWebAssembly platformsfor more information.

os.initgroups(username,gid,/)

Call the system initgroups() to initialize the group access list with all of the groups of which the specified username is a member, plus the specified group id.

Availability:Unix, not WASI, not Android.

Added in version 3.2.

os.putenv(key,value,/)

Set the environment variable namedkeyto the stringvalue.Such changes to the environment affect subprocesses started withos.system(), popen()orfork()andexecv().

Assignments to items inos.environare automatically translated into corresponding calls toputenv();however, calls toputenv() don’t updateos.environ,so it is actually preferable to assign to items ofos.environ.This also applies togetenv()andgetenvb(),which respectively useos.environandos.environbin their implementations.

Note

On some platforms, including FreeBSD and macOS, settingenvironmay cause memory leaks. Refer to the system documentation forputenv().

Raises anauditing eventos.putenvwith argumentskey,value.

Changed in version 3.9:The function is now always available.

os.setegid(egid,/)

Set the current process’s effective group id.

Availability:Unix, not WASI, not Android.

os.seteuid(euid,/)

Set the current process’s effective user id.

Availability:Unix, not WASI, not Android.

os.setgid(gid,/)

Set the current process’ group id.

Availability:Unix, not WASI, not Android.

os.setgroups(groups,/)

Set the list of supplemental group ids associated with the current process to groups.groupsmust be a sequence, and each element must be an integer identifying a group. This operation is typically available only to the superuser.

Availability:Unix, not WASI.

Note

On macOS, the length ofgroupsmay not exceed the system-defined maximum number of effective group ids, typically 16. See the documentation forgetgroups()for cases where it may not return the same group list set by calling setgroups().

os.setns(fd,nstype=0)

Reassociate the current thread with a Linux namespace. See thesetns(2)andnamespaces(7)man pages for more details.

Iffdrefers to a/proc/pid/ns/link,setns()reassociates the calling thread with the namespace associated with that link, andnstypemay be set to one of the CLONE_NEW* constants to impose constraints on the operation (0means no constraints).

Since Linux 5.8,fdmay refer to a PID file descriptor obtained from pidfd_open().In this case,setns()reassociates the calling thread into one or more of the same namespaces as the thread referred to byfd. This is subject to any constraints imposed bynstype, which is a bit mask combining one or more of the CLONE_NEW* constants, e.g.setns(fd,os.CLONE_NEWUTS|os.CLONE_NEWPID). The caller’s memberships in unspecified namespaces are left unchanged.

fdcan be any object with afileno()method, or a raw file descriptor.

This example reassociates the thread with theinitprocess’s network namespace:

fd=os.open("/proc/1/ns/net",os.O_RDONLY)
os.setns(fd,os.CLONE_NEWNET)
os.close(fd)

Availability:Linux >= 3.0 with glibc >= 2.14.

Added in version 3.12.

See also

Theunshare()function.

os.setpgrp()

Call the system callsetpgrp()orsetpgrp(0,0)depending on which version is implemented (if any). See the Unix manual for the semantics.

Availability:Unix, not WASI.

os.setpgid(pid,pgrp,/)

Call the system callsetpgid()to set the process group id of the process with idpidto the process group with idpgrp.See the Unix manual for the semantics.

Availability:Unix, not WASI.

os.setpriority(which,who,priority)

Set program scheduling priority. The valuewhichis one of PRIO_PROCESS,PRIO_PGRP,orPRIO_USER,andwho is interpreted relative towhich(a process identifier for PRIO_PROCESS,process group identifier forPRIO_PGRP,and a user ID forPRIO_USER). A zero value forwhodenotes (respectively) the calling process, the process group of the calling process, or the real user ID of the calling process. priorityis a value in the range -20 to 19. The default priority is 0; lower priorities cause more favorable scheduling.

Availability:Unix, not WASI.

Added in version 3.3.

os.setregid(rgid,egid,/)

Set the current process’s real and effective group ids.

Availability:Unix, not WASI, not Android.

os.setresgid(rgid,egid,sgid,/)

Set the current process’s real, effective, and saved group ids.

Availability:Unix, not WASI, not Android.

Added in version 3.2.

os.setresuid(ruid,euid,suid,/)

Set the current process’s real, effective, and saved user ids.

Availability:Unix, not WASI, not Android.

Added in version 3.2.

os.setreuid(ruid,euid,/)

Set the current process’s real and effective user ids.

Availability:Unix, not WASI, not Android.

os.getsid(pid,/)

Call the system callgetsid().See the Unix manual for the semantics.

Availability:Unix, not WASI.

os.setsid()

Call the system callsetsid().See the Unix manual for the semantics.

Availability:Unix, not WASI.

os.setuid(uid,/)

Set the current process’s user id.

Availability:Unix, not WASI, not Android.

os.strerror(code,/)

Return the error message corresponding to the error code incode. On platforms wherestrerror()returnsNULLwhen given an unknown error number,ValueErroris raised.

os.supports_bytes_environ

Trueif the native OS type of the environment is bytes (eg.Falseon Windows).

Added in version 3.2.

os.umask(mask,/)

Set the current numeric umask and return the previous umask.

The function is a stub on WASI, seeWebAssembly platformsfor more information.

os.uname()

Returns information identifying the current operating system. The return value is an object with five attributes:

• sysname- operating system name

• nodename- name of machine on network (implementation-defined)

• release- operating system release

• version- operating system version

• machine- hardware identifier

For backwards compatibility, this object is also iterable, behaving like a five-tuple containingsysname,nodename, release,version,andmachine in that order.

Some systems truncatenodenameto 8 characters or to the leading component; a better way to get the hostname is socket.gethostname()or even socket.gethostbyaddr(socket.gethostname()).

On macOS, iOS and Android, this returns thekernelname and version (i.e., 'Darwin'on macOS and iOS;'Linux'on Android).platform.uname() can be used to get the user-facing operating system name and version on iOS and Android.

Availability:Unix.

Changed in version 3.3:Return type changed from a tuple to a tuple-like object with named attributes.

os.unsetenv(key,/)

Unset (delete) the environment variable namedkey.Such changes to the environment affect subprocesses started withos.system(),popen()or fork()andexecv().

Deletion of items inos.environis automatically translated into a corresponding call tounsetenv();however, calls tounsetenv() don’t updateos.environ,so it is actually preferable to delete items of os.environ.

Raises anauditing eventos.unsetenvwith argumentkey.

Changed in version 3.9:The function is now always available and is also available on Windows.

os.unshare(flags)

Disassociate parts of the process execution context, and move them into a newly created namespace. See theunshare(2) man page for more details. Theflagsargument is a bit mask, combining zero or more of the CLONE_* constants, that specifies which parts of the execution context should be unshared from their existing associations and moved to a new namespace. If theflagsargument is0,no changes are made to the calling process’s execution context.

Availability:Linux >= 2.6.16.

Added in version 3.12.

See also

Thesetns()function.

Flags to theunshare()function, if the implementation supports them. Seeunshare(2)in the Linux manual for their exact effect and availability.

os.CLONE_FILES

os.CLONE_FS

os.CLONE_NEWCGROUP

os.CLONE_NEWIPC

os.CLONE_NEWNET

os.CLONE_NEWNS

os.CLONE_NEWPID

os.CLONE_NEWTIME

os.CLONE_NEWUSER

os.CLONE_NEWUTS

os.CLONE_SIGHAND

os.CLONE_SYSVSEM

os.CLONE_THREAD

os.CLONE_VM

File Object Creation

These functions create newfile objects.(See also open()for opening file descriptors.)

os.fdopen(fd,*args,**kwargs)

Return an open file object connected to the file descriptorfd.This is an alias of theopen()built-in function and accepts the same arguments. The only difference is that the first argument offdopen()must always be an integer.

File Descriptor Operations

These functions operate on I/O streams referenced using file descriptors.

File descriptors are small integers corresponding to a file that has been opened by the current process. For example, standard input is usually file descriptor 0, standard output is 1, and standard error is 2. Further files opened by a process will then be assigned 3, 4, 5, and so forth. The name “file descriptor” is slightly deceptive; on Unix platforms, sockets and pipes are also referenced by file descriptors.

Thefileno()method can be used to obtain the file descriptor associated with afile objectwhen required. Note that using the file descriptor directly will bypass the file object methods, ignoring aspects such as internal buffering of data.

os.close(fd)

Close file descriptorfd.

Note

This function is intended for low-level I/O and must be applied to a file descriptor as returned byos.open()orpipe().To close a “file object” returned by the built-in functionopen()or bypopen()or fdopen(),use itsclose()method.

os.closerange(fd_low,fd_high,/)

Close all file descriptors fromfd_low(inclusive) tofd_high(exclusive), ignoring errors. Equivalent to (but much faster than):

forfdinrange(fd_low,fd_high):
try:
os.close(fd)
exceptOSError:
pass

os.copy_file_range(src,dst,count,offset_src=None,offset_dst=None)

Copycountbytes from file descriptorsrc,starting from offset offset_src,to file descriptordst,starting from offsetoffset_dst. Ifoffset_srcisNone,thensrcis read from the current position; respectively foroffset_dst.

In Linux kernel older than 5.3, the files pointed to bysrcanddst must reside in the same filesystem, otherwise anOSErroris raised witherrnoset toerrno.EXDEV.

This copy is done without the additional cost of transferring data from the kernel to user space and then back into the kernel. Additionally, some filesystems could implement extra optimizations, such as the use of reflinks (i.e., two or more inodes that share pointers to the same copy-on-write disk blocks; supported file systems include btrfs and XFS) and server-side copy (in the case of NFS).

The function copies bytes between two file descriptors. Text options, like the encoding and the line ending, are ignored.

The return value is the amount of bytes copied. This could be less than the amount requested.

Note

On Linux,os.copy_file_range()should not be used for copying a range of a pseudo file from a special filesystem like procfs and sysfs. It will always copy no bytes and return 0 as if the file was empty because of a known Linux kernel issue.

Availability:Linux >= 4.5 with glibc >= 2.27.

Added in version 3.8.

os.device_encoding(fd)

Return a string describing the encoding of the device associated withfd if it is connected to a terminal; else returnNone.

On Unix, if thePython UTF-8 Modeis enabled, return 'UTF-8'rather than the device encoding.

Changed in version 3.10:On Unix, the function now implements the Python UTF-8 Mode.

os.dup(fd,/)

Return a duplicate of file descriptorfd.The new file descriptor is non-inheritable.

On Windows, when duplicating a standard stream (0: stdin, 1: stdout, 2: stderr), the new file descriptor isinheritable.

Availability:not WASI.

Changed in version 3.4:The new file descriptor is now non-inheritable.

os.dup2(fd,fd2,inheritable=True)

Duplicate file descriptorfdtofd2,closing the latter first if necessary. Returnfd2.The new file descriptor isinheritableby default or non-inheritable ifinheritable isFalse.

Availability:not WASI.

Changed in version 3.4:Add the optionalinheritableparameter.

Changed in version 3.7:Returnfd2on success. Previously,Nonewas always returned.

os.fchmod(fd,mode)

Change the mode of the file given byfdto the numericmode.See the docs forchmod()for possible values ofmode.As of Python 3.3, this is equivalent toos.chmod(fd,mode).

Raises anauditing eventos.chmodwith argumentspath,mode,dir_fd.

Availability:Unix, Windows.

The function is limited on WASI, seeWebAssembly platformsfor more information.

Changed in version 3.13:Added support on Windows.

os.fchown(fd,uid,gid)

Change the owner and group id of the file given byfdto the numericuid andgid.To leave one of the ids unchanged, set it to -1. See chown().As of Python 3.3, this is equivalent toos.chown(fd,uid, gid).

Raises anauditing eventos.chownwith argumentspath,uid,gid,dir_fd.

Availability:Unix.

The function is limited on WASI, seeWebAssembly platformsfor more information.

os.fdatasync(fd)

Force write of file with filedescriptorfdto disk. Does not force update of metadata.

Availability:Unix.

Note

This function is not available on MacOS.

os.fpathconf(fd,name,/)

Return system configuration information relevant to an open file.name specifies the configuration value to retrieve; it may be a string which is the name of a defined system value; these names are specified in a number of standards (POSIX.1, Unix 95, Unix 98, and others). Some platforms define additional names as well. The names known to the host operating system are given in thepathconf_namesdictionary. For configuration variables not included in that mapping, passing an integer fornameis also accepted.

Ifnameis a string and is not known,ValueErroris raised. If a specific value fornameis not supported by the host system, even if it is included inpathconf_names,anOSErroris raised with errno.EINVALfor the error number.

As of Python 3.3, this is equivalent toos.pathconf(fd,name).

Availability:Unix.

os.fstat(fd)

Get the status of the file descriptorfd.Return astat_result object.

As of Python 3.3, this is equivalent toos.stat(fd).

See also

Thestat()function.

os.fstatvfs(fd,/)

Return information about the filesystem containing the file associated with file descriptorfd,likestatvfs().As of Python 3.3, this is equivalent toos.statvfs(fd).

Availability:Unix.

os.fsync(fd)

Force write of file with filedescriptorfdto disk. On Unix, this calls the nativefsync()function; on Windows, the MS_commit()function.

If you’re starting with a buffered Pythonfile objectf,first do f.flush(),and then doos.fsync(f.fileno()),to ensure that all internal buffers associated withfare written to disk.

Availability:Unix, Windows.

os.ftruncate(fd,length,/)

Truncate the file corresponding to file descriptorfd,so that it is at mostlengthbytes in size. As of Python 3.3, this is equivalent to os.truncate(fd,length).

Raises anauditing eventos.truncatewith argumentsfd,length.

Availability:Unix, Windows.

Changed in version 3.5:Added support for Windows

os.get_blocking(fd,/)

Get the blocking mode of the file descriptor:Falseif the O_NONBLOCKflag is set,Trueif the flag is cleared.

See alsoset_blocking()andsocket.socket.setblocking().

Availability:Unix, Windows.

The function is limited on WASI, seeWebAssembly platformsfor more information.

On Windows, this function is limited to pipes.

Added in version 3.5.

Changed in version 3.12:Added support for pipes on Windows.

os.grantpt(fd,/)

Grant access to the slave pseudo-terminal device associated with the master pseudo-terminal device to which the file descriptorfdrefers. The file descriptorfdis not closed upon failure.

Calls the C standard library functiongrantpt().

Availability:Unix, not WASI.

Added in version 3.13.

os.isatty(fd,/)

ReturnTrueif the file descriptorfdis open and connected to a tty(-like) device, elseFalse.

os.lockf(fd,cmd,len,/)

Apply, test or remove a POSIX lock on an open file descriptor. fdis an open file descriptor. cmdspecifies the command to use - one ofF_LOCK,F_TLOCK, F_ULOCKorF_TEST. lenspecifies the section of the file to lock.

Raises anauditing eventos.lockfwith argumentsfd,cmd,len.

Availability:Unix.

Added in version 3.3.

os.F_LOCK

os.F_TLOCK

os.F_ULOCK

os.F_TEST

Flags that specify what actionlockf()will take.

Availability:Unix.

Added in version 3.3.

os.login_tty(fd,/)

Prepare the tty of which fd is a file descriptor for a new login session. Make the calling process a session leader; make the tty the controlling tty, the stdin, the stdout, and the stderr of the calling process; close fd.

Availability:Unix, not WASI.

Added in version 3.11.

os.lseek(fd,pos,whence,/)

Set the current position of file descriptorfdto positionpos,modified bywhence,and return the new position in bytes relative to the start of the file. Valid values forwhenceare:

• SEEK_SETor0– setposrelative to the beginning of the file

• SEEK_CURor1– setposrelative to the current file position

• SEEK_ENDor2– setposrelative to the end of the file

• SEEK_HOLE– setposto the next data location, relative topos

• SEEK_DATA– setposto the next data hole, relative topos

Changed in version 3.3:Add support forSEEK_HOLEandSEEK_DATA.

os.SEEK_SET

os.SEEK_CUR

os.SEEK_END

Parameters to thelseek()function and theseek() method onfile-like objects, for whence to adjust the file position indicator.

SEEK_SET

Adjust the file position relative to the beginning of the file.

SEEK_CUR

Adjust the file position relative to the current file position.

SEEK_END

Adjust the file position relative to the end of the file.

Their values are 0, 1, and 2, respectively.

os.SEEK_HOLE

os.SEEK_DATA

Parameters to thelseek()function and theseek() method onfile-like objects, for seeking file data and holes on sparsely allocated files.

SEEK_DATA

Adjust the file offset to the next location containing data, relative to the seek position.

SEEK_HOLE

Adjust the file offset to the next location containing a hole, relative to the seek position. A hole is defined as a sequence of zeros.

Note

These operations only make sense for filesystems that support them.

Availability:Linux >= 3.1, macOS, Unix

Added in version 3.3.

os.open(path,flags,mode=0o777,*,dir_fd=None)

Open the filepathand set various flags according toflagsand possibly its mode according tomode.When computingmode,the current umask value is first masked out. Return the file descriptor for the newly opened file. The new file descriptor isnon-inheritable.

For a description of the flag and mode values, see the C run-time documentation; flag constants (likeO_RDONLYandO_WRONLY) are defined in theosmodule. In particular, on Windows adding O_BINARYis needed to open files in binary mode.

This function can supportpaths relative to directory descriptorswith thedir_fdparameter.

Raises anauditing eventopenwith argumentspath,mode,flags.

Changed in version 3.4:The new file descriptor is now non-inheritable.

Note

This function is intended for low-level I/O. For normal usage, use the built-in functionopen(),which returns afile objectwith read()andwrite()methods (and many more). To wrap a file descriptor in a file object, usefdopen().

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.5:If the system call is interrupted and the signal handler does not raise an exception, the function now retries the system call instead of raising an InterruptedErrorexception (seePEP 475for the rationale).

Changed in version 3.6:Accepts apath-like object.

The following constants are options for theflagsparameter to the open()function. They can be combined using the bitwise OR operator |.Some of them are not available on all platforms. For descriptions of their availability and use, consult theopen(2)manual page on Unix orthe MSDNon Windows.

os.O_RDONLY

os.O_WRONLY

os.O_RDWR

os.O_APPEND

os.O_CREAT

os.O_EXCL

os.O_TRUNC

The above constants are available on Unix and Windows.

os.O_DSYNC

os.O_RSYNC

os.O_SYNC

os.O_NDELAY

os.O_NONBLOCK

os.O_NOCTTY

os.O_CLOEXEC

The above constants are only available on Unix.

Changed in version 3.3:AddO_CLOEXECconstant.

os.O_BINARY

os.O_NOINHERIT

os.O_SHORT_LIVED

os.O_TEMPORARY

os.O_RANDOM

os.O_SEQUENTIAL

os.O_TEXT

The above constants are only available on Windows.

os.O_EVTONLY

os.O_FSYNC

os.O_SYMLINK

os.O_NOFOLLOW_ANY

The above constants are only available on macOS.

Changed in version 3.10:AddO_EVTONLY,O_FSYNC,O_SYMLINK andO_NOFOLLOW_ANYconstants.

os.O_ASYNC

os.O_DIRECT

os.O_DIRECTORY

os.O_NOFOLLOW

os.O_NOATIME

os.O_PATH

os.O_TMPFILE

os.O_SHLOCK

os.O_EXLOCK

The above constants are extensions and not present if they are not defined by the C library.

Changed in version 3.4:AddO_PATHon systems that support it. AddO_TMPFILE,only available on Linux Kernel 3.11 or newer.

os.openpty()

Open a new pseudo-terminal pair. Return a pair of file descriptors (master,slave)for the pty and the tty, respectively. The new file descriptors arenon-inheritable.For a (slightly) more portable approach, use theptymodule.

Availability:Unix, not WASI.

Changed in version 3.4:The new file descriptors are now non-inheritable.

os.pipe()

Create a pipe. Return a pair of file descriptors(r,w)usable for reading and writing, respectively. The new file descriptor is non-inheritable.

Availability:Unix, Windows.

Changed in version 3.4:The new file descriptors are now non-inheritable.

os.pipe2(flags,/)

Create a pipe withflagsset atomically. flagscan be constructed by ORing together one or more of these values: O_NONBLOCK,O_CLOEXEC. Return a pair of file descriptors(r,w)usable for reading and writing, respectively.

Availability:Unix, not WASI.

Added in version 3.3.

os.posix_fallocate(fd,offset,len,/)

Ensures that enough disk space is allocated for the file specified byfd starting fromoffsetand continuing forlenbytes.

Availability:Unix.

Added in version 3.3.

os.posix_fadvise(fd,offset,len,advice,/)

Announces an intention to access data in a specific pattern thus allowing the kernel to make optimizations. The advice applies to the region of the file specified byfdstarting at offsetand continuing forlenbytes. adviceis one ofPOSIX_FADV_NORMAL,POSIX_FADV_SEQUENTIAL, POSIX_FADV_RANDOM,POSIX_FADV_NOREUSE, POSIX_FADV_WILLNEEDorPOSIX_FADV_DONTNEED.

Availability:Unix.

Added in version 3.3.

os.POSIX_FADV_NORMAL

os.POSIX_FADV_SEQUENTIAL

os.POSIX_FADV_RANDOM

os.POSIX_FADV_NOREUSE

os.POSIX_FADV_WILLNEED

os.POSIX_FADV_DONTNEED

Flags that can be used inadviceinposix_fadvise()that specify the access pattern that is likely to be used.

Availability:Unix.

Added in version 3.3.

os.pread(fd,n,offset,/)

Read at mostnbytes from file descriptorfdat a position ofoffset, leaving the file offset unchanged.

Return a bytestring containing the bytes read. If the end of the file referred to byfdhas been reached, an empty bytes object is returned.

Availability:Unix.

Added in version 3.3.

os.posix_openpt(oflag,/)

Open and return a file descriptor for a master pseudo-terminal device.

Calls the C standard library functionposix_openpt().Theoflag argument is used to set file status flags and file access modes as specified in the manual page ofposix_openpt()of your system.

The returned file descriptor isnon-inheritable. If the valueO_CLOEXECis available on the system, it is added to oflag.

Availability:Unix, not WASI.

Added in version 3.13.

os.preadv(fd,buffers,offset,flags=0,/)

Read from a file descriptorfdat a position ofoffsetinto mutable bytes-like objectsbuffers,leaving the file offset unchanged. Transfer data into each buffer until it is full and then move on to the next buffer in the sequence to hold the rest of the data.

The flags argument contains a bitwise OR of zero or more of the following flags:

• RWF_HIPRI

• RWF_NOWAIT

Return the total number of bytes actually read which can be less than the total capacity of all the objects.

The operating system may set a limit (sysconf()value 'SC_IOV_MAX') on the number of buffers that can be used.

Combine the functionality ofos.readv()andos.pread().

Availability:Linux >= 2.6.30, FreeBSD >= 6.0, OpenBSD >= 2.7, AIX >= 7.1.

Using flags requires Linux >= 4.6.

Added in version 3.7.

os.RWF_NOWAIT

Do not wait for data which is not immediately available. If this flag is specified, the system call will return instantly if it would have to read data from the backing storage or wait for a lock.

If some data was successfully read, it will return the number of bytes read. If no bytes were read, it will return-1and set errno to errno.EAGAIN.

Availability:Linux >= 4.14.

Added in version 3.7.

os.RWF_HIPRI

High priority read/write. Allows block-based filesystems to use polling of the device, which provides lower latency, but may use additional resources.

Currently, on Linux, this feature is usable only on a file descriptor opened using theO_DIRECTflag.

Availability:Linux >= 4.6.

Added in version 3.7.

os.ptsname(fd,/)

Return the name of the slave pseudo-terminal device associated with the master pseudo-terminal device to which the file descriptorfdrefers. The file descriptorfdis not closed upon failure.

Calls the reentrant C standard library functionptsname_r()if it is available; otherwise, the C standard library function ptsname(),which is not guaranteed to be thread-safe, is called.

Availability:Unix, not WASI.

Added in version 3.13.

os.pwrite(fd,str,offset,/)

Write the bytestring instrto file descriptorfdat position of offset,leaving the file offset unchanged.

Return the number of bytes actually written.

Availability:Unix.

Added in version 3.3.

os.pwritev(fd,buffers,offset,flags=0,/)

Write thebufferscontents to file descriptorfdat an offsetoffset, leaving the file offset unchanged.buffersmust be a sequence of bytes-like objects.Buffers are processed in array order. Entire contents of the first buffer is written before proceeding to the second, and so on.

The flags argument contains a bitwise OR of zero or more of the following flags:

• RWF_DSYNC

• RWF_SYNC

• RWF_APPEND

Return the total number of bytes actually written.

The operating system may set a limit (sysconf()value 'SC_IOV_MAX') on the number of buffers that can be used.

Combine the functionality ofos.writev()andos.pwrite().

Availability:Linux >= 2.6.30, FreeBSD >= 6.0, OpenBSD >= 2.7, AIX >= 7.1.

Using flags requires Linux >= 4.6.

Added in version 3.7.

os.RWF_DSYNC

Provide a per-write equivalent of theO_DSYNCos.open()flag. This flag effect applies only to the data range written by the system call.

Availability:Linux >= 4.7.

Added in version 3.7.

os.RWF_SYNC

Provide a per-write equivalent of theO_SYNCos.open()flag. This flag effect applies only to the data range written by the system call.

Availability:Linux >= 4.7.

Added in version 3.7.

os.RWF_APPEND

Provide a per-write equivalent of theO_APPENDos.open() flag. This flag is meaningful only foros.pwritev(),and its effect applies only to the data range written by the system call. The offsetargument does not affect the write operation; the data is always appended to the end of the file. However, if theoffsetargument is -1,the current fileoffsetis updated.

Availability:Linux >= 4.16.

Added in version 3.10.

os.read(fd,n,/)

Read at mostnbytes from file descriptorfd.

Return a bytestring containing the bytes read. If the end of the file referred to byfdhas been reached, an empty bytes object is returned.

Note

This function is intended for low-level I/O and must be applied to a file descriptor as returned byos.open()orpipe().To read a “file object” returned by the built-in functionopen()or by popen()orfdopen(),orsys.stdin,use its read()orreadline()methods.

Changed in version 3.5:If the system call is interrupted and the signal handler does not raise an exception, the function now retries the system call instead of raising an InterruptedErrorexception (seePEP 475for the rationale).

os.sendfile(out_fd,in_fd,offset,count)

os.sendfile(out_fd,in_fd,offset,count,headers=(),trailers=(),flags=0)

Copycountbytes from file descriptorin_fdto file descriptorout_fd starting atoffset. Return the number of bytes sent. When EOF is reached return0.

The first function notation is supported by all platforms that define sendfile().

On Linux, ifoffsetis given asNone,the bytes are read from the current position ofin_fdand the position ofin_fdis updated.

The second case may be used on macOS and FreeBSD whereheadersand trailersare arbitrary sequences of buffers that are written before and after the data fromin_fdis written. It returns the same as the first case.

On macOS and FreeBSD, a value of0forcountspecifies to send until the end ofin_fdis reached.

All platforms support sockets asout_fdfile descriptor, and some platforms allow other types (e.g. regular file, pipe) as well.

Cross-platform applications should not useheaders,trailersandflags arguments.

Availability:Unix, not WASI.

Note

For a higher-level wrapper ofsendfile(),see socket.socket.sendfile().

Added in version 3.3.

Changed in version 3.9:Parametersoutandinwas renamed toout_fdandin_fd.

os.SF_NODISKIO

os.SF_MNOWAIT

os.SF_SYNC

Parameters to thesendfile()function, if the implementation supports them.

Availability:Unix, not WASI.

Added in version 3.3.

os.SF_NOCACHE

Parameter to thesendfile()function, if the implementation supports it. The data won’t be cached in the virtual memory and will be freed afterwards.

Availability:Unix, not WASI.

Added in version 3.11.

os.set_blocking(fd,blocking,/)

Set the blocking mode of the specified file descriptor. Set the O_NONBLOCKflag if blocking isFalse,clear the flag otherwise.

See alsoget_blocking()andsocket.socket.setblocking().

Availability:Unix, Windows.

The function is limited on WASI, seeWebAssembly platformsfor more information.

On Windows, this function is limited to pipes.

Added in version 3.5.

Changed in version 3.12:Added support for pipes on Windows.

os.splice(src,dst,count,offset_src=None,offset_dst=None)

Transfercountbytes from file descriptorsrc,starting from offset offset_src,to file descriptordst,starting from offsetoffset_dst. At least one of the file descriptors must refer to a pipe. Ifoffset_src isNone,thensrcis read from the current position; respectively for offset_dst.The offset associated to the file descriptor that refers to a pipe must beNone.The files pointed to bysrcanddstmust reside in the same filesystem, otherwise anOSErroris raised with errnoset toerrno.EXDEV.

This copy is done without the additional cost of transferring data from the kernel to user space and then back into the kernel. Additionally, some filesystems could implement extra optimizations. The copy is done as if both files are opened as binary.

Upon successful completion, returns the number of bytes spliced to or from the pipe. A return value of 0 means end of input. Ifsrcrefers to a pipe, then this means that there was no data to transfer, and it would not make sense to block because there are no writers connected to the write end of the pipe.

Availability:Linux >= 2.6.17 with glibc >= 2.5

Added in version 3.10.

os.SPLICE_F_MOVE

os.SPLICE_F_NONBLOCK

os.SPLICE_F_MORE

Added in version 3.10.

os.readv(fd,buffers,/)

Read from a file descriptorfdinto a number of mutablebytes-like objectsbuffers.Transfer data into each buffer until it is full and then move on to the next buffer in the sequence to hold the rest of the data.

Return the total number of bytes actually read which can be less than the total capacity of all the objects.

The operating system may set a limit (sysconf()value 'SC_IOV_MAX') on the number of buffers that can be used.

Availability:Unix.

Added in version 3.3.

os.tcgetpgrp(fd,/)

Return the process group associated with the terminal given byfd(an open file descriptor as returned byos.open()).

Availability:Unix, not WASI.

os.tcsetpgrp(fd,pg,/)

Set the process group associated with the terminal given byfd(an open file descriptor as returned byos.open()) topg.

Availability:Unix, not WASI.

os.ttyname(fd,/)

Return a string which specifies the terminal device associated with file descriptorfd.Iffdis not associated with a terminal device, an exception is raised.

Availability:Unix.

os.unlockpt(fd,/)

Unlock the slave pseudo-terminal device associated with the master pseudo-terminal device to which the file descriptorfdrefers. The file descriptorfdis not closed upon failure.

Calls the C standard library functionunlockpt().

Availability:Unix, not WASI.

Added in version 3.13.

os.write(fd,str,/)

Write the bytestring instrto file descriptorfd.

Return the number of bytes actually written.

Note

This function is intended for low-level I/O and must be applied to a file descriptor as returned byos.open()orpipe().To write a “file object” returned by the built-in functionopen()or bypopen()or fdopen(),orsys.stdoutorsys.stderr,use its write()method.

Changed in version 3.5:If the system call is interrupted and the signal handler does not raise an exception, the function now retries the system call instead of raising an InterruptedErrorexception (seePEP 475for the rationale).

os.writev(fd,buffers,/)

Write the contents ofbuffersto file descriptorfd.buffersmust be a sequence ofbytes-like objects.Buffers are processed in array order. Entire contents of the first buffer is written before proceeding to the second, and so on.

Returns the total number of bytes actually written.

The operating system may set a limit (sysconf()value 'SC_IOV_MAX') on the number of buffers that can be used.

Availability:Unix.

Added in version 3.3.

Querying the size of a terminal

Added in version 3.3.

os.get_terminal_size(fd=STDOUT_FILENO,/)

Return the size of the terminal window as(columns,lines), tuple of typeterminal_size.

The optional argumentfd(defaultSTDOUT_FILENO,or standard output) specifies which file descriptor should be queried.

If the file descriptor is not connected to a terminal, anOSError is raised.

shutil.get_terminal_size()is the high-level function which should normally be used,os.get_terminal_sizeis the low-level implementation.

Availability:Unix, Windows.

classos.terminal_size

A subclass of tuple, holding(columns,lines)of the terminal window size.

columns

Width of the terminal window in characters.

lines

Height of the terminal window in characters.

Inheritance of File Descriptors

Added in version 3.4.

A file descriptor has an “inheritable” flag which indicates if the file descriptor can be inherited by child processes. Since Python 3.4, file descriptors created by Python are non-inheritable by default.

On UNIX, non-inheritable file descriptors are closed in child processes at the execution of a new program, other file descriptors are inherited.

On Windows, non-inheritable handles and file descriptors are closed in child processes, except for standard streams (file descriptors 0, 1 and 2: stdin, stdout and stderr), which are always inherited. Usingspawn*functions, all inheritable handles and all inheritable file descriptors are inherited. Using thesubprocessmodule, all file descriptors except standard streams are closed, and inheritable handles are only inherited if the close_fdsparameter isFalse.

On WebAssembly platforms, the file descriptor cannot be modified.

os.get_inheritable(fd,/)

Get the “inheritable” flag of the specified file descriptor (a boolean).

os.set_inheritable(fd,inheritable,/)

Set the “inheritable” flag of the specified file descriptor.

os.get_handle_inheritable(handle,/)

Get the “inheritable” flag of the specified handle (a boolean).

Availability:Windows.

os.set_handle_inheritable(handle,inheritable,/)

Set the “inheritable” flag of the specified handle.

Availability:Windows.

Files and Directories

On some Unix platforms, many of these functions support one or more of these features:

• specifying a file descriptor: Normally thepathargument provided to functions in theosmodule must be a string specifying a file path. However, some functions now alternatively accept an open file descriptor for theirpathargument. The function will then operate on the file referred to by the descriptor. (For POSIX systems, Python will call the variant of the function prefixed withf(e.g. callfchdirinstead ofchdir).)

  You can check whether or notpathcan be specified as a file descriptor for a particular function on your platform usingos.supports_fd. If this functionality is unavailable, using it will raise a NotImplementedError.

  If the function also supportsdir_fdorfollow_symlinksarguments, it’s an error to specify one of those when supplyingpathas a file descriptor.

• paths relative to directory descriptors:Ifdir_fdis notNone,it should be a file descriptor referring to a directory, and the path to operate on should be relative; path will then be relative to that directory. If the path is absolute,dir_fdis ignored. (For POSIX systems, Python will call the variant of the function with anatsuffix and possibly prefixed with f(e.g. callfaccessatinstead ofaccess).

  You can check whether or notdir_fdis supported for a particular function on your platform usingos.supports_dir_fd.If it’s unavailable, using it will raise aNotImplementedError.

• not following symlinks:Iffollow_symlinksis False,and the last element of the path to operate on is a symbolic link, the function will operate on the symbolic link itself rather than the file pointed to by the link. (For POSIX systems, Python will call thel... variant of the function.)

  You can check whether or notfollow_symlinksis supported for a particular function on your platform usingos.supports_follow_symlinks. If it’s unavailable, using it will raise aNotImplementedError.

os.access(path,mode,*,dir_fd=None,effective_ids=False,follow_symlinks=True)

Use the real uid/gid to test for access topath.Note that most operations will use the effective uid/gid, therefore this routine can be used in a suid/sgid environment to test if the invoking user has the specified access to path.modeshould beF_OKto test the existence ofpath,or it can be the inclusive OR of one or more ofR_OK,W_OK,and X_OKto test permissions. ReturnTrueif access is allowed, Falseif not. See the Unix man pageaccess(2)for more information.

This function can support specifyingpaths relative to directory descriptorsandnot following symlinks.

Ifeffective_idsisTrue,access()will perform its access checks using the effective uid/gid instead of the real uid/gid. effective_idsmay not be supported on your platform; you can check whether or not it is available usingos.supports_effective_ids.If it is unavailable, using it will raise aNotImplementedError.

Note

Usingaccess()to check if a user is authorized to e.g. open a file before actually doing so usingopen()creates a security hole, because the user might exploit the short time interval between checking and opening the file to manipulate it. It’s preferable to useEAFP techniques. For example:

ifos.access("myfile",os.R_OK):
withopen("myfile")asfp:
returnfp.read()
return"some default data"

is better written as:

try:
fp=open("myfile")
exceptPermissionError:
return"some default data"
else:
withfp:
returnfp.read()

Note

I/O operations may fail even whenaccess()indicates that they would succeed, particularly for operations on network filesystems which may have permissions semantics beyond the usual POSIX permission-bit model.

Changed in version 3.3:Added thedir_fd,effective_ids,andfollow_symlinksparameters.

Changed in version 3.6:Accepts apath-like object.

os.F_OK

os.R_OK

os.W_OK

os.X_OK

Values to pass as themodeparameter ofaccess()to test the existence, readability, writability and executability ofpath, respectively.

os.chdir(path)

Change the current working directory topath.

This function can supportspecifying a file descriptor.The descriptor must refer to an opened directory, not an open file.

This function can raiseOSErrorand subclasses such as FileNotFoundError,PermissionError,andNotADirectoryError.

Raises anauditing eventos.chdirwith argumentpath.

Changed in version 3.3:Added support for specifyingpathas a file descriptor on some platforms.

Changed in version 3.6:Accepts apath-like object.

os.chflags(path,flags,*,follow_symlinks=True)

Set the flags ofpathto the numericflags.flagsmay take a combination (bitwise OR) of the following values (as defined in thestatmodule):

• stat.UF_NODUMP

• stat.UF_IMMUTABLE

• stat.UF_APPEND

• stat.UF_OPAQUE

• stat.UF_NOUNLINK

• stat.UF_COMPRESSED

• stat.UF_HIDDEN

• stat.SF_ARCHIVED

• stat.SF_IMMUTABLE

• stat.SF_APPEND

• stat.SF_NOUNLINK

• stat.SF_SNAPSHOT

This function can supportnot following symlinks.

Raises anauditing eventos.chflagswith argumentspath,flags.

Availability:Unix, not WASI.

Changed in version 3.3:Added thefollow_symlinksparameter.

Changed in version 3.6:Accepts apath-like object.

os.chmod(path,mode,*,dir_fd=None,follow_symlinks=True)

Change the mode ofpathto the numericmode.modemay take one of the following values (as defined in thestatmodule) or bitwise ORed combinations of them:

• stat.S_ISUID

• stat.S_ISGID

• stat.S_ENFMT

• stat.S_ISVTX

• stat.S_IREAD

• stat.S_IWRITE

• stat.S_IEXEC

• stat.S_IRWXU

• stat.S_IRUSR

• stat.S_IWUSR

• stat.S_IXUSR

• stat.S_IRWXG

• stat.S_IRGRP

• stat.S_IWGRP

• stat.S_IXGRP

• stat.S_IRWXO

• stat.S_IROTH

• stat.S_IWOTH

• stat.S_IXOTH

This function can supportspecifying a file descriptor, paths relative to directory descriptorsandnot following symlinks.

Note

Although Windows supportschmod(),you can only set the file’s read-only flag with it (via thestat.S_IWRITEandstat.S_IREAD constants or a corresponding integer value). All other bits are ignored. The default value offollow_symlinksisFalseon Windows.

The function is limited on WASI, seeWebAssembly platformsfor more information.

Raises anauditing eventos.chmodwith argumentspath,mode,dir_fd.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor, and thedir_fdandfollow_symlinksarguments.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.13:Added support for a file descriptor and thefollow_symlinksargument on Windows.

os.chown(path,uid,gid,*,dir_fd=None,follow_symlinks=True)

Change the owner and group id ofpathto the numericuidandgid.To leave one of the ids unchanged, set it to -1.

This function can supportspecifying a file descriptor, paths relative to directory descriptorsandnot following symlinks.

Seeshutil.chown()for a higher-level function that accepts names in addition to numeric ids.

Raises anauditing eventos.chownwith argumentspath,uid,gid,dir_fd.

Availability:Unix.

The function is limited on WASI, seeWebAssembly platformsfor more information.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor, and thedir_fdandfollow_symlinksarguments.

Changed in version 3.6:Supports apath-like object.

os.chroot(path)

Change the root directory of the current process topath.

Availability:Unix, not WASI, not Android.

Changed in version 3.6:Accepts apath-like object.

os.fchdir(fd)

Change the current working directory to the directory represented by the file descriptorfd.The descriptor must refer to an opened directory, not an open file. As of Python 3.3, this is equivalent toos.chdir(fd).

Raises anauditing eventos.chdirwith argumentpath.

Availability:Unix.

os.getcwd()

Return a string representing the current working directory.

os.getcwdb()

Return a bytestring representing the current working directory.

Changed in version 3.8:The function now uses the UTF-8 encoding on Windows, rather than the ANSI code page: seePEP 529for the rationale. The function is no longer deprecated on Windows.

os.lchflags(path,flags)

Set the flags ofpathto the numericflags,likechflags(),but do not follow symbolic links. As of Python 3.3, this is equivalent to os.chflags(path,flags,follow_symlinks=False).

Raises anauditing eventos.chflagswith argumentspath,flags.

Availability:Unix, not WASI.

Changed in version 3.6:Accepts apath-like object.

os.lchmod(path,mode)

Change the mode ofpathto the numericmode.If path is a symlink, this affects the symlink rather than the target. See the docs forchmod() for possible values ofmode.As of Python 3.3, this is equivalent to os.chmod(path,mode,follow_symlinks=False).

lchmod()is not part of POSIX, but Unix implementations may have it if changing the mode of symbolic links is supported.

Raises anauditing eventos.chmodwith argumentspath,mode,dir_fd.

Availability:Unix, Windows, not Linux, FreeBSD >= 1.3, NetBSD >= 1.3, not OpenBSD

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.13:Added support on Windows.

os.lchown(path,uid,gid)

Change the owner and group id ofpathto the numericuidandgid.This function will not follow symbolic links. As of Python 3.3, this is equivalent toos.chown(path,uid,gid,follow_symlinks=False).

Raises anauditing eventos.chownwith argumentspath,uid,gid,dir_fd.

Availability:Unix.

Changed in version 3.6:Accepts apath-like object.

os.link(src,dst,*,src_dir_fd=None,dst_dir_fd=None,follow_symlinks=True)

Create a hard link pointing tosrcnameddst.

This function can support specifyingsrc_dir_fdand/ordst_dir_fdto supplypaths relative to directory descriptors,andnot following symlinks.

Raises anauditing eventos.linkwith argumentssrc,dst,src_dir_fd,dst_dir_fd.

Availability:Unix, Windows.

Changed in version 3.2:Added Windows support.

Changed in version 3.3:Added thesrc_dir_fd,dst_dir_fd,andfollow_symlinksparameters.

Changed in version 3.6:Accepts apath-like objectforsrcanddst.

os.listdir(path='.')

Return a list containing the names of the entries in the directory given by path.The list is in arbitrary order, and does not include the special entries'.'and'..'even if they are present in the directory. If a file is removed from or added to the directory during the call of this function, whether a name for that file be included is unspecified.

pathmay be apath-like object.Ifpathis of typebytes (directly or indirectly through thePathLikeinterface), the filenames returned will also be of typebytes; in all other circumstances, they will be of typestr.

This function can also supportspecifying a file descriptor;the file descriptor must refer to a directory.

Raises anauditing eventos.listdirwith argumentpath.

Note

To encodestrfilenames tobytes,usefsencode().

See also

Thescandir()function returns directory entries along with file attribute information, giving better performance for many common use cases.

Changed in version 3.2:Thepathparameter became optional.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor.

Changed in version 3.6:Accepts apath-like object.

os.listdrives()

Return a list containing the names of drives on a Windows system.

A drive name typically looks like'C:\\'.Not every drive name will be associated with a volume, and some may be inaccessible for a variety of reasons, including permissions, network connectivity or missing media. This function does not test for access.

May raiseOSErrorif an error occurs collecting the drive names.

Raises anauditing eventos.listdriveswith no arguments.

Availability:Windows

Added in version 3.12.

os.listmounts(volume)

Return a list containing the mount points for a volume on a Windows system.

volumemust be represented as a GUID path, like those returned by os.listvolumes().Volumes may be mounted in multiple locations or not at all. In the latter case, the list will be empty. Mount points that are not associated with a volume will not be returned by this function.

The mount points return by this function will be absolute paths, and may be longer than the drive name.

RaisesOSErrorif the volume is not recognized or if an error occurs collecting the paths.

Raises anauditing eventos.listmountswith argumentvolume.

Availability:Windows

Added in version 3.12.

os.listvolumes()

Return a list containing the volumes in the system.

Volumes are typically represented as a GUID path that looks like \\?\Volume{xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx}\.Files can usually be accessed through a GUID path, permissions allowing. However, users are generally not familiar with them, and so the recommended use of this function is to retrieve mount points usingos.listmounts().

May raiseOSErrorif an error occurs collecting the volumes.

Raises anauditing eventos.listvolumeswith no arguments.

Availability:Windows

Added in version 3.12.

os.lstat(path,*,dir_fd=None)

Perform the equivalent of anlstat()system call on the given path. Similar tostat(),but does not follow symbolic links. Return a stat_resultobject.

On platforms that do not support symbolic links, this is an alias for stat().

As of Python 3.3, this is equivalent toos.stat(path,dir_fd=dir_fd, follow_symlinks=False).

This function can also supportpaths relative to directory descriptors.

See also

Thestat()function.

Changed in version 3.2:Added support for Windows 6.0 (Vista) symbolic links.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.8:On Windows, now opens reparse points that represent another path (name surrogates), including symbolic links and directory junctions. Other kinds of reparse points are resolved by the operating system as forstat().

os.mkdir(path,mode=0o777,*,dir_fd=None)

Create a directory namedpathwith numeric modemode.

If the directory already exists,FileExistsErroris raised. If a parent directory in the path does not exist,FileNotFoundErroris raised.

On some systems,modeis ignored. Where it is used, the current umask value is first masked out. If bits other than the last 9 (i.e. the last 3 digits of the octal representation of themode) are set, their meaning is platform-dependent. On some platforms, they are ignored and you should call chmod()explicitly to set them.

On Windows, amodeof0o700is specifically handled to apply access control to the new directory such that only the current user and administrators have access. Other values ofmodeare ignored.

This function can also supportpaths relative to directory descriptors.

It is also possible to create temporary directories; see the tempfilemodule’stempfile.mkdtemp()function.

Raises anauditing eventos.mkdirwith argumentspath,mode,dir_fd.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.13:Windows now handles amodeof0o700.

os.makedirs(name,mode=0o777,exist_ok=False)

Recursive directory creation function. Likemkdir(),but makes all intermediate-level directories needed to contain the leaf directory.

Themodeparameter is passed tomkdir()for creating the leaf directory; seethe mkdir() descriptionfor how it is interpreted. To set the file permission bits of any newly created parent directories you can set the umask before invokingmakedirs().The file permission bits of existing parent directories are not changed.

Ifexist_okisFalse(the default), aFileExistsErroris raised if the target directory already exists.

Note

makedirs()will become confused if the path elements to create includepardir(eg. “..” on UNIX systems).

This function handles UNC paths correctly.

Raises anauditing eventos.mkdirwith argumentspath,mode,dir_fd.

Changed in version 3.2:Added theexist_okparameter.

Changed in version 3.4.1:Before Python 3.4.1, ifexist_okwasTrueand the directory existed, makedirs()would still raise an error ifmodedid not match the mode of the existing directory. Since this behavior was impossible to implement safely, it was removed in Python 3.4.1. Seebpo-21082.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.7:Themodeargument no longer affects the file permission bits of newly created intermediate-level directories.

os.mkfifo(path,mode=0o666,*,dir_fd=None)

Create a FIFO (a named pipe) namedpathwith numeric modemode. The current umask value is first masked out from the mode.

This function can also supportpaths relative to directory descriptors.

FIFOs are pipes that can be accessed like regular files. FIFOs exist until they are deleted (for example withos.unlink()). Generally, FIFOs are used as rendezvous between “client” and “server” type processes: the server opens the FIFO for reading, and the client opens it for writing. Note thatmkfifo() doesn’t open the FIFO — it just creates the rendezvous point.

Availability:Unix, not WASI.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

os.mknod(path,mode=0o600,device=0,*,dir_fd=None)

Create a filesystem node (file, device special file or named pipe) named path.modespecifies both the permissions to use and the type of node to be created, being combined (bitwise OR) with one ofstat.S_IFREG, stat.S_IFCHR,stat.S_IFBLK,andstat.S_IFIFO(those constants are available instat). Forstat.S_IFCHRandstat.S_IFBLK, devicedefines the newly created device special file (probably using os.makedev()), otherwise it is ignored.

This function can also supportpaths relative to directory descriptors.

Availability:Unix, not WASI.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

os.major(device,/)

Extract the device major number from a raw device number (usually the st_devorst_rdevfield fromstat).

os.minor(device,/)

Extract the device minor number from a raw device number (usually the st_devorst_rdevfield fromstat).

os.makedev(major,minor,/)

Compose a raw device number from the major and minor device numbers.

os.pathconf(path,name)

Return system configuration information relevant to a named file.name specifies the configuration value to retrieve; it may be a string which is the name of a defined system value; these names are specified in a number of standards (POSIX.1, Unix 95, Unix 98, and others). Some platforms define additional names as well. The names known to the host operating system are given in thepathconf_namesdictionary. For configuration variables not included in that mapping, passing an integer fornameis also accepted.

Ifnameis a string and is not known,ValueErroris raised. If a specific value fornameis not supported by the host system, even if it is included inpathconf_names,anOSErroris raised with errno.EINVALfor the error number.

This function can supportspecifying a file descriptor.

Availability:Unix.

Changed in version 3.6:Accepts apath-like object.

os.pathconf_names

Dictionary mapping names accepted bypathconf()andfpathconf()to the integer values defined for those names by the host operating system. This can be used to determine the set of names known to the system.

Availability:Unix.

os.readlink(path,*,dir_fd=None)

Return a string representing the path to which the symbolic link points. The result may be either an absolute or relative pathname; if it is relative, it may be converted to an absolute pathname using os.path.join(os.path.dirname(path),result).

If thepathis a string object (directly or indirectly through a PathLikeinterface), the result will also be a string object, and the call may raise a UnicodeDecodeError. If thepathis a bytes object (direct or indirectly), the result will be a bytes object.

This function can also supportpaths relative to directory descriptors.

When trying to resolve a path that may contain links, use realpath()to properly handle recursion and platform differences.

Availability:Unix, Windows.

Changed in version 3.2:Added support for Windows 6.0 (Vista) symbolic links.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like objecton Unix.

Changed in version 3.8:Accepts apath-like objectand a bytes object on Windows.

Added support for directory junctions, and changed to return the substitution path (which typically includes\\?\prefix) rather than the optional “print name” field that was previously returned.

os.remove(path,*,dir_fd=None)

Remove (delete) the filepath.Ifpathis a directory, an OSErroris raised. Usermdir()to remove directories. If the file does not exist, aFileNotFoundErroris raised.

This function can supportpaths relative to directory descriptors.

On Windows, attempting to remove a file that is in use causes an exception to be raised; on Unix, the directory entry is removed but the storage allocated to the file is not made available until the original file is no longer in use.

This function is semantically identical tounlink().

Raises anauditing eventos.removewith argumentspath,dir_fd.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

os.removedirs(name)

Remove directories recursively. Works likermdir()except that, if the leaf directory is successfully removed,removedirs()tries to successively remove every parent directory mentioned inpathuntil an error is raised (which is ignored, because it generally means that a parent directory is not empty). For example,os.removedirs('foo/bar/baz')will first remove the directory'foo/bar/baz',and then remove'foo/bar'and'foo'if they are empty. RaisesOSErrorif the leaf directory could not be successfully removed.

Raises anauditing eventos.removewith argumentspath,dir_fd.

Changed in version 3.6:Accepts apath-like object.

os.rename(src,dst,*,src_dir_fd=None,dst_dir_fd=None)

Rename the file or directorysrctodst.Ifdstexists, the operation will fail with anOSErrorsubclass in a number of cases:

On Windows, ifdstexists aFileExistsErroris always raised. The operation may fail ifsrcanddstare on different filesystems. Use shutil.move()to support moves to a different filesystem.

On Unix, ifsrcis a file anddstis a directory or vice-versa, an IsADirectoryErroror aNotADirectoryErrorwill be raised respectively. If both are directories anddstis empty,dstwill be silently replaced. Ifdstis a non-empty directory, anOSError is raised. If both are files,dstwill be replaced silently if the user has permission. The operation may fail on some Unix flavors ifsrcand dstare on different filesystems. If successful, the renaming will be an atomic operation (this is a POSIX requirement).

This function can support specifyingsrc_dir_fdand/ordst_dir_fdto supplypaths relative to directory descriptors.

If you want cross-platform overwriting of the destination, usereplace().

Raises anauditing eventos.renamewith argumentssrc,dst,src_dir_fd,dst_dir_fd.

Changed in version 3.3:Added thesrc_dir_fdanddst_dir_fdparameters.

Changed in version 3.6:Accepts apath-like objectforsrcanddst.

os.renames(old,new)

Recursive directory or file renaming function. Works likerename(),except creation of any intermediate directories needed to make the new pathname good is attempted first. After the rename, directories corresponding to rightmost path segments of the old name will be pruned away usingremovedirs().

Note

This function can fail with the new directory structure made if you lack permissions needed to remove the leaf directory or file.

Raises anauditing eventos.renamewith argumentssrc,dst,src_dir_fd,dst_dir_fd.

Changed in version 3.6:Accepts apath-like objectforoldandnew.

os.replace(src,dst,*,src_dir_fd=None,dst_dir_fd=None)

Rename the file or directorysrctodst.Ifdstis a non-empty directory, OSErrorwill be raised. Ifdstexists and is a file, it will be replaced silently if the user has permission. The operation may fail ifsrcanddstare on different filesystems. If successful, the renaming will be an atomic operation (this is a POSIX requirement).

This function can support specifyingsrc_dir_fdand/ordst_dir_fdto supplypaths relative to directory descriptors.

Raises anauditing eventos.renamewith argumentssrc,dst,src_dir_fd,dst_dir_fd.

Added in version 3.3.

Changed in version 3.6:Accepts apath-like objectforsrcanddst.

os.rmdir(path,*,dir_fd=None)

Remove (delete) the directorypath.If the directory does not exist or is not empty, aFileNotFoundErroror anOSErroris raised respectively. In order to remove whole directory trees, shutil.rmtree()can be used.

This function can supportpaths relative to directory descriptors.

Raises anauditing eventos.rmdirwith argumentspath,dir_fd.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

os.scandir(path='.')

Return an iterator ofos.DirEntryobjects corresponding to the entries in the directory given bypath.The entries are yielded in arbitrary order, and the special entries'.'and'..'are not included. If a file is removed from or added to the directory after creating the iterator, whether an entry for that file be included is unspecified.

Usingscandir()instead oflistdir()can significantly increase the performance of code that also needs file type or file attribute information, becauseos.DirEntryobjects expose this information if the operating system provides it when scanning a directory. Allos.DirEntrymethods may perform a system call, but is_dir()andis_file()usually only require a system call for symbolic links;os.DirEntry.stat() always requires a system call on Unix but only requires one for symbolic links on Windows.

pathmay be apath-like object.Ifpathis of typebytes (directly or indirectly through thePathLikeinterface), the type of thenameandpath attributes of eachos.DirEntrywill bebytes;in all other circumstances, they will be of typestr.

This function can also supportspecifying a file descriptor;the file descriptor must refer to a directory.

Raises anauditing eventos.scandirwith argumentpath.

Thescandir()iterator supports thecontext managerprotocol and has the following method:

scandir.close()

Close the iterator and free acquired resources.

This is called automatically when the iterator is exhausted or garbage collected, or when an error happens during iterating. However it is advisable to call it explicitly or use thewith statement.

Added in version 3.6.

The following example shows a simple use ofscandir()to display all the files (excluding directories) in the givenpaththat don’t start with '.'.Theentry.is_file()call will generally not make an additional system call:

withos.scandir(path)asit:
forentryinit:
ifnotentry.name.startswith('.')andentry.is_file():
print(entry.name)

Note

On Unix-based systems,scandir()uses the system’s opendir() and readdir() functions. On Windows, it uses the Win32 FindFirstFileW and FindNextFileW functions.

Added in version 3.5.

Changed in version 3.6:Added support for thecontext managerprotocol and the close()method. If ascandir()iterator is neither exhausted nor explicitly closed aResourceWarningwill be emitted in its destructor.

The function accepts apath-like object.

Changed in version 3.7:Added support forfile descriptorson Unix.

classos.DirEntry

Object yielded byscandir()to expose the file path and other file attributes of a directory entry.

scandir()will provide as much of this information as possible without making additional system calls. When astat()orlstat()system call is made, theos.DirEntryobject will cache the result.

os.DirEntryinstances are not intended to be stored in long-lived data structures; if you know the file metadata has changed or if a long time has elapsed since callingscandir(),callos.stat(entry.path)to fetch up-to-date information.

Because theos.DirEntrymethods can make operating system calls, they may also raiseOSError.If you need very fine-grained control over errors, you can catchOSErrorwhen calling one of the os.DirEntrymethods and handle as appropriate.

To be directly usable as apath-like object,os.DirEntry implements thePathLikeinterface.

Attributes and methods on aos.DirEntryinstance are as follows:

name

The entry’s base filename, relative to thescandir()path argument.

Thenameattribute will bebytesif thescandir() pathargument is of typebytesandstrotherwise. Use fsdecode()to decode byte filenames.

path

The entry’s full path name: equivalent toos.path.join(scandir_path, entry.name)wherescandir_pathis thescandir()path argument. The path is only absolute if thescandir()path argument was absolute. If thescandir()path argument was afile descriptor,thepath attribute is the same as thenameattribute.

Thepathattribute will bebytesif thescandir() pathargument is of typebytesandstrotherwise. Use fsdecode()to decode byte filenames.

inode()

Return the inode number of the entry.

The result is cached on theos.DirEntryobject. Use os.stat(entry.path,follow_symlinks=False).st_inoto fetch up-to-date information.

On the first, uncached call, a system call is required on Windows but not on Unix.

is_dir(*,follow_symlinks=True)

ReturnTrueif this entry is a directory or a symbolic link pointing to a directory; returnFalseif the entry is or points to any other kind of file, or if it doesn’t exist anymore.

Iffollow_symlinksisFalse,returnTrueonly if this entry is a directory (without following symlinks); returnFalseif the entry is any other kind of file or if it doesn’t exist anymore.

The result is cached on theos.DirEntryobject, with a separate cache forfollow_symlinksTrueandFalse.Callos.stat()along withstat.S_ISDIR()to fetch up-to-date information.

On the first, uncached call, no system call is required in most cases. Specifically, for non-symlinks, neither Windows or Unix require a system call, except on certain Unix file systems, such as network file systems, that returndirent.d_type==DT_UNKNOWN.If the entry is a symlink, a system call will be required to follow the symlink unless follow_symlinksisFalse.

This method can raiseOSError,such asPermissionError, butFileNotFoundErroris caught and not raised.

is_file(*,follow_symlinks=True)

ReturnTrueif this entry is a file or a symbolic link pointing to a file; returnFalseif the entry is or points to a directory or other non-file entry, or if it doesn’t exist anymore.

Iffollow_symlinksisFalse,returnTrueonly if this entry is a file (without following symlinks); returnFalseif the entry is a directory or other non-file entry, or if it doesn’t exist anymore.

The result is cached on theos.DirEntryobject. Caching, system calls made, and exceptions raised are as peris_dir().

is_symlink()

ReturnTrueif this entry is a symbolic link (even if broken); returnFalseif the entry points to a directory or any kind of file, or if it doesn’t exist anymore.

The result is cached on theos.DirEntryobject. Call os.path.islink()to fetch up-to-date information.

On the first, uncached call, no system call is required in most cases. Specifically, neither Windows or Unix require a system call, except on certain Unix file systems, such as network file systems, that return dirent.d_type==DT_UNKNOWN.

This method can raiseOSError,such asPermissionError, butFileNotFoundErroris caught and not raised.

is_junction()

ReturnTrueif this entry is a junction (even if broken); returnFalseif the entry points to a regular directory, any kind of file, a symlink, or if it doesn’t exist anymore.

The result is cached on theos.DirEntryobject. Call os.path.isjunction()to fetch up-to-date information.

Added in version 3.12.

stat(*,follow_symlinks=True)

Return astat_resultobject for this entry. This method follows symbolic links by default; to stat a symbolic link add the follow_symlinks=Falseargument.

On Unix, this method always requires a system call. On Windows, it only requires a system call iffollow_symlinksisTrueand the entry is a reparse point (for example, a symbolic link or directory junction).

On Windows, thest_ino,st_devandst_nlinkattributes of the stat_resultare always set to zero. Callos.stat()to get these attributes.

The result is cached on theos.DirEntryobject, with a separate cache forfollow_symlinksTrueandFalse.Callos.stat()to fetch up-to-date information.

Note that there is a nice correspondence between several attributes and methods ofos.DirEntryand ofpathlib.Path.In particular, thenameattribute has the same meaning, as do theis_dir(),is_file(),is_symlink(), is_junction(),andstat()methods.

Added in version 3.5.

Changed in version 3.6:Added support for thePathLikeinterface. Added support forbytespaths on Windows.

Changed in version 3.12:Thest_ctimeattribute of a stat result is deprecated on Windows. The file creation time is properly available asst_birthtime,and in the futurest_ctimemay be changed to return zero or the metadata change time, if available.

os.stat(path,*,dir_fd=None,follow_symlinks=True)

Get the status of a file or a file descriptor. Perform the equivalent of a stat()system call on the given path.pathmay be specified as either a string or bytes – directly or indirectly through thePathLike interface – or as an open file descriptor. Return astat_result object.

This function normally follows symlinks; to stat a symlink add the argument follow_symlinks=False,or uselstat().

This function can supportspecifying a file descriptorand not following symlinks.

On Windows, passingfollow_symlinks=Falsewill disable following all name-surrogate reparse points, which includes symlinks and directory junctions. Other types of reparse points that do not resemble links or that the operating system is unable to follow will be opened directly. When following a chain of multiple links, this may result in the original link being returned instead of the non-link that prevented full traversal. To obtain stat results for the final path in this case, use the os.path.realpath()function to resolve the path name as far as possible and calllstat()on the result. This does not apply to dangling symlinks or junction points, which will raise the usual exceptions.

Example:

>>>importos
>>>statinfo=os.stat('somefile.txt')
>>>statinfo
os.stat_result(st_mode=33188, st_ino=7876932, st_dev=234881026,
st_nlink=1, st_uid=501, st_gid=501, st_size=264, st_atime=1297230295,
st_mtime=1297230027, st_ctime=1297230027)
>>>statinfo.st_size
264

See also

fstat()andlstat()functions.

Changed in version 3.3:Added thedir_fdandfollow_symlinksparameters, specifying a file descriptor instead of a path.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.8:On Windows, all reparse points that can be resolved by the operating system are now followed, and passingfollow_symlinks=False disables following all name surrogate reparse points. If the operating system reaches a reparse point that it is not able to follow,statnow returns the information for the original path as if follow_symlinks=Falsehad been specified instead of raising an error.

classos.stat_result

Object whose attributes correspond roughly to the members of the statstructure. It is used for the result ofos.stat(), os.fstat()andos.lstat().

Attributes:

st_mode

File mode: file type and file mode bits (permissions).

st_ino

Platform dependent, but if non-zero, uniquely identifies the file for a given value ofst_dev.Typically:

• the inode number on Unix,

• thefile indexon Windows

st_dev

Identifier of the device on which this file resides.

st_nlink

Number of hard links.

st_uid

User identifier of the file owner.

st_gid

Group identifier of the file owner.

st_size

Size of the file in bytes, if it is a regular file or a symbolic link. The size of a symbolic link is the length of the pathname it contains, without a terminating null byte.

Timestamps:

st_atime

Time of most recent access expressed in seconds.

st_mtime

Time of most recent content modification expressed in seconds.

st_ctime

Time of most recent metadata change expressed in seconds.

Changed in version 3.12:st_ctimeis deprecated on Windows. Usest_birthtimefor the file creation time. In the future,st_ctimewill contain the time of the most recent metadata change, as for other platforms.

st_atime_ns

Time of most recent access expressed in nanoseconds as an integer.

Added in version 3.3.

st_mtime_ns

Time of most recent content modification expressed in nanoseconds as an integer.

Added in version 3.3.

st_ctime_ns

Time of most recent metadata change expressed in nanoseconds as an integer.

Added in version 3.3.

Changed in version 3.12:st_ctime_nsis deprecated on Windows. Usest_birthtime_ns for the file creation time. In the future,st_ctimewill contain the time of the most recent metadata change, as for other platforms.

st_birthtime

Time of file creation expressed in seconds. This attribute is not always available, and may raiseAttributeError.

Changed in version 3.12:st_birthtimeis now available on Windows.

st_birthtime_ns

Time of file creation expressed in nanoseconds as an integer. This attribute is not always available, and may raise AttributeError.

Added in version 3.12.

Note

The exact meaning and resolution of thest_atime, st_mtime,st_ctimeandst_birthtimeattributes depend on the operating system and the file system. For example, on Windows systems using the FAT32 file systems,st_mtimehas 2-second resolution, andst_atimehas only 1-day resolution. See your operating system documentation for details.

Similarly, althoughst_atime_ns,st_mtime_ns, st_ctime_nsandst_birthtime_nsare always expressed in nanoseconds, many systems do not provide nanosecond precision. On systems that do provide nanosecond precision, the floating-point object used to storest_atime,st_mtime,st_ctimeand st_birthtimecannot preserve all of it, and as such will be slightly inexact. If you need the exact timestamps you should always use st_atime_ns,st_mtime_ns,st_ctime_nsand st_birthtime_ns.

On some Unix systems (such as Linux), the following attributes may also be available:

st_blocks

Number of 512-byte blocks allocated for file. This may be smaller thanst_size/512 when the file has holes.

st_blksize

“Preferred” blocksize for efficient file system I/O. Writing to a file in smaller chunks may cause an inefficient read-modify-rewrite.

st_rdev

Type of device if an inode device.

st_flags

User defined flags for file.

On other Unix systems (such as FreeBSD), the following attributes may be available (but may be only filled out if root tries to use them):

st_gen

File generation number.

On Solaris and derivatives, the following attributes may also be available:

st_fstype

String that uniquely identifies the type of the filesystem that contains the file.

On macOS systems, the following attributes may also be available:

st_rsize

Real size of the file.

st_creator

Creator of the file.

st_type

File type.

On Windows systems, the following attributes are also available:

st_file_attributes

Windows file attributes:dwFileAttributesmember of the BY_HANDLE_FILE_INFORMATIONstructure returned by GetFileInformationByHandle(). See theFILE_ATTRIBUTE_*<stat.FILE_ATTRIBUTE_ARCHIVE> constants in thestatmodule.

Added in version 3.5.

st_reparse_tag

Whenst_file_attributeshas theFILE_ATTRIBUTE_REPARSE_POINT set, this field contains the tag identifying the type of reparse point. See theIO_REPARSE_TAG_* constants in thestatmodule.

The standard modulestatdefines functions and constants that are useful for extracting information from astatstructure. (On Windows, some items are filled with dummy values.)

For backward compatibility, astat_resultinstance is also accessible as a tuple of at least 10 integers giving the most important (and portable) members of thestatstructure, in the order st_mode,st_ino,st_dev,st_nlink, st_uid,st_gid,st_size,st_atime, st_mtime,st_ctime.More items may be added at the end by some implementations. For compatibility with older Python versions, accessingstat_resultas a tuple always returns integers.

Changed in version 3.5:Windows now returns the file index asst_inowhen available.

Changed in version 3.7:Added thest_fstypemember to Solaris/derivatives.

Changed in version 3.8:Added thest_reparse_tagmember on Windows.

Changed in version 3.8:On Windows, thest_modemember now identifies special files asS_IFCHR,S_IFIFOorS_IFBLK as appropriate.

Changed in version 3.12:On Windows,st_ctimeis deprecated. Eventually, it will contain the last metadata change time, for consistency with other platforms, but for now still contains creation time. Usest_birthtimefor the creation time.

On Windows,st_inomay now be up to 128 bits, depending on the file system. Previously it would not be above 64 bits, and larger file identifiers would be arbitrarily packed.

On Windows,st_rdevno longer returns a value. Previously it would contain the same asst_dev,which was incorrect.

Added thest_birthtimemember on Windows.

os.statvfs(path)

Perform astatvfs()system call on the given path. The return value is an object whose attributes describe the filesystem on the given path, and correspond to the members of thestatvfsstructure, namely: f_bsize,f_frsize,f_blocks,f_bfree, f_bavail,f_files,f_ffree,f_favail, f_flag,f_namemax,f_fsid.

Two module-level constants are defined for thef_flagattribute’s bit-flags: ifST_RDONLYis set, the filesystem is mounted read-only, and ifST_NOSUIDis set, the semantics of setuid/setgid bits are disabled or not supported.

Additional module-level constants are defined for GNU/glibc based systems. These areST_NODEV(disallow access to device special files), ST_NOEXEC(disallow program execution),ST_SYNCHRONOUS (writes are synced at once),ST_MANDLOCK(allow mandatory locks on an FS), ST_WRITE(write on file/directory/symlink),ST_APPEND (append-only file),ST_IMMUTABLE(immutable file),ST_NOATIME (do not update access times),ST_NODIRATIME(do not update directory access times),ST_RELATIME(update atime relative to mtime/ctime).

This function can supportspecifying a file descriptor.

Availability:Unix.

Changed in version 3.2:TheST_RDONLYandST_NOSUIDconstants were added.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor.

Changed in version 3.4:TheST_NODEV,ST_NOEXEC,ST_SYNCHRONOUS, ST_MANDLOCK,ST_WRITE,ST_APPEND, ST_IMMUTABLE,ST_NOATIME,ST_NODIRATIME, andST_RELATIMEconstants were added.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.7:Added thef_fsidattribute.

os.supports_dir_fd

Asetobject indicating which functions in theos module accept an open file descriptor for theirdir_fdparameter. Different platforms provide different features, and the underlying functionality Python uses to implement thedir_fdparameter is not available on all platforms Python supports. For consistency’s sake, functions that may supportdir_fdalways allow specifying the parameter, but will throw an exception if the functionality is used when it’s not locally available. (SpecifyingNonefordir_fd is always supported on all platforms.)

To check whether a particular function accepts an open file descriptor for itsdir_fdparameter, use theinoperator onsupports_dir_fd. As an example, this expression evaluates toTrueifos.stat() accepts open file descriptors fordir_fdon the local platform:

os.statinos.supports_dir_fd

Currentlydir_fdparameters only work on Unix platforms; none of them work on Windows.

Added in version 3.3.

os.supports_effective_ids

Asetobject indicating whetheros.access()permits specifyingTruefor itseffective_idsparameter on the local platform. (SpecifyingFalseforeffective_idsis always supported on all platforms.) If the local platform supports it, the collection will contain os.access();otherwise it will be empty.

This expression evaluates toTrueifos.access()supports effective_ids=Trueon the local platform:

os.accessinos.supports_effective_ids

Currentlyeffective_idsis only supported on Unix platforms; it does not work on Windows.

Added in version 3.3.

os.supports_fd

Asetobject indicating which functions in the osmodule permit specifying theirpathparameter as an open file descriptor on the local platform. Different platforms provide different features, and the underlying functionality Python uses to accept open file descriptors aspatharguments is not available on all platforms Python supports.

To determine whether a particular function permits specifying an open file descriptor for itspathparameter, use theinoperator on supports_fd.As an example, this expression evaluates toTrueif os.chdir()accepts open file descriptors forpathon your local platform:

os.chdirinos.supports_fd

Added in version 3.3.

os.supports_follow_symlinks

Asetobject indicating which functions in theosmodule acceptFalsefor theirfollow_symlinksparameter on the local platform. Different platforms provide different features, and the underlying functionality Python uses to implementfollow_symlinksis not available on all platforms Python supports. For consistency’s sake, functions that may supportfollow_symlinksalways allow specifying the parameter, but will throw an exception if the functionality is used when it’s not locally available. (SpecifyingTrueforfollow_symlinksis always supported on all platforms.)

To check whether a particular function acceptsFalsefor its follow_symlinksparameter, use theinoperator on supports_follow_symlinks.As an example, this expression evaluates toTrueif you may specifyfollow_symlinks=Falsewhen calling os.stat()on the local platform:

os.statinos.supports_follow_symlinks

Added in version 3.3.

os.symlink(src,dst,target_is_directory=False,*,dir_fd=None)

Create a symbolic link pointing tosrcnameddst.

On Windows, a symlink represents either a file or a directory, and does not morph to the target dynamically. If the target is present, the type of the symlink will be created to match. Otherwise, the symlink will be created as a directory iftarget_is_directoryisTrueor a file symlink (the default) otherwise. On non-Windows platforms,target_is_directoryis ignored.

This function can supportpaths relative to directory descriptors.

Note

On newer versions of Windows 10, unprivileged accounts can create symlinks if Developer Mode is enabled. When Developer Mode is not available/enabled, theSeCreateSymbolicLinkPrivilegeprivilege is required, or the process must be run as an administrator.

OSErroris raised when the function is called by an unprivileged user.

Raises anauditing eventos.symlinkwith argumentssrc,dst,dir_fd.

Availability:Unix, Windows.

The function is limited on WASI, seeWebAssembly platformsfor more information.

Changed in version 3.2:Added support for Windows 6.0 (Vista) symbolic links.

Changed in version 3.3:Added thedir_fdparameter, and now allowtarget_is_directory on non-Windows platforms.

Changed in version 3.6:Accepts apath-like objectforsrcanddst.

Changed in version 3.8:Added support for unelevated symlinks on Windows with Developer Mode.

os.sync()

Force write of everything to disk.

Availability:Unix.

Added in version 3.3.

os.truncate(path,length)

Truncate the file corresponding topath,so that it is at most lengthbytes in size.

This function can supportspecifying a file descriptor.

Raises anauditing eventos.truncatewith argumentspath,length.

Availability:Unix, Windows.

Added in version 3.3.

Changed in version 3.5:Added support for Windows

Changed in version 3.6:Accepts apath-like object.

os.unlink(path,*,dir_fd=None)

Remove (delete) the filepath.This function is semantically identical toremove();theunlinkname is its traditional Unix name. Please see the documentation for remove()for further information.

Raises anauditing eventos.removewith argumentspath,dir_fd.

Changed in version 3.3:Added thedir_fdparameter.

Changed in version 3.6:Accepts apath-like object.

os.utime(path,times=None,*,[ns,]dir_fd=None,follow_symlinks=True)

Set the access and modified times of the file specified bypath.

utime()takes two optional parameters,timesandns. These specify the times set onpathand are used as follows:

• Ifnsis specified, it must be a 2-tuple of the form(atime_ns,mtime_ns) where each member is an int expressing nanoseconds.

• Iftimesis notNone, it must be a 2-tuple of the form(atime,mtime) where each member is an int or float expressing seconds.

• IftimesisNoneandnsis unspecified, this is equivalent to specifyingns=(atime_ns,mtime_ns) where both times are the current time.

It is an error to specify tuples for bothtimesandns.

Note that the exact times you set here may not be returned by a subsequent stat()call, depending on the resolution with which your operating system records access and modification times; seestat().The best way to preserve exact times is to use thest_atime_nsandst_mtime_ns fields from theos.stat()result object with thensparameter to utime().

This function can supportspecifying a file descriptor, paths relative to directory descriptorsandnot following symlinks.

Raises anauditing eventos.utimewith argumentspath,times,ns,dir_fd.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor, and thedir_fd,follow_symlinks,andnsparameters.

Changed in version 3.6:Accepts apath-like object.

os.walk(top,topdown=True,onerror=None,followlinks=False)

Generate the file names in a directory tree by walking the tree either top-down or bottom-up. For each directory in the tree rooted at directory top(includingtopitself), it yields a 3-tuple(dirpath,dirnames, filenames).

dirpathis a string, the path to the directory.dirnamesis a list of the names of the subdirectories indirpath(including symlinks to directories, and excluding'.'and'..'). filenamesis a list of the names of the non-directory files indirpath. Note that the names in the lists contain no path components. To get a full path (which begins withtop) to a file or directory indirpath,do os.path.join(dirpath,name).Whether or not the lists are sorted depends on the file system. If a file is removed from or added to the dirpathdirectory during generating the lists, whether a name for that file be included is unspecified.

If optional argumenttopdownisTrueor not specified, the triple for a directory is generated before the triples for any of its subdirectories (directories are generated top-down). IftopdownisFalse,the triple for a directory is generated after the triples for all of its subdirectories (directories are generated bottom-up). No matter the value oftopdown,the list of subdirectories is retrieved before the tuples for the directory and its subdirectories are generated.

WhentopdownisTrue,the caller can modify thedirnameslist in-place (perhaps usingdelor slice assignment), andwalk()will only recurse into the subdirectories whose names remain indirnames;this can be used to prune the search, impose a specific order of visiting, or even to inform walk()about directories the caller creates or renames before it resumes walk()again. ModifyingdirnameswhentopdownisFalsehas no effect on the behavior of the walk, because in bottom-up mode the directories indirnamesare generated beforedirpathitself is generated.

By default, errors from thescandir()call are ignored. If optional argumentonerroris specified, it should be a function; it will be called with one argument, anOSErrorinstance. It can report the error to continue with the walk, or raise the exception to abort the walk. Note that the filename is available as thefilenameattribute of the exception object.

By default,walk()will not walk down into symbolic links that resolve to directories. SetfollowlinkstoTrueto visit directories pointed to by symlinks, on systems that support them.

Note

Be aware that settingfollowlinkstoTruecan lead to infinite recursion if a link points to a parent directory of itself.walk() does not keep track of the directories it visited already.

Note

If you pass a relative pathname, don’t change the current working directory between resumptions ofwalk().walk()never changes the current directory, and assumes that its caller doesn’t either.

This example displays the number of bytes taken by non-directory files in each directory under the starting directory, except that it doesn’t look under any CVS subdirectory:

importos
fromos.pathimportjoin,getsize
forroot,dirs,filesinos.walk('python/Lib/email'):
print(root,"consumes",end="")
print(sum(getsize(join(root,name))fornameinfiles),end="")
print("bytes in",len(files),"non-directory files")
if'CVS'indirs:
dirs.remove('CVS')# don't visit CVS directories

In the next example (simple implementation ofshutil.rmtree()), walking the tree bottom-up is essential,rmdir()doesn’t allow deleting a directory before the directory is empty:

# Delete everything reachable from the directory named in "top",
# assuming there are no symbolic links.
# CAUTION: This is dangerous! For example, if top == '/', it
# could delete all your disk files.
importos
forroot,dirs,filesinos.walk(top,topdown=False):
fornameinfiles:
os.remove(os.path.join(root,name))
fornameindirs:
os.rmdir(os.path.join(root,name))
os.rmdir(top)

Raises anauditing eventos.walkwith argumentstop,topdown,onerror,followlinks.

Changed in version 3.5:This function now callsos.scandir()instead ofos.listdir(), making it faster by reducing the number of calls toos.stat().

Changed in version 3.6:Accepts apath-like object.

os.fwalk(top='.',topdown=True,onerror=None,*,follow_symlinks=False,dir_fd=None)

This behaves exactly likewalk(),except that it yields a 4-tuple (dirpath,dirnames,filenames,dirfd),and it supportsdir_fd.

dirpath,dirnamesandfilenamesare identical towalk()output, anddirfdis a file descriptor referring to the directorydirpath.

This function always supportspaths relative to directory descriptorsandnot following symlinks.Note however that, unlike other functions, thefwalk()default value for follow_symlinksisFalse.

Note

Sincefwalk()yields file descriptors, those are only valid until the next iteration step, so you should duplicate them (e.g. with dup()) if you want to keep them longer.

This example displays the number of bytes taken by non-directory files in each directory under the starting directory, except that it doesn’t look under any CVS subdirectory:

importos
forroot,dirs,files,rootfdinos.fwalk('python/Lib/email'):
print(root,"consumes",end="")
print(sum([os.stat(name,dir_fd=rootfd).st_sizefornameinfiles]),
end="")
print("bytes in",len(files),"non-directory files")
if'CVS'indirs:
dirs.remove('CVS')# don't visit CVS directories

In the next example, walking the tree bottom-up is essential: rmdir()doesn’t allow deleting a directory before the directory is empty:

# Delete everything reachable from the directory named in "top",
# assuming there are no symbolic links.
# CAUTION: This is dangerous! For example, if top == '/', it
# could delete all your disk files.
importos
forroot,dirs,files,rootfdinos.fwalk(top,topdown=False):
fornameinfiles:
os.unlink(name,dir_fd=rootfd)
fornameindirs:
os.rmdir(name,dir_fd=rootfd)

Raises anauditing eventos.fwalkwith argumentstop,topdown,onerror,follow_symlinks,dir_fd.

Availability:Unix.

Added in version 3.3.

Changed in version 3.6:Accepts apath-like object.

Changed in version 3.7:Added support forbytespaths.

os.memfd_create(name[,flags=os.MFD_CLOEXEC])

Create an anonymous file and return a file descriptor that refers to it. flagsmust be one of theos.MFD_*constants available on the system (or a bitwise ORed combination of them). By default, the new file descriptor isnon-inheritable.

The name supplied innameis used as a filename and will be displayed as the target of the corresponding symbolic link in the directory /proc/self/fd/.The displayed name is always prefixed withmemfd: and serves only for debugging purposes. Names do not affect the behavior of the file descriptor, and as such multiple files can have the same name without any side effects.

Availability:Linux >= 3.17 with glibc >= 2.27.

Added in version 3.8.

os.MFD_CLOEXEC

os.MFD_ALLOW_SEALING

os.MFD_HUGETLB

os.MFD_HUGE_SHIFT

os.MFD_HUGE_MASK

os.MFD_HUGE_64KB

os.MFD_HUGE_512KB

os.MFD_HUGE_1MB

os.MFD_HUGE_2MB

os.MFD_HUGE_8MB

os.MFD_HUGE_16MB

os.MFD_HUGE_32MB

os.MFD_HUGE_256MB

os.MFD_HUGE_512MB

os.MFD_HUGE_1GB

os.MFD_HUGE_2GB

os.MFD_HUGE_16GB

These flags can be passed tomemfd_create().

Availability:Linux >= 3.17 with glibc >= 2.27

TheMFD_HUGE*flags are only available since Linux 4.14.

Added in version 3.8.

os.eventfd(initval[,flags=os.EFD_CLOEXEC])

Create and return an event file descriptor. The file descriptors supports rawread()andwrite()with a buffer size of 8, select(),poll()and similar. See man page eventfd(2)for more information. By default, the new file descriptor isnon-inheritable.

initvalis the initial value of the event counter. The initial value must be a 32 bit unsigned integer. Please note that the initial value is limited to a 32 bit unsigned int although the event counter is an unsigned 64 bit integer with a maximum value of 2⁶⁴-2.

flagscan be constructed fromEFD_CLOEXEC, EFD_NONBLOCK,andEFD_SEMAPHORE.

IfEFD_SEMAPHOREis specified and the event counter is non-zero, eventfd_read()returns 1 and decrements the counter by one.

IfEFD_SEMAPHOREis not specified and the event counter is non-zero,eventfd_read()returns the current event counter value and resets the counter to zero.

If the event counter is zero andEFD_NONBLOCKis not specified,eventfd_read()blocks.

eventfd_write()increments the event counter. Write blocks if the write operation would increment the counter to a value larger than 2⁶⁴-2.

Example:

importos

# semaphore with start value '1'
fd=os.eventfd(1,os.EFD_SEMAPHORE|os.EFC_CLOEXEC)
try:
# acquire semaphore
v=os.eventfd_read(fd)
try:
do_work()
finally:
# release semaphore
os.eventfd_write(fd,v)
finally:
os.close(fd)

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.10.

os.eventfd_read(fd)

Read value from aneventfd()file descriptor and return a 64 bit unsigned int. The function does not verify thatfdis aneventfd().

Availability:Linux >= 2.6.27

Added in version 3.10.

os.eventfd_write(fd,value)

Add value to aneventfd()file descriptor.valuemust be a 64 bit unsigned int. The function does not verify thatfdis aneventfd().

Availability:Linux >= 2.6.27

Added in version 3.10.

os.EFD_CLOEXEC

Set close-on-exec flag for neweventfd()file descriptor.

Availability:Linux >= 2.6.27

Added in version 3.10.

os.EFD_NONBLOCK

SetO_NONBLOCKstatus flag for neweventfd()file descriptor.

Availability:Linux >= 2.6.27

Added in version 3.10.

os.EFD_SEMAPHORE

Provide semaphore-like semantics for reads from aneventfd()file descriptor. On read the internal counter is decremented by one.

Availability:Linux >= 2.6.30

Added in version 3.10.

Timer File Descriptors

Added in version 3.13.

These functions provide support for Linux’stimer file descriptorAPI. Naturally, they are all only available on Linux.

os.timerfd_create(clockid,/,*,flags=0)

Create and return a timer file descriptor (timerfd).

The file descriptor returned bytimerfd_create()supports:

• read()

• select()

• poll()

The file descriptor’sread()method can be called with a buffer size of 8. If the timer has already expired one or more times,read() returns the number of expirations with the host’s endianness, which may be converted to anintbyint.from_bytes(x,byteorder=sys.byteorder).

select()andpoll()can be used to wait until timer expires and the file descriptor is readable.

clockidmust be a validclock ID, as defined in thetimemodule:

• time.CLOCK_REALTIME

• time.CLOCK_MONOTONIC

• time.CLOCK_BOOTTIME(Since Linux 3.15 for timerfd_create)

Ifclockidistime.CLOCK_REALTIME,a settable system-wide real-time clock is used. If system clock is changed, timer setting need to be updated. To cancel timer when system clock is changed, see TFD_TIMER_CANCEL_ON_SET.

Ifclockidistime.CLOCK_MONOTONIC,a non-settable monotonically increasing clock is used. Even if the system clock is changed, the timer setting will not be affected.

Ifclockidistime.CLOCK_BOOTTIME,same astime.CLOCK_MONOTONIC except it includes any time that the system is suspended.

The file descriptor’s behaviour can be modified by specifying aflagsvalue. Any of the following variables may used, combined using bitwise OR (the|operator):

• TFD_NONBLOCK

• TFD_CLOEXEC

IfTFD_NONBLOCKis not set as a flag,read()blocks until the timer expires. If it is set as a flag,read()doesn’t block, but If there hasn’t been an expiration since the last call to read, read()raisesOSErrorwitherrnois set to errno.EAGAIN.

TFD_CLOEXECis always set by Python automatically.

The file descriptor must be closed withos.close()when it is no longer needed, or else the file descriptor will be leaked.

See also

Thetimerfd_create(2)man page.

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.timerfd_settime(fd,/,*,flags=flags,initial=0.0,interval=0.0)

Alter a timer file descriptor’s internal timer. This function operates the same interval timer astimerfd_settime_ns().

fdmust be a valid timer file descriptor.

The timer’s behaviour can be modified by specifying aflagsvalue. Any of the following variables may used, combined using bitwise OR (the|operator):

• TFD_TIMER_ABSTIME

• TFD_TIMER_CANCEL_ON_SET

The timer is disabled by settinginitialto zero (0). Ifinitialis equal to or greater than zero, the timer is enabled. Ifinitialis less than zero, it raises anOSErrorexception witherrnoset toerrno.EINVAL

By default the timer will fire wheninitialseconds have elapsed. (Ifinitialis zero, timer will fire immediately.)

However, if theTFD_TIMER_ABSTIMEflag is set, the timer will fire when the timer’s clock (set byclockidintimerfd_create()) reachesinitialseconds.

The timer’s interval is set by theintervalfloat. Ifintervalis zero, the timer only fires once, on the initial expiration. Ifintervalis greater than zero, the timer fires every timeinterval seconds have elapsed since the previous expiration. Ifintervalis less than zero, it raisesOSErrorwitherrno set toerrno.EINVAL

If theTFD_TIMER_CANCEL_ON_SETflag is set along with TFD_TIMER_ABSTIMEand the clock for this timer is time.CLOCK_REALTIME,the timer is marked as cancelable if the real-time clock is changed discontinuously. Reading the descriptor is aborted with the error ECANCELED.

Linux manages system clock as UTC. A daylight-savings time transition is done by changing time offset only and doesn’t cause discontinuous system clock change.

Discontinuous system clock change will be caused by the following events:

• settimeofday

• clock_settime

• set the system date and time bydatecommand

Return a two-item tuple of (next_expiration,interval) from the previous timer state, before this function executed.

See also

timerfd_create(2),timerfd_settime(2), settimeofday(2),clock_settime(2), anddate(1).

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.timerfd_settime_ns(fd,/,*,flags=0,initial=0,interval=0)

Similar totimerfd_settime(),but use time as nanoseconds. This function operates the same interval timer astimerfd_settime().

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.timerfd_gettime(fd,/)

Return a two-item tuple of floats (next_expiration,interval).

next_expirationdenotes the relative time until next the timer next fires, regardless of if theTFD_TIMER_ABSTIMEflag is set.

intervaldenotes the timer’s interval. If zero, the timer will only fire once, afternext_expirationseconds have elapsed.

See also

timerfd_gettime(2)

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.timerfd_gettime_ns(fd,/)

Similar totimerfd_gettime(),but return time as nanoseconds.

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.TFD_NONBLOCK

A flag for thetimerfd_create()function, which sets theO_NONBLOCKstatus flag for the new timer file descriptor. IfTFD_NONBLOCKis not set as a flag,read()blocks.

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.TFD_CLOEXEC

A flag for thetimerfd_create()function, IfTFD_CLOEXECis set as a flag, set close-on-exec flag for new file descriptor.

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.TFD_TIMER_ABSTIME

A flag for thetimerfd_settime()andtimerfd_settime_ns()functions. If this flag is set,initialis interpreted as an absolute value on the timer’s clock (in UTC seconds or nanoseconds since the Unix Epoch).

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

os.TFD_TIMER_CANCEL_ON_SET

A flag for thetimerfd_settime()andtimerfd_settime_ns() functions along withTFD_TIMER_ABSTIME. The timer is cancelled when the time of the underlying clock changes discontinuously.

Availability:Linux >= 2.6.27 with glibc >= 2.8

Added in version 3.13.

Linux extended attributes

Added in version 3.3.

These functions are all available on Linux only.

os.getxattr(path,attribute,*,follow_symlinks=True)

Return the value of the extended filesystem attributeattributefor path.attributecan be bytes or str (directly or indirectly through the PathLikeinterface). If it is str, it is encoded with the filesystem encoding.

This function can supportspecifying a file descriptorand not following symlinks.

Raises anauditing eventos.getxattrwith argumentspath,attribute.

Changed in version 3.6:Accepts apath-like objectforpathandattribute.

os.listxattr(path=None,*,follow_symlinks=True)

Return a list of the extended filesystem attributes onpath.The attributes in the list are represented as strings decoded with the filesystem encoding. IfpathisNone,listxattr()will examine the current directory.

This function can supportspecifying a file descriptorand not following symlinks.

Raises anauditing eventos.listxattrwith argumentpath.

Changed in version 3.6:Accepts apath-like object.

os.removexattr(path,attribute,*,follow_symlinks=True)

Removes the extended filesystem attributeattributefrompath. attributeshould be bytes or str (directly or indirectly through the PathLikeinterface). If it is a string, it is encoded with thefilesystem encoding and error handler.

This function can supportspecifying a file descriptorand not following symlinks.

Raises anauditing eventos.removexattrwith argumentspath,attribute.

Changed in version 3.6:Accepts apath-like objectforpathandattribute.

os.setxattr(path,attribute,value,flags=0,*,follow_symlinks=True)

Set the extended filesystem attributeattributeonpathtovalue. attributemust be a bytes or str with no embedded NULs (directly or indirectly through thePathLikeinterface). If it is a str, it is encoded with thefilesystem encoding and error handler.flagsmay be XATTR_REPLACEorXATTR_CREATE.IfXATTR_REPLACEis given and the attribute does not exist,ENODATAwill be raised. IfXATTR_CREATEis given and the attribute already exists, the attribute will not be created andEEXISTSwill be raised.

This function can supportspecifying a file descriptorand not following symlinks.

Note

A bug in Linux kernel versions less than 2.6.39 caused the flags argument to be ignored on some filesystems.

Raises anauditing eventos.setxattrwith argumentspath,attribute,value,flags.

Changed in version 3.6:Accepts apath-like objectforpathandattribute.

os.XATTR_SIZE_MAX

The maximum size the value of an extended attribute can be. Currently, this is 64 KiB on Linux.

os.XATTR_CREATE

This is a possible value for the flags argument insetxattr().It indicates the operation must create an attribute.

os.XATTR_REPLACE

This is a possible value for the flags argument insetxattr().It indicates the operation must replace an existing attribute.

Process Management

These functions may be used to create and manage processes.

The variousexec*functions take a list of arguments for the new program loaded into the process. In each case, the first of these arguments is passed to the new program as its own name rather than as an argument a user may have typed on a command line. For the C programmer, this is theargv[0] passed to a program’smain().For example,os.execv('/bin/echo', ['foo','bar'])will only printbaron standard output;foowill seem to be ignored.

os.abort()

Generate aSIGABRTsignal to the current process. On Unix, the default behavior is to produce a core dump; on Windows, the process immediately returns an exit code of3.Be aware that calling this function will not call the Python signal handler registered forSIGABRTwith signal.signal().

os.add_dll_directory(path)

Add a path to the DLL search path.

This search path is used when resolving dependencies for imported extension modules (the module itself is resolved through sys.path), and also byctypes.

Remove the directory by callingclose()on the returned object or using it in awithstatement.

See theMicrosoft documentation for more information about how DLLs are loaded.

Raises anauditing eventos.add_dll_directorywith argumentpath.

Availability:Windows.

Added in version 3.8:Previous versions of CPython would resolve DLLs using the default behavior for the current process. This led to inconsistencies, such as only sometimes searchingPATHor the current working directory, and OS functions such asAddDllDirectory having no effect.

In 3.8, the two primary ways DLLs are loaded now explicitly override the process-wide behavior to ensure consistency. See the porting notesfor information on updating libraries.

os.execl(path,arg0,arg1,...)

os.execle(path,arg0,arg1,...,env)

os.execlp(file,arg0,arg1,...)

os.execlpe(file,arg0,arg1,...,env)

os.execv(path,args)

os.execve(path,args,env)

os.execvp(file,args)

os.execvpe(file,args,env)

These functions all execute a new program, replacing the current process; they do not return. On Unix, the new executable is loaded into the current process, and will have the same process id as the caller. Errors will be reported as OSErrorexceptions.

The current process is replaced immediately. Open file objects and descriptors are not flushed, so if there may be data buffered on these open files, you should flush them using sys.stdout.flush()oros.fsync()before calling an exec*function.

The “l” and “v” variants of theexec*functions differ in how command-line arguments are passed. The “l” variants are perhaps the easiest to work with if the number of parameters is fixed when the code is written; the individual parameters simply become additional parameters to theexecl*() functions. The “v” variants are good when the number of parameters is variable, with the arguments being passed in a list or tuple as theargs parameter. In either case, the arguments to the child process should start with the name of the command being run, but this is not enforced.

The variants which include a “p” near the end (execlp(), execlpe(),execvp(),andexecvpe()) will use the PATHenvironment variable to locate the programfile.When the environment is being replaced (using one of theexec*evariants, discussed in the next paragraph), the new environment is used as the source of thePATHvariable. The other variants,execl(),execle(), execv(),andexecve(),will not use thePATHvariable to locate the executable;pathmust contain an appropriate absolute or relative path. Relative paths must include at least one slash, even on Windows, as plain names will not be resolved.

Forexecle(),execlpe(),execve(),andexecvpe()(note that these all end in “e” ), theenvparameter must be a mapping which is used to define the environment variables for the new process (these are used instead of the current process’ environment); the functionsexecl(), execlp(),execv(),andexecvp()all cause the new process to inherit the environment of the current process.

Forexecve()on some platforms,pathmay also be specified as an open file descriptor. This functionality may not be supported on your platform; you can check whether or not it is available usingos.supports_fd. If it is unavailable, using it will raise aNotImplementedError.

Raises anauditing eventos.execwith argumentspath,args,env.

Availability:Unix, Windows, not WASI, not Android, not iOS.

Changed in version 3.3:Added support for specifyingpathas an open file descriptor forexecve().

Changed in version 3.6:Accepts apath-like object.

os._exit(n)

Exit the process with statusn,without calling cleanup handlers, flushing stdio buffers, etc.

Note

The standard way to exit issys.exit(n)._exit()should normally only be used in the child process after afork().

The following exit codes are defined and can be used with_exit(), although they are not required. These are typically used for system programs written in Python, such as a mail server’s external command delivery program.

Note

Some of these may not be available on all Unix platforms, since there is some variation. These constants are defined where they are defined by the underlying platform.

os.EX_OK

Exit code that means no error occurred. May be taken from the defined value of EXIT_SUCCESSon some platforms. Generally has a value of zero.

Availability:Unix, Windows.

os.EX_USAGE

Exit code that means the command was used incorrectly, such as when the wrong number of arguments are given.

Availability:Unix, not WASI.

os.EX_DATAERR

Exit code that means the input data was incorrect.

Availability:Unix, not WASI.

os.EX_NOINPUT

Exit code that means an input file did not exist or was not readable.

Availability:Unix, not WASI.

os.EX_NOUSER

Exit code that means a specified user did not exist.

Availability:Unix, not WASI.

os.EX_NOHOST

Exit code that means a specified host did not exist.

Availability:Unix, not WASI.

os.EX_UNAVAILABLE

Exit code that means that a required service is unavailable.

Availability:Unix, not WASI.

os.EX_SOFTWARE

Exit code that means an internal software error was detected.

Availability:Unix, not WASI.

os.EX_OSERR

Exit code that means an operating system error was detected, such as the inability to fork or create a pipe.

Availability:Unix, not WASI.

os.EX_OSFILE

Exit code that means some system file did not exist, could not be opened, or had some other kind of error.

Availability:Unix, not WASI.

os.EX_CANTCREAT

Exit code that means a user specified output file could not be created.

Availability:Unix, not WASI.

os.EX_IOERR

Exit code that means that an error occurred while doing I/O on some file.

Availability:Unix, not WASI.

os.EX_TEMPFAIL

Exit code that means a temporary failure occurred. This indicates something that may not really be an error, such as a network connection that couldn’t be made during a retryable operation.

Availability:Unix, not WASI.

os.EX_PROTOCOL

Exit code that means that a protocol exchange was illegal, invalid, or not understood.

Availability:Unix, not WASI.

os.EX_NOPERM

Exit code that means that there were insufficient permissions to perform the operation (but not intended for file system problems).

Availability:Unix, not WASI.

os.EX_CONFIG

Exit code that means that some kind of configuration error occurred.

Availability:Unix, not WASI.

os.EX_NOTFOUND

Exit code that means something like “an entry was not found”.

Availability:Unix, not WASI.

os.fork()

Fork a child process. Return0in the child and the child’s process id in the parent. If an error occursOSErroris raised.

Note that some platforms including FreeBSD <= 6.3 and Cygwin have known issues when usingfork()from a thread.

Raises anauditing eventos.forkwith no arguments.

Warning

If you use TLS sockets in an application callingfork(),see the warning in thessldocumentation.

Warning

On macOS the use of this function is unsafe when mixed with using higher-level system APIs, and that includes usingurllib.request.

Changed in version 3.8:Callingfork()in a subinterpreter is no longer supported (RuntimeErroris raised).

Changed in version 3.12:If Python is able to detect that your process has multiple threads,os.fork()now raises aDeprecationWarning.

We chose to surface this as a warning, when detectable, to better inform developers of a design problem that the POSIX platform specifically notes as not supported. Even in code that appearsto work, it has never been safe to mix threading with os.fork()on POSIX platforms. The CPython runtime itself has always made API calls that are not safe for use in the child process when threads existed in the parent (such asmallocand free).

Users of macOS or users of libc or malloc implementations other than those typically found in glibc to date are among those already more likely to experience deadlocks running such code.

Seethis discussion on fork being incompatible with threads for technical details of why we’re surfacing this longstanding platform compatibility problem to developers.

Availability:POSIX, not WASI, not Android, not iOS.

os.forkpty()

Fork a child process, using a new pseudo-terminal as the child’s controlling terminal. Return a pair of(pid,fd),wherepidis0in the child, the new child’s process id in the parent, andfdis the file descriptor of the master end of the pseudo-terminal. For a more portable approach, use the ptymodule. If an error occursOSErroris raised.

Raises anauditing eventos.forkptywith no arguments.

Warning

On macOS the use of this function is unsafe when mixed with using higher-level system APIs, and that includes usingurllib.request.

Changed in version 3.8:Callingforkpty()in a subinterpreter is no longer supported (RuntimeErroris raised).

Changed in version 3.12:If Python is able to detect that your process has multiple threads, this now raises aDeprecationWarning.See the longer explanation onos.fork().

Availability:Unix, not WASI, not Android, not iOS.

os.kill(pid,sig,/)

Send signalsigto the processpid.Constants for the specific signals available on the host platform are defined in thesignalmodule.

Windows: Thesignal.CTRL_C_EVENTand signal.CTRL_BREAK_EVENTsignals are special signals which can only be sent to console processes which share a common console window, e.g., some subprocesses. Any other value forsigwill cause the process to be unconditionally killed by the TerminateProcess API, and the exit code will be set tosig.The Windows version ofkill()additionally takes process handles to be killed.

See alsosignal.pthread_kill().

Raises anauditing eventos.killwith argumentspid,sig.

Availability:Unix, Windows, not WASI, not iOS.

Changed in version 3.2:Added Windows support.

os.killpg(pgid,sig,/)

Send the signalsigto the process grouppgid.

Raises anauditing eventos.killpgwith argumentspgid,sig.

Availability:Unix, not WASI, not iOS.

os.nice(increment,/)

Addincrementto the process’s “niceness”. Return the new niceness.

Availability:Unix, not WASI.

os.pidfd_open(pid,flags=0)

Return a file descriptor referring to the processpidwithflagsset. This descriptor can be used to perform process management without races and signals.

See thepidfd_open(2)man page for more details.

Availability:Linux >= 5.3, Android >=build-timeAPI level 31

Added in version 3.9.

os.PIDFD_NONBLOCK

This flag indicates that the file descriptor will be non-blocking. If the process referred to by the file descriptor has not yet terminated, then an attempt to wait on the file descriptor usingwaitid(2) will immediately return the errorEAGAINrather than blocking.

Availability:Linux >= 5.10

Added in version 3.12.

os.plock(op,/)

Lock program segments into memory. The value ofop(defined in <sys/lock.h>) determines which segments are locked.

Availability:Unix, not WASI, not iOS.

os.popen(cmd,mode='r',buffering=-1)

Open a pipe to or from commandcmd. The return value is an open file object connected to the pipe, which can be read or written depending on whethermode is'r'(default) or'w'. Thebufferingargument have the same meaning as the corresponding argument to the built-inopen()function. The returned file object reads or writes text strings rather than bytes.

Theclosemethod returnsNoneif the subprocess exited successfully, or the subprocess’s return code if there was an error. On POSIX systems, if the return code is positive it represents the return value of the process left-shifted by one byte. If the return code is negative, the process was terminated by the signal given by the negated value of the return code. (For example, the return value might be-signal.SIGKILLif the subprocess was killed.) On Windows systems, the return value contains the signed integer return code from the child process.

On Unix,waitstatus_to_exitcode()can be used to convert theclose method result (exit status) into an exit code if it is notNone.On Windows, theclosemethod result is directly the exit code (orNone).

This is implemented usingsubprocess.Popen;see that class’s documentation for more powerful ways to manage and communicate with subprocesses.

Availability:not WASI, not Android, not iOS.

Note

ThePython UTF-8 Modeaffects encodings used forcmdand pipe contents.

popen()is a simple wrapper aroundsubprocess.Popen. Usesubprocess.Popenorsubprocess.run()to control options like encodings.

os.posix_spawn(path,argv,env,*,file_actions=None,setpgroup=None,resetids=False,setsid=False,setsigmask=(),setsigdef=(),scheduler=None)

Wraps theposix_spawn()C library API for use from Python.

Most users should usesubprocess.run()instead ofposix_spawn().

The positional-only argumentspath,args,andenvare similar to execve().envis allowed to beNone,in which case current process’ environment is used.

Thepathparameter is the path to the executable file. Thepathshould contain a directory. Useposix_spawnp()to pass an executable file without directory.

Thefile_actionsargument may be a sequence of tuples describing actions to take on specific file descriptors in the child process between the C library implementation’sfork()andexec()steps. The first item in each tuple must be one of the three type indicator listed below describing the remaining tuple elements:

os.POSIX_SPAWN_OPEN

(os.POSIX_SPAWN_OPEN,fd,path,flags,mode)

Performsos.dup2(os.open(path,flags,mode),fd).

os.POSIX_SPAWN_CLOSE

(os.POSIX_SPAWN_CLOSE,fd)

Performsos.close(fd).

os.POSIX_SPAWN_DUP2

(os.POSIX_SPAWN_DUP2,fd,new_fd)

Performsos.dup2(fd,new_fd).

os.POSIX_SPAWN_CLOSEFROM

(os.POSIX_SPAWN_CLOSEFROM,fd)

Performsos.closerange(fd,INF).

These tuples correspond to the C library posix_spawn_file_actions_addopen(), posix_spawn_file_actions_addclose(), posix_spawn_file_actions_adddup2(),and posix_spawn_file_actions_addclosefrom_np()API calls used to prepare for theposix_spawn()call itself.

Thesetpgroupargument will set the process group of the child to the value specified. If the value specified is 0, the child’s process group ID will be made the same as its process ID. If the value ofsetpgroupis not set, the child will inherit the parent’s process group ID. This argument corresponds to the C libraryPOSIX_SPAWN_SETPGROUPflag.

If theresetidsargument isTrueit will reset the effective UID and GID of the child to the real UID and GID of the parent process. If the argument isFalse,then the child retains the effective UID and GID of the parent. In either case, if the set-user-ID and set-group-ID permission bits are enabled on the executable file, their effect will override the setting of the effective UID and GID. This argument corresponds to the C libraryPOSIX_SPAWN_RESETIDSflag.

If thesetsidargument isTrue,it will create a new session ID forposix_spawn.setsidrequiresPOSIX_SPAWN_SETSID orPOSIX_SPAWN_SETSID_NPflag. Otherwise,NotImplementedError is raised.

Thesetsigmaskargument will set the signal mask to the signal set specified. If the parameter is not used, then the child inherits the parent’s signal mask. This argument corresponds to the C library POSIX_SPAWN_SETSIGMASKflag.

Thesigdefargument will reset the disposition of all signals in the set specified. This argument corresponds to the C library POSIX_SPAWN_SETSIGDEFflag.

Theschedulerargument must be a tuple containing the (optional) scheduler policy and an instance ofsched_paramwith the scheduler parameters. A value ofNonein the place of the scheduler policy indicates that is not being provided. This argument is a combination of the C library POSIX_SPAWN_SETSCHEDPARAMandPOSIX_SPAWN_SETSCHEDULER flags.

Raises anauditing eventos.posix_spawnwith argumentspath,argv,env.

Added in version 3.8.

Changed in version 3.13:envparameter acceptsNone. os.POSIX_SPAWN_CLOSEFROMis available on platforms where posix_spawn_file_actions_addclosefrom_np()exists.

Availability:Unix, not WASI, not Android, not iOS.

os.posix_spawnp(path,argv,env,*,file_actions=None,setpgroup=None,resetids=False,setsid=False,setsigmask=(),setsigdef=(),scheduler=None)

Wraps theposix_spawnp()C library API for use from Python.

Similar toposix_spawn()except that the system searches for theexecutablefile in the list of directories specified by the PATHenvironment variable (in the same way as forexecvp(3)).

Raises anauditing eventos.posix_spawnwith argumentspath,argv,env.

Added in version 3.8.

Availability:POSIX, not WASI, not Android, not iOS.

Seeposix_spawn()documentation.

os.register_at_fork(*,before=None,after_in_parent=None,after_in_child=None)

Register callables to be executed when a new child process is forked usingos.fork()or similar process cloning APIs. The parameters are optional and keyword-only. Each specifies a different call point.

• beforeis a function called before forking a child process.

• after_in_parentis a function called from the parent process after forking a child process.

• after_in_childis a function called from the child process.

These calls are only made if control is expected to return to the Python interpreter. A typicalsubprocesslaunch will not trigger them as the child is not going to re-enter the interpreter.

Functions registered for execution before forking are called in reverse registration order. Functions registered for execution after forking (either in the parent or in the child) are called in registration order.

Note thatfork()calls made by third-party C code may not call those functions, unless it explicitly callsPyOS_BeforeFork(), PyOS_AfterFork_Parent()andPyOS_AfterFork_Child().

There is no way to unregister a function.

Availability:Unix, not WASI, not Android, not iOS.

Added in version 3.7.

os.spawnl(mode,path,...)

os.spawnle(mode,path,...,env)

os.spawnlp(mode,file,...)

os.spawnlpe(mode,file,...,env)

os.spawnv(mode,path,args)

os.spawnve(mode,path,args,env)

os.spawnvp(mode,file,args)

os.spawnvpe(mode,file,args,env)

Execute the programpathin a new process.

(Note that thesubprocessmodule provides more powerful facilities for spawning new processes and retrieving their results; using that module is preferable to using these functions. Check especially the Replacing Older Functions with the subprocess Modulesection.)

IfmodeisP_NOWAIT,this function returns the process id of the new process; ifmodeisP_WAIT,returns the process’s exit code if it exits normally, or-signal,wheresignalis the signal that killed the process. On Windows, the process id will actually be the process handle, so can be used with thewaitpid()function.

Note on VxWorks, this function doesn’t return-signalwhen the new process is killed. Instead it raises OSError exception.

The “l” and “v” variants of thespawn*functions differ in how command-line arguments are passed. The “l” variants are perhaps the easiest to work with if the number of parameters is fixed when the code is written; the individual parameters simply become additional parameters to the spawnl*()functions. The “v” variants are good when the number of parameters is variable, with the arguments being passed in a list or tuple as theargsparameter. In either case, the arguments to the child process must start with the name of the command being run.

The variants which include a second “p” near the end (spawnlp(), spawnlpe(),spawnvp(),andspawnvpe()) will use the PATHenvironment variable to locate the programfile.When the environment is being replaced (using one of thespawn*evariants, discussed in the next paragraph), the new environment is used as the source of thePATHvariable. The other variants,spawnl(), spawnle(),spawnv(),andspawnve(),will not use the PATHvariable to locate the executable;pathmust contain an appropriate absolute or relative path.

Forspawnle(),spawnlpe(),spawnve(),andspawnvpe() (note that these all end in “e” ), theenvparameter must be a mapping which is used to define the environment variables for the new process (they are used instead of the current process’ environment); the functions spawnl(),spawnlp(),spawnv(),andspawnvp()all cause the new process to inherit the environment of the current process. Note that keys and values in theenvdictionary must be strings; invalid keys or values will cause the function to fail, with a return value of127.

As an example, the following calls tospawnlp()andspawnvpe()are equivalent:

importos
os.spawnlp(os.P_WAIT,'cp','cp','index.html','/dev/null')

L=['cp','index.html','/dev/null']
os.spawnvpe(os.P_WAIT,'cp',L,os.environ)

Raises anauditing eventos.spawnwith argumentsmode,path,args,env.

Availability:Unix, Windows, not WASI, not Android, not iOS.

spawnlp(),spawnlpe(),spawnvp() andspawnvpe()are not available on Windows.spawnle()and spawnve()are not thread-safe on Windows; we advise you to use the subprocessmodule instead.

Changed in version 3.6:Accepts apath-like object.

os.P_NOWAIT

os.P_NOWAITO

Possible values for themodeparameter to thespawn*family of functions. If either of these values is given, thespawn*functions will return as soon as the new process has been created, with the process id as the return value.

Availability:Unix, Windows.

os.P_WAIT

Possible value for themodeparameter to thespawn*family of functions. If this is given asmode,thespawn*functions will not return until the new process has run to completion and will return the exit code of the process the run is successful, or-signalif a signal kills the process.

Availability:Unix, Windows.

os.P_DETACH

os.P_OVERLAY

Possible values for themodeparameter to thespawn*family of functions. These are less portable than those listed above.P_DETACH is similar toP_NOWAIT,but the new process is detached from the console of the calling process. IfP_OVERLAYis used, the current process will be replaced; thespawn*function will not return.

Availability:Windows.

os.startfile(path[,operation][,arguments][,cwd][,show_cmd])

Start a file with its associated application.

Whenoperationis not specified, this acts like double-clicking the file in Windows Explorer, or giving the file name as an argument to the startcommand from the interactive command shell: the file is opened with whatever application (if any) its extension is associated.

When anotheroperationis given, it must be a “command verb” that specifies what should be done with the file. Common verbs documented by Microsoft are'open', 'print'and'edit'(to be used on files) as well as'explore'and 'find'(to be used on directories).

When launching an application, specifyargumentsto be passed as a single string. This argument may have no effect when using this function to launch a document.

The default working directory is inherited, but may be overridden by thecwd argument. This should be an absolute path. A relativepathwill be resolved against this argument.

Useshow_cmdto override the default window style. Whether this has any effect will depend on the application being launched. Values are integers as supported by the Win32ShellExecute()function.

startfile()returns as soon as the associated application is launched. There is no option to wait for the application to close, and no way to retrieve the application’s exit status. Thepathparameter is relative to the current directory orcwd.If you want to use an absolute path, make sure the first character is not a slash ('/') Usepathlibor the os.path.normpath()function to ensure that paths are properly encoded for Win32.

To reduce interpreter startup overhead, the Win32ShellExecute() function is not resolved until this function is first called. If the function cannot be resolved,NotImplementedErrorwill be raised.

Raises anauditing eventos.startfilewith argumentspath,operation.

Raises anauditing eventos.startfile/2with argumentspath,operation,arguments,cwd,show_cmd.

Availability:Windows.

Changed in version 3.10:Added thearguments,cwdandshow_cmdarguments, and the os.startfile/2audit event.

os.system(command)

Execute the command (a string) in a subshell. This is implemented by calling the Standard C functionsystem(),and has the same limitations. Changes tosys.stdin,etc. are not reflected in the environment of the executed command. Ifcommandgenerates any output, it will be sent to the interpreter standard output stream. The C standard does not specify the meaning of the return value of the C function, so the return value of the Python function is system-dependent.

On Unix, the return value is the exit status of the process encoded in the format specified forwait().

On Windows, the return value is that returned by the system shell after runningcommand.The shell is given by the Windows environment variable COMSPEC:it is usuallycmd.exe,which returns the exit status of the command run; on systems using a non-native shell, consult your shell documentation.

Thesubprocessmodule provides more powerful facilities for spawning new processes and retrieving their results; using that module is preferable to using this function. See theReplacing Older Functions with the subprocess Modulesection in thesubprocessdocumentation for some helpful recipes.

On Unix,waitstatus_to_exitcode()can be used to convert the result (exit status) into an exit code. On Windows, the result is directly the exit code.

Raises anauditing eventos.systemwith argumentcommand.

Availability:Unix, Windows, not WASI, not Android, not iOS.

os.times()

Returns the current global process times. The return value is an object with five attributes:

• user- user time

• system- system time

• children_user- user time of all child processes

• children_system- system time of all child processes

• elapsed- elapsed real time since a fixed point in the past

For backwards compatibility, this object also behaves like a five-tuple containinguser,system,children_user, children_system,andelapsedin that order.

See the Unix manual page times(2)andtimes(3)manual page on Unix orthe GetProcessTimes MSDN on Windows. On Windows, onlyuserandsystemare known; the other attributes are zero.

Availability:Unix, Windows.

Changed in version 3.3:Return type changed from a tuple to a tuple-like object with named attributes.

os.wait()

Wait for completion of a child process, and return a tuple containing its pid and exit status indication: a 16-bit number, whose low byte is the signal number that killed the process, and whose high byte is the exit status (if the signal number is zero); the high bit of the low byte is set if a core file was produced.

If there are no children that could be waited for,ChildProcessError is raised.

waitstatus_to_exitcode()can be used to convert the exit status into an exit code.

Availability:Unix, not WASI, not Android, not iOS.

See also

The otherwait*()functions documented below can be used to wait for the completion of a specific child process and have more options. waitpid()is the only one also available on Windows.

os.waitid(idtype,id,options,/)

Wait for the completion of a child process.

idtypecan beP_PID,P_PGID,P_ALL,or (on Linux)P_PIDFD. The interpretation ofiddepends on it; see their individual descriptions.

optionsis an OR combination of flags. At least one ofWEXITED, WSTOPPEDorWCONTINUEDis required; WNOHANGandWNOWAITare additional optional flags.

The return value is an object representing the data contained in the siginfo_tstructure with the following attributes:

• si_pid(process ID)

• si_uid(real user ID of the child)

• si_signo(alwaysSIGCHLD)

• si_status(the exit status or signal number, depending onsi_code)

• si_code(seeCLD_EXITEDfor possible values)

IfWNOHANGis specified and there are no matching children in the requested state,Noneis returned. Otherwise, if there are no matching children that could be waited for,ChildProcessErroris raised.

Availability:Unix, not WASI, not Android, not iOS.

Added in version 3.3.

Changed in version 3.13:This function is now available on macOS as well.

os.waitpid(pid,options,/)

The details of this function differ on Unix and Windows.

On Unix: Wait for completion of a child process given by process idpid,and return a tuple containing its process id and exit status indication (encoded as forwait()). The semantics of the call are affected by the value of the integeroptions,which should be0for normal operation.

Ifpidis greater than0,waitpid()requests status information for that specific process. Ifpidis0,the request is for the status of any child in the process group of the current process. Ifpidis-1,the request pertains to any child of the current process. Ifpidis less than -1,status is requested for any process in the process group-pid(the absolute value ofpid).

optionsis an OR combination of flags. If it containsWNOHANGand there are no matching children in the requested state,(0,0)is returned. Otherwise, if there are no matching children that could be waited for,ChildProcessErroris raised. Other options that can be used are WUNTRACEDandWCONTINUED.

On Windows: Wait for completion of a process given by process handlepid,and return a tuple containingpid,and its exit status shifted left by 8 bits (shifting makes cross-platform use of the function easier). Apidless than or equal to0has no special meaning on Windows, and raises an exception. The value of integeroptionshas no effect.pidcan refer to any process whose id is known, not necessarily a child process. Thespawn* functions called withP_NOWAITreturn suitable process handles.

waitstatus_to_exitcode()can be used to convert the exit status into an exit code.

Availability:Unix, Windows, not WASI, not Android, not iOS.

Changed in version 3.5:If the system call is interrupted and the signal handler does not raise an exception, the function now retries the system call instead of raising an InterruptedErrorexception (seePEP 475for the rationale).

os.wait3(options)

Similar towaitpid(),except no process id argument is given and a 3-element tuple containing the child’s process id, exit status indication, and resource usage information is returned. Refer to resource.getrusage()for details on resource usage information. The optionsargument is the same as that provided towaitpid()and wait4().

waitstatus_to_exitcode()can be used to convert the exit status into an exitcode.

Availability:Unix, not WASI, not Android, not iOS.

os.wait4(pid,options)

Similar towaitpid(),except a 3-element tuple, containing the child’s process id, exit status indication, and resource usage information is returned. Refer toresource.getrusage()for details on resource usage information. The arguments towait4()are the same as those provided towaitpid().

waitstatus_to_exitcode()can be used to convert the exit status into an exitcode.

Availability:Unix, not WASI, not Android, not iOS.

os.P_PID

os.P_PGID

os.P_ALL

os.P_PIDFD

These are the possible values foridtypeinwaitid().They affect howidis interpreted:

• P_PID- wait for the child whose PID isid.

• P_PGID- wait for any child whose progress group ID isid.

• P_ALL- wait for any child;idis ignored.

• P_PIDFD- wait for the child identified by the file descriptor id(a process file descriptor created withpidfd_open()).

Availability:Unix, not WASI, not Android, not iOS.

Note

P_PIDFDis only available on Linux >= 5.4.

Added in version 3.3.

Added in version 3.9:TheP_PIDFDconstant.

os.WCONTINUED

Thisoptionsflag forwaitpid(),wait3(),wait4(),and waitid()causes child processes to be reported if they have been continued from a job control stop since they were last reported.

Availability:Unix, not WASI, not Android, not iOS.

os.WEXITED

Thisoptionsflag forwaitid()causes child processes that have terminated to be reported.

The otherwait*functions always report children that have terminated, so this option is not available for them.

Availability:Unix, not WASI, not Android, not iOS.

Added in version 3.3.

os.WSTOPPED

Thisoptionsflag forwaitid()causes child processes that have been stopped by the delivery of a signal to be reported.

This option is not available for the otherwait*functions.

Availability:Unix, not WASI, not Android, not iOS.

Added in version 3.3.

os.WUNTRACED

Thisoptionsflag forwaitpid(),wait3(),andwait4()causes child processes to also be reported if they have been stopped but their current state has not been reported since they were stopped.

This option is not available forwaitid().

Availability:Unix, not WASI, not Android, not iOS.

os.WNOHANG

Thisoptionsflag causeswaitpid(),wait3(),wait4(),and waitid()to return right away if no child process status is available immediately.

Availability:Unix, not WASI, not Android, not iOS.

os.WNOWAIT

Thisoptionsflag causeswaitid()to leave the child in a waitable state, so that a laterwait*()call can be used to retrieve the child status information again.

This option is not available for the otherwait*functions.

Availability:Unix, not WASI, not Android, not iOS.

os.CLD_EXITED

os.CLD_KILLED

os.CLD_DUMPED

os.CLD_TRAPPED

os.CLD_STOPPED

os.CLD_CONTINUED

These are the possible values forsi_codein the result returned by waitid().

Availability:Unix, not WASI, not Android, not iOS.

Added in version 3.3.

Changed in version 3.9:AddedCLD_KILLEDandCLD_STOPPEDvalues.

os.waitstatus_to_exitcode(status)

Convert a wait status to an exit code.

On Unix:

• If the process exited normally (ifWIFEXITED(status)is true), return the process exit status (returnWEXITSTATUS(status)): result greater than or equal to 0.

• If the process was terminated by a signal (ifWIFSIGNALED(status)is true), return-signumwheresignumis the number of the signal that caused the process to terminate (return-WTERMSIG(status)): result less than 0.

• Otherwise, raise aValueError.

On Windows, returnstatusshifted right by 8 bits.

On Unix, if the process is being traced or ifwaitpid()was called withWUNTRACEDoption, the caller must first check if WIFSTOPPED(status)is true. This function must not be called if WIFSTOPPED(status)is true.

See also

WIFEXITED(),WEXITSTATUS(),WIFSIGNALED(), WTERMSIG(),WIFSTOPPED(),WSTOPSIG()functions.

Availability:Unix, Windows, not WASI, not Android, not iOS.

Added in version 3.9.

The following functions take a process status code as returned by system(),wait(),orwaitpid()as a parameter. They may be used to determine the disposition of a process.

os.WCOREDUMP(status,/)

ReturnTrueif a core dump was generated for the process, otherwise returnFalse.

This function should be employed only ifWIFSIGNALED()is true.

Availability:Unix, not WASI, not Android, not iOS.

os.WIFCONTINUED(status)

ReturnTrueif a stopped child has been resumed by delivery of SIGCONT(if the process has been continued from a job control stop), otherwise returnFalse.

SeeWCONTINUEDoption.

Availability:Unix, not WASI, not Android, not iOS.

os.WIFSTOPPED(status)

ReturnTrueif the process was stopped by delivery of a signal, otherwise returnFalse.

WIFSTOPPED()only returnsTrueif thewaitpid()call was done usingWUNTRACEDoption or when the process is being traced (see ptrace(2)).

Availability:Unix, not WASI, not Android, not iOS.

os.WIFSIGNALED(status)

ReturnTrueif the process was terminated by a signal, otherwise return False.

Availability:Unix, not WASI, not Android, not iOS.

os.WIFEXITED(status)

ReturnTrueif the process exited terminated normally, that is, by callingexit()or_exit(),or by returning frommain(); otherwise returnFalse.

Availability:Unix, not WASI, not Android, not iOS.

os.WEXITSTATUS(status)

Return the process exit status.

This function should be employed only ifWIFEXITED()is true.

Availability:Unix, not WASI, not Android, not iOS.

os.WSTOPSIG(status)

Return the signal which caused the process to stop.

This function should be employed only ifWIFSTOPPED()is true.

Availability:Unix, not WASI, not Android, not iOS.

os.WTERMSIG(status)

Return the number of the signal that caused the process to terminate.

This function should be employed only ifWIFSIGNALED()is true.

Availability:Unix, not WASI, not Android, not iOS.

Interface to the scheduler

These functions control how a process is allocated CPU time by the operating system. They are only available on some Unix platforms. For more detailed information, consult your Unix manpages.

Added in version 3.3.

The following scheduling policies are exposed if they are supported by the operating system.

os.SCHED_OTHER

The default scheduling policy.

os.SCHED_BATCH

Scheduling policy for CPU-intensive processes that tries to preserve interactivity on the rest of the computer.

os.SCHED_IDLE

Scheduling policy for extremely low priority background tasks.

os.SCHED_SPORADIC

Scheduling policy for sporadic server programs.

os.SCHED_FIFO

A First In First Out scheduling policy.

os.SCHED_RR

A round-robin scheduling policy.

os.SCHED_RESET_ON_FORK

This flag can be OR’ed with any other scheduling policy. When a process with this flag set forks, its child’s scheduling policy and priority are reset to the default.

classos.sched_param(sched_priority)

This class represents tunable scheduling parameters used in sched_setparam(),sched_setscheduler(),and sched_getparam().It is immutable.

At the moment, there is only one possible parameter:

sched_priority

The scheduling priority for a scheduling policy.

os.sched_get_priority_min(policy)

Get the minimum priority value forpolicy.policyis one of the scheduling policy constants above.

os.sched_get_priority_max(policy)

Get the maximum priority value forpolicy.policyis one of the scheduling policy constants above.

os.sched_setscheduler(pid,policy,param,/)

Set the scheduling policy for the process with PIDpid.Apidof 0 means the calling process.policyis one of the scheduling policy constants above.paramis asched_paraminstance.

os.sched_getscheduler(pid,/)

Return the scheduling policy for the process with PIDpid.Apidof 0 means the calling process. The result is one of the scheduling policy constants above.

os.sched_setparam(pid,param,/)

Set the scheduling parameters for the process with PIDpid.Apidof 0 means the calling process.paramis asched_paraminstance.

os.sched_getparam(pid,/)

Return the scheduling parameters as asched_paraminstance for the process with PIDpid.Apidof 0 means the calling process.

os.sched_rr_get_interval(pid,/)

Return the round-robin quantum in seconds for the process with PIDpid.A pidof 0 means the calling process.

os.sched_yield()

Voluntarily relinquish the CPU.

os.sched_setaffinity(pid,mask,/)

Restrict the process with PIDpid(or the current process if zero) to a set of CPUs.maskis an iterable of integers representing the set of CPUs to which the process should be restricted.

os.sched_getaffinity(pid,/)

Return the set of CPUs the process with PIDpidis restricted to.

Ifpidis zero, return the set of CPUs the calling thread of the current process is restricted to.

See also theprocess_cpu_count()function.

Miscellaneous System Information

os.confstr(name,/)

Return string-valued system configuration values.namespecifies the configuration value to retrieve; it may be a string which is the name of a defined system value; these names are specified in a number of standards (POSIX, Unix 95, Unix 98, and others). Some platforms define additional names as well. The names known to the host operating system are given as the keys of the confstr_namesdictionary. For configuration variables not included in that mapping, passing an integer fornameis also accepted.

If the configuration value specified bynameisn’t defined,Noneis returned.

Ifnameis a string and is not known,ValueErroris raised. If a specific value fornameis not supported by the host system, even if it is included inconfstr_names,anOSErroris raised with errno.EINVALfor the error number.

Availability:Unix.

os.confstr_names

Dictionary mapping names accepted byconfstr()to the integer values defined for those names by the host operating system. This can be used to determine the set of names known to the system.

Availability:Unix.

os.cpu_count()

Return the number of logical CPUs in thesystem.ReturnsNoneif undetermined.

Theprocess_cpu_count()function can be used to get the number of logical CPUs usable by the calling thread of thecurrent process.

Added in version 3.4.

Changed in version 3.13:If-Xcpu_countis given orPYTHON_CPU_COUNTis set, cpu_count()returns the overridden valuen.

os.getloadavg()

Return the number of processes in the system run queue averaged over the last 1, 5, and 15 minutes or raisesOSErrorif the load average was unobtainable.

Availability:Unix.

os.process_cpu_count()

Get the number of logical CPUs usable by the calling thread of thecurrent process.ReturnsNoneif undetermined. It can be less than cpu_count()depending on the CPU affinity.

Thecpu_count()function can be used to get the number of logical CPUs in thesystem.

If-Xcpu_countis given orPYTHON_CPU_COUNTis set, process_cpu_count()returns the overridden valuen.

See also thesched_getaffinity()function.

Added in version 3.13.

os.sysconf(name,/)

Return integer-valued system configuration values. If the configuration value specified bynameisn’t defined,-1is returned. The comments regarding thenameparameter forconfstr()apply here as well; the dictionary that provides information on the known names is given bysysconf_names.

Availability:Unix.

os.sysconf_names

Dictionary mapping names accepted bysysconf()to the integer values defined for those names by the host operating system. This can be used to determine the set of names known to the system.

Availability:Unix.

Changed in version 3.11:Add'SC_MINSIGSTKSZ'name.

The following data values are used to support path manipulation operations. These are defined for all platforms.

Higher-level operations on pathnames are defined in theos.pathmodule.

os.curdir

The constant string used by the operating system to refer to the current directory. This is'.'for Windows and POSIX. Also available via os.path.

os.pardir

The constant string used by the operating system to refer to the parent directory. This is'..'for Windows and POSIX. Also available via os.path.

os.sep

The character used by the operating system to separate pathname components. This is'/'for POSIX and'\\'for Windows. Note that knowing this is not sufficient to be able to parse or concatenate pathnames — use os.path.split()andos.path.join()— but it is occasionally useful. Also available viaos.path.

os.altsep

An alternative character used by the operating system to separate pathname components, orNoneif only one separator character exists. This is set to '/'on Windows systems wheresepis a backslash. Also available via os.path.

os.extsep

The character which separates the base filename from the extension; for example, the'.'inos.py.Also available viaos.path.

os.pathsep

The character conventionally used by the operating system to separate search path components (as inPATH), such as':'for POSIX or';'for Windows. Also available viaos.path.

os.defpath

The default search path used byexec*p*and spawn*p*if the environment doesn’t have a'PATH' key. Also available viaos.path.

os.linesep

The string used to separate (or, rather, terminate) lines on the current platform. This may be a single character, such as'\n'for POSIX, or multiple characters, for example,'\r\n'for Windows. Do not use os.linesepas a line terminator when writing files opened in text mode (the default); use a single'\n'instead, on all platforms.

os.devnull

The file path of the null device. For example:'/dev/null'for POSIX,'nul'for Windows. Also available viaos.path.

os.RTLD_LAZY

os.RTLD_NOW

os.RTLD_GLOBAL

os.RTLD_LOCAL

os.RTLD_NODELETE

os.RTLD_NOLOAD

os.RTLD_DEEPBIND

Flags for use with thesetdlopenflags()and getdlopenflags()functions. See the Unix manual page dlopen(3)for what the different flags mean.

Added in version 3.3.

Random numbers

os.getrandom(size,flags=0)

Get up tosizerandom bytes. The function can return less bytes than requested.

These bytes can be used to seed user-space random number generators or for cryptographic purposes.

getrandom()relies on entropy gathered from device drivers and other sources of environmental noise. Unnecessarily reading large quantities of data will have a negative impact on other users of the/dev/randomand /dev/urandomdevices.

The flags argument is a bit mask that can contain zero or more of the following values ORed together:os.GRND_RANDOMand GRND_NONBLOCK.

See also theLinux getrandom() manual page.

Availability:Linux >= 3.17.

Added in version 3.6.

os.urandom(size,/)

Return a bytestring ofsizerandom bytes suitable for cryptographic use.

This function returns random bytes from an OS-specific randomness source. The returned data should be unpredictable enough for cryptographic applications, though its exact quality depends on the OS implementation.

On Linux, if thegetrandom()syscall is available, it is used in blocking mode: block until the system urandom entropy pool is initialized (128 bits of entropy are collected by the kernel). See thePEP 524for the rationale. On Linux, thegetrandom()function can be used to get random bytes in non-blocking mode (using theGRND_NONBLOCKflag) or to poll until the system urandom entropy pool is initialized.

On a Unix-like system, random bytes are read from the/dev/urandom device. If the/dev/urandomdevice is not available or not readable, the NotImplementedErrorexception is raised.

On Windows, it will useBCryptGenRandom().

See also

Thesecretsmodule provides higher level functions. For an easy-to-use interface to the random number generator provided by your platform, please seerandom.SystemRandom.

Changed in version 3.5:On Linux 3.17 and newer, thegetrandom()syscall is now used when available. On OpenBSD 5.6 and newer, the Cgetentropy() function is now used. These functions avoid the usage of an internal file descriptor.

Changed in version 3.5.2:On Linux, if thegetrandom()syscall blocks (the urandom entropy pool is not initialized yet), fall back on reading/dev/urandom.

Changed in version 3.6:On Linux,getrandom()is now used in blocking mode to increase the security.

Changed in version 3.11:On Windows,BCryptGenRandom()is used instead ofCryptGenRandom() which is deprecated.

os.GRND_NONBLOCK

By default, when reading from/dev/random,getrandom()blocks if no random bytes are available, and when reading from/dev/urandom,it blocks if the entropy pool has not yet been initialized.

If theGRND_NONBLOCKflag is set, thengetrandom()does not block in these cases, but instead immediately raisesBlockingIOError.

Added in version 3.6.

os.GRND_RANDOM

If this bit is set, then random bytes are drawn from the /dev/randompool instead of the/dev/urandompool.

Added in version 3.6.