18
$\begingroup$

Is there a way to work with pathnames that are longer than the typical 256 in Mathematica?

For example run the following in Cygwin

echo 1 > 034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890

Then run

FilePrint["034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890"]

General::noopen: Cannot open ...

$\endgroup$
  • 6
    $\begingroup$ The problem arises because NT supports paths up to 32768 (UTF-16) characters long and containing any characters except NUL, but Win32 imposes many more restrictions than this. Cygwin is probably calling the NT function NtCreateFile or ZwCreateFile, while Win32 programs (including Mathematica) normally call CreateFile. $\endgroup$ – Oleksandr R. Oct 24 '15 at 21:58
19
$\begingroup$

With a few rare exceptions, Mathematica is generally unable to work with long path names on Windows. This response presents two strategies to work around this difficulty: extended-length path syntax and short path names (documented in Naming Files, Paths, and Namespaces from the Windows Dev Center).

Let's start with a big file name:

$big = "034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890345678903456789034567890";

Extended-length Path Syntax

The Windows echo command cannot handle this big name:

Run["cmd /c echo hello >", $big]
(* 1 *)   (* i.e. "error" *)

It can, however, handle extended-length path syntax (in recent versions of Windows). This entails tacking the prefix \\?\ onto the front of an absolute pathname. We will use some helper functions:

$longPrefix = "\\\\?\\";

toLongPath[filename_] := $longPrefix <> ExpandFileName[filename]

fromLongPath[filename_] := StringReplace[filename, StartOfString~~$longPrefix -> ""]

Usage:

toLongPath[$big]
(* \\?\C:\Users\wreach\Documents\03456789...9034567890 *)

fromLongPath[%]
(* C:\Users\wreach\Documents\03456789...9034567890 *)

The Windows echo command accepts this syntax:

Run["cmd /c echo hello >", toLongPath[$big]]
(* 0 *)   (* i.e. "success" *)

FirstCase[FileNames[], n_ /; StringLength[n] > 200]
(* "03456789034567890345...789034567890" *)

The Mathematica CopyFile function will also accept this syntax:

CopyFile[toLongPath[$big], "zot"];
FilePrint["zot"]
(* hello *)

DeleteFile["zot"]

Short Path Names

Unfortunately, FilePrint does not support extended-length path syntax:

FilePrint[toLongPath[$big]]
(* Cannot open \\?\C:\Users\wreach\Documents\03456...890. >> *)

We will switch to the second strategy: short path names. These are alternative names that Windows provides for use by DOS-like components (read: most of the Windows ecosystem). We can view short path names for files by using the Windows dir /x command:

Import["!cmd /c dir /x 03456*", "Text"]

(* ...
   Directory of C:\Users\wreach\Documents
   ...
   2015-10-24  14:33     8    034567~1     034567890...3456789034567890
   ...
*)

The directory listing shows us that the short file name for $big is 034567~1.

FilePrint["034567~1"]
(* hello *)

It would be nice to obtain such names programmatically. Alas, Mathematica does not have a built-in function to obtain short file names. We will use NETLink to call the Win32 API function GetShortPathName:

Needs["NETLink`"]
InstallNET[];

getLastError = DefineDLLFunction["GetLastError", "kernel32.dll", "DWORD", {}];

getShortPathNameW =
  DefineDLLFunction["GetShortPathNameW"
  , "kernel32.dll", "DWORD", {"LPCTSTR", "System.Text.StringBuilder", "DWORD"}
  , MarshalStringsAs -> "Unicode"
  ];

shortPath[name_String] :=
  NETBlock @ Module[{result = NETNew["System.Text.StringBuilder", 260]}
  , getShortPathNameW[toLongPath@name, result, result@Capacity]
  ; getLastError[] /. 0 :> fromLongPath@result@ToString[]
  ]

Here is our helper function in action:

shortPath[$big]
(* C:\Users\wreach\DOCUME~1\034567~1 *)

This short path can be used everywhere in Mathematica, including FilePrint:

FilePrint[shortPath[$big]]
(* hello *)

Such short paths can also be used by all of the components that support Mathematica, whether they be DLLs, Java, command line tools or anything else.

Note that short paths can only be obtained for files that already exist. This mechanism cannot be used to create new files with big names.


Incidentally, DeleteFile is one of the few functions that accepts extended-length path syntax:

DeleteFile[toLongPath[$big]]
$\endgroup$
  • 7
    $\begingroup$ Note that short names do not necessarily exist. Their creation can be switched off as an NTFS option. (It is a manifestation of the general truth that an NTFS file can have multiple names, or a single name can have multiple files associated with it.) $\endgroup$ – Oleksandr R. Oct 24 '15 at 21:58
  • $\begingroup$ When running Import["!cmd /c dir /x 03456*", "Text"] I get 2(yes it appears to be the number 2) as the short path name. Alas they must disabled or something here is the link Oleksandr references. Here is my M output $\endgroup$ – William Nov 4 '15 at 2:31
  • 1
    $\begingroup$ So it would seem. The "2" you see is the file byte count. If you have control over such policy, I would recommend turning short names back on. Many Microsoft-supplied components rely upon them, and some 3rd party applications do as well (especially cross-platform ones). $\endgroup$ – WReach Nov 4 '15 at 17:12
2
$\begingroup$

(An addition to the great answer by WReach above.)

At least starting from version 11.3 Mathematica has built-in way to obtain short filenames using undocumented function FileInformation:

Export["Тест.txt", ""]
FileInformation["Тест.txt", "AbsoluteShortFileName"]
"Тест.txt"

"D:\\DOCUME~1\\47BB~1.TXT"

(thanks to GenericAccountName for the info).

Also starting from version 12.0.0 there is an under-documented (but "official") way to obtain the short path:

Information[File["Тест.txt"], "Properties"] // Shallow
{"ObjectType", "FileName", "DirectoryName", "AbsoluteFileName", "AbsoluteShortFileName", "ShortFileName", "FileBaseName", "FileExtension", "CreationDate", "LastAccessDate", <<33>>}
Information[File["Тест.txt"], "AbsoluteShortFileName"]
"D:\\DOCUME~1\\47BB~1.TXT"
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.