What characters are forbidden in Windows and Linux directory names?

A “comprehensive guide” of forbidden filename characters is not going to work on Windows because it reserves filenames as well as characters. Yes, characters like * " ? and others are forbidden, but there are a infinite number of names composed only of valid characters that are forbidden. For example, spaces and dots are valid filename characters, but names composed only of those characters are forbidden.

Windows does not distinguish between upper-case and lower-case characters, so you cannot create a folder named A if one named a already exists. Worse, seemingly-allowed names like PRN and CON, and many others, are reserved and not allowed. Windows also has several length restrictions; a filename valid in one folder may become invalid if moved to another folder. The rules for naming files and folders are on the Microsoft docs.

You cannot, in general, use user-generated text to create Windows directory names. If you want to allow users to name anything they want, you have to create safe names like A, AB, A2 et al., store user-generated names and their path equivalents in an application data file, and perform path mapping in your application.

If you absolutely must allow user-generated folder names, the only way to tell if they are invalid is to catch exceptions and assume the name is invalid. Even that is fraught with peril, as the exceptions thrown for denied access, offline drives, and out of drive space overlap with those that can be thrown for invalid names. You are opening up one huge can of hurt.

Under Linux and other Unix-related systems, there are only two characters that cannot appear in the name of a file or directory, and those are NUL '\0' and slash '/'. The slash, of course, can appear in a pathname, separating directory components.

Rumour¹ has it that Steven Bourne (of 'shell' fame) had a directory containing 254 files, one for every single letter (character code) that can appear in a file name (excluding /, '\0'; the name . was the current directory, of course). It was used to test the Bourne shell and routinely wrought havoc on unwary programs such as backup programs.

Other people have covered the rules for Windows filenames, with links to Microsoft and Wikipedia on the topic.

Note that MacOS X has a case-insensitive file system. Current versions of it appear to allow colon : in file names, though historically that was not always the case:

$ echo a:b > a:b
$ ls -l a:b
-rw-r--r--  1 jonathanleffler  staff  4 Nov 12 07:38 a:b
$

POSIX defines a Portable Filename Character Set consisting of:

A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
a b c d e f g h i j k l m n o p q r s t u v w x y z
0 1 2 3 4 5 6 7 8 9 . _ -

Sticking with names formed solely from those characters avoids most of the problems, though Windows still adds some complications.

When Steve Bourne was writing his Unix shell (which came to be known as the Bourne shell), he made a directory of 254 files with one-character names, one for each byte value except '\0' and slash, the two characters that cannot appear in Unix file names. He used that directory for all manner of tests of pattern-matching and tokenization. (The test directory was of course created by a program.) For years afterwards, that directory was the bane of file-tree-walking programs; it tested them to destruction.

^{Note that the directory must have contained entries . and .., so it was arguably 253 files (and 2 directories), or 255 name entries, rather than 254 files. This doesn't affect the effectiveness of the anecdote, or the careful testing it describes.}

^{TPOP was previously at http://plan9.bell-labs.com/cm/cs/tpop and http://cm.bell-labs.com/cm/cs/tpop but both are now (2021-11-12) broken. See also Wikipedia on TPOP.}

Well, if only for research purposes, then your best bet is to look at this Wikipedia entry on Filenames.

If you want to write a portable function to validate user input and create filenames based on that, the short answer is don't. Take a look at a portable module like Perl's File::Spec to have a glimpse to all the hops needed to accomplish such a "simple" task.

Discussing different possible approaches

Difficulties with defining, what's legal and not were already adressed and whitelists were suggested. But Windows supports more-than-8-bit characters. Wikipedia states, that (for example) the

modifier letter colon [(See 7. below) is] sometimes used in Windows filenames as it is identical to the colon in the Segoe UI font used for filenames. The [inherited ASCII] colon itself is not permitted.

Therefore, I want to present a much more liberal approach using Unicode Homoglyph characters to replace the "illegal" ones. I found the result in my comparable use-case by far more readable. Plus you can even restore the original content from the replacements.

Possible choices and research notes

To keep things organized, I will always give the character, it's name and the hexadecimal number representation. The latter is is not case sensitive and leading zeroes can be added or ommitted freely, so for example U+002A and u+2a are equivalent. If available, I'll try to point to more info or alternatives - feel free to show me more or better ones.

1. Instead of * (U+2A * ASTERISK), you can use one of the many listed, for example U+2217 ∗ (ASTERISK OPERATOR) or the Full Width Asterisk U+FF0A ＊. u+20f0 ⃰ combining asterisk above from combining diacritical marks for symbols might also be a valid choice.

2. Instead of . (U+2E . full stop), one of these could be a good option, for example ⋅ U+22C5 dot operator

3. Instead of " (U+22 " quotation mark), you can use “ U+201C english leftdoublequotemark, more alternatives see here. I also included some of the good suggestions of Wally Brockway's answer, in this case u+2036 ‶ reversed double prime and u+2033 ″ double prime - I will from now on denote ideas from that source by ⁷.

4. Instead of / (U+2F / SOLIDUS), you can use ∕ DIVISION SLASH U+2215 (others here), ̸ U+0338 COMBINING LONG SOLIDUS OVERLAY, ̷ COMBINING SHORT SOLIDUS OVERLAY U+0337 or u+2044 ⁄ fraction slash⁷. Be aware about spacing for some characters, including the combining or overlay ones, as they have no width and can produce something like -> ̸th̷is which is ̸th̷is. With added spaces you get -> ̸ th ̷ is, which is ̸ th ̷ is. The second one (COMBINING SHORT SOLIDUS OVERLAY) looks bad in the stackoverflow-font.

5. Instead of \ (U+5C Reverse solidus), you can use ⧵ U+29F5 Reverse solidus operator (more) or u+20E5 ⃥ combining reverse solidus overlay⁷

6. To replace [ (U+5B [ Left square bracket) and ] (U+005D ] Right square bracket), you can use for example U+FF3B［ FULLWIDTH LEFT SQUARE BRACKET and U+FF3D ］FULLWIDTH RIGHT SQUARE BRACKET (from here, more possibilities here)

7. Instead of : (u+3a : colon), you can use U+2236 ∶ RATIO (for mathematical usage) or U+A789 ꞉ MODIFIER LETTER COLON, (see colon (letter), sometimes used in Windows filenames as it is identical to the colon in the Segoe UI font used for filenames. The colon itself is not permitted ... source and more replacements see here). Another alternative is this one: u+1361 ፡ ethiopic wordspace⁷

8. Instead of ; (u+3b ; semicolon), you can use U+037E ; GREEK QUESTION MARK (see here)

9. For | (u+7c | vertical line), there are some good substitutes such as: U+2223 ∣ DIVIDES, U+0964 । DEVANAGARI DANDA, U+01C0 ǀ LATIN LETTER DENTAL CLICK (the last ones from Wikipedia) or U+2D4F ⵏ Tifinagh Letter Yan. Also the box drawing characters contain various other options.

10. Instead of , (, U+002C COMMA), you can use for example ‚ U+201A SINGLE LOW-9 QUOTATION MARK (see here)

11. For ? (U+003F ? QUESTION MARK), these are good candidates: U+FF1F ？ FULLWIDTH QUESTION MARK or U+FE56 ﹖ SMALL QUESTION MARK (from here and here). There are also two more from the Dingbats Block (search for "question") and the u+203d ‽ interrobang⁷

12. While my machine seems to accept it unchanged, I still want to include > (u+3e greater-than sign) and < (u+3c less-than sign) for the sake of completeness. The best replacement here is probably also from the quotation block, such as u+203a › single right-pointing angle quotation mark and u+2039 ‹ single left-pointing angle quotation mark respectively. The tifinagh block only contains ⵦ (u+2D66)⁷ to replace <. The last notion is ⋖ less-than with dot u+22D6 and ⋗ greater-than with dot u+22D7.

For additional ideas, you can also look for example into this block. You still want more ideas? You can try to draw your desired character and look at the suggestions here.

How do you type these characters

Say you want to type ⵏ (Tifinagh Letter Yan). To get its information, you can always search for this character on a suited platform such as the Unicode Lookup (add 0x when you search for hex) or this Unicode Table (that only allows to search for the name, in this case "Tifinagh Letter Yan"). You should obtain its Unicode number U+2D4F and the HTML-code ⵏ ^{(note that 2D4F is hexadecimal for 11599)}. With this knowledge, you have several options to produce these special characters including the use of

• code points to unicode converter or again the Unicode Lookup to reversely, convert the numerical representation into the unicode character - set the code point base below to decimal or hexadecimal respectively
• a one-liner makro in Autohotkey: :?*:altpipe::{U+2D4F} to type ⵏ instead of the string altpipe - this is the way I input those special characters, my Autohotkey script can be shared if there is common interest
• Alt Characters or alt-codes by pressing and holding alt, followed by the decimal number for the desired character (more info for example here, look at a table here or there). For the example, that would be Alt+11599. Be aware, that many programs do not fully support this windows feature for all of unicode (as of time writing). Microsoft Office is an exception where it usually works, some other OSes provide similar functionality. Typing these chars with Alt-combinations into MS Word is also the way Wally Brockway suggests in his answer⁷ that was already mentionted - if you don't want to transfer all the hexadecimal values to the decimal asc, you can find some of them there⁷.
• in MS Office, you can also use ALT + X as described in this MS article to produce the chars
• if you rarely need it, you can of course still just copy-paste the special character of your choice instead of typing it

For anyone looking for a regex:

const BLACKLIST = /[<>:"\/\\|?*]/g;

In Windows 10 (2019), the following characters are forbidden by an error when you try to type them:

A file name can't contain any of the following characters:

\ / : * ? " < > |

Here's a c# implementation for windows based on Christopher Oezbek's answer

It was made more complex by the containsFolder boolean, but hopefully covers everything

/// <summary>
/// This will replace invalid chars with underscores, there are also some reserved words that it adds underscore to
/// </summary>
/// <remarks>
/// https://stackoverflow.com/questions/1976007/what-characters-are-forbidden-in-windows-and-linux-directory-names
/// </remarks>
/// <param name="containsFolder">Pass in true if filename represents a folder\file (passing true will allow slash)</param>
public static string EscapeFilename_Windows(string filename, bool containsFolder = false)
{
    StringBuilder builder = new StringBuilder(filename.Length + 12);

    int index = 0;

    // Allow colon if it's part of the drive letter
    if (containsFolder)
    {
        Match match = Regex.Match(filename, @"^\s*[A-Z]:\\", RegexOptions.IgnoreCase);
        if (match.Success)
        {
            builder.Append(match.Value);
            index = match.Length;
        }
    }

    // Character substitutions
    for (int cntr = index; cntr < filename.Length; cntr++)
    {
        char c = filename[cntr];

        switch (c)
        {
            case '\u0000':
            case '\u0001':
            case '\u0002':
            case '\u0003':
            case '\u0004':
            case '\u0005':
            case '\u0006':
            case '\u0007':
            case '\u0008':
            case '\u0009':
            case '\u000A':
            case '\u000B':
            case '\u000C':
            case '\u000D':
            case '\u000E':
            case '\u000F':
            case '\u0010':
            case '\u0011':
            case '\u0012':
            case '\u0013':
            case '\u0014':
            case '\u0015':
            case '\u0016':
            case '\u0017':
            case '\u0018':
            case '\u0019':
            case '\u001A':
            case '\u001B':
            case '\u001C':
            case '\u001D':
            case '\u001E':
            case '\u001F':

            case '<':
            case '>':
            case ':':
            case '"':
            case '/':
            case '|':
            case '?':
            case '*':
                builder.Append('_');
                break;

            case '\\':
                builder.Append(containsFolder ? c : '_');
                break;

            default:
                builder.Append(c);
                break;
        }
    }

    string built = builder.ToString();

    if (built == "")
    {
        return "_";
    }

    if (built.EndsWith(" ") || built.EndsWith("."))
    {
        built = built.Substring(0, built.Length - 1) + "_";
    }

    // These are reserved names, in either the folder or file name, but they are fine if following a dot
    // CON, PRN, AUX, NUL, COM0 .. COM9, LPT0 .. LPT9
    builder = new StringBuilder(built.Length + 12);
    index = 0;
    foreach (Match match in Regex.Matches(built, @"(^|\\)\s*(?<bad>CON|PRN|AUX|NUL|COM\d|LPT\d)\s*(\.|\\|$)", RegexOptions.IgnoreCase))
    {
        Group group = match.Groups["bad"];
        if (group.Index > index)
        {
            builder.Append(built.Substring(index, match.Index - index + 1));
        }

        builder.Append(group.Value);
        builder.Append("_");        // putting an underscore after this keyword is enough to make it acceptable

        index = group.Index + group.Length;
    }

    if (index == 0)
    {
        return built;
    }

    if (index < built.Length - 1)
    {
        builder.Append(built.Substring(index));
    }

    return builder.ToString();
}

The .NET Framework System.IO provides the following functions for invalid file system characters:

• Path.GetInvalidFileNameChars
• Path.GetInvalidPathChars

Those functions should return appropriate results depending on the platform the .NET runtime is running in. That said, the Remarks in the documentation pages for those functions say:

The array returned from this method is not guaranteed to contain the complete set of characters that are invalid in file and directory names. The full set of invalid characters can vary by file system.

I always assumed that banned characters in Windows filenames meant that all exotic characters would also be outlawed. The inability to use ?, / and : in particular irked me. One day I discovered that it was virtually only those chars which were banned. Other Unicode characters may be used. So the nearest Unicode characters to the banned ones I could find were identified and MS Word macros were made for them as Alt+?, Alt+: etc. Now I form the filename in Word, using the substitute chars, and copy it to the Windows filename. So far I have had no problems.

Here are the substitute chars (Alt + the decimal Unicode) :

• ⃰ ⇔ Alt8432
• ⁄ ⇔ Alt8260
• ⃥ ⇔ Alt8421
• ∣ ⇔ Alt8739
• ⵦ ⇔ Alt11622
• ⮚ ⇔ Alt11162
• ‽ ⇔ Alt8253
• ፡ ⇔ Alt4961
• ‶ ⇔ Alt8246
• ″ ⇔ Alt8243

As a test I formed a filename using all of those chars and Windows accepted it.

This is good enough for me in Python:

def fix_filename(name, max_length=255):
    """
    Replace invalid characters on Linux/Windows/MacOS with underscores.
    List from https://stackoverflow.com/a/31976060/819417
    Trailing spaces & periods are ignored on Windows.
    >>> fix_filename("  COM1  ")
    '_ COM1 _'
    >>> fix_filename("COM10")
    'COM10'
    >>> fix_filename("COM1,")
    'COM1,'
    >>> fix_filename("COM1.txt")
    '_.txt'
    >>> all('_' == fix_filename(chr(i)) for i in list(range(32)))
    True
    """
    return re.sub(r'[/\\:|<>"?*\0-\x1f]|^(AUX|COM[1-9]|CON|LPT[1-9]|NUL|PRN)(?![^.])|^\s|[\s.]$', "_", name[:max_length], flags=re.IGNORECASE)

See also this outdated list for additional legacy stuff like = in FAT32.