3  Using Unicode in Erlang

3 Using Unicode in Erlang

Implementing support for Unicode character sets is an ongoing process. The Erlang Enhancement Proposal (EEP) 10 outlined the basics of Unicode support and specified a default encoding in binaries that all Unicode-aware modules are to handle in the future.

Here is an overview what has been done so far:

  • The functionality described in EEP10 was implemented in Erlang/OTP R13A.

  • Erlang/OTP R14B01 added support for Unicode filenames, but it was not complete and was by default disabled on platforms where no guarantee was given for the filename encoding.

  • With Erlang/OTP R16A came support for UTF-8 encoded source code, with enhancements to many of the applications to support both Unicode encoded filenames and support for UTF-8 encoded files in many circumstances. Most notable is the support for UTF-8 in files read by file:consult/1, release handler support for UTF-8, and more support for Unicode character sets in the I/O system.

  • In Erlang/OTP 17.0, the encoding default for Erlang source files was switched to UTF-8.

  • In Erlang/OTP 20.0, atoms and function can contain Unicode characters. Module names, application names, and node names are still restricted to the ISO Latin-1 range.

    Support was added for normalizations forms in unicode and the string module now handles utf8-encoded binaries.

This section outlines the current Unicode support and gives some recipes for working with Unicode data.

Experience with the Unicode support in Erlang has made it clear that understanding Unicode characters and encodings is not as easy as one would expect. The complexity of the field and the implications of the standard require thorough understanding of concepts rarely before thought of.

Also, the Erlang implementation requires understanding of concepts that were never an issue for many (Erlang) programmers. To understand and use Unicode characters requires that you study the subject thoroughly, even if you are an experienced programmer.

As an example, contemplate the issue of converting between upper and lower case letters. Reading the standard makes you realize that there is not a simple one to one mapping in all scripts, for example:

  • In German, the letter "ß" (sharp s) is in lower case, but the uppercase equivalent is "SS".

  • In Greek, the letter "Σ" has two different lowercase forms, "ς" in word-final position and "σ" elsewhere.

  • In Turkish, both dotted and dotless "i" exist in lower case and upper case forms.

  • Cyrillic "I" has usually no lowercase form.

  • Languages with no concept of upper case (or lower case).

So, a conversion function must know not only one character at a time, but possibly the whole sentence, the natural language to translate to, the differences in input and output string length, and so on. Erlang/OTP has currently no Unicode uppercase/lowercase functionality with language specific handling, but publicly available libraries address these issues.

Another example is the accented characters, where the same glyph has two different representations. The Swedish letter "ö" is one example. The Unicode standard has a code point for it, but you can also write it as "o" followed by "U+0308" (Combining Diaeresis, with the simplified meaning that the last letter is to have "¨" above). They have the same glyph, user perceived character. They are for most purposes the same, but have different representations. For example, MacOS X converts all filenames to use Combining Diaeresis, while most other programs (including Erlang) try to hide that by doing the opposite when, for example, listing directories. However it is done, it is usually important to normalize such characters to avoid confusion.

The list of examples can be made long. One need a kind of knowledge that was not needed when programs only considered one or two languages. The complexity of human languages and scripts has certainly made this a challenge when constructing a universal standard. Supporting Unicode properly in your program will require effort.

Unicode is a standard defining code points (numbers) for all known, living or dead, scripts. In principle, every symbol used in any language has a Unicode code point. Unicode code points are defined and published by the Unicode Consortium, which is a non-profit organization.

Support for Unicode is increasing throughout the world of computing, as the benefits of one common character set are overwhelming when programs are used in a global environment. Along with the base of the standard, the code points for all the scripts, some encoding standards are available.

It is vital to understand the difference between encodings and Unicode characters. Unicode characters are code points according to the Unicode standard, while the encodings are ways to represent such code points. An encoding is only a standard for representation. UTF-8 can, for example, be used to represent a very limited part of the Unicode character set (for example ISO-Latin-1) or the full Unicode range. It is only an encoding format.

As long as all character sets were limited to 256 characters, each character could be stored in one single byte, so there was more or less only one practical encoding for the characters. Encoding each character in one byte was so common that the encoding was not even named. With the Unicode system there are much more than 256 characters, so a common way is needed to represent these. The common ways of representing the code points are the encodings. This means a whole new concept to the programmer, the concept of character representation, which was a non-issue earlier.

Different operating systems and tools support different encodings. For example, Linux and MacOS X have chosen the UTF-8 encoding, which is backward compatible with 7-bit ASCII and therefore affects programs written in plain English the least. Windows supports a limited version of UTF-16, namely all the code planes where the characters can be stored in one single 16-bit entity, which includes most living languages.

The following are the most widely spread encodings:

This is not a proper Unicode representation, but the representation used for characters before the Unicode standard. It can still be used to represent character code points in the Unicode standard with numbers < 256, which exactly corresponds to the ISO Latin-1 character set. In Erlang, this is commonly denoted latin1 encoding, which is slightly misleading as ISO Latin-1 is a character code range, not an encoding.

Each character is stored in one to four bytes depending on code point. The encoding is backward compatible with bytewise representation of 7-bit ASCII, as all 7-bit characters are stored in one single byte in UTF-8. The characters beyond code point 127 are stored in more bytes, letting the most significant bit in the first character indicate a multi-byte character. For details on the encoding, the RFC is publicly available.

Notice that UTF-8 is not compatible with bytewise representation for code points from 128 through 255, so an ISO Latin-1 bytewise representation is generally incompatible with UTF-8.

This encoding has many similarities to UTF-8, but the basic unit is a 16-bit number. This means that all characters occupy at least two bytes, and some high numbers four bytes. Some programs, libraries, and operating systems claiming to use UTF-16 only allow for characters that can be stored in one 16-bit entity, which is usually sufficient to handle living languages. As the basic unit is more than one byte, byte-order issues occur, which is why UTF-16 exists in both a big-endian and a little-endian variant.

In Erlang, the full UTF-16 range is supported when applicable, like in the unicode module and in the bit syntax.

The most straightforward representation. Each character is stored in one single 32-bit number. There is no need for escapes or any variable number of entities for one character. All Unicode code points can be stored in one single 32-bit entity. As with UTF-16, there are byte-order issues. UTF-32 can be both big-endian and little-endian.

Basically the same as UTF-32, but without some Unicode semantics, defined by IEEE, and has little use as a separate encoding standard. For all normal (and possibly abnormal) use, UTF-32 and UCS-4 are interchangeable.

Certain number ranges are unused in the Unicode standard and certain ranges are even deemed invalid. The most notable invalid range is 16#D800-16#DFFF, as the UTF-16 encoding does not allow for encoding of these numbers. This is possibly because the UTF-16 encoding standard, from the beginning, was expected to be able to hold all Unicode characters in one 16-bit entity, but was then extended, leaving a hole in the Unicode range to handle backward compatibility.

Code point 16#FEFF is used for Byte Order Marks (BOMs) and use of that character is not encouraged in other contexts. It is valid though, as the character "ZWNBS" (Zero Width Non Breaking Space). BOMs are used to identify encodings and byte order for programs where such parameters are not known in advance. BOMs are more seldom used than expected, but can become more widely spread as they provide the means for programs to make educated guesses about the Unicode format of a certain file.

To support Unicode in Erlang, problems in various areas have been addressed. This section describes each area briefly and more thoroughly later in this User's Guide.

To handle Unicode characters in Erlang, a common representation in both lists and binaries is needed. EEP (10) and the subsequent initial implementation in Erlang/OTP R13A settled a standard representation of Unicode characters in Erlang.

The Unicode characters need to be processed by the Erlang program, which is why library functions must be able to handle them. In some cases functionality has been added to already existing interfaces (as the string module now can handle strings with any code points). In some cases new functionality or options have been added (as in the io module, the file handling, the unicode module, and the bit syntax). Today most modules in Kernel and STDLIB, as well as the VM are Unicode-aware.

I/O is by far the most problematic area for Unicode. A file is an entity where bytes are stored, and the lore of programming has been to treat characters and bytes as interchangeable. With Unicode characters, you must decide on an encoding when you want to store the data in a file. In Erlang, you can open a text file with an encoding option, so that you can read characters from it rather than bytes, but you can also open a file for bytewise I/O.

The Erlang I/O-system has been designed (or at least used) in a way where you expect any I/O server to handle any string data. That is, however, no longer the case when working with Unicode characters. The Erlang programmer must now know the capabilities of the device where the data ends up. Also, ports in Erlang are byte-oriented, so an arbitrary string of (Unicode) characters cannot be sent to a port without first converting it to an encoding of choice.

Terminal I/O is slightly easier than file I/O. The output is meant for human reading and is usually Erlang syntax (for example, in the shell). There exists syntactic representation of any Unicode character without displaying the glyph (instead written as \x{HHH}). Unicode data can therefore usually be displayed even if the terminal as such does not support the whole Unicode range.

Filenames can be stored as Unicode strings in different ways depending on the underlying operating system and file system. This can be handled fairly easy by a program. The problems arise when the file system is inconsistent in its encodings. For example, Linux allows files to be named with any sequence of bytes, leaving to each program to interpret those bytes. On systems where these "transparent" filenames are used, Erlang must be informed about the filename encoding by a startup flag. The default is bytewise interpretation, which is usually wrong, but allows for interpretation of all filenames.

The concept of "raw filenames" can be used to handle wrongly encoded filenames if one enables Unicode filename translation (+fnu) on platforms where this is not the default.

The Erlang source code has support for the UTF-8 encoding and bytewise encoding. The default in Erlang/OTP R16B was bytewise (latin1) encoding. It was changed to UTF-8 in Erlang/OTP 17.0. You can control the encoding by a comment like the following in the beginning of the file:

%% -*- coding: utf-8 -*-

This of course requires your editor to support UTF-8 as well. The same comment is also interpreted by functions like file:consult/1, the release handler, and so on, so that you can have all text files in your source directories in UTF-8 encoding.

Having the source code in UTF-8 also allows you to write string literals, function names, and atoms containing Unicode characters with code points > 255. Module names, application names, and node names are still restricted to the ISO Latin-1 range. Binary literals, where you use type /utf8, can also be expressed using Unicode characters > 255. Having module names or application names using characters other than 7-bit ASCII can cause trouble on operating systems with inconsistent file naming schemes, and can hurt portability, so it is not recommended.

EEP 40 suggests that the language is also to allow for Unicode characters > 255 in variable names. Whether to implement that EEP is yet to be decided.

In Erlang, strings are lists of integers. A string was until Erlang/OTP R13 defined to be encoded in the ISO Latin-1 (ISO 8859-1) character set, which is, code point by code point, a subrange of the Unicode character set.

The standard list encoding for strings was therefore easily extended to handle the whole Unicode range. A Unicode string in Erlang is a list containing integers, where each integer is a valid Unicode code point and represents one character in the Unicode character set.

Erlang strings in ISO Latin-1 are a subset of Unicode strings.

Only if a string contains code points < 256, can it be directly converted to a binary by using, for example, erlang:iolist_to_binary/1 or can be sent directly to a port. If the string contains Unicode characters > 255, an encoding must be decided upon and the string is to be converted to a binary in the preferred encoding using unicode:characters_to_binary/1,2,3. Strings are not generally lists of bytes, as they were before Erlang/OTP R13, they are lists of characters. Characters are not generally bytes, they are Unicode code points.

Binaries are more troublesome. For performance reasons, programs often store textual data in binaries instead of lists, mainly because they are more compact (one byte per character instead of two words per character, as is the case with lists). Using erlang:list_to_binary/1, an ISO Latin-1 Erlang string can be converted into a binary, effectively using bytewise encoding: one byte per character. This was convenient for those limited Erlang strings, but cannot be done for arbitrary Unicode lists.

As the UTF-8 encoding is widely spread and provides some backward compatibility in the 7-bit ASCII range, it is selected as the standard encoding for Unicode characters in binaries for Erlang.

The standard binary encoding is used whenever a library function in Erlang is to handle Unicode data in binaries, but is of course not enforced when communicating externally. Functions and bit syntax exist to encode and decode both UTF-8, UTF-16, and UTF-32 in binaries. However, library functions dealing with binaries and Unicode in general only deal with the default encoding.

Character data can be combined from many sources, sometimes available in a mix of strings and binaries. Erlang has for long had the concept of iodata or iolists, where binaries and lists can be combined to represent a sequence of bytes. In the same way, the Unicode-aware modules often allow for combinations of binaries and lists, where the binaries have characters encoded in UTF-8 and the lists contain such binaries or numbers representing Unicode code points:

unicode_binary() = binary() with characters encoded in UTF-8 coding standard

chardata() = charlist() | unicode_binary()

charlist() = maybe_improper_list(char() | unicode_binary() | charlist(),
  unicode_binary() | nil())

The module unicode even supports similar mixes with binaries containing other encodings than UTF-8, but that is a special case to allow for conversions to and from external data:

external_unicode_binary() = binary() with characters coded in a user-specified
  Unicode encoding other than UTF-8 (UTF-16 or UTF-32)

external_chardata() = external_charlist() | external_unicode_binary()

external_charlist() = maybe_improper_list(char() | external_unicode_binary() |
  external_charlist(), external_unicode_binary() | nil())

As from Erlang/OTP R16, Erlang source files can be written in UTF-8 or bytewise (latin1) encoding. For information about how to state the encoding of an Erlang source file, see the epp(3) module. As from Erlang/OTP R16, strings and comments can be written using Unicode. As from Erlang/OTP 20, also atoms and functions can be written using Unicode. Modules, applications, and nodes must still be named using characters from the ISO Latin-1 character set. (These restrictions in the language are independent of the encoding of the source file.)

The bit syntax contains types for handling binary data in the three main encodings. The types are named utf8, utf16, and utf32. The utf16 and utf32 types can be in a big-endian or a little-endian variant:

<<Ch/utf8,_/binary>> = Bin1,
<<Ch/utf16-little,_/binary>> = Bin2,
Bin3 = <<$H/utf32-little, $e/utf32-little, $l/utf32-little, $l/utf32-little,
$o/utf32-little>>,

For convenience, literal strings can be encoded with a Unicode encoding in binaries using the following (or similar) syntax:

Bin4 = <<"Hello"/utf16>>,

For source code, there is an extension to syntax \OOO (backslash followed by three octal numbers) and \xHH (backslash followed by x, followed by two hexadecimal characters), namely \x{H ...} (backslash followed by x, followed by left curly bracket, any number of hexadecimal digits, and a terminating right curly bracket). This allows for entering characters of any code point literally in a string even when the encoding of the source file is bytewise (latin1).

In the shell, if using a Unicode input device, or in source code stored in UTF-8, $ can be followed directly by a Unicode character producing an integer. In the following example, the code point of a Cyrillic с is output:

7> $с.
1089

In certain output functions and in the output of return values in the shell, Erlang tries to detect string data in lists and binaries heuristically. Typically you will see heuristic detection in a situation like this:

1> [97,98,99].
"abc"
2> <<97,98,99>>.
<<"abc">>    
3> <<195,165,195,164,195,182>>.
<<"åäö"/utf8>>

Here the shell detects lists containing printable characters or binaries containing printable characters in bytewise or UTF-8 encoding. But what is a printable character? One view is that anything the Unicode standard thinks is printable, is also printable according to the heuristic detection. The result is then that almost any list of integers are deemed a string, and all sorts of characters are printed, maybe also characters that your terminal lacks in its font set (resulting in some unappreciated generic output). Another way is to keep it backward compatible so that only the ISO Latin-1 character set is used to detect a string. A third way is to let the user decide exactly what Unicode ranges that are to be viewed as characters.

As from Erlang/OTP R16B you can select the ISO Latin-1 range or the whole Unicode range by supplying startup flag +pc latin1 or +pc unicode, respectively. For backward compatibility, latin1 is default. This only controls how heuristic string detection is done. More ranges are expected to be added in the future, enabling tailoring of the heuristics to the language and region relevant to the user.

The following examples show the two startup options:

$ erl +pc latin1
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)  
1> [1024].
[1024]
2> [1070,1085,1080,1082,1086,1076].
[1070,1085,1080,1082,1086,1076]
3> [229,228,246].
"åäö"
4> <<208,174,208,189,208,184,208,186,208,190,208,180>>.
<<208,174,208,189,208,184,208,186,208,190,208,180>>
5> <<229/utf8,228/utf8,246/utf8>>.
<<"åäö"/utf8>>
$ erl +pc unicode
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)  
1> [1024].
"Ѐ"
2> [1070,1085,1080,1082,1086,1076].
"Юникод"
3> [229,228,246].
"åäö"
4> <<208,174,208,189,208,184,208,186,208,190,208,180>>.
<<"Юникод"/utf8>>
5> <<229/utf8,228/utf8,246/utf8>>.
<<"åäö"/utf8>>

In the examples, you can see that the default Erlang shell interprets only characters from the ISO Latin1 range as printable and only detects lists or binaries with those "printable" characters as containing string data. The valid UTF-8 binary containing the Russian word "Юникод", is not printed as a string. When started with all Unicode characters printable (+pc unicode), the shell outputs anything containing printable Unicode data (in binaries, either UTF-8 or bytewise encoded) as string data.

These heuristics are also used by io:format/2, io_lib:format/2, and friends when modifier t is used with ~p or ~P:

$ erl +pc latin1
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)  
1> io:format("~tp~n",[{<<"åäö">>, <<"åäö"/utf8>>, <<208,174,208,189,208,184,208,186,208,190,208,180>>}]).
{<<"åäö">>,<<"åäö"/utf8>>,<<208,174,208,189,208,184,208,186,208,190,208,180>>}
ok
$ erl +pc unicode
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)  
1> io:format("~tp~n",[{<<"åäö">>, <<"åäö"/utf8>>, <<208,174,208,189,208,184,208,186,208,190,208,180>>}]).
{<<"åäö">>,<<"åäö"/utf8>>,<<"Юникод"/utf8>>}
ok

Notice that this only affects heuristic interpretation of lists and binaries on output. For example, the ~ts format sequence always outputs a valid list of characters, regardless of the +pc setting, as the programmer has explicitly requested string output.

The interactive Erlang shell can support Unicode input and output.

On Windows, proper operation requires that a suitable font is installed and selected for the Erlang application to use. If no suitable font is available on your system, try installing the DejaVu fonts, which are freely available, and then select that font in the Erlang shell application.

On Unix-like operating systems, the terminal is to be able to handle UTF-8 on input and output (this is done by, for example, modern versions of XTerm, KDE Konsole, and the Gnome terminal) and your locale settings must be proper. As an example, a LANG environment variable can be set as follows:

$ echo $LANG
en_US.UTF-8

Most systems handle variable LC_CTYPE before LANG, so if that is set, it must be set to UTF-8:

$ echo $LC_CTYPE
en_US.UTF-8

The LANG or LC_CTYPE setting are to be consistent with what the terminal is capable of. There is no portable way for Erlang to ask the terminal about its UTF-8 capacity, we have to rely on the language and character type settings.

To investigate what Erlang thinks about the terminal, the call io:getopts() can be used when the shell is started:

$ LC_CTYPE=en_US.ISO-8859-1 erl
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)
1> lists:keyfind(encoding, 1, io:getopts()).
{encoding,latin1}
2> q().
ok
$ LC_CTYPE=en_US.UTF-8 erl
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)
1> lists:keyfind(encoding, 1, io:getopts()).
{encoding,unicode}
2>

When (finally?) everything is in order with the locale settings, fonts. and the terminal emulator, you have probably found a way to input characters in the script you desire. For testing, the simplest way is to add some keyboard mappings for other languages, usually done with some applet in your desktop environment.

In a KDE environment, select KDE Control Center (Personal Settings) > Regional and Accessibility > Keyboard Layout.

On Windows XP, select Control Panel > Regional and Language Options, select tab Language, and click button Details... in the square named Text Services and Input Languages.

Your environment probably provides similar means of changing the keyboard layout. Ensure that you have a way to switch back and forth between keyboards easily if you are not used to this. For example, entering commands using a Cyrillic character set is not easily done in the Erlang shell.

Now you are set up for some Unicode input and output. The simplest thing to do is to enter a string in the shell:

$ erl
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)
1> lists:keyfind(encoding, 1, io:getopts()).
{encoding,unicode}
2> "Юникод".
"Юникод"
3> io:format("~ts~n", [v(2)]).
Юникод
ok
4>

While strings can be input as Unicode characters, the language elements are still limited to the ISO Latin-1 character set. Only character constants and strings are allowed to be beyond that range:

$ erl
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)
1> $ξ.
958
2> Юникод.
* 1: illegal character
2> 

When Erlang is started without an interactive shell (-noshell, -noinput or as an escript) the unicode support is identified using environment variables just as for interactive shells. Working with unicode in non-interactive sessions works just the same as for interactive sessions.

In some situations you may need to be able to read and write raw bytes from standard_io. If that is the case, then you want to set the standard_io_encoding configuration parameter to latin1 and use the file API to read and write data (as explained in Unicode Data in Files).

In the example below we first read the character ξ from standard_io and then print the charlist() represented by it.

#!/usr/bin/env escript
%%! -kernel standard_io_encoding latin1

main(_) ->
  {ok, Char} = file:read_line(standard_io),
  ok = file:write(standard_io, string:trim(Char)),
  ok = file:write(standard_io, io_lib:format(": ~w~n",[string:trim(Char)])),
  ok.
$ escript test.es
ξ
ξ: [206,190]

ξ would normally be represented as the integer 958, but since we are using bytewise encoding (latin1), it is represented by 206 and 190, which is the utf-8 bytes representing ξ. When we echo those bytes back to standard_io, the terminal will see the bytes as utf-8 and show the correct value even though in Erlang we never knew that it was indeed a unicode string.

Most modern operating systems support Unicode filenames in some way. There are many different ways to do this and Erlang by default treats the different approaches differently:

Windows, Android and, for most cases, MacOS X enforce Unicode support for filenames. All files created in the file system have names that can consistently be interpreted. In MacOS X and Android, all filenames are retrieved in UTF-8 encoding. In Windows, each system call handling filenames has a special Unicode-aware variant, giving much the same effect. There are no filenames on these systems that are not Unicode filenames. So, the default behavior of the Erlang VM is to work in "Unicode filename translation mode". This means that a filename can be specified as a Unicode list, which is automatically translated to the proper name encoding for the underlying operating system and file system.

Doing, for example, a file:list_dir/1 on one of these systems can return Unicode lists with code points > 255, depending on the content of the file system.

Most Unix operating systems have adopted a simpler approach, namely that Unicode file naming is not enforced, but by convention. Those systems usually use UTF-8 encoding for Unicode filenames, but do not enforce it. On such a system, a filename containing characters with code points from 128 through 255 can be named as plain ISO Latin-1 or use UTF-8 encoding. As no consistency is enforced, the Erlang VM cannot do consistent translation of all filenames.

By default on such systems, Erlang starts in utf8 filename mode if the terminal supports UTF-8, otherwise in latin1 mode.

In latin1 mode, filenames are bytewise encoded. This allows for list representation of all filenames in the system. However, a a file named "Östersund.txt", appears in file:list_dir/1 either as "Östersund.txt" (if the filename was encoded in bytewise ISO Latin-1 by the program creating the file) or more probably as [195,150,115,116,101,114,115,117,110,100], which is a list containing UTF-8 bytes (not what you want). If you use Unicode filename translation on such a system, non-UTF-8 filenames are ignored by functions like file:list_dir/1. They can be retrieved with function file:list_dir_all/1, but wrongly encoded filenames appear as "raw filenames".

The Unicode file naming support was introduced in Erlang/OTP R14B01. A VM operating in Unicode filename translation mode can work with files having names in any language or character set (as long as it is supported by the underlying operating system and file system). The Unicode character list is used to denote filenames or directory names. If the file system content is listed, you also get Unicode lists as return value. The support lies in the Kernel and STDLIB modules, which is why most applications (that do not explicitly require the filenames to be in the ISO Latin-1 range) benefit from the Unicode support without change.

On operating systems with mandatory Unicode filenames, this means that you more easily conform to the filenames of other (non-Erlang) applications. You can also process filenames that, at least on Windows, were inaccessible (because of having names that could not be represented in ISO Latin-1). Also, you avoid creating incomprehensible filenames on MacOS X, as the vfs layer of the operating system accepts all your filenames as UTF-8 does not rewrite them.

For most systems, turning on Unicode filename translation is no problem even if it uses transparent file naming. Very few systems have mixed filename encodings. A consistent UTF-8 named system works perfectly in Unicode filename mode. It was still, however, considered experimental in Erlang/OTP R14B01 and is still not the default on such systems.

Unicode filename translation is turned on with switch +fnu. On Linux, a VM started without explicitly stating the filename translation mode defaults to latin1 as the native filename encoding. On Windows, MacOS X and Android, the default behavior is that of Unicode filename translation. Therefore file:native_name_encoding/0 by default returns utf8 on those systems (Windows does not use UTF-8 on the file system level, but this can safely be ignored by the Erlang programmer). The default behavior can, as stated earlier, be changed using option +fnu or +fnl to the VM, see the erl program. If the VM is started in Unicode filename translation mode, file:native_name_encoding/0 returns atom utf8. Switch +fnu can be followed by w, i, or e to control how wrongly encoded filenames are to be reported.

  • w means that a warning is sent to the error_logger whenever a wrongly encoded filename is "skipped" in directory listings. w is the default.

  • i means that wrongly encoded filenames are silently ignored.

  • e means that the API function returns an error whenever a wrongly encoded filename (or directory name) is encountered.

Notice that file:read_link/1 always returns an error if the link points to an invalid filename.

In Unicode filename mode, filenames given to BIF open_port/2 with option {spawn_executable,...} are also interpreted as Unicode. So is the parameter list specified in option args available when using spawn_executable. The UTF-8 translation of arguments can be avoided using binaries, see section Notes About Raw Filenames.

Notice that the file encoding options specified when opening a file has nothing to do with the filename encoding convention. You can very well open files containing data encoded in UTF-8, but having filenames in bytewise (latin1) encoding or conversely.

Note

Erlang drivers and NIF-shared objects still cannot be named with names containing code points > 127. This limitation will be removed in a future release. However, Erlang modules can, but it is definitely not a good idea and is still considered experimental.

Note

Note that raw filenames not necessarily are encoded the same way as on the OS level.

Raw filenames were introduced together with Unicode filename support in ERTS 5.8.2 (Erlang/OTP R14B01). The reason "raw filenames" were introduced in the system was to be able to represent filenames, specified in different encodings on the same system, consistently. It can seem practical to have the VM automatically translate a filename that is not in UTF-8 to a list of Unicode characters, but this would open up for both duplicate filenames and other inconsistent behavior.

Consider a directory containing a file named "björn" in ISO Latin-1, while the Erlang VM is operating in Unicode filename mode (and therefore expects UTF-8 file naming). The ISO Latin-1 name is not valid UTF-8 and one can be tempted to think that automatic conversion in, for example, file:list_dir/1 is a good idea. But what would happen if we later tried to open the file and have the name as a Unicode list (magically converted from the ISO Latin-1 filename)? The VM converts the filename to UTF-8, as this is the encoding expected. Effectively this means trying to open the file named <<"björn"/utf8>>. This file does not exist, and even if it existed it would not be the same file as the one that was listed. We could even create two files named "björn", one named in UTF-8 encoding and one not. If file:list_dir/1 would automatically convert the ISO Latin-1 filename to a list, we would get two identical filenames as the result. To avoid this, we must differentiate between filenames that are properly encoded according to the Unicode file naming convention (that is, UTF-8) and filenames that are invalid under the encoding. By the common function file:list_dir/1, the wrongly encoded filenames are ignored in Unicode filename translation mode, but by function file:list_dir_all/1 the filenames with invalid encoding are returned as "raw" filenames, that is, as binaries.

The file module accepts raw filenames as input. open_port({spawn_executable, ...} ...) also accepts them. As mentioned earlier, the arguments specified in the option list to open_port({spawn_executable, ...} ...) undergo the same conversion as the filenames, meaning that the executable is provided with arguments in UTF-8 as well. This translation is avoided consistently with how the filenames are treated, by giving the argument as a binary.

To force Unicode filename translation mode on systems where this is not the default was considered experimental in Erlang/OTP R14B01. This was because the initial implementation did not ignore wrongly encoded filenames, so that raw filenames could spread unexpectedly throughout the system. As from Erlang/OTP R16B, the wrongly encoded filenames are only retrieved by special functions (such as file:list_dir_all/1). Since the impact on existing code is therefore much lower it is now supported. Unicode filename translation is expected to be default in future releases.

Even if you are operating without Unicode file naming translation automatically done by the VM, you can access and create files with names in UTF-8 encoding by using raw filenames encoded as UTF-8. Enforcing the UTF-8 encoding regardless of the mode the Erlang VM is started in can in some circumstances be a good idea, as the convention of using UTF-8 filenames is spreading.

The vfs layer of MacOS X enforces UTF-8 filenames in an aggressive way. Older versions did this by refusing to create non-UTF-8 conforming filenames, while newer versions replace offending bytes with the sequence "%HH", where HH is the original character in hexadecimal notation. As Unicode translation is enabled by default on MacOS X, the only way to come up against this is to either start the VM with flag +fnl or to use a raw filename in bytewise (latin1) encoding. If using a raw filename, with a bytewise encoding containing characters from 127 through 255, to create a file, the file cannot be opened using the same name as the one used to create it. There is no remedy for this behavior, except keeping the filenames in the correct encoding.

MacOS X reorganizes the filenames so that the representation of accents, and so on, uses the "combining characters". For example, character ö is represented as code points [111,776], where 111 is character o and 776 is the special accent character "Combining Diaeresis". This way of normalizing Unicode is otherwise very seldom used. Erlang normalizes those filenames in the opposite way upon retrieval, so that filenames using combining accents are not passed up to the Erlang application. In Erlang, filename "björn" is retrieved as [98,106,246,114,110], not as [98,106,117,776,114,110], although the file system can think differently. The normalization into combining accents is redone when accessing files, so this can usually be ignored by the Erlang programmer.

Environment variables and their interpretation are handled much in the same way as filenames. If Unicode filenames are enabled, environment variables as well as parameters to the Erlang VM are expected to be in Unicode.

If Unicode filenames are enabled, the calls to os:getenv/0,1, os:putenv/2, and os:unsetenv/1 handle Unicode strings. On Unix-like platforms, the built-in functions translate environment variables in UTF-8 to/from Unicode strings, possibly with code points > 255. On Windows, the Unicode versions of the environment system API are used, and code points > 255 are allowed.

On Unix-like operating systems, parameters are expected to be UTF-8 without translation if Unicode filenames are enabled.

Most of the modules in Erlang/OTP are Unicode-unaware in the sense that they have no notion of Unicode and should not have. Typically they handle non-textual or byte-oriented data (such as gen_tcp).

Modules handling textual data (such as io_lib and string are sometimes subject to conversion or extension to be able to handle Unicode characters.

Fortunately, most textual data has been stored in lists and range checking has been sparse, so modules like string work well for Unicode strings with little need for conversion or extension.

Some modules are, however, changed to be explicitly Unicode-aware. These modules include:

The unicode module is clearly Unicode-aware. It contains functions for conversion between different Unicode formats and some utilities for identifying byte order marks. Few programs handling Unicode data survive without this module.

The io module has been extended along with the actual I/O protocol to handle Unicode data. This means that many functions require binaries to be in UTF-8, and there are modifiers to format control sequences to allow for output of Unicode strings.

I/O-servers throughout the system can handle Unicode data and have options for converting data upon output or input to/from the device. As shown earlier, the shell module has support for Unicode terminals and the file module allows for translation to and from various Unicode formats on disk.

Reading and writing of files with Unicode data is, however, not best done with the file module, as its interface is byte-oriented. A file opened with a Unicode encoding (like UTF-8) is best read or written using the io module.

The re module allows for matching Unicode strings as a special option. As the library is centered on matching in binaries, the Unicode support is UTF-8-centered.

The graphical library wx has extensive support for Unicode text.

The string module works perfectly for Unicode strings and ISO Latin-1 strings, except the language-dependent functions string:uppercase/1 and string:lowercase/1. These two functions can never function correctly for Unicode characters in their current form, as there are language and locale issues to consider when converting text between cases. Converting case in an international environment is a large subject not yet addressed in OTP.

Although Erlang can handle Unicode data in many forms does not automatically mean that the content of any file can be Unicode text. The external entities, such as ports and I/O servers, are not generally Unicode capable.

Ports are always byte-oriented, so before sending data that you are not sure is bytewise-encoded to a port, ensure to encode it in a proper Unicode encoding. Sometimes this means that only part of the data must be encoded as, for example, UTF-8. Some parts can be binary data (like a length indicator) or something else that must not undergo character encoding, so no automatic translation is present.

I/O servers behave a little differently. The I/O servers connected to terminals (or stdout) can usually cope with Unicode data regardless of the encoding option. This is convenient when one expects a modern environment but do not want to crash when writing to an archaic terminal or pipe.

A file can have an encoding option that makes it generally usable by the io module (for example {encoding,utf8}), but is by default opened as a byte-oriented file. The file module is byte-oriented, so only ISO Latin-1 characters can be written using that module. Use the io module if Unicode data is to be output to a file with other encoding than latin1 (bytewise encoding). It is slightly confusing that a file opened with, for example, file:open(Name,[read,{encoding,utf8}]) cannot be properly read using file:read(File,N), but using the io module to retrieve the Unicode data from it. The reason is that file:read and file:write (and friends) are purely byte-oriented, and should be, as that is the way to access files other than text files, byte by byte. As with ports, you can write encoded data into a file by "manually" converting the data to the encoding of choice (using the unicode module or the bit syntax) and then output it on a bytewise (latin1) encoded file.

Recommendations:

  • Use the file module for files opened for bytewise access ({encoding,latin1}).

  • Use the io module when accessing files with any other encoding (for example {encoding,utf8}).

Functions reading Erlang syntax from files recognize the coding: comment and can therefore handle Unicode data on input. When writing Erlang terms to a file, you are advised to insert such comments when applicable:

$ erl +fna +pc unicode
Erlang R16B (erts-5.10.1) [source]  [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1  (abort with ^G)
1> file:write_file("test.term",<<"%% coding: utf-8\n[{\"Юникод\",4711}].\n"/utf8>>).
ok
2> file:consult("test.term").   
{ok,[[{"Юникод",4711}]]}

The Unicode support is controlled by both command-line switches, some standard environment variables, and the OTP version you are using. Most options affect mainly how Unicode data is displayed, not the functionality of the APIs in the standard libraries. This means that Erlang programs usually do not need to concern themselves with these options, they are more for the development environment. An Erlang program can be written so that it works well regardless of the type of system or the Unicode options that are in effect.

Here follows a summary of the settings affecting Unicode:

The language setting in the operating system mainly affects the shell. The terminal (that is, the group leader) operates with {encoding, unicode} only if the environment tells it that UTF-8 is allowed. This setting is to correspond to the terminal you are using.

The environment can also affect filename interpretation, if Erlang is started with flag +fna (which is default from Erlang/OTP 17.0).

You can check the setting of this by calling io:getopts(), which gives you an option list containing {encoding,unicode} or {encoding,latin1}.

This flag affects what is interpreted as string data when doing heuristic string detection in the shell and in io/ io_lib:format with the "~tp" and ~tP formatting instructions, as described earlier.

You can check this option by calling io:printable_range/0, which returns unicode or latin1. To be compatible with future (expected) extensions to the settings, rather use io_lib:printable_list/1 to check if a list is printable according to the setting. That function takes into account new possible settings returned from io:printable_range/0.

This flag affects how the filenames are to be interpreted. On operating systems with transparent file naming, this must be specified to allow for file naming in Unicode characters (and for correct interpretation of filenames containing characters > 255).

  • +fnl means bytewise interpretation of filenames, which was the usual way to represent ISO Latin-1 filenames before UTF-8 file naming got widespread.

  • +fnu means that filenames are encoded in UTF-8, which is nowadays the common scheme (although not enforced).

  • +fna means that you automatically select between +fnl and +fnu, based on environment variables LANG and LC_CTYPE. This is optimistic heuristics indeed, nothing enforces a user to have a terminal with the same encoding as the file system, but this is usually the case. This is the default on all Unix-like operating systems, except MacOS X.

The filename translation mode can be read with function file:native_name_encoding/0, which returns latin1 (bytewise encoding) or utf8.

This function returns the default encoding for Erlang source files (if no encoding comment is present) in the currently running release. In Erlang/OTP R16B, latin1 (bytewise encoding) was returned. As from Erlang/OTP 17.0, utf8 is returned.

The encoding of each file can be specified using comments as described in the epp(3) module.

When Erlang is started the encoding for standard_io is by default set to what the locale settings indicate. You can override the default by setting the kernel configuration parameter standard_io_encoding to the desired encoding.

You can set the encoding of a file or other I/O server with function io:setopts/2. This can also be set when opening a file. Setting the terminal (or other standard_io server) unconditionally to option {encoding,utf8} implies that UTF-8 encoded characters are written to the device, regardless of how Erlang was started or the user's environment.

Note

If you use io:setopts/2 to change the encoding of standard_io the I/O server may already have read some data using the default encoding. To avoid this you should set the encoding using standard_io_encoding.

Opening files with option encoding is convenient when writing or reading text files in a known encoding.

You can retrieve the encoding setting for an I/O server with function io:getopts().

When starting with Unicode, one often stumbles over some common issues. This section describes some methods of dealing with Unicode data.

A common method of identifying encoding in text files is to put a Byte Order Mark (BOM) first in the file. The BOM is the code point 16#FEFF encoded in the same way as the remaining file. If such a file is to be read, the first few bytes (depending on encoding) are not part of the text. This code outlines how to open a file that is believed to have a BOM, and sets the files encoding and position for further sequential reading (preferably using the io module).

Notice that error handling is omitted from the code:

open_bom_file_for_reading(File) ->
    {ok,F} = file:open(File,[read,binary]),
    {ok,Bin} = file:read(F,4),
    {Type,Bytes} = unicode:bom_to_encoding(Bin),
    file:position(F,Bytes),
    io:setopts(F,[{encoding,Type}]),
    {ok,F}.

Function unicode:bom_to_encoding/1 identifies the encoding from a binary of at least four bytes. It returns, along with a term suitable for setting the encoding of the file, the byte length of the BOM, so that the file position can be set accordingly. Notice that function file:position/2 always works on byte-offsets, so that the byte length of the BOM is needed.

To open a file for writing and place the BOM first is even simpler:

open_bom_file_for_writing(File,Encoding) ->
    {ok,F} = file:open(File,[write,binary]),
    ok = file:write(File,unicode:encoding_to_bom(Encoding)),
    io:setopts(F,[{encoding,Encoding}]),
    {ok,F}.

The file is in both these cases then best processed using the io module, as the functions in that module can handle code points beyond the ISO Latin-1 range.

When reading and writing to Unicode-aware entities, like a file opened for Unicode translation, you probably want to format text strings using the functions in the io module or the io_lib module. For backward compatibility reasons, these functions do not accept any list as a string, but require a special translation modifier when working with Unicode texts. The modifier is t. When applied to control character s in a formatting string, it accepts all Unicode code points and expects binaries to be in UTF-8:

1> io:format("~ts~n",[<<"åäö"/utf8>>]).
åäö
ok
2> io:format("~s~n",[<<"åäö"/utf8>>]).
åäö
ok

Clearly, the second io:format/2 gives undesired output, as the UTF-8 binary is not in latin1. For backward compatibility, the non-prefixed control character s expects bytewise-encoded ISO Latin-1 characters in binaries and lists containing only code points < 256.

As long as the data is always lists, modifier t can be used for any string, but when binary data is involved, care must be taken to make the correct choice of formatting characters. A bytewise-encoded binary is also interpreted as a string, and printed even when using ~ts, but it can be mistaken for a valid UTF-8 string. Avoid therefore using the ~ts control if the binary contains bytewise-encoded characters and not UTF-8.

Function io_lib:format/2 behaves similarly. It is defined to return a deep list of characters and the output can easily be converted to binary data for outputting on any device by a simple erlang:list_to_binary/1. When the translation modifier is used, the list can, however, contain characters that cannot be stored in one byte. The call to erlang:list_to_binary/1 then fails. However, if the I/O server you want to communicate with is Unicode-aware, the returned list can still be used directly:

$ erl +pc unicode
Erlang R16B (erts-5.10.1) [source] [async-threads:0] [hipe] [kernel-poll:false]

Eshell V5.10.1 (abort with ^G)
1> io_lib:format("~ts~n", ["Γιούνικοντ"]).
["Γιούνικοντ","\n"]
2> io:put_chars(io_lib:format("~ts~n", ["Γιούνικοντ"])).
Γιούνικοντ
ok

The Unicode string is returned as a Unicode list, which is recognized as such, as the Erlang shell uses the Unicode encoding (and is started with all Unicode characters considered printable). The Unicode list is valid input to function io:put_chars/2, so data can be output on any Unicode-capable device. If the device is a terminal, characters are output in format \x{H...} if encoding is latin1. Otherwise in UTF-8 (for the non-interactive terminal: "oldshell" or "noshell") or whatever is suitable to show the character properly (for an interactive terminal: the regular shell).

So, you can always send Unicode data to the standard_io device. Files, however, accept only Unicode code points beyond ISO Latin-1 if encoding is set to something else than latin1.

While it is strongly encouraged that the encoding of characters in binary data is known before processing, that is not always possible. On a typical Linux system, there is a mix of UTF-8 and ISO Latin-1 text files, and there are seldom any BOMs in the files to identify them.

UTF-8 is designed so that ISO Latin-1 characters with numbers beyond the 7-bit ASCII range are seldom considered valid when decoded as UTF-8. Therefore one can usually use heuristics to determine if a file is in UTF-8 or if it is encoded in ISO Latin-1 (one byte per character). The unicode module can be used to determine if data can be interpreted as UTF-8:

heuristic_encoding_bin(Bin) when is_binary(Bin) ->
    case unicode:characters_to_binary(Bin,utf8,utf8) of
	Bin ->
	    utf8;
	_ ->
	    latin1
    end.

If you do not have a complete binary of the file content, you can instead chunk through the file and check part by part. The return-tuple {incomplete,Decoded,Rest} from function unicode:characters_to_binary/1,2,3 comes in handy. The incomplete rest from one chunk of data read from the file is prepended to the next chunk and we therefore avoid the problem of character boundaries when reading chunks of bytes in UTF-8 encoding:

heuristic_encoding_file(FileName) ->
    {ok,F} = file:open(FileName,[read,binary]),
    loop_through_file(F,<<>>,file:read(F,1024)).

loop_through_file(_,<<>>,eof) ->
    utf8;
loop_through_file(_,_,eof) ->
    latin1;
loop_through_file(F,Acc,{ok,Bin}) when is_binary(Bin) ->
    case unicode:characters_to_binary([Acc,Bin]) of
	{error,_,_} ->
	    latin1;
	{incomplete,_,Rest} ->
	    loop_through_file(F,Rest,file:read(F,1024));
	Res when is_binary(Res) ->
	    loop_through_file(F,<<>>,file:read(F,1024))
    end.

Another option is to try to read the whole file in UTF-8 encoding and see if it fails. Here we need to read the file using function io:get_chars/3, as we have to read characters with a code point > 255:

heuristic_encoding_file2(FileName) ->
    {ok,F} = file:open(FileName,[read,binary,{encoding,utf8}]),
    loop_through_file2(F,io:get_chars(F,'',1024)).

loop_through_file2(_,eof) ->
    utf8;
loop_through_file2(_,{error,_Err}) ->
    latin1;
loop_through_file2(F,Bin) when is_binary(Bin) ->
    loop_through_file2(F,io:get_chars(F,'',1024)).

For various reasons, you can sometimes have a list of UTF-8 bytes. This is not a regular string of Unicode characters, as each list element does not contain one character. Instead you get the "raw" UTF-8 encoding that you have in binaries. This is easily converted to a proper Unicode string by first converting byte per byte into a binary, and then converting the binary of UTF-8 encoded characters back to a Unicode string:

utf8_list_to_string(StrangeList) ->
  unicode:characters_to_list(list_to_binary(StrangeList)).

When working with binaries, you can get the horrible "double UTF-8 encoding", where strange characters are encoded in your binaries or files. In other words, you can get a UTF-8 encoded binary that for the second time is encoded as UTF-8. A common situation is where you read a file, byte by byte, but the content is already UTF-8. If you then convert the bytes to UTF-8, using, for example, the unicode module, or by writing to a file opened with option {encoding,utf8}, you have each byte in the input file encoded as UTF-8, not each character of the original text (one character can have been encoded in many bytes). There is no real remedy for this other than to be sure of which data is encoded in which format, and never convert UTF-8 data (possibly read byte by byte from a file) into UTF-8 again.

By far the most common situation where this occurs, is when you get lists of UTF-8 instead of proper Unicode strings, and then convert them to UTF-8 in a binary or on a file:

wrong_thing_to_do() ->
  {ok,Bin} = file:read_file("an_utf8_encoded_file.txt"),
  MyList = binary_to_list(Bin), %% Wrong! It is an utf8 binary!
  {ok,C} = file:open("catastrophe.txt",[write,{encoding,utf8}]), 
  io:put_chars(C,MyList), %% Expects a Unicode string, but get UTF-8
                          %% bytes in a list!
  file:close(C). %% The file catastrophe.txt contains more or less unreadable
                 %% garbage!

Ensure you know what a binary contains before converting it to a string. If no other option exists, try heuristics:

if_you_can_not_know() ->
  {ok,Bin} = file:read_file("maybe_utf8_encoded_file.txt"),
  MyList = case unicode:characters_to_list(Bin) of
    L when is_list(L) ->
      L;
    _ ->
      binary_to_list(Bin) %% The file was bytewise encoded
  end,
  %% Now we know that the list is a Unicode string, not a list of UTF-8 bytes
  {ok,G} = file:open("greatness.txt",[write,{encoding,utf8}]), 
  io:put_chars(G,MyList), %% Expects a Unicode string, which is what it gets!
  file:close(G). %% The file contains valid UTF-8 encoded Unicode characters!