Non-blocking streaming Unicode codec.
Uutf is a non-blocking streaming codec to decode and encode the UTF-8, UTF-16, UTF-16LE and UTF-16BE encoding schemes. It can efficiently work character by character without blocking on IO. Decoders perform character position tracking and support newline normalization.
Functions are also provided to fold over the characters of UTF encoded OCaml string values and to directly encode characters in OCaml
Stdlib.Buffer.t values. Note that since OCaml 4.14, that functionality can be found in
Stdlib.Buffer and you are encouraged to migrate to it.
See examples of use.
- The Unicode Consortium. The Unicode Standard. (latest version)
val u_bom : Uchar.t
The type for Unicode encoding schemes.
type decoder_encoding = [
val encoding_of_string : string -> decoder_encoding option
encoding_of_string s converts a (case insensitive) IANA character set name to an encoding.
val encoding_to_string : [< decoder_encoding ] -> string
encoding_to_string e is a IANA character set name for
type src = [
`Channel of in_channel
`String of string
The type for input sources. With a
`Manual source the client must provide input with
The type for newline normalizations. The variant argument is the normalization character.
`ASCII, normalizes CR (
U+000D), LF (
U+000A) and CRLF (<
`NLF, normalizes the Unicode newline function (NLF). This is NEL (
U+0085) and the normalizations of
`Readline, normalizes for a Unicode readline function. This is FF (
U+000C), LS (
U+2028), PS (
U+2029), and the normalizations of
Used with an appropriate normalization character the
`Readline normalizations allow to implement all the different recommendations of Unicode's newline guidelines (section 5.8 in Unicode 9.0.0).
decoder nln encoding src is a decoder that inputs from
Byte order mark. Byte order mark (BOM) constraints are application dependent and prone to misunderstandings (see the FAQ). Hence,
Uutf decoders have a simple rule: an initial BOM is always removed from the input and not counted in character position tracking. The function
decoder_removed_bom does however return
true if a BOM was removed so that all the information can be recovered if needed.
For UTF-16BE and UTF-16LE the above rule is a violation of conformance D96 and D97 of the standard.
Uutf favors the idea that if there's a BOM, decoding with
`UTF_16 or the
`UTF_16XX corresponding to the BOM should decode the same character sequence (this is not the case if you stick to the standard). The client can however regain conformance by consulting the result of
decoder_removed_bom and take appropriate action.
encoding specifies the decoded encoding scheme. If
`UTF_16 is used the endianness is determined according to the standard: from a BOM if there is one,
encoding is unspecified it is guessed. The result of a guess can only be
`UTF_16LE. The heuristic looks at the first three bytes of input (or less if impossible) and takes the first matching byte pattern in the table below.
xx = any byte .. = any byte or no byte (input too small) pp = positive byte uu = valid UTF-8 first byte Bytes | Guess | Rationale ---------+-----------+----------------------------------------------- EF BB BF | `UTF_8 | UTF-8 BOM FE FF .. | `UTF_16BE | UTF-16BE BOM FF FE .. | `UTF_16LE | UTF-16LE BOM 00 pp .. | `UTF_16BE | ASCII UTF-16BE and U+0000 is often forbidden pp 00 .. | `UTF_16LE | ASCII UTF-16LE and U+0000 is often forbidden uu .. .. | `UTF_8 | ASCII UTF-8 or valid UTF-8 first byte. xx xx .. | `UTF_16BE | Not UTF-8 => UTF-16, no BOM => UTF-16BE .. .. .. | `UTF_8 | Single malformed UTF-8 byte or no input.
This heuristic is compatible both with BOM based recognitition and JSON-like encoding recognition that relies on ASCII being present at the beginning of the stream. Also,
decoder_removed_bom will tell the client if the guess was BOM based.
Newline normalization. If
nln is specified, the given newline normalization is performed, see
nln. Otherwise all newlines are returned as found in the input.
Character position. The line number, column number, byte count and character count of the last decoded character (including
`Malformed ones) are respectively returned by
decoder_count. Before the first call to
decode the line number is
1 and the column is
`Malformed increments the column until a newline. On a newline, the line number is incremented and the column set to zero. For example the line is
2 and column
0 after the first newline was decoded. This can be understood as if
decode was moving an insertion point to the right in the data. A newline is anything normalized by
Uutf assumes that each Unicode scalar value has a column width of 1. The same assumption may not be made by the display program (e.g. for
emacs' compilation mode you need to set
nil). The problem is in general difficult to solve without interaction or convention with the display program's rendering engine. Depending on the context better column increments can be implemented by using
Uucp.Break.tty_width_hint or grapheme cluster boundaries (see
decode d is:
`Manualinput source and awaits for more input. The client must use
Manual.srcto provide it.
`Uchar uif a Unicode scalar value
`Endif the end of input was reached.
`Malformed bytesif the
bytessequence is malformed according to the decoded encoding scheme. If you are interested in a best-effort decoding you can still continue to decode after an error until the decoder synchronizes again on valid bytes. It may however be a good idea to signal the malformed characters by adding an
u_repcharacter to the parsed data, see the examples.
Note. Repeated invocation always eventually returns
`End, even in case of errors.
decoder_encoding d is
d's the decoded encoding scheme of
Warning. If the decoder guesses the encoding or uses
`UTF_16, rely on this value only after the first
`Uchar was decoded.
val decoder_line : decoder -> int
decoder_line d is the line number of the last decoded (or malformed) character. See
decoder for details.
val decoder_col : decoder -> int
decoder_col d is the column number of the last decoded (or malformed) character. See
decoder for details.
val decoder_byte_count : decoder -> int
decoder_byte_count d is the number of bytes already decoded on
d (including malformed ones). This is the last
decode's end byte offset counting from the beginning of the stream.
val decoder_count : decoder -> int
decoder_count d is the number of characters already decoded on
d (including malformed ones). See
decoder for details.
val decoder_removed_bom : decoder -> bool
val pp_decode : Format.formatter -> [< `Await | `Uchar of Uchar.t | `End | `Malformed of string ] -> unit
pp_decode ppf v prints an unspecified representation of
The type for output destinations. With a
`Manual destination the client must provide output storage with
encoder encoding dst is an encoder for
encoding that outputs to
Note. No initial BOM is encoded. If needed, this duty is left to the client.
encode e v is :
`Manualdestination and needs more output storage. The client must use
Manual.dstto provide a new buffer and then call
`Okwhen the encoder is ready to encode a new
`Manual destination, encoding
`End always returns
`Partial, the client should continue as usual with
`Ok is returned at which point
e is guaranteed to be the size of the last provided buffer (i.e. nothing was written).
Invalid_argument if an
`End is encoded after a
module Manual : sig ... end
Manual sources and destinations.
module String : sig ... end
Fold over the characters of UTF encoded OCaml
module Buffer : sig ... end
UTF encode characters in OCaml
The value of
lines src is the list of lines in
src as UTF-8 encoded OCaml strings. Line breaks are determined according to the recommendation R4 for a
readline function in section 5.8 of Unicode 9.0.0. If a decoding error occurs we silently replace the malformed sequence by the replacement character
u_rep and continue.
let lines ?encoding (src : [`Channel of in_channel | `String of string]) = let rec loop d buf acc = match Uutf.decode d with | `Uchar u -> begin match Uchar.to_int u with | 0x000A -> let line = Buffer.contents buf in Buffer.clear buf; loop d buf (line :: acc) | _ -> Uutf.Buffer.add_utf_8 buf u; loop d buf acc end | `End -> List.rev (Buffer.contents buf :: acc) | `Malformed _ -> Uutf.Buffer.add_utf_8 buf Uutf.u_rep; loop d buf acc | `Await -> assert false in let nln = `Readline (Uchar.of_int 0x000A) in loop (Uutf.decoder ~nln ?encoding src) (Buffer.create 512) 
lines_fd does the same but on a Unix file descriptor.
let lines_fd ?encoding (fd : Unix.file_descr) = let rec loop fd s d buf acc = match Uutf.decode d with | `Uchar u -> begin match Uchar.to_int u with | 0x000A -> let line = Buffer.contents buf in Buffer.clear buf; loop fd s d buf (line :: acc) | _ -> Uutf.Buffer.add_utf_8 buf u; loop fd s d buf acc end | `End -> List.rev (Buffer.contents buf :: acc) | `Malformed _ -> Uutf.Buffer.add_utf_8 buf Uutf.u_rep; loop fd s d buf acc | `Await -> let rec unix_read fd s j l = try Unix.read fd s j l with | Unix.Unix_error (Unix.EINTR, _, _) -> unix_read fd s j l in let rc = unix_read fd s 0 (Bytes.length s) in Uutf.Manual.src d s 0 rc; loop fd s d buf acc in let s = Bytes.create 65536 (* UNIX_BUFFER_SIZE in 4.0.0 *) in let nln = `Readline (Uchar.of_int 0x000A) in loop fd s (Uutf.decoder ~nln ?encoding `Manual) (Buffer.create 512) 
The result of
recode src out_encoding dst has the characters of
src written on
dst with encoding
out_encoding. If a decoding error occurs we silently replace the malformed sequence by the replacement character
u_rep and continue. Note that we don't add an initial BOM to
dst, recoding will thus loose the initial BOM
src may have. Whether this is a problem or not depends on the context.
let recode ?nln ?encoding out_encoding (src : [`Channel of in_channel | `String of string]) (dst : [`Channel of out_channel | `Buffer of Buffer.t]) = let rec loop d e = match Uutf.decode d with | `Uchar _ as u -> ignore (Uutf.encode e u); loop d e | `End -> ignore (Uutf.encode e `End) | `Malformed _ -> ignore (Uutf.encode e (`Uchar Uutf.u_rep)); loop d e | `Await -> assert false in let d = Uutf.decoder ?nln ?encoding src in let e = Uutf.encoder out_encoding dst in loop d e
recode_fd does the same but between Unix file descriptors.
let recode_fd ?nln ?encoding out_encoding (fdi : Unix.file_descr) (fdo : Unix.file_descr) = let rec encode fd s e v = match Uutf.encode e v with `Ok -> () | `Partial -> let rec unix_write fd s j l = let rec write fd s j l = try Unix.single_write fd s j l with | Unix.Unix_error (Unix.EINTR, _, _) -> write fd s j l in let wc = write fd s j l in if wc < l then unix_write fd s (j + wc) (l - wc) else () in unix_write fd s 0 (Bytes.length s - Uutf.Manual.dst_rem e); Uutf.Manual.dst e s 0 (Bytes.length s); encode fd s e `Await in let rec loop fdi fdo ds es d e = match Uutf.decode d with | `Uchar _ as u -> encode fdo es e u; loop fdi fdo ds es d e | `End -> encode fdo es e `End | `Malformed _ -> encode fdo es e (`Uchar Uutf.u_rep); loop fdi fdo ds es d e | `Await -> let rec unix_read fd s j l = try Unix.read fd s j l with | Unix.Unix_error (Unix.EINTR, _, _) -> unix_read fd s j l in let rc = unix_read fdi ds 0 (Bytes.length ds) in Uutf.Manual.src d ds 0 rc; loop fdi fdo ds es d e in let ds = Bytes.create 65536 (* UNIX_BUFFER_SIZE in 4.0.0 *) in let es = Bytes.create 65536 (* UNIX_BUFFER_SIZE in 4.0.0 *) in let d = Uutf.decoder ?nln ?encoding `Manual in let e = Uutf.encoder out_encoding `Manual in Uutf.Manual.dst e es 0 (Bytes.length es); loop fdi fdo ds es d e