detect-byte-array ( bytes -- encoding )

Vocabulary

io.encodings.detect

Inputs

bytes

an object

Outputs

encoding

an object

Definition

USING: combinators io.encodings.binary
io.encodings.detect.private io.encodings.utf16
io.encodings.utf32 io.encodings.utf8 kernel namespaces sequences
;

IN: io.encodings.detect

: detect-byte-array ( bytes -- encoding )
    {
        { [ dup B{ 0 0 254 255 } head? ] [ drop utf32be ] }
        { [ dup B{ 255 254 0 0 } head? ] [ drop utf32le ] }
        { [ dup B{ 254 255 } head? ] [ drop utf16be ] }
        { [ dup B{ 255 254 } head? ] [ drop utf16le ] }
        { [ dup B{ 239 187 191 } head? ] [ drop utf8 ] }
        {
            [ dup B{ 60 63 120 109 108 } head? ]
            [ detect-xml-prolog ]
        }
        { [ 0 over member? ] [ drop binary ] }
        { [ dup empty? ] [ drop utf8 ] }
        { [ dup valid-utf8? ] [ drop utf8 ] }
        [ drop default-encoding get ]
    } cond ;