;; utf8 conversion red from win1252 codepage

_utf8.red

Red[]
;; read win1252 textfile , convert to utf8 text string / lines with _myread %filename
;;#include %../inc/_utf8.red
_m: []
insert/dup _m #{} 255

;-----------------------------------------
_ini: function [/extern _m ][
;-----------------------------------------
;; erzeugen ut8 map dec -> utf8
bin: debase {4oKswoHigJrGkuKAnuKApuKAoOKAocuG4oCwxaDigLnFksKNxb3Cj8KQ4oCY4oCZ4oCc4oCd4oCi4oCT4oCUy5zihKLFoeKAusWTwp3FvsW4wqDCocKiwqPCpMKlwqbCp8KowqnCqsKrwqzCrcKuwq/CsMKxwrLCs8K0wrXCtsK3wrjCucK6wrvCvMK9wr7Cv8OAw4HDgsODw4TDhcOGw4fDiMOJw4rDi8OMw43DjsOPw5DDkcOSw5PDlMOVw5bDl8OYw5nDmsObw5zDncOew5/DoMOhw6LDo8Okw6XDpsOnw6jDqcOqw6vDrMOtw67Dr8Oww7HDssOzw7TDtcO2w7fDuMO5w7rDu8O8w73DvsO/}

ut8: append #{} first bin
ind: 128

foreach ele (skip bin 1) [
 if (ele >= to-integer #{c0}) [
    _m/:ind: copy ut8
    ind: ind + 1
    clear ut8
  ]
  append ut8 ele 
]
    _m/:ind: ut8
]

;-----------------------------------------
_myread: function [fi /lines /extern _m  ][
;-----------------------------------------
  M40: 1024 * 1024 * 40
  if M40 < (size? fi) [print "Error: file > 40M !!" halt]
  bin: read/binary fi
  bout: #{}
      
  foreach ele bin [  either ele < 128 [append bout ele][append bout _m/:ele]   ]
  bin: to-string bout
  
  if lines [return split bin newline]
  return bin
]
_ini