enc-dos.lisp

开源跨平台Lisp编译器

  (#xF6 #x00F7) ; DIVISION SIGN
  (#xF7 #x00B8) ; CEDILLA
  (#xF8 #x00B0) ; DEGREE SIGN
  (#xF9 #x00A8) ; DIAERESIS
  (#xFA #x00B7) ; MIDDLE DOT
  (#xFB #x00B9) ; SUPERSCRIPT ONE
  (#xFC #x00B3) ; SUPERSCRIPT THREE
  (#xFD #x00B2) ; SUPERSCRIPT TWO
  (#xFE #x25A0) ; BLACK SQUARE
  (#xFF #x00A0) ; NO-BREAK SPACE
)

(declaim (inline get-cp850-bytes))
(defun get-cp850-bytes (string pos)
  (declare (optimize speed (safety 0))
           (type simple-string string)
           (type array-range pos))
  (get-latin-bytes #'code->cp850-mapper :cp850 string pos))

(defun string->cp850 (string sstart send null-padding)
  (declare (optimize speed (safety 0))
           (type simple-string string)
           (type array-range sstart send))
  (values (string->latin% string sstart send #'get-cp850-bytes null-padding)))

(defmacro define-cp850->string* (accessor type)
  (declare (ignore type))
  (let ((name (make-od-name 'cp850->string* accessor)))
    `(progn
      (defun ,name (string sstart send array astart aend)
        (,(make-od-name 'latin->string* accessor) string sstart send array astart aend #'cp850->code-mapper)))))

(instantiate-octets-definition define-cp850->string*)

(defmacro define-cp850->string (accessor type)
  (declare (ignore type))
  `(defun ,(make-od-name 'cp850->string accessor) (array astart aend)
    (,(make-od-name 'latin->string accessor) array astart aend #'cp850->code-mapper)))

(instantiate-octets-definition define-cp850->string)

(add-external-format-funs '(:cp850 :|cp850|)
                          '(cp850->string-aref string->cp850))

(define-external-format (:cp850 :|cp850|)
    1 t
    (let ((cp850-byte (code->cp850-mapper bits)))
      (if cp850-byte
          (setf (sap-ref-8 sap tail) cp850-byte)
          (external-format-encoding-error stream bits)))
    (let ((code (cp850->code-mapper byte)))
      (if code
          (code-char code)
          (external-format-decoding-error stream byte)))) ;; TODO -- error check

(define-unibyte-mapper cp852->code-mapper code->cp852-mapper
  (#x80 #x00C7) ; LATIN CAPITAL LETTER C WITH CEDILLA
  (#x81 #x00FC) ; LATIN SMALL LETTER U WITH DIAERESIS
  (#x82 #x00E9) ; LATIN SMALL LETTER E WITH ACUTE
  (#x83 #x00E2) ; LATIN SMALL LETTER A WITH CIRCUMFLEX
  (#x84 #x00E4) ; LATIN SMALL LETTER A WITH DIAERESIS
  (#x85 #x016F) ; LATIN SMALL LETTER U WITH RING ABOVE
  (#x86 #x0107) ; LATIN SMALL LETTER C WITH ACUTE
  (#x87 #x00E7) ; LATIN SMALL LETTER C WITH CEDILLA
  (#x88 #x0142) ; LATIN SMALL LETTER L WITH STROKE
  (#x89 #x00EB) ; LATIN SMALL LETTER E WITH DIAERESIS
  (#x8A #x0150) ; LATIN CAPITAL LETTER O WITH DOUBLE ACUTE
  (#x8B #x0151) ; LATIN SMALL LETTER O WITH DOUBLE ACUTE
  (#x8C #x00EE) ; LATIN SMALL LETTER I WITH CIRCUMFLEX
  (#x8D #x0179) ; LATIN CAPITAL LETTER Z WITH ACUTE
  (#x8E #x00C4) ; LATIN CAPITAL LETTER A WITH DIAERESIS
  (#x8F #x0106) ; LATIN CAPITAL LETTER C WITH ACUTE
  (#x90 #x00C9) ; LATIN CAPITAL LETTER E WITH ACUTE
  (#x91 #x0139) ; LATIN CAPITAL LETTER L WITH ACUTE
  (#x92 #x013A) ; LATIN SMALL LETTER L WITH ACUTE
  (#x93 #x00F4) ; LATIN SMALL LETTER O WITH CIRCUMFLEX
  (#x94 #x00F6) ; LATIN SMALL LETTER O WITH DIAERESIS
  (#x95 #x013D) ; LATIN CAPITAL LETTER L WITH CARON
  (#x96 #x013E) ; LATIN SMALL LETTER L WITH CARON
  (#x97 #x015A) ; LATIN CAPITAL LETTER S WITH ACUTE
  (#x98 #x015B) ; LATIN SMALL LETTER S WITH ACUTE
  (#x99 #x00D6) ; LATIN CAPITAL LETTER O WITH DIAERESIS
  (#x9A #x00DC) ; LATIN CAPITAL LETTER U WITH DIAERESIS
  (#x9B #x0164) ; LATIN CAPITAL LETTER T WITH CARON
  (#x9C #x0165) ; LATIN SMALL LETTER T WITH CARON
  (#x9D #x0141) ; LATIN CAPITAL LETTER L WITH STROKE
  (#x9E #x00D7) ; MULTIPLICATION SIGN
  (#x9F #x010D) ; LATIN SMALL LETTER C WITH CARON
  (#xA0 #x00E1) ; LATIN SMALL LETTER A WITH ACUTE
  (#xA1 #x00ED) ; LATIN SMALL LETTER I WITH ACUTE
  (#xA2 #x00F3) ; LATIN SMALL LETTER O WITH ACUTE
  (#xA3 #x00FA) ; LATIN SMALL LETTER U WITH ACUTE
  (#xA4 #x0104) ; LATIN CAPITAL LETTER A WITH OGONEK
  (#xA5 #x0105) ; LATIN SMALL LETTER A WITH OGONEK
  (#xA6 #x017D) ; LATIN CAPITAL LETTER Z WITH CARON
  (#xA7 #x017E) ; LATIN SMALL LETTER Z WITH CARON
  (#xA8 #x0118) ; LATIN CAPITAL LETTER E WITH OGONEK
  (#xA9 #x0119) ; LATIN SMALL LETTER E WITH OGONEK
  (#xAA #x00AC) ; NOT SIGN
  (#xAB #x017A) ; LATIN SMALL LETTER Z WITH ACUTE
  (#xAC #x010C) ; LATIN CAPITAL LETTER C WITH CARON
  (#xAD #x015F) ; LATIN SMALL LETTER S WITH CEDILLA
  (#xAE #x00AB) ; LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
  (#xAF #x00BB) ; RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
  (#xB0 #x2591) ; LIGHT SHADE
  (#xB1 #x2592) ; MEDIUM SHADE
  (#xB2 #x2593) ; DARK SHADE
  (#xB3 #x2502) ; BOX DRAWINGS LIGHT VERTICAL
  (#xB4 #x2524) ; BOX DRAWINGS LIGHT VERTICAL AND LEFT
  (#xB5 #x00C1) ; LATIN CAPITAL LETTER A WITH ACUTE
  (#xB6 #x00C2) ; LATIN CAPITAL LETTER A WITH CIRCUMFLEX
  (#xB7 #x011A) ; LATIN CAPITAL LETTER E WITH CARON
  (#xB8 #x015E) ; LATIN CAPITAL LETTER S WITH CEDILLA
  (#xB9 #x2563) ; BOX DRAWINGS DOUBLE VERTICAL AND LEFT
  (#xBA #x2551) ; BOX DRAWINGS DOUBLE VERTICAL
  (#xBB #x2557) ; BOX DRAWINGS DOUBLE DOWN AND LEFT
  (#xBC #x255D) ; BOX DRAWINGS DOUBLE UP AND LEFT
  (#xBD #x017B) ; LATIN CAPITAL LETTER Z WITH DOT ABOVE
  (#xBE #x017C) ; LATIN SMALL LETTER Z WITH DOT ABOVE
  (#xBF #x2510) ; BOX DRAWINGS LIGHT DOWN AND LEFT
  (#xC0 #x2514) ; BOX DRAWINGS LIGHT UP AND RIGHT
  (#xC1 #x2534) ; BOX DRAWINGS LIGHT UP AND HORIZONTAL
  (#xC2 #x252C) ; BOX DRAWINGS LIGHT DOWN AND HORIZONTAL
  (#xC3 #x251C) ; BOX DRAWINGS LIGHT VERTICAL AND RIGHT
  (#xC4 #x2500) ; BOX DRAWINGS LIGHT HORIZONTAL
  (#xC5 #x253C) ; BOX DRAWINGS LIGHT VERTICAL AND HORIZONTAL
  (#xC6 #x0102) ; LATIN CAPITAL LETTER A WITH BREVE
  (#xC7 #x0103) ; LATIN SMALL LETTER A WITH BREVE
  (#xC8 #x255A) ; BOX DRAWINGS DOUBLE UP AND RIGHT
  (#xC9 #x2554) ; BOX DRAWINGS DOUBLE DOWN AND RIGHT
  (#xCA #x2569) ; BOX DRAWINGS DOUBLE UP AND HORIZONTAL
  (#xCB #x2566) ; BOX DRAWINGS DOUBLE DOWN AND HORIZONTAL
  (#xCC #x2560) ; BOX DRAWINGS DOUBLE VERTICAL AND RIGHT
  (#xCD #x2550) ; BOX DRAWINGS DOUBLE HORIZONTAL
  (#xCE #x256C) ; BOX DRAWINGS DOUBLE VERTICAL AND HORIZONTAL
  (#xCF #x00A4) ; CURRENCY SIGN
  (#xD0 #x0111) ; LATIN SMALL LETTER D WITH STROKE
  (#xD1 #x0110) ; LATIN CAPITAL LETTER D WITH STROKE
  (#xD2 #x010E) ; LATIN CAPITAL LETTER D WITH CARON
  (#xD3 #x00CB) ; LATIN CAPITAL LETTER E WITH DIAERESIS
  (#xD4 #x010F) ; LATIN SMALL LETTER D WITH CARON
  (#xD5 #x0147) ; LATIN CAPITAL LETTER N WITH CARON
  (#xD6 #x00CD) ; LATIN CAPITAL LETTER I WITH ACUTE
  (#xD7 #x00CE) ; LATIN CAPITAL LETTER I WITH CIRCUMFLEX
  (#xD8 #x011B) ; LATIN SMALL LETTER E WITH CARON
  (#xD9 #x2518) ; BOX DRAWINGS LIGHT UP AND LEFT
  (#xDA #x250C) ; BOX DRAWINGS LIGHT DOWN AND RIGHT
  (#xDB #x2588) ; FULL BLOCK
  (#xDC #x2584) ; LOWER HALF BLOCK
  (#xDD #x0162) ; LATIN CAPITAL LETTER T WITH CEDILLA
  (#xDE #x016E) ; LATIN CAPITAL LETTER U WITH RING ABOVE
  (#xDF #x2580) ; UPPER HALF BLOCK
  (#xE0 #x00D3) ; LATIN CAPITAL LETTER O WITH ACUTE
  (#xE1 #x00DF) ; LATIN SMALL LETTER SHARP S
  (#xE2 #x00D4) ; LATIN CAPITAL LETTER O WITH CIRCUMFLEX
  (#xE3 #x0143) ; LATIN CAPITAL LETTER N WITH ACUTE
  (#xE4 #x0144) ; LATIN SMALL LETTER N WITH ACUTE
  (#xE5 #x0148) ; LATIN SMALL LETTER N WITH CARON
  (#xE6 #x0160) ; LATIN CAPITAL LETTER S WITH CARON
  (#xE7 #x0161) ; LATIN SMALL LETTER S WITH CARON
  (#xE8 #x0154) ; LATIN CAPITAL LETTER R WITH ACUTE
  (#xE9 #x00DA) ; LATIN CAPITAL LETTER U WITH ACUTE
  (#xEA #x0155) ; LATIN SMALL LETTER R WITH ACUTE
  (#xEB #x0170) ; LATIN CAPITAL LETTER U WITH DOUBLE ACUTE
  (#xEC #x00FD) ; LATIN SMALL LETTER Y WITH ACUTE
  (#xED #x00DD) ; LATIN CAPITAL LETTER Y WITH ACUTE
  (#xEE #x0163) ; LATIN SMALL LETTER T WITH CEDILLA
  (#xEF #x00B4) ; ACUTE ACCENT
  (#xF0 #x00AD) ; SOFT HYPHEN
  (#xF1 #x02DD) ; DOUBLE ACUTE ACCENT
  (#xF2 #x02DB) ; OGONEK
  (#xF3 #x02C7) ; CARON
  (#xF4 #x02D8) ; BREVE
  (#xF5 #x00A7) ; SECTION SIGN
  (#xF6 #x00F7) ; DIVISION SIGN
  (#xF7 #x00B8) ; CEDILLA
  (#xF8 #x00B0) ; DEGREE SIGN
  (#xF9 #x00A8) ; DIAERESIS
  (#xFA #x02D9) ; DOT ABOVE
  (#xFB #x0171) ; LATIN SMALL LETTER U WITH DOUBLE ACUTE
  (#xFC #x0158) ; LATIN CAPITAL LETTER R WITH CARON
  (#xFD #x0159) ; LATIN SMALL LETTER R WITH CARON
  (#xFE #x25A0) ; BLACK SQUARE
  (#xFF #x00A0) ; NO-BREAK SPACE
)

(declaim (inline get-cp852-bytes))
(defun get-cp852-bytes (string pos)
  (declare (optimize speed (safety 0))
           (type simple-string string)
           (type array-range pos))
  (get-latin-bytes #'code->cp852-mapper :cp852 string pos))

(defun string->cp852 (string sstart send null-padding)
  (declare (optimize speed (safety 0))
           (type simple-string string)
           (type array-range sstart send))
  (values (string->latin% string sstart send #'get-cp852-bytes null-padding)))

(defmacro define-cp852->string* (accessor type)
  (declare (ignore type))
  (let ((name (make-od-name 'cp852->string* accessor)))
    `(progn
      (defun ,name (string sstart send array astart aend)
        (,(make-od-name 'latin->string* accessor) string sstart send array astart aend #'cp852->code-mapper)))))

(instantiate-octets-definition define-cp852->string*)

(defmacro define-cp852->string (accessor type)
  (declare (ignore type))
  `(defun ,(make-od-name 'cp852->string accessor) (array astart aend)
    (,(make-od-name 'latin->string accessor) array astart aend #'cp852->code-mapper)))

(instantiate-octets-definition define-cp852->string)

(add-external-format-funs '(:cp852 :|cp852|)
                          '(cp852->string-aref string->cp852))

(define-external-format (:cp852 :|cp852|)
    1 t
    (let ((cp852-byte (code->cp852-mapper bits)))
      (if cp852-byte
          (setf (sap-ref-8 sap tail) cp852-byte)
          (external-format-encoding-error stream bits)))
    (let ((code (cp852->code-mapper byte)))
      (if code
          (code-char code)
          (external-format-decoding-error stream byte)))) ;; TODO -- error check

(define-unibyte-mapper cp855->code-mapper code->cp855-mapper
  (#x80 #x0452) ; CYRILLIC SMALL LETTER DJE
  (#x81 #x0402) ; CYRILLIC CAPITAL LETTER DJE
  (#x82 #x0453) ; CYRILLIC SMALL LETTER GJE
  (#x83 #x0403) ; CYRILLIC CAPITAL LETTER GJE
  (#x84 #x0451) ; CYRILLIC SMALL LETTER IO
  (#x85 #x0401) ; CYRILLIC CAPITAL LETTER IO
  (#x86 #x0454) ; CYRILLIC SMALL LETTER UKRAINIAN IE
  (#x87 #x0404) ; CYRILLIC CAPITAL LETTER UKRAINIAN IE
  (#x88 #x0455) ; CYRILLIC SMALL LETTER DZE
  (#x89 #x0405) ; CYRILLIC CAPITAL LETTER DZE
  (#x8A #x0456) ; CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
  (#x8B #x0406) ; CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
  (#x8C #x0457) ; CYRILLIC SMALL LETTER YI
  (#x8D #x0407) ; CYRILLIC CAPITAL LETTER YI
  (#x8E #x0458) ; CYRILLIC SMALL LETTER JE
  (#x8F #x0408) ; CYRILLIC CAPITAL LETTER JE
  (#x90 #x0459) ; CYRILLIC SMALL LETTER LJE
  (#x91 #x0409) ; CYRILLIC CAPITAL LETTER LJE
  (#x92 #x045A) ; CYRILLIC SMALL LETTER NJE
  (#x93 #x040A) ; CYRILLIC CAPITAL LETTER NJE
  (#x94 #x045B) ; CYRILLIC SMALL LETTER TSHE
  (#x95 #x040B) ; CYRILLIC CAPITAL LETTER TSHE
  (#x96 #x045C) ; CYRILLIC SMALL LETTER KJE
  (#x97 #x040C) ; CYRILLIC CAPITAL LETTER KJE
  (#x98 #x045E) ; CYRILLIC SMALL LETTER SHORT U
  (#x99 #x040E) ; CYRILLIC CAPITAL LETTER SHORT U
  (#x9A #x045F) ; CYRILLIC SMALL LETTER DZHE
  (#x9B #x040F) ; CYRILLIC CAPITAL LETTER DZHE
  (#x9C #x044E) ; CYRILLIC SMALL LETTER YU
  (#x9D #x042E) ; CYRILLIC CAPITAL LETTER YU
  (#x9E #x044A) ; CYRILLIC SMALL LETTER HARD SIGN
  (#x9F #x042A) ; CYRILLIC CAPITAL LETTER HARD SIGN
  (#xA0 #x0430) ; CYRILLIC SMALL LETTER A
  (#xA1 #x0410) ; CYRILLIC CAPITAL LETTER A
  (#xA2 #x0431) ; CYRILLIC SMALL LETTER BE
  (#xA3 #x0411) ; CYRILLIC CAPITAL LETTER BE
  (#xA4 #x0446) ; CYRILLIC SMALL LETTER TSE
  (#xA5 #x0426) ; CYRILLIC CAPITAL LETTER TSE
  (#xA6 #x0434) ; CYRILLIC SMALL LETTER DE
  (#xA7 #x0414) ; CYRILLIC CAPITAL LETTER DE
  (#xA8 #x0435) ; CYRILLIC SMALL LETTER IE
  (#xA9 #x0415) ; CYRILLIC CAPITAL LETTER IE
  (#xAA #x0444) ; CYRILLIC SMALL LETTER EF
  (#xAB #x0424) ; CYRILLIC CAPITAL LETTER EF
  (#xAC #x0433) ; CYRILLIC SMALL LETTER GHE
  (#xAD #x0413) ; CYRILLIC CAPITAL LETTER GHE
  (#xAE #x00AB) ; LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
  (#xAF #x00BB) ; RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
  (#xB0 #x2591) ; LIGHT SHADE
  (#xB1 #x2592) ; MEDIUM SHADE
  (#xB2 #x2593) ; DARK SHADE
  (#xB3 #x2502) ; BOX DRAWINGS LIGHT VERTICAL
  (#xB4 #x2524) ; BOX DRAWINGS LIGHT VERTICAL AND LEFT
  (#xB5 #x0445) ; CYRILLIC SMALL LETTER HA
  (#xB6 #x0425) ; CYRILLIC CAPITAL LETTER HA
  (#xB7 #x0438) ; CYRILLIC SMALL LETTER I
  (#xB8 #x0418) ; CYRILLIC CAPITAL LETTER I
  (#xB9 #x2563) ; BOX DRAWINGS DOUBLE VERTICAL AND LEFT
  (#xBA #x2551) ; BOX DRAWINGS DOUBLE VERTICAL
  (#xBB #x2557) ; BOX DRAWINGS DOUBLE DOWN AND LEFT
  (#xBC #x255D) ; BOX DRAWINGS DOUBLE UP AND LEFT
  (#xBD #x0439) ; CYRILLIC SMALL LETTER SHORT I
  (#xBE #x0419) ; CYRILLIC CAPITAL LETTER SHORT I
  (#xBF #x2510) ; BOX DRAWINGS LIGHT DOWN AND LEFT
  (#xC0 #x2514) ; BOX DRAWINGS LIGHT UP AND RIGHT