| 1 |
;;; |
| 2 |
;;; Generate Unicode <-> Latin-N tables |
| 3 |
;;; |
| 4 |
|
| 5 |
;; The tables can be obtained from |
| 6 |
;; https://unicode.org/Public/MAPPINGS/ISO8859/ |
| 7 |
|
| 8 |
(use file.util) |
| 9 |
(use gauche.dictionary) |
| 10 |
(use gauche.collection) |
| 11 |
(use gauche.cgen) |
| 12 |
(use util.match) |
| 13 |
(use srfi.42) |
| 14 |
|
| 15 |
(define-class <latin-ces> () |
| 16 |
((n :init-keyword :n) ; iso8859-N |
| 17 |
(bimap :init-keyword :bimap) ; code <-> ucs |
| 18 |
)) |
| 19 |
|
| 20 |
;; Returns bimap between iso8859-N (upper plane) <-> UCS |
| 21 |
(define (parse-txt file) |
| 22 |
(rlet1 b (make-bimap (make-hash-table 'eqv?) (make-hash-table 'eqv?)) |
| 23 |
(dolist [line (file->string-list file)] |
| 24 |
(rxmatch-case line |
| 25 |
[#/^0x([[:xdigit:]]+)\s+0x([[:xdigit:]]+)/ (_ scode sucs) |
| 26 |
(let ([code (string->number scode 16)] |
| 27 |
[ucs (string->number sucs 16)]) |
| 28 |
(when (>= code #xa0) |
| 29 |
(bimap-put! b code ucs)))])))) |
| 30 |
|
| 31 |
(define (load-ces-data dir n) |
| 32 |
(make <latin-ces> :n n :bimap (parse-txt #"~|dir|/8859-~|n|.TXT"))) |
| 33 |
|
| 34 |
(define (latN->utf8 ces) |
| 35 |
(define procname #"lat~(~ ces'n)_utf8") |
| 36 |
(cgen-extern #"static ScmConvProc ~|procname|;") |
| 37 |
|
| 38 |
(cgen-body "" |
| 39 |
#"static ScmSize ~|procname|(ScmConvInfo *cinfo," |
| 40 |
#" const char *inptr, ScmSize inroom SCM_UNUSED," |
| 41 |
#" char *outptr, ScmSize outroom, ScmSize *outchars)" |
| 42 |
#"{" |
| 43 |
#" static u_char tab[] = {") |
| 44 |
(do-ec (: code #xa0 #x100) |
| 45 |
(let* ([ucs (hash-table-get (bimap-left (~ ces'bimap)) code 0)] |
| 46 |
[utf8 (take* (ucs4->utf8 ucs) 3 #t 0)]) |
| 47 |
(cgen-body |
| 48 |
(format " 0x~2,'0x,0x~2,'0x,0x~2,'0x, /* ~2,'0x */" |
| 49 |
(car utf8) (cadr utf8) (caddr utf8) |
| 50 |
code)))) |
| 51 |
(cgen-body #" };" |
| 52 |
#" u_char c = *inptr;" |
| 53 |
#" if (c < 0x80) {" |
| 54 |
#" *outptr = c;" |
| 55 |
#" *outchars = 1;" |
| 56 |
#" } else if (c < 0xa0) {" |
| 57 |
#" OUTCHK(2);" |
| 58 |
#" outptr[0] = 0xc2;" |
| 59 |
#" outptr[1] = c;" |
| 60 |
#" *outchars = 2;" |
| 61 |
#" } else {" |
| 62 |
#" int index = (c - 0xa0) * 3;" |
| 63 |
#" if (tab[index] == 0) {" |
| 64 |
#" DO_SUBST;" |
| 65 |
#" } else if (tab[index + 2] == 0) {" |
| 66 |
#" OUTCHK(2);" |
| 67 |
#" outptr[0] = tab[index];" |
| 68 |
#" outptr[1] = tab[index+1];" |
| 69 |
#" *outchars = 2;" |
| 70 |
#" } else {" |
| 71 |
#" OUTCHK(3);" |
| 72 |
#" outptr[0] = tab[index];" |
| 73 |
#" outptr[1] = tab[index+1];" |
| 74 |
#" outptr[2] = tab[index+2];" |
| 75 |
#" *outchars = 3;" |
| 76 |
#" }" |
| 77 |
#" }" |
| 78 |
#" return 1;" |
| 79 |
#"}")) |
| 80 |
|
| 81 |
(define (utf8->latN ces) |
| 82 |
(define procname #"utf8_lat~(~ ces'n)") |
| 83 |
;; grouped :: ((first-octet branch ...) ...) |
| 84 |
;; branch :: (second-octet . code) | (second-octet third-octet . code) |
| 85 |
(define grouped |
| 86 |
(sort (map (^g (cons (caar g) (sort (map cdr g) < car))) |
| 87 |
(group-collection |
| 88 |
(map (^p (match-let1 (ucs . code) p |
| 89 |
(append (ucs4->utf8 ucs) code))) |
| 90 |
(hash-table->alist (bimap-right (~ ces'bimap)))) |
| 91 |
:key car)) |
| 92 |
< car)) |
| 93 |
|
| 94 |
;; If branch has more than 8 entries and all of them are 2-octet sequences, |
| 95 |
;; we use lookup table. |
| 96 |
(define (gen-lookup pfx branches) |
| 97 |
(receive (lo hi) (find-min&max (map car branches)) |
| 98 |
(cgen-decl "" |
| 99 |
#"/* offset ~|lo|*/" |
| 100 |
#"static u_char ~|procname|_~|pfx|[] = {") |
| 101 |
(do-ec (: n lo (+ hi 1)) |
| 102 |
(:let e (assv n branches)) |
| 103 |
(cgen-decl (format " 0x~2,'0x, /* [~a ~2,'0x] */" |
| 104 |
(if e (cdr e) 0) pfx n))) |
| 105 |
(cgen-decl "};") |
| 106 |
(cgen-body #" INCHK(2);" |
| 107 |
#" u_char u1 = inptr[1];" |
| 108 |
#" u_char out = 0;" |
| 109 |
#" if (u1 >= ~|lo| && u1 <= ~|hi|) {" |
| 110 |
#" out = utf8_lat~(~ ces'n)_~|pfx|[u1 - ~|lo|];" |
| 111 |
#" }" |
| 112 |
#" if (out == 0) DO_SUBST;" |
| 113 |
#" else outptr[0] = out;"))) |
| 114 |
|
| 115 |
;; Else we do linear search |
| 116 |
(define (gen-linear pfx branches) |
| 117 |
(cgen-body " u_char u1 = inptr[1];") |
| 118 |
(dolist [br branches] |
| 119 |
(if (not (pair? (cdr br))) |
| 120 |
(cgen-body (format " if (u1 == 0x~2,'0x) {" (car br)) |
| 121 |
(format " *outptr = 0x~2,'0x;" (cdr br))) |
| 122 |
(cgen-body (format " if (u1 == 0x~2,'0x &&" (car br)) |
| 123 |
(format " (u_char)inptr[2] == 0x~2,'0x) {" (cadr br)) |
| 124 |
(format " *outptr = 0x~2,'0x;" (cddr br)))) |
| 125 |
(cgen-body " } else")) |
| 126 |
(cgen-body " { DO_SUBST; }")) |
| 127 |
|
| 128 |
(cgen-extern #"static ScmConvProc ~|procname|;") |
| 129 |
(cgen-body #"static ScmSize ~|procname|(ScmConvInfo *cinfo," |
| 130 |
#" const char *inptr, ScmSize inroom," |
| 131 |
#" char *outptr, ScmSize outroom, ScmSize *outchars)" |
| 132 |
#"{" |
| 133 |
#" u_char u0 = inptr[0];" |
| 134 |
#" ScmSize nread;" |
| 135 |
#" if (u0 < 0x80) {" |
| 136 |
#" outptr[0] = u0;" |
| 137 |
#" *outchars = 1;" |
| 138 |
#" return 1;" |
| 139 |
#" }" |
| 140 |
#" else if (u0 < 0xc0) { return ILLEGAL_SEQUENCE; }" |
| 141 |
#" else if (u0 < 0xe0) { nread = 2; }" |
| 142 |
#" else if (u0 < 0xf0) { nread = 3; }" |
| 143 |
#" else if (u0 < 0xf8) { nread = 4; }" |
| 144 |
#" else if (u0 < 0xfc) { nread = 5; }" |
| 145 |
#" else if (u0 < 0xfe) { nread = 6; }" |
| 146 |
#" else { return ILLEGAL_SEQUENCE; }" |
| 147 |
#" INCHK(nread);") |
| 148 |
|
| 149 |
(cgen-body " if (u0 == 0xc2 && (u_char)inptr[1] >= 0x80 && (u_char)inptr[1] < 0xa0) {") |
| 150 |
(cgen-body " outptr[0] = inptr[1]; /* 0x80 - 0x9f */") |
| 151 |
(cgen-body " } else") |
| 152 |
(dolist [group grouped] |
| 153 |
(define pfx (format "~2,'0x" (car group))) |
| 154 |
(cgen-body #" if (u0 == 0x~|pfx|) {") |
| 155 |
(if (or (length<? (cdr group) 8) |
| 156 |
(any (^e (pair? (cdr e))) (cdr group))) |
| 157 |
(gen-linear pfx (cdr group)) |
| 158 |
(gen-lookup pfx (cdr group))) |
| 159 |
(cgen-body " } else")) |
| 160 |
(cgen-body " { DO_SUBST; }") |
| 161 |
(cgen-body " *outchars = 1;") |
| 162 |
(cgen-body " return nread;") |
| 163 |
(cgen-body "}")) |
| 164 |
|
| 165 |
;; This is in gauche.unicode, but we can't depend on it yet during |
| 166 |
;; building charconv. |
| 167 |
;; For Latin-N set, we only need to take care up to 3-octet ranges. |
| 168 |
(define (ucs4->utf8 code) |
| 169 |
(cond [(< code #x80) `(,code)] |
| 170 |
[(< code #x800) `(,(logior (ash code -6) #xc0) |
| 171 |
,(logior (logand code #x3f) #x80))] |
| 172 |
[else `(,(logior (ash code -12) #xe0) |
| 173 |
,(logior (logand (ash code -6) #x3f) #x80) |
| 174 |
,(logior (logand code #x3f) #x80))])) |
| 175 |
|
| 176 |
(define (main args) |
| 177 |
(match (cdr args) |
| 178 |
[(datadir) |
| 179 |
(parameterize ([cgen-current-unit (make <cgen-unit> |
| 180 |
:name "latin_tab" |
| 181 |
:preamble "/* Generated by gen-lattab.scm */" |
| 182 |
:init-prologue "" |
| 183 |
:init-epilogue "")]) |
| 184 |
(dolist [n '(2 3 4 5 6 7 8 9 10 11 13 14 15 16)] |
| 185 |
(let1 ces (load-ces-data datadir n) |
| 186 |
(latN->utf8 ces) |
| 187 |
(utf8->latN ces))) |
| 188 |
(cgen-emit-c (cgen-current-unit)) |
| 189 |
(cgen-emit-h (cgen-current-unit)))] |
| 190 |
) |
| 191 |
0) |