Mercurial > hg > xemacs-beta
annotate lisp/mule/ccl.el @ 5473:ac37a5f7e5be
Merge with trunk.
| author | Mats Lidell <matsl@xemacs.org> |
|---|---|
| date | Thu, 17 Mar 2011 23:42:59 +0100 |
| parents | 308d34e9f07d f00192e1cd49 |
| children | 4dee0387b9de |
| rev | line source |
|---|---|
| 4080 | 1 ;;; ccl.el --- CCL (Code Conversion Language) compiler -*- coding: iso-2022-7bit; -*- |
| 2 | |
| 3 ;; Copyright (C) 1995 Electrotechnical Laboratory, JAPAN. | |
| 4 ;; Licensed to the Free Software Foundation. | |
| 5 ;; Copyright (C) 2002, 2007 Free Software Foundation, Inc. | |
| 6 | |
| 7 ;; Keywords: CCL, mule, multilingual, character set, coding-system | |
| 8 | |
| 9 ;; This file is part of XEmacs. | |
| 10 | |
|
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
11 ;; XEmacs is free software: you can redistribute it and/or modify it |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
12 ;; under the terms of the GNU General Public License as published by the |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
13 ;; Free Software Foundation, either version 3 of the License, or (at your |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
14 ;; option) any later version. |
| 4080 | 15 |
|
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
16 ;; XEmacs is distributed in the hope that it will be useful, but WITHOUT |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
17 ;; ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
18 ;; FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
|
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
19 ;; for more details. |
| 4080 | 20 |
| 21 ;; You should have received a copy of the GNU General Public License | |
|
5402
308d34e9f07d
Changed bulk of GPLv2 or later files identified by script
Mats Lidell <matsl@xemacs.org>
parents:
4080
diff
changeset
|
22 ;; along with XEmacs. If not, see <http://www.gnu.org/licenses/>. |
| 4080 | 23 |
| 24 ;; Synched up with: FSF 21.0.90 | |
| 25 | |
| 26 ;;; Commentary: | |
| 27 | |
| 28 ;; CCL (Code Conversion Language) is a simple programming language to | |
| 29 ;; be used for various kind of code conversion. CCL program is | |
| 30 ;; compiled to CCL code (vector of integers) and executed by CCL | |
| 31 ;; interpreter of Emacs. | |
| 32 ;; | |
| 33 ;; CCL is used for code conversion at process I/O and file I/O for | |
| 34 ;; non-standard coding-system. In addition, it is used for | |
| 35 ;; calculating a code point of X's font from a character code. | |
| 36 ;; However, since CCL is designed as a powerful programming language, | |
| 37 ;; it can be used for more generic calculation. For instance, | |
| 38 ;; combination of three or more arithmetic operations can be | |
| 39 ;; calculated faster than Emacs Lisp. | |
| 40 ;; | |
| 41 ;; Syntax and semantics of CCL program is described in the | |
| 42 ;; documentation of `define-ccl-program'. | |
| 43 | |
| 44 ;;; Code: | |
| 45 | |
| 46 (defconst ccl-command-table | |
| 47 [if branch loop break repeat write-repeat write-read-repeat | |
| 48 read read-if read-branch write call end | |
| 49 read-multibyte-character write-multibyte-character | |
| 50 translate-character mule-to-unicode unicode-to-mule | |
| 51 iterate-multiple-map map-multiple map-single lookup-integer | |
| 52 lookup-character] | |
| 53 "Vector of CCL commands (symbols).") | |
| 54 | |
| 55 ;; Put a property to each symbol of CCL commands for the compiler. | |
| 56 (let (op (i 0) (len (length ccl-command-table))) | |
| 57 (while (< i len) | |
| 58 (setq op (aref ccl-command-table i)) | |
| 59 (put op 'ccl-compile-function (intern (format "ccl-compile-%s" op))) | |
| 60 (setq i (1+ i)))) | |
| 61 | |
| 62 (defconst ccl-code-table | |
| 63 [set-register | |
| 64 set-short-const | |
| 65 set-const | |
| 66 set-array | |
| 67 jump | |
| 68 jump-cond | |
| 69 write-register-jump | |
| 70 write-register-read-jump | |
| 71 write-const-jump | |
| 72 write-const-read-jump | |
| 73 write-string-jump | |
| 74 write-array-read-jump | |
| 75 read-jump | |
| 76 branch | |
| 77 read-register | |
| 78 write-expr-const | |
| 79 read-branch | |
| 80 write-register | |
| 81 write-expr-register | |
| 82 call | |
| 83 write-const-string | |
| 84 write-array | |
| 85 end | |
| 86 set-assign-expr-const | |
| 87 set-assign-expr-register | |
| 88 set-expr-const | |
| 89 set-expr-register | |
| 90 jump-cond-expr-const | |
| 91 jump-cond-expr-register | |
| 92 read-jump-cond-expr-const | |
| 93 read-jump-cond-expr-register | |
| 94 ex-cmd | |
| 95 ] | |
| 96 "Vector of CCL compiled codes (symbols).") | |
| 97 | |
| 98 (defconst ccl-extended-code-table | |
| 99 [read-multibyte-character | |
| 100 write-multibyte-character | |
| 101 translate-character | |
| 102 translate-character-const-tbl | |
| 103 mule-to-unicode | |
| 104 unicode-to-mule | |
| 105 nil nil nil nil nil nil nil nil nil nil ; 0x06-0x0f | |
| 106 iterate-multiple-map | |
| 107 map-multiple | |
| 108 map-single | |
| 109 lookup-int-const-tbl | |
| 110 lookup-char-const-tbl | |
| 111 ] | |
| 112 "Vector of CCL extended compiled codes (symbols).") | |
| 113 | |
| 114 ;; Put a property to each symbol of CCL codes for the disassembler. | |
| 115 (let (code (i 0) (len (length ccl-code-table))) | |
| 116 (while (< i len) | |
| 117 (setq code (aref ccl-code-table i)) | |
| 118 (put code 'ccl-code i) | |
| 119 (put code 'ccl-dump-function (intern (format "ccl-dump-%s" code))) | |
| 120 (setq i (1+ i)))) | |
| 121 | |
| 122 (let (code (i 0) (len (length ccl-extended-code-table))) | |
| 123 (while (< i len) | |
| 124 (setq code (aref ccl-extended-code-table i)) | |
| 125 (if code | |
| 126 (progn | |
| 127 (put code 'ccl-ex-code i) | |
| 128 (put code 'ccl-dump-function (intern (format "ccl-dump-%s" code))))) | |
| 129 (setq i (1+ i)))) | |
| 130 | |
| 131 (defconst ccl-jump-code-list | |
| 132 '(jump jump-cond write-register-jump write-register-read-jump | |
| 133 write-const-jump write-const-read-jump write-string-jump | |
| 134 write-array-read-jump read-jump)) | |
| 135 | |
| 136 ;; Put a property `jump-flag' to each CCL code which execute jump in | |
| 137 ;; some way. | |
| 138 (let ((l ccl-jump-code-list)) | |
| 139 (while l | |
| 140 (put (car l) 'jump-flag t) | |
| 141 (setq l (cdr l)))) | |
| 142 | |
| 143 (defconst ccl-register-table | |
| 144 [r0 r1 r2 r3 r4 r5 r6 r7] | |
| 145 "Vector of CCL registers (symbols).") | |
| 146 | |
| 147 ;; Put a property to indicate register number to each symbol of CCL. | |
| 148 ;; registers. | |
| 149 (let (reg (i 0) (len (length ccl-register-table))) | |
| 150 (while (< i len) | |
| 151 (setq reg (aref ccl-register-table i)) | |
| 152 (put reg 'ccl-register-number i) | |
| 153 (setq i (1+ i)))) | |
| 154 | |
| 155 (defconst ccl-arith-table | |
| 156 [+ - * / % & | ^ << >> <8 >8 // nil nil nil | |
| 157 < > == <= >= != de-sjis en-sjis] | |
| 158 "Vector of CCL arithmetic/logical operators (symbols).") | |
| 159 | |
| 160 ;; Put a property to each symbol of CCL operators for the compiler. | |
| 161 (let (arith (i 0) (len (length ccl-arith-table))) | |
| 162 (while (< i len) | |
| 163 (setq arith (aref ccl-arith-table i)) | |
| 164 (if arith (put arith 'ccl-arith-code i)) | |
| 165 (setq i (1+ i)))) | |
| 166 | |
| 167 (defconst ccl-assign-arith-table | |
| 168 [+= -= *= /= %= &= |= ^= <<= >>= <8= >8= //=] | |
| 169 "Vector of CCL assignment operators (symbols).") | |
| 170 | |
| 171 ;; Put a property to each symbol of CCL assignment operators for the compiler. | |
| 172 (let (arith (i 0) (len (length ccl-assign-arith-table))) | |
| 173 (while (< i len) | |
| 174 (setq arith (aref ccl-assign-arith-table i)) | |
| 175 (put arith 'ccl-self-arith-code i) | |
| 176 (setq i (1+ i)))) | |
| 177 | |
| 178 (defvar ccl-program-vector nil | |
| 179 "Working vector of CCL codes produced by CCL compiler.") | |
| 180 (defvar ccl-current-ic 0 | |
| 181 "The current index for `ccl-program-vector'.") | |
| 182 | |
| 183 ;; Embed integer DATA in `ccl-program-vector' at `ccl-current-ic' and | |
| 184 ;; increment it. If IC is specified, embed DATA at IC. | |
| 185 (defun ccl-embed-data (data &optional ic) | |
| 186 ;; XEmacs: Embed characters as characters, since their integer values vary at | |
| 187 ;; runtime. | |
| 188 ; (if (characterp data) | |
| 189 ; (setq data (char-int data))) | |
| 190 (if ic | |
| 191 (aset ccl-program-vector ic data) | |
| 192 (let ((len (length ccl-program-vector))) | |
| 193 (if (>= ccl-current-ic len) | |
| 194 (let ((new (make-vector (* len 2) nil))) | |
| 195 (while (> len 0) | |
| 196 (setq len (1- len)) | |
| 197 (aset new len (aref ccl-program-vector len))) | |
| 198 (setq ccl-program-vector new)))) | |
| 199 (aset ccl-program-vector ccl-current-ic data) | |
| 200 (setq ccl-current-ic (1+ ccl-current-ic)))) | |
| 201 | |
| 202 ;; Embed pair of SYMBOL and PROP where (get SYMBOL PROP) should give | |
| 203 ;; proper index number for SYMBOL. PROP should be | |
| 204 ;; `translation-table-id', `translation-hash-table-id' | |
| 205 ;; `code-conversion-map-id', or `ccl-program-idx'. | |
| 206 (defun ccl-embed-symbol (symbol prop) | |
| 207 (ccl-embed-data (cons symbol prop))) | |
| 208 | |
| 209 ;; Embed string STR of length LEN in `ccl-program-vector' at | |
| 210 ;; `ccl-current-ic'. | |
| 211 (defun ccl-embed-string (len str) | |
| 212 (let ((i 0)) | |
| 213 (while (< i len) | |
| 214 (ccl-embed-data (logior (ash (aref str i) 16) | |
| 215 (if (< (1+ i) len) | |
| 216 (ash (aref str (1+ i)) 8) | |
| 217 0) | |
| 218 (if (< (+ i 2) len) | |
| 219 (aref str (+ i 2)) | |
| 220 0))) | |
| 221 (setq i (+ i 3))))) | |
| 222 | |
| 223 ;; Embed a relative jump address to `ccl-current-ic' in | |
| 224 ;; `ccl-program-vector' at IC without altering the other bit field. | |
| 225 (defun ccl-embed-current-address (ic) | |
| 226 (let ((relative (- ccl-current-ic (1+ ic)))) | |
| 227 (aset ccl-program-vector ic | |
| 228 (logior (aref ccl-program-vector ic) (ash relative 8))))) | |
| 229 | |
| 230 ;; Embed CCL code for the operation OP and arguments REG and DATA in | |
| 231 ;; `ccl-program-vector' at `ccl-current-ic' in the following format. | |
| 232 ;; |----------------- integer (28-bit) ------------------| | |
| 233 ;; |------------ 20-bit ------------|- 3-bit --|- 5-bit -| | |
| 234 ;; |------------- DATA -------------|-- REG ---|-- OP ---| | |
| 235 ;; If REG2 is specified, embed a code in the following format. | |
| 236 ;; |------- 17-bit ------|- 3-bit --|- 3-bit --|- 5-bit -| | |
| 237 ;; |-------- DATA -------|-- REG2 --|-- REG ---|-- OP ---| | |
| 238 | |
| 239 ;; If REG is a CCL register symbol (e.g. r0, r1...), the register | |
| 240 ;; number is embedded. If OP is one of unconditional jumps, DATA is | |
| 241 ;; changed to an relative jump address. | |
| 242 | |
| 243 (defun ccl-embed-code (op reg data &optional reg2) | |
| 244 (if (and (> data 0) (get op 'jump-flag)) | |
| 245 ;; DATA is an absolute jump address. Make it relative to the | |
| 246 ;; next of jump code. | |
| 247 (setq data (- data (1+ ccl-current-ic)))) | |
| 248 (let ((code (logior (get op 'ccl-code) | |
| 249 (ash | |
| 250 (if (symbolp reg) (get reg 'ccl-register-number) reg) 5) | |
| 251 (if reg2 | |
| 252 (logior (ash (get reg2 'ccl-register-number) 8) | |
| 253 (ash data 11)) | |
| 254 (ash data 8))))) | |
| 255 (ccl-embed-data code))) | |
| 256 | |
| 257 ;; extended ccl command format | |
| 258 ;; |- 14-bit -|- 3-bit --|- 3-bit --|- 3-bit --|- 5-bit -| | |
| 259 ;; |- EX-OP --|-- REG3 --|-- REG2 --|-- REG ---|-- OP ---| | |
| 260 (defun ccl-embed-extended-command (ex-op reg reg2 reg3) | |
| 261 (let ((data (logior (ash (get ex-op 'ccl-ex-code) 3) | |
| 262 (if (symbolp reg3) | |
| 263 (get reg3 'ccl-register-number) | |
| 264 0)))) | |
| 265 (ccl-embed-code 'ex-cmd reg data reg2))) | |
| 266 | |
| 267 ;; Just advance `ccl-current-ic' by INC. | |
| 268 (defun ccl-increment-ic (inc) | |
| 269 (setq ccl-current-ic (+ ccl-current-ic inc))) | |
| 270 | |
| 271 ;; If non-nil, index of the start of the current loop. | |
| 272 (defvar ccl-loop-head nil) | |
| 273 ;; If non-nil, list of absolute addresses of the breaking points of | |
| 274 ;; the current loop. | |
| 275 (defvar ccl-breaks nil) | |
| 276 | |
| 277 ;;;###autoload | |
| 278 (defun ccl-compile (ccl-program) | |
| 279 "Return a compiled code of CCL-PROGRAM as a vector of integer." | |
| 280 (if (or (null (consp ccl-program)) | |
| 281 (null (integerp (car ccl-program))) | |
| 282 (null (listp (car (cdr ccl-program))))) | |
| 283 (error "CCL: Invalid CCL program: %s" ccl-program)) | |
| 284 (if (null (vectorp ccl-program-vector)) | |
| 285 (setq ccl-program-vector (make-vector 8192 0))) | |
| 286 (setq ccl-loop-head nil ccl-breaks nil) | |
| 287 (setq ccl-current-ic 0) | |
| 288 | |
| 289 ;; The first element is the buffer magnification. | |
| 290 (ccl-embed-data (car ccl-program)) | |
| 291 | |
| 292 ;; The second element is the address of the start CCL code for | |
| 293 ;; processing end of input buffer (we call it eof-processor). We | |
| 294 ;; set it later. | |
| 295 (ccl-increment-ic 1) | |
| 296 | |
| 297 ;; Compile the main body of the CCL program. | |
| 298 (ccl-compile-1 (car (cdr ccl-program))) | |
| 299 | |
| 300 ;; Embed the address of eof-processor. | |
| 301 (ccl-embed-data ccl-current-ic 1) | |
| 302 | |
| 303 ;; Then compile eof-processor. | |
| 304 (if (nth 2 ccl-program) | |
| 305 (ccl-compile-1 (nth 2 ccl-program))) | |
| 306 | |
| 307 ;; At last, embed termination code. | |
| 308 (ccl-embed-code 'end 0 0) | |
| 309 | |
| 310 (let ((vec (make-vector ccl-current-ic 0)) | |
| 311 (i 0)) | |
| 312 (while (< i ccl-current-ic) | |
| 313 (aset vec i (aref ccl-program-vector i)) | |
| 314 (setq i (1+ i))) | |
| 315 vec)) | |
| 316 | |
| 317 ;; Signal syntax error. | |
| 318 (defun ccl-syntax-error (cmd) | |
| 319 (error "CCL: Syntax error: %s" cmd)) | |
| 320 | |
| 321 ;; Check if ARG is a valid CCL register. | |
| 322 (defun ccl-check-register (arg cmd) | |
| 323 (if (get arg 'ccl-register-number) | |
| 324 arg | |
| 325 (error "CCL: Invalid register %s in %s." arg cmd))) | |
| 326 | |
| 327 ;; Check if ARG is a valid CCL command. | |
| 328 (defun ccl-check-compile-function (arg cmd) | |
| 329 (or (get arg 'ccl-compile-function) | |
| 330 (error "CCL: Invalid command: %s" cmd))) | |
| 331 | |
| 332 ;; In the following code, most ccl-compile-XXXX functions return t if | |
| 333 ;; they end with unconditional jump, else return nil. | |
| 334 | |
| 335 ;; Compile CCL-BLOCK (see the syntax above). | |
| 336 (defun ccl-compile-1 (ccl-block) | |
| 337 (let (unconditional-jump | |
| 338 cmd) | |
| 339 (if (or (integer-or-char-p ccl-block) | |
| 340 (stringp ccl-block) | |
| 341 (and ccl-block (symbolp (car ccl-block)))) | |
| 342 ;; This block consists of single statement. | |
| 343 (setq ccl-block (list ccl-block))) | |
| 344 | |
| 345 ;; Now CCL-BLOCK is a list of statements. Compile them one by | |
| 346 ;; one. | |
| 347 (while ccl-block | |
| 348 (setq cmd (car ccl-block)) | |
| 349 (setq unconditional-jump | |
| 350 (cond ((integer-or-char-p cmd) | |
| 351 ;; SET statement for the register 0. | |
| 352 (ccl-compile-set (list 'r0 '= cmd))) | |
| 353 | |
| 354 ((stringp cmd) | |
| 355 ;; WRITE statement of string argument. | |
| 356 (ccl-compile-write-string cmd)) | |
| 357 | |
| 358 ((listp cmd) | |
| 359 ;; The other statements. | |
| 360 (cond ((eq (nth 1 cmd) '=) | |
| 361 ;; SET statement of the form `(REG = EXPRESSION)'. | |
| 362 (ccl-compile-set cmd)) | |
| 363 | |
| 364 ((and (symbolp (nth 1 cmd)) | |
| 365 (get (nth 1 cmd) 'ccl-self-arith-code)) | |
| 366 ;; SET statement with an assignment operation. | |
| 367 (ccl-compile-self-set cmd)) | |
| 368 | |
| 369 (t | |
| 370 (funcall (ccl-check-compile-function (car cmd) cmd) | |
| 371 cmd)))) | |
| 372 | |
| 373 (t | |
| 374 (ccl-syntax-error cmd)))) | |
| 375 (setq ccl-block (cdr ccl-block))) | |
| 376 unconditional-jump)) | |
| 377 | |
| 378 (defconst ccl-max-short-const (ash 1 19)) | |
| 379 (defconst ccl-min-short-const (ash -1 19)) | |
| 380 | |
| 381 ;; Compile SET statement. | |
| 382 (defun ccl-compile-set (cmd) | |
| 383 (let ((rrr (ccl-check-register (car cmd) cmd)) | |
| 384 (right (nth 2 cmd))) | |
| 385 (cond ((listp right) | |
| 386 ;; CMD has the form `(RRR = (XXX OP YYY))'. | |
| 387 (ccl-compile-expression rrr right)) | |
| 388 | |
| 389 ((integer-or-char-p right) | |
| 390 ;; CMD has the form `(RRR = integer)'. | |
| 391 (if (and (<= right ccl-max-short-const) | |
| 392 (>= right ccl-min-short-const)) | |
| 393 (ccl-embed-code 'set-short-const rrr right) | |
| 394 (ccl-embed-code 'set-const rrr 0) | |
| 395 (ccl-embed-data right))) | |
| 396 | |
| 397 (t | |
| 398 ;; CMD has the form `(RRR = rrr [ array ])'. | |
| 399 (ccl-check-register right cmd) | |
| 400 (let ((ary (nth 3 cmd))) | |
| 401 (if (vectorp ary) | |
| 402 (let ((i 0) (len (length ary))) | |
| 403 (ccl-embed-code 'set-array rrr len right) | |
| 404 (while (< i len) | |
| 405 (ccl-embed-data (aref ary i)) | |
| 406 (setq i (1+ i)))) | |
| 407 (ccl-embed-code 'set-register rrr 0 right)))))) | |
| 408 nil) | |
| 409 | |
| 410 ;; Compile SET statement with ASSIGNMENT_OPERATOR. | |
| 411 (defun ccl-compile-self-set (cmd) | |
| 412 (let ((rrr (ccl-check-register (car cmd) cmd)) | |
| 413 (right (nth 2 cmd))) | |
| 414 (if (listp right) | |
| 415 ;; CMD has the form `(RRR ASSIGN_OP (XXX OP YYY))', compile | |
| 416 ;; the right hand part as `(r7 = (XXX OP YYY))' (note: the | |
| 417 ;; register 7 can be used for storing temporary value). | |
| 418 (progn | |
| 419 (ccl-compile-expression 'r7 right) | |
| 420 (setq right 'r7))) | |
| 421 ;; Now CMD has the form `(RRR ASSIGN_OP ARG)'. Compile it as | |
| 422 ;; `(RRR = (RRR OP ARG))'. | |
| 423 (ccl-compile-expression | |
| 424 rrr | |
| 425 (list rrr (intern (substring (symbol-name (nth 1 cmd)) 0 -1)) right))) | |
| 426 nil) | |
| 427 | |
| 428 ;; Compile SET statement of the form `(RRR = EXPR)'. | |
| 429 (defun ccl-compile-expression (rrr expr) | |
| 430 (let ((left (car expr)) | |
| 431 (op (get (nth 1 expr) 'ccl-arith-code)) | |
| 432 (right (nth 2 expr))) | |
| 433 (if (listp left) | |
| 434 (progn | |
| 435 ;; EXPR has the form `((EXPR2 OP2 ARG) OP RIGHT)'. Compile | |
| 436 ;; the first term as `(r7 = (EXPR2 OP2 ARG)).' | |
| 437 (ccl-compile-expression 'r7 left) | |
| 438 (setq left 'r7))) | |
| 439 | |
| 440 ;; Now EXPR has the form (LEFT OP RIGHT). | |
| 441 (if (and (eq rrr left) | |
| 442 (< op (length ccl-assign-arith-table))) | |
| 443 ;; Compile this SET statement as `(RRR OP= RIGHT)'. | |
| 444 (if (integer-or-char-p right) | |
| 445 (progn | |
| 446 (ccl-embed-code 'set-assign-expr-const rrr (ash op 3) 'r0) | |
| 447 (ccl-embed-data right)) | |
| 448 (ccl-check-register right expr) | |
| 449 (ccl-embed-code 'set-assign-expr-register rrr (ash op 3) right)) | |
| 450 | |
| 451 ;; Compile this SET statement as `(RRR = (LEFT OP RIGHT))'. | |
| 452 (if (integer-or-char-p right) | |
| 453 (progn | |
| 454 (ccl-embed-code 'set-expr-const rrr (ash op 3) left) | |
| 455 (ccl-embed-data right)) | |
| 456 (ccl-check-register right expr) | |
| 457 (ccl-embed-code 'set-expr-register | |
| 458 rrr | |
| 459 (logior (ash op 3) (get right 'ccl-register-number)) | |
| 460 left))))) | |
| 461 | |
| 462 ;; Compile WRITE statement with string argument. | |
| 463 (defun ccl-compile-write-string (str) | |
| 464 (setq str (encode-coding-string str 'binary)) | |
| 465 (let ((len (length str))) | |
| 466 (ccl-embed-code 'write-const-string 1 len) | |
| 467 (ccl-embed-string len str)) | |
| 468 nil) | |
| 469 | |
| 470 ;; Compile IF statement of the form `(if CONDITION TRUE-PART FALSE-PART)'. | |
| 471 ;; If READ-FLAG is non-nil, this statement has the form | |
| 472 ;; `(read-if (REG OPERATOR ARG) TRUE-PART FALSE-PART)'. | |
| 473 (defun ccl-compile-if (cmd &optional read-flag) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
474 (if (and (not (<= 3 (length cmd) 4))) |
| 4080 | 475 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 476 (let ((condition (nth 1 cmd)) | |
| 477 (true-cmds (nth 2 cmd)) | |
| 478 (false-cmds (nth 3 cmd)) | |
| 479 jump-cond-address) | |
| 480 (if (and (listp condition) | |
| 481 (listp (car condition))) | |
| 482 ;; If CONDITION is a nested expression, the inner expression | |
| 483 ;; should be compiled at first as SET statement, i.e.: | |
| 484 ;; `(if ((X OP2 Y) OP Z) ...)' is compiled into two statements: | |
| 485 ;; `(r7 = (X OP2 Y)) (if (r7 OP Z) ...)'. | |
| 486 (progn | |
| 487 (ccl-compile-expression 'r7 (car condition)) | |
| 488 (setq condition (cons 'r7 (cdr condition))) | |
| 489 (setq cmd (cons (car cmd) | |
| 490 (cons condition (cdr (cdr cmd))))))) | |
| 491 | |
| 492 (setq jump-cond-address ccl-current-ic) | |
| 493 ;; Compile CONDITION. | |
| 494 (if (symbolp condition) | |
| 495 ;; CONDITION is a register. | |
| 496 (progn | |
| 497 (ccl-check-register condition cmd) | |
| 498 (ccl-embed-code 'jump-cond condition 0)) | |
| 499 ;; CONDITION is a simple expression of the form (RRR OP ARG). | |
| 500 (let ((rrr (car condition)) | |
| 501 (op (get (nth 1 condition) 'ccl-arith-code)) | |
| 502 (arg (nth 2 condition))) | |
| 503 (ccl-check-register rrr cmd) | |
| 504 (if (integer-or-char-p arg) | |
| 505 (progn | |
| 506 (ccl-embed-code (if read-flag 'read-jump-cond-expr-const | |
| 507 'jump-cond-expr-const) | |
| 508 rrr 0) | |
| 509 (ccl-embed-data op) | |
| 510 (ccl-embed-data arg)) | |
| 511 (ccl-check-register arg cmd) | |
| 512 (ccl-embed-code (if read-flag 'read-jump-cond-expr-register | |
| 513 'jump-cond-expr-register) | |
| 514 rrr 0) | |
| 515 (ccl-embed-data op) | |
| 516 (ccl-embed-data (get arg 'ccl-register-number))))) | |
| 517 | |
| 518 ;; Compile TRUE-PART. | |
| 519 (let ((unconditional-jump (ccl-compile-1 true-cmds))) | |
| 520 (if (null false-cmds) | |
| 521 ;; This is the place to jump to if condition is false. | |
| 522 (progn | |
| 523 (ccl-embed-current-address jump-cond-address) | |
| 524 (setq unconditional-jump nil)) | |
| 525 (let (end-true-part-address) | |
| 526 (if (not unconditional-jump) | |
| 527 (progn | |
| 528 ;; If TRUE-PART does not end with unconditional jump, we | |
| 529 ;; have to jump to the end of FALSE-PART from here. | |
| 530 (setq end-true-part-address ccl-current-ic) | |
| 531 (ccl-embed-code 'jump 0 0))) | |
| 532 ;; This is the place to jump to if CONDITION is false. | |
| 533 (ccl-embed-current-address jump-cond-address) | |
| 534 ;; Compile FALSE-PART. | |
| 535 (setq unconditional-jump | |
| 536 (and (ccl-compile-1 false-cmds) unconditional-jump)) | |
| 537 (if end-true-part-address | |
| 538 ;; This is the place to jump to after the end of TRUE-PART. | |
| 539 (ccl-embed-current-address end-true-part-address)))) | |
| 540 unconditional-jump))) | |
| 541 | |
| 542 ;; Compile BRANCH statement. | |
| 543 (defun ccl-compile-branch (cmd) | |
| 544 (if (< (length cmd) 3) | |
| 545 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 546 (ccl-compile-branch-blocks 'branch | |
| 547 (ccl-compile-branch-expression (nth 1 cmd) cmd) | |
| 548 (cdr (cdr cmd)))) | |
| 549 | |
| 550 ;; Compile READ statement of the form `(read-branch EXPR BLOCK0 BLOCK1 ...)'. | |
| 551 (defun ccl-compile-read-branch (cmd) | |
| 552 (if (< (length cmd) 3) | |
| 553 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 554 (ccl-compile-branch-blocks 'read-branch | |
| 555 (ccl-compile-branch-expression (nth 1 cmd) cmd) | |
| 556 (cdr (cdr cmd)))) | |
| 557 | |
| 558 ;; Compile EXPRESSION part of BRANCH statement and return register | |
| 559 ;; which holds a value of the expression. | |
| 560 (defun ccl-compile-branch-expression (expr cmd) | |
| 561 (if (listp expr) | |
| 562 ;; EXPR has the form `(EXPR2 OP ARG)'. Compile it as SET | |
| 563 ;; statement of the form `(r7 = (EXPR2 OP ARG))'. | |
| 564 (progn | |
| 565 (ccl-compile-expression 'r7 expr) | |
| 566 'r7) | |
| 567 (ccl-check-register expr cmd))) | |
| 568 | |
| 569 ;; Compile BLOCKs of BRANCH statement. CODE is 'branch or 'read-branch. | |
| 570 ;; REG is a register which holds a value of EXPRESSION part. BLOCKs | |
| 571 ;; is a list of CCL-BLOCKs. | |
| 572 (defun ccl-compile-branch-blocks (code rrr blocks) | |
| 573 (let ((branches (length blocks)) | |
| 574 branch-idx | |
| 575 jump-table-head-address | |
| 576 empty-block-indexes | |
| 577 block-tail-addresses | |
| 578 block-unconditional-jump) | |
| 579 (ccl-embed-code code rrr branches) | |
| 580 (setq jump-table-head-address ccl-current-ic) | |
| 581 ;; The size of jump table is the number of blocks plus 1 (for the | |
| 582 ;; case RRR is out of range). | |
| 583 (ccl-increment-ic (1+ branches)) | |
| 584 (setq empty-block-indexes (list branches)) | |
| 585 ;; Compile each block. | |
| 586 (setq branch-idx 0) | |
| 587 (while blocks | |
| 588 (if (null (car blocks)) | |
| 589 ;; This block is empty. | |
| 590 (setq empty-block-indexes (cons branch-idx empty-block-indexes) | |
| 591 block-unconditional-jump t) | |
| 592 ;; This block is not empty. | |
| 593 (ccl-embed-data (- ccl-current-ic jump-table-head-address) | |
| 594 (+ jump-table-head-address branch-idx)) | |
| 595 (setq block-unconditional-jump (ccl-compile-1 (car blocks))) | |
| 596 (if (not block-unconditional-jump) | |
| 597 (progn | |
| 598 ;; Jump address of the end of branches are embedded later. | |
| 599 ;; For the moment, just remember where to embed them. | |
| 600 (setq block-tail-addresses | |
| 601 (cons ccl-current-ic block-tail-addresses)) | |
| 602 (ccl-embed-code 'jump 0 0)))) | |
| 603 (setq branch-idx (1+ branch-idx)) | |
| 604 (setq blocks (cdr blocks))) | |
| 605 (if (not block-unconditional-jump) | |
| 606 ;; We don't need jump code at the end of the last block. | |
| 607 (setq block-tail-addresses (cdr block-tail-addresses) | |
| 608 ccl-current-ic (1- ccl-current-ic))) | |
| 609 ;; Embed jump address at the tailing jump commands of blocks. | |
| 610 (while block-tail-addresses | |
| 611 (ccl-embed-current-address (car block-tail-addresses)) | |
| 612 (setq block-tail-addresses (cdr block-tail-addresses))) | |
| 613 ;; For empty blocks, make entries in the jump table point directly here. | |
| 614 (while empty-block-indexes | |
| 615 (ccl-embed-data (- ccl-current-ic jump-table-head-address) | |
| 616 (+ jump-table-head-address (car empty-block-indexes))) | |
| 617 (setq empty-block-indexes (cdr empty-block-indexes)))) | |
| 618 ;; Branch command ends by unconditional jump if RRR is out of range. | |
| 619 nil) | |
| 620 | |
| 621 ;; Compile LOOP statement. | |
| 622 (defun ccl-compile-loop (cmd) | |
| 623 (if (< (length cmd) 2) | |
| 624 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 625 (let* ((ccl-loop-head ccl-current-ic) | |
| 626 (ccl-breaks nil) | |
| 627 unconditional-jump) | |
| 628 (setq cmd (cdr cmd)) | |
| 629 (if cmd | |
| 630 (progn | |
| 631 (setq unconditional-jump t) | |
| 632 (while cmd | |
| 633 (setq unconditional-jump | |
| 634 (and (ccl-compile-1 (car cmd)) unconditional-jump)) | |
| 635 (setq cmd (cdr cmd))) | |
| 636 (if (not ccl-breaks) | |
| 637 unconditional-jump | |
| 638 ;; Embed jump address for break statements encountered in | |
| 639 ;; this loop. | |
| 640 (while ccl-breaks | |
| 641 (ccl-embed-current-address (car ccl-breaks)) | |
| 642 (setq ccl-breaks (cdr ccl-breaks)))) | |
| 643 nil)))) | |
| 644 | |
| 645 ;; Compile BREAK statement. | |
| 646 (defun ccl-compile-break (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
647 (if (not (eql (length cmd) 1)) |
| 4080 | 648 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 649 (if (null ccl-loop-head) | |
| 650 (error "CCL: No outer loop: %s" cmd)) | |
| 651 (setq ccl-breaks (cons ccl-current-ic ccl-breaks)) | |
| 652 (ccl-embed-code 'jump 0 0) | |
| 653 t) | |
| 654 | |
| 655 ;; Compile REPEAT statement. | |
| 656 (defun ccl-compile-repeat (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
657 (if (not (eql (length cmd) 1)) |
| 4080 | 658 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 659 (if (null ccl-loop-head) | |
| 660 (error "CCL: No outer loop: %s" cmd)) | |
| 661 (ccl-embed-code 'jump 0 ccl-loop-head) | |
| 662 t) | |
| 663 | |
| 664 ;; Compile WRITE-REPEAT statement. | |
| 665 (defun ccl-compile-write-repeat (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
666 (if (not (eql (length cmd) 2)) |
| 4080 | 667 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 668 (if (null ccl-loop-head) | |
| 669 (error "CCL: No outer loop: %s" cmd)) | |
| 670 (let ((arg (nth 1 cmd))) | |
| 671 (cond ((integer-or-char-p arg) | |
| 672 (ccl-embed-code 'write-const-jump 0 ccl-loop-head) | |
| 673 (ccl-embed-data arg)) | |
| 674 ((stringp arg) | |
| 675 (setq arg (encode-coding-string arg 'binary)) | |
| 676 (let ((len (length arg))) | |
| 677 (ccl-embed-code 'write-string-jump 0 ccl-loop-head) | |
| 678 (ccl-embed-data len) | |
| 679 (ccl-embed-string len arg))) | |
| 680 (t | |
| 681 (ccl-check-register arg cmd) | |
| 682 (ccl-embed-code 'write-register-jump arg ccl-loop-head)))) | |
| 683 t) | |
| 684 | |
| 685 ;; Compile WRITE-READ-REPEAT statement. | |
| 686 (defun ccl-compile-write-read-repeat (cmd) | |
| 687 (if (or (< (length cmd) 2) (> (length cmd) 3)) | |
| 688 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 689 (if (null ccl-loop-head) | |
| 690 (error "CCL: No outer loop: %s" cmd)) | |
| 691 (let ((rrr (ccl-check-register (nth 1 cmd) cmd)) | |
| 692 (arg (nth 2 cmd))) | |
| 693 (cond ((null arg) | |
| 694 (ccl-embed-code 'write-register-read-jump rrr ccl-loop-head)) | |
| 695 ((integer-or-char-p arg) | |
| 696 (ccl-embed-code 'write-const-read-jump rrr arg ccl-loop-head)) | |
| 697 ((vectorp arg) | |
| 698 (let ((len (length arg)) | |
| 699 (i 0)) | |
| 700 (ccl-embed-code 'write-array-read-jump rrr ccl-loop-head) | |
| 701 (ccl-embed-data len) | |
| 702 (while (< i len) | |
| 703 (ccl-embed-data (aref arg i)) | |
| 704 (setq i (1+ i))))) | |
| 705 (t | |
| 706 (error "CCL: Invalid argument %s: %s" arg cmd))) | |
| 707 (ccl-embed-code 'read-jump rrr ccl-loop-head)) | |
| 708 t) | |
| 709 | |
| 710 ;; Compile READ statement. | |
| 711 (defun ccl-compile-read (cmd) | |
| 712 (if (< (length cmd) 2) | |
| 713 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 714 (let* ((args (cdr cmd)) | |
| 715 (i (1- (length args)))) | |
| 716 (while args | |
| 717 (let ((rrr (ccl-check-register (car args) cmd))) | |
| 718 (ccl-embed-code 'read-register rrr i) | |
| 719 (setq args (cdr args) i (1- i))))) | |
| 720 nil) | |
| 721 | |
| 722 ;; Compile READ-IF statement. | |
| 723 (defun ccl-compile-read-if (cmd) | |
| 724 (ccl-compile-if cmd 'read)) | |
| 725 | |
| 726 ;; Compile WRITE statement. | |
| 727 (defun ccl-compile-write (cmd) | |
| 728 (if (< (length cmd) 2) | |
| 729 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 730 (let ((rrr (nth 1 cmd))) | |
| 731 (cond ((integer-or-char-p rrr) | |
| 732 (ccl-embed-code 'write-const-string 0 rrr)) | |
| 733 ((stringp rrr) | |
| 734 (ccl-compile-write-string rrr)) | |
| 735 ((and (symbolp rrr) (vectorp (nth 2 cmd))) | |
| 736 (ccl-check-register rrr cmd) | |
| 737 ;; CMD has the form `(write REG ARRAY)'. | |
| 738 (let* ((arg (nth 2 cmd)) | |
| 739 (len (length arg)) | |
| 740 (i 0)) | |
| 741 (ccl-embed-code 'write-array rrr len) | |
| 742 (while (< i len) | |
| 743 (if (not (integer-or-char-p (aref arg i))) | |
| 744 (error "CCL: Invalid argument %s: %s" arg cmd)) | |
| 745 (ccl-embed-data (aref arg i)) | |
| 746 (setq i (1+ i))))) | |
| 747 | |
| 748 ((symbolp rrr) | |
| 749 ;; CMD has the form `(write REG ...)'. | |
| 750 (let* ((args (cdr cmd)) | |
| 751 (i (1- (length args)))) | |
| 752 (while args | |
| 753 (setq rrr (ccl-check-register (car args) cmd)) | |
| 754 (ccl-embed-code 'write-register rrr i) | |
| 755 (setq args (cdr args) i (1- i))))) | |
| 756 | |
| 757 ((listp rrr) | |
| 758 ;; CMD has the form `(write (LEFT OP RIGHT))'. | |
| 759 (let ((left (car rrr)) | |
| 760 (op (get (nth 1 rrr) 'ccl-arith-code)) | |
| 761 (right (nth 2 rrr))) | |
| 762 (if (listp left) | |
| 763 (progn | |
| 764 ;; RRR has the form `((EXPR OP2 ARG) OP RIGHT)'. | |
| 765 ;; Compile the first term as `(r7 = (EXPR OP2 ARG))'. | |
| 766 (ccl-compile-expression 'r7 left) | |
| 767 (setq left 'r7))) | |
| 768 ;; Now RRR has the form `(ARG OP RIGHT)'. | |
| 769 (if (integer-or-char-p right) | |
| 770 (progn | |
| 771 (ccl-embed-code 'write-expr-const 0 (ash op 3) left) | |
| 772 (ccl-embed-data right)) | |
| 773 (ccl-check-register right rrr) | |
| 774 (ccl-embed-code 'write-expr-register 0 | |
| 775 (logior (ash op 3) | |
| 776 (get right 'ccl-register-number)))))) | |
| 777 | |
| 778 (t | |
| 779 (error "CCL: Invalid argument: %s" cmd)))) | |
| 780 nil) | |
| 781 | |
| 782 ;; Compile CALL statement. | |
| 783 (defun ccl-compile-call (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
784 (if (not (eql (length cmd) 2)) |
| 4080 | 785 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 786 (if (not (symbolp (nth 1 cmd))) | |
| 787 (error "CCL: Subroutine should be a symbol: %s" cmd)) | |
| 788 (ccl-embed-code 'call 1 0) | |
| 789 (ccl-embed-symbol (nth 1 cmd) 'ccl-program-idx) | |
| 790 nil) | |
| 791 | |
| 792 ;; Compile END statement. | |
| 793 (defun ccl-compile-end (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
794 (if (not (eql (length cmd) 1)) |
| 4080 | 795 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 796 (ccl-embed-code 'end 0 0) | |
| 797 t) | |
| 798 | |
| 799 ;; Compile read-multibyte-character | |
| 800 (defun ccl-compile-read-multibyte-character (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
801 (if (not (eql (length cmd) 3)) |
| 4080 | 802 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 803 (let ((RRR (nth 1 cmd)) | |
| 804 (rrr (nth 2 cmd))) | |
| 805 (ccl-check-register rrr cmd) | |
| 806 (ccl-check-register RRR cmd) | |
| 807 (ccl-embed-extended-command 'read-multibyte-character rrr RRR 0)) | |
| 808 nil) | |
| 809 | |
| 810 ;; Compile write-multibyte-character | |
| 811 (defun ccl-compile-write-multibyte-character (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
812 (if (not (eql (length cmd) 3)) |
| 4080 | 813 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 814 (let ((RRR (nth 1 cmd)) | |
| 815 (rrr (nth 2 cmd))) | |
| 816 (ccl-check-register rrr cmd) | |
| 817 (ccl-check-register RRR cmd) | |
| 818 (ccl-embed-extended-command 'write-multibyte-character rrr RRR 0)) | |
| 819 nil) | |
| 820 | |
| 821 ;; Compile translate-character | |
| 822 (defun ccl-compile-translate-character (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
823 (if (not (eql (length cmd) 4)) |
| 4080 | 824 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 825 (let ((Rrr (nth 1 cmd)) | |
| 826 (RRR (nth 2 cmd)) | |
| 827 (rrr (nth 3 cmd))) | |
| 828 (ccl-check-register rrr cmd) | |
| 829 (ccl-check-register RRR cmd) | |
| 830 (cond ((and (symbolp Rrr) (not (get Rrr 'ccl-register-number))) | |
| 831 (ccl-embed-extended-command 'translate-character-const-tbl | |
| 832 rrr RRR 0) | |
| 833 (ccl-embed-symbol Rrr 'translation-table-id)) | |
| 834 (t | |
| 835 (ccl-check-register Rrr cmd) | |
| 836 (ccl-embed-extended-command 'translate-character rrr RRR Rrr)))) | |
| 837 nil) | |
| 838 | |
| 839 ;; Compile mule-to-unicode | |
| 840 (defun ccl-compile-mule-to-unicode (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
841 (if (not (eql (length cmd) 3)) |
| 4080 | 842 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 843 (let ((RRR (nth 1 cmd)) | |
| 844 (rrr (nth 2 cmd))) | |
| 845 (ccl-check-register RRR cmd) | |
| 846 (ccl-check-register rrr cmd) | |
| 847 (ccl-embed-extended-command 'mule-to-unicode RRR rrr 0)) | |
| 848 nil) | |
| 849 | |
| 850 ;; Given a Unicode code point in register rrr, write the charset ID of the | |
| 851 ;; corresponding character in RRR, and the Mule-CCL form of its code in rrr. | |
| 852 (defun ccl-compile-unicode-to-mule (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
853 (if (not (eql (length cmd) 3)) |
| 4080 | 854 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 855 (let ((rrr (nth 1 cmd)) | |
| 856 (RRR (nth 2 cmd))) | |
| 857 (ccl-check-register rrr cmd) | |
| 858 (ccl-check-register RRR cmd) | |
| 859 (ccl-embed-extended-command 'unicode-to-mule rrr RRR 0)) | |
| 860 nil) | |
| 861 | |
| 862 ;; Compile lookup-integer | |
| 863 (defun ccl-compile-lookup-integer (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
864 (if (not (eql (length cmd) 4)) |
| 4080 | 865 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 866 (let ((Rrr (nth 1 cmd)) | |
| 867 (RRR (nth 2 cmd)) | |
| 868 (rrr (nth 3 cmd))) | |
| 869 (ccl-check-register RRR cmd) | |
| 870 (ccl-check-register rrr cmd) | |
| 871 (cond ((and (symbolp Rrr) (not (get Rrr 'ccl-register-number))) | |
| 872 (ccl-embed-extended-command 'lookup-int-const-tbl | |
| 873 rrr RRR 0) | |
| 874 (ccl-embed-symbol Rrr 'translation-hash-table-id)) | |
| 875 (t | |
| 876 (error "CCL: non-constant table: %s" cmd) | |
| 877 ;; not implemented: | |
| 878 (ccl-check-register Rrr cmd) | |
| 879 (ccl-embed-extended-command 'lookup-int rrr RRR 0)))) | |
| 880 nil) | |
| 881 | |
| 882 ;; Compile lookup-character | |
| 883 (defun ccl-compile-lookup-character (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
884 (if (not (eql (length cmd) 4)) |
| 4080 | 885 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 886 (let ((Rrr (nth 1 cmd)) | |
| 887 (RRR (nth 2 cmd)) | |
| 888 (rrr (nth 3 cmd))) | |
| 889 (ccl-check-register RRR cmd) | |
| 890 (ccl-check-register rrr cmd) | |
| 891 (cond ((and (symbolp Rrr) (not (get Rrr 'ccl-register-number))) | |
| 892 (ccl-embed-extended-command 'lookup-char-const-tbl | |
| 893 rrr RRR 0) | |
| 894 (ccl-embed-symbol Rrr 'translation-hash-table-id)) | |
| 895 (t | |
| 896 (error "CCL: non-constant table: %s" cmd) | |
| 897 ;; not implemented: | |
| 898 (ccl-check-register Rrr cmd) | |
| 899 (ccl-embed-extended-command 'lookup-char rrr RRR 0)))) | |
| 900 nil) | |
| 901 | |
| 902 (defun ccl-compile-iterate-multiple-map (cmd) | |
| 903 (ccl-compile-multiple-map-function 'iterate-multiple-map cmd) | |
| 904 nil) | |
| 905 | |
| 906 (defun ccl-compile-map-multiple (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
907 (if (not (eql (length cmd) 4)) |
| 4080 | 908 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 909 (let (func arg) | |
| 910 (setq func | |
| 911 (lambda (arg mp) | |
| 912 (let ((len 0) result add) | |
| 913 (while arg | |
| 914 (if (consp (car arg)) | |
| 915 (setq add (funcall func (car arg) t) | |
| 916 result (append result add) | |
| 917 add (+ (- (car add)) 1)) | |
| 918 (setq result | |
| 919 (append result | |
| 920 (list (car arg))) | |
| 921 add 1)) | |
| 922 (setq arg (cdr arg) | |
| 923 len (+ len add))) | |
| 924 (if mp | |
| 925 (cons (- len) result) | |
| 926 result)))) | |
| 927 (setq arg (append (list (nth 0 cmd) (nth 1 cmd) (nth 2 cmd)) | |
| 928 (funcall func (nth 3 cmd) nil))) | |
| 929 (ccl-compile-multiple-map-function 'map-multiple arg)) | |
| 930 nil) | |
| 931 | |
| 932 (defun ccl-compile-map-single (cmd) | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
933 (if (not (eql (length cmd) 4)) |
| 4080 | 934 (error "CCL: Invalid number of arguments: %s" cmd)) |
| 935 (let ((RRR (nth 1 cmd)) | |
| 936 (rrr (nth 2 cmd)) | |
| 937 (map (nth 3 cmd))) | |
| 938 (ccl-check-register rrr cmd) | |
| 939 (ccl-check-register RRR cmd) | |
| 940 (ccl-embed-extended-command 'map-single rrr RRR 0) | |
| 941 (cond ((symbolp map) | |
| 942 (if (get map 'code-conversion-map) | |
| 943 (ccl-embed-symbol map 'code-conversion-map-id) | |
| 944 (error "CCL: Invalid map: %s" map))) | |
| 945 (t | |
| 946 (error "CCL: Invalid type of arguments: %s" cmd)))) | |
| 947 nil) | |
| 948 | |
| 949 (defun ccl-compile-multiple-map-function (command cmd) | |
| 950 (if (< (length cmd) 4) | |
| 951 (error "CCL: Invalid number of arguments: %s" cmd)) | |
| 952 (let ((RRR (nth 1 cmd)) | |
| 953 (rrr (nth 2 cmd)) | |
| 954 (args (nthcdr 3 cmd)) | |
| 955 map) | |
| 956 (ccl-check-register rrr cmd) | |
| 957 (ccl-check-register RRR cmd) | |
| 958 (ccl-embed-extended-command command rrr RRR 0) | |
| 959 (ccl-embed-data (length args)) | |
| 960 (while args | |
| 961 (setq map (car args)) | |
| 962 (cond ((symbolp map) | |
| 963 (if (get map 'code-conversion-map) | |
| 964 (ccl-embed-symbol map 'code-conversion-map-id) | |
| 965 (error "CCL: Invalid map: %s" map))) | |
| 966 ((numberp map) | |
| 967 (ccl-embed-data map)) | |
| 968 (t | |
| 969 (error "CCL: Invalid type of arguments: %s" cmd))) | |
| 970 (setq args (cdr args))))) | |
| 971 | |
| 972 | |
| 973 ;;; CCL dump staffs | |
| 974 | |
| 975 ;; To avoid byte-compiler warning. | |
| 976 (defvar ccl-code) | |
| 977 | |
| 978 ;;;###autoload | |
| 979 (defun ccl-dump (ccl-code) | |
| 980 "Disassemble compiled CCL-CODE." | |
| 981 (let ((len (length ccl-code)) | |
| 982 (buffer-mag (aref ccl-code 0))) | |
| 983 (cond ((= buffer-mag 0) | |
| 984 (insert "Don't output anything.\n")) | |
| 985 ((= buffer-mag 1) | |
| 986 (insert "Out-buffer must be as large as in-buffer.\n")) | |
| 987 (t | |
| 988 (insert | |
| 989 (format "Out-buffer must be %d times bigger than in-buffer.\n" | |
| 990 buffer-mag)))) | |
| 991 (insert "Main-body:\n") | |
| 992 (setq ccl-current-ic 2) | |
| 993 (if (> (aref ccl-code 1) 0) | |
| 994 (progn | |
| 995 (while (< ccl-current-ic (aref ccl-code 1)) | |
| 996 (ccl-dump-1)) | |
| 997 (insert "At EOF:\n"))) | |
| 998 (while (< ccl-current-ic len) | |
| 999 (ccl-dump-1)) | |
| 1000 )) | |
| 1001 | |
| 1002 ;; Return a CCL code in `ccl-code' at `ccl-current-ic'. | |
| 1003 (defun ccl-get-next-code () | |
| 1004 (prog1 | |
| 1005 (aref ccl-code ccl-current-ic) | |
| 1006 (setq ccl-current-ic (1+ ccl-current-ic)))) | |
| 1007 | |
| 1008 (defun ccl-dump-1 () | |
| 1009 (let* ((code (ccl-get-next-code)) | |
| 1010 (cmd (aref ccl-code-table (logand code 31))) | |
| 1011 (rrr (ash (logand code 255) -5)) | |
| 1012 (cc (ash code -8))) | |
| 1013 (insert (format "%5d:[%s] " (1- ccl-current-ic) cmd)) | |
| 1014 (funcall (get cmd 'ccl-dump-function) rrr cc))) | |
| 1015 | |
| 1016 (defun ccl-dump-set-register (rrr cc) | |
| 1017 (insert (format "r%d = r%d\n" rrr cc))) | |
| 1018 | |
| 1019 (defun ccl-dump-set-short-const (rrr cc) | |
| 1020 (insert (format "r%d = %d\n" rrr cc))) | |
| 1021 | |
| 1022 (defun ccl-dump-set-const (rrr ignore) | |
| 1023 (insert (format "r%d = %d\n" rrr (ccl-get-next-code)))) | |
| 1024 | |
| 1025 (defun ccl-dump-set-array (rrr cc) | |
| 1026 (let ((rrr2 (logand cc 7)) | |
| 1027 (len (ash cc -3)) | |
| 1028 (i 0)) | |
| 1029 (insert (format "r%d = array[r%d] of length %d\n\t" | |
| 1030 rrr rrr2 len)) | |
| 1031 (while (< i len) | |
| 1032 (insert (format "%d " (ccl-get-next-code))) | |
| 1033 (setq i (1+ i))) | |
| 1034 (insert "\n"))) | |
| 1035 | |
| 1036 (defun ccl-dump-jump (ignore cc &optional address) | |
| 1037 (insert (format "jump to %d(" (+ (or address ccl-current-ic) cc))) | |
| 1038 (if (>= cc 0) | |
| 1039 (insert "+")) | |
| 1040 (insert (format "%d)\n" (1+ cc)))) | |
| 1041 | |
| 1042 (defun ccl-dump-jump-cond (rrr cc) | |
| 1043 (insert (format "if (r%d == 0), " rrr)) | |
| 1044 (ccl-dump-jump nil cc)) | |
| 1045 | |
| 1046 (defun ccl-dump-write-register-jump (rrr cc) | |
| 1047 (insert (format "write r%d, " rrr)) | |
| 1048 (ccl-dump-jump nil cc)) | |
| 1049 | |
| 1050 (defun ccl-dump-write-register-read-jump (rrr cc) | |
| 1051 (insert (format "write r%d, read r%d, " rrr rrr)) | |
| 1052 (ccl-dump-jump nil cc) | |
| 1053 (ccl-get-next-code) ; Skip dummy READ-JUMP | |
| 1054 ) | |
| 1055 | |
| 1056 (defun ccl-extract-arith-op (cc) | |
| 1057 (aref ccl-arith-table (ash cc -6))) | |
| 1058 | |
| 1059 (defun ccl-dump-write-expr-const (ignore cc) | |
| 1060 (insert (format "write (r%d %s %d)\n" | |
| 1061 (logand cc 7) | |
| 1062 (ccl-extract-arith-op cc) | |
| 1063 (ccl-get-next-code)))) | |
| 1064 | |
| 1065 (defun ccl-dump-write-expr-register (ignore cc) | |
| 1066 (insert (format "write (r%d %s r%d)\n" | |
| 1067 (logand cc 7) | |
| 1068 (ccl-extract-arith-op cc) | |
| 1069 (logand (ash cc -3) 7)))) | |
| 1070 | |
| 1071 (defun ccl-dump-insert-char (cc) | |
| 1072 (cond ((= cc ?\t) (insert " \"^I\"")) | |
| 1073 ((= cc ?\n) (insert " \"^J\"")) | |
| 1074 (t (insert (format " \"%c\"" cc))))) | |
| 1075 | |
| 1076 (defun ccl-dump-write-const-jump (ignore cc) | |
| 1077 (let ((address ccl-current-ic)) | |
| 1078 (insert "write char") | |
| 1079 (ccl-dump-insert-char (ccl-get-next-code)) | |
| 1080 (insert ", ") | |
| 1081 (ccl-dump-jump nil cc address))) | |
| 1082 | |
| 1083 (defun ccl-dump-write-const-read-jump (rrr cc) | |
| 1084 (let ((address ccl-current-ic)) | |
| 1085 (insert "write char") | |
| 1086 (ccl-dump-insert-char (ccl-get-next-code)) | |
| 1087 (insert (format ", read r%d, " rrr)) | |
| 1088 (ccl-dump-jump cc address) | |
| 1089 (ccl-get-next-code) ; Skip dummy READ-JUMP | |
| 1090 )) | |
| 1091 | |
| 1092 (defun ccl-dump-write-string-jump (ignore cc) | |
| 1093 (let ((address ccl-current-ic) | |
| 1094 (len (ccl-get-next-code)) | |
| 1095 (i 0)) | |
| 1096 (insert "write \"") | |
| 1097 (while (< i len) | |
| 1098 (let ((code (ccl-get-next-code))) | |
| 1099 (insert (ash code -16)) | |
| 1100 (if (< (1+ i) len) (insert (logand (ash code -8) 255))) | |
| 1101 (if (< (+ i 2) len) (insert (logand code 255)))) | |
| 1102 (setq i (+ i 3))) | |
| 1103 (insert "\", ") | |
| 1104 (ccl-dump-jump nil cc address))) | |
| 1105 | |
| 1106 (defun ccl-dump-write-array-read-jump (rrr cc) | |
| 1107 (let ((address ccl-current-ic) | |
| 1108 (len (ccl-get-next-code)) | |
| 1109 (i 0)) | |
| 1110 (insert (format "write array[r%d] of length %d,\n\t" rrr len)) | |
| 1111 (while (< i len) | |
| 1112 (ccl-dump-insert-char (ccl-get-next-code)) | |
| 1113 (setq i (1+ i))) | |
| 1114 (insert (format "\n\tthen read r%d, " rrr)) | |
| 1115 (ccl-dump-jump nil cc address) | |
| 1116 (ccl-get-next-code) ; Skip dummy READ-JUMP. | |
| 1117 )) | |
| 1118 | |
| 1119 (defun ccl-dump-read-jump (rrr cc) | |
| 1120 (insert (format "read r%d, " rrr)) | |
| 1121 (ccl-dump-jump nil cc)) | |
| 1122 | |
| 1123 (defun ccl-dump-branch (rrr len) | |
| 1124 (let ((jump-table-head ccl-current-ic) | |
| 1125 (i 0)) | |
| 1126 (insert (format "jump to array[r%d] of length %d\n\t" rrr len)) | |
| 1127 (while (<= i len) | |
| 1128 (insert (format "%d " (+ jump-table-head (ccl-get-next-code)))) | |
| 1129 (setq i (1+ i))) | |
| 1130 (insert "\n"))) | |
| 1131 | |
| 1132 (defun ccl-dump-read-register (rrr cc) | |
| 1133 (insert (format "read r%d (%d remaining)\n" rrr cc))) | |
| 1134 | |
| 1135 (defun ccl-dump-read-branch (rrr len) | |
| 1136 (insert (format "read r%d, " rrr)) | |
| 1137 (ccl-dump-branch rrr len)) | |
| 1138 | |
| 1139 (defun ccl-dump-write-register (rrr cc) | |
| 1140 (insert (format "write r%d (%d remaining)\n" rrr cc))) | |
| 1141 | |
| 1142 (defun ccl-dump-call (ignore cc) | |
| 1143 (insert (format "call subroutine #%d\n" cc))) | |
| 1144 | |
| 1145 (defun ccl-dump-write-const-string (rrr cc) | |
| 1146 (if (= rrr 0) | |
| 1147 (progn | |
| 1148 (insert "write char") | |
| 1149 (ccl-dump-insert-char cc) | |
| 1150 (newline)) | |
| 1151 (let ((len cc) | |
| 1152 (i 0)) | |
| 1153 (insert "write \"") | |
| 1154 (while (< i len) | |
| 1155 (let ((code (ccl-get-next-code))) | |
| 1156 (insert (format "%c" (lsh code -16))) | |
| 1157 (if (< (1+ i) len) | |
| 1158 (insert (format "%c" (logand (lsh code -8) 255)))) | |
| 1159 (if (< (+ i 2) len) | |
| 1160 (insert (format "%c" (logand code 255)))) | |
| 1161 (setq i (+ i 3)))) | |
| 1162 (insert "\"\n")))) | |
| 1163 | |
| 1164 (defun ccl-dump-write-array (rrr cc) | |
| 1165 (let ((i 0)) | |
| 1166 (insert (format "write array[r%d] of length %d\n\t" rrr cc)) | |
| 1167 (while (< i cc) | |
| 1168 (ccl-dump-insert-char (ccl-get-next-code)) | |
| 1169 (setq i (1+ i))) | |
| 1170 (insert "\n"))) | |
| 1171 | |
| 1172 (defun ccl-dump-end (&rest ignore) | |
| 1173 (insert "end\n")) | |
| 1174 | |
| 1175 (defun ccl-dump-set-assign-expr-const (rrr cc) | |
| 1176 (insert (format "r%d %s= %d\n" | |
| 1177 rrr | |
| 1178 (ccl-extract-arith-op cc) | |
| 1179 (ccl-get-next-code)))) | |
| 1180 | |
| 1181 (defun ccl-dump-set-assign-expr-register (rrr cc) | |
| 1182 (insert (format "r%d %s= r%d\n" | |
| 1183 rrr | |
| 1184 (ccl-extract-arith-op cc) | |
| 1185 (logand cc 7)))) | |
| 1186 | |
| 1187 (defun ccl-dump-set-expr-const (rrr cc) | |
| 1188 (insert (format "r%d = r%d %s %d\n" | |
| 1189 rrr | |
| 1190 (logand cc 7) | |
| 1191 (ccl-extract-arith-op cc) | |
| 1192 (ccl-get-next-code)))) | |
| 1193 | |
| 1194 (defun ccl-dump-set-expr-register (rrr cc) | |
| 1195 (insert (format "r%d = r%d %s r%d\n" | |
| 1196 rrr | |
| 1197 (logand cc 7) | |
| 1198 (ccl-extract-arith-op cc) | |
| 1199 (logand (ash cc -3) 7)))) | |
| 1200 | |
| 1201 (defun ccl-dump-jump-cond-expr-const (rrr cc) | |
| 1202 (let ((address ccl-current-ic)) | |
| 1203 (insert (format "if !(r%d %s %d), " | |
| 1204 rrr | |
| 1205 (aref ccl-arith-table (ccl-get-next-code)) | |
| 1206 (ccl-get-next-code))) | |
| 1207 (ccl-dump-jump nil cc address))) | |
| 1208 | |
| 1209 (defun ccl-dump-jump-cond-expr-register (rrr cc) | |
| 1210 (let ((address ccl-current-ic)) | |
| 1211 (insert (format "if !(r%d %s r%d), " | |
| 1212 rrr | |
| 1213 (aref ccl-arith-table (ccl-get-next-code)) | |
| 1214 (ccl-get-next-code))) | |
| 1215 (ccl-dump-jump nil cc address))) | |
| 1216 | |
| 1217 (defun ccl-dump-read-jump-cond-expr-const (rrr cc) | |
| 1218 (insert (format "read r%d, " rrr)) | |
| 1219 (ccl-dump-jump-cond-expr-const rrr cc)) | |
| 1220 | |
| 1221 (defun ccl-dump-read-jump-cond-expr-register (rrr cc) | |
| 1222 (insert (format "read r%d, " rrr)) | |
| 1223 (ccl-dump-jump-cond-expr-register rrr cc)) | |
| 1224 | |
| 1225 (defun ccl-dump-binary (ccl-code) | |
| 1226 (let ((len (length ccl-code)) | |
| 1227 (i 2)) | |
| 1228 (while (< i len) | |
| 1229 (let ((code (aref ccl-code i)) | |
| 1230 (j 27)) | |
| 1231 (while (>= j 0) | |
| 1232 (insert (if (= (logand code (ash 1 j)) 0) ?0 ?1)) | |
| 1233 (setq j (1- j))) | |
| 1234 (setq code (logand code 31)) | |
| 1235 (if (< code (length ccl-code-table)) | |
| 1236 (insert (format ":%s" (aref ccl-code-table code)))) | |
| 1237 (insert "\n")) | |
| 1238 (setq i (1+ i))))) | |
| 1239 | |
| 1240 (defun ccl-dump-ex-cmd (rrr cc) | |
| 1241 (let* ((RRR (logand cc #x7)) | |
| 1242 (Rrr (logand (ash cc -3) #x7)) | |
| 1243 (ex-op (aref ccl-extended-code-table (logand (ash cc -6) #x3fff)))) | |
| 1244 (insert (format "<%s> " ex-op)) | |
| 1245 (funcall (get ex-op 'ccl-dump-function) rrr RRR Rrr))) | |
| 1246 | |
| 1247 (defun ccl-dump-read-multibyte-character (rrr RRR Rrr) | |
| 1248 (insert (format "read-multibyte-character r%d r%d\n" RRR rrr))) | |
| 1249 | |
| 1250 (defun ccl-dump-write-multibyte-character (rrr RRR Rrr) | |
| 1251 (insert (format "write-multibyte-character r%d r%d\n" RRR rrr))) | |
| 1252 | |
| 1253 (defun ccl-dump-translate-character (rrr RRR Rrr) | |
| 1254 (insert (format "translation table(r%d) r%d r%d\n" Rrr RRR rrr))) | |
| 1255 | |
| 1256 (defun ccl-dump-translate-character-const-tbl (rrr RRR Rrr) | |
| 1257 (let ((tbl (ccl-get-next-code))) | |
| 1258 (insert (format "translation table(%S) r%d r%d\n" tbl RRR rrr)))) | |
| 1259 | |
| 1260 (defun ccl-dump-lookup-int-const-tbl (rrr RRR Rrr) | |
| 1261 (let ((tbl (ccl-get-next-code))) | |
| 1262 (insert (format "hash table(%S) r%d r%d\n" tbl RRR rrr)))) | |
| 1263 | |
| 1264 (defun ccl-dump-lookup-char-const-tbl (rrr RRR Rrr) | |
| 1265 (let ((tbl (ccl-get-next-code))) | |
| 1266 (insert (format "hash table(%S) r%d r%d\n" tbl RRR rrr)))) | |
| 1267 | |
| 1268 (defun ccl-dump-mule-to-unicode (rrr RRR Rrr) | |
| 1269 (insert (format "change chars in r%d and r%d to unicode\n" RRR rrr))) | |
| 1270 | |
| 1271 (defun ccl-dump-unicode-to-mule (rrr RRR Rrr) | |
| 1272 (insert (format "converter UCS code %d to a Mule char\n" rrr))) | |
| 1273 | |
| 1274 (defun ccl-dump-iterate-multiple-map (rrr RRR Rrr) | |
| 1275 (let ((notbl (ccl-get-next-code)) | |
| 1276 (i 0) id) | |
| 1277 (insert (format "iterate-multiple-map r%d r%d\n" RRR rrr)) | |
| 1278 (insert (format "\tnumber of maps is %d .\n\t [" notbl)) | |
| 1279 (while (< i notbl) | |
| 1280 (setq id (ccl-get-next-code)) | |
| 1281 (insert (format "%S" id)) | |
| 1282 (setq i (1+ i))) | |
| 1283 (insert "]\n"))) | |
| 1284 | |
| 1285 (defun ccl-dump-map-multiple (rrr RRR Rrr) | |
| 1286 (let ((notbl (ccl-get-next-code)) | |
| 1287 (i 0) id) | |
| 1288 (insert (format "map-multiple r%d r%d\n" RRR rrr)) | |
| 1289 (insert (format "\tnumber of maps and separators is %d\n\t [" notbl)) | |
| 1290 (while (< i notbl) | |
| 1291 (setq id (ccl-get-next-code)) | |
| 1292 (if (= id -1) | |
| 1293 (insert "]\n\t [") | |
| 1294 (insert (format "%S " id))) | |
| 1295 (setq i (1+ i))) | |
| 1296 (insert "]\n"))) | |
| 1297 | |
| 1298 (defun ccl-dump-map-single (rrr RRR Rrr) | |
| 1299 (let ((id (ccl-get-next-code))) | |
| 1300 (insert (format "map-single r%d r%d map(%S)\n" RRR rrr id)))) | |
| 1301 | |
| 1302 | |
| 1303 ;; CCL emulation staffs | |
| 1304 | |
| 1305 ;; Not yet implemented. | |
| 1306 | |
| 1307 ;; Auto-loaded functions. | |
| 1308 | |
| 1309 ;;;###autoload | |
| 1310 (defmacro declare-ccl-program (name &optional vector) | |
| 1311 "Declare NAME as a name of CCL program. | |
| 1312 | |
| 1313 This macro exists for backward compatibility. In the old version of | |
| 1314 Emacs, to compile a CCL program which calls another CCL program not | |
| 1315 yet defined, it must be declared as a CCL program in advance. But, | |
| 1316 now CCL program names are resolved not at compile time but before | |
| 1317 execution. | |
| 1318 | |
| 1319 Optional arg VECTOR is a compiled CCL code of the CCL program." | |
| 1320 `(put ',name 'ccl-program-idx (register-ccl-program ',name ,vector))) | |
| 1321 | |
| 1322 ;;;###autoload | |
| 1323 (defmacro define-ccl-program (name ccl-program &optional doc) | |
| 1324 "Set NAME to be the compiled CCL code of CCL-PROGRAM. | |
| 1325 | |
| 1326 CCL-PROGRAM has this form: | |
| 1327 (BUFFER_MAGNIFICATION | |
| 1328 CCL_MAIN_CODE | |
| 1329 [ CCL_EOF_CODE ]) | |
| 1330 | |
| 1331 BUFFER_MAGNIFICATION is an integer value specifying the approximate | |
| 1332 output buffer magnification size compared with the bytes of input data | |
| 1333 text. If the value is zero, the CCL program can't execute `read' and | |
| 1334 `write' commands. | |
| 1335 | |
| 1336 CCL_MAIN_CODE and CCL_EOF_CODE are CCL program codes. CCL_MAIN_CODE is | |
| 1337 executed first. If there are no more input data when a `read' command is | |
| 1338 executed in CCL_MAIN_CODE, CCL_EOF_CODE is executed. If CCL_MAIN_CODE is | |
| 1339 terminated, CCL_EOF_CODE is not executed. | |
| 1340 | |
| 1341 Here's the syntax of CCL program code in BNF notation. The lines starting | |
| 1342 with two semicolons (and optional leading spaces) describe the semantics. | |
| 1343 | |
| 1344 CCL_MAIN_CODE := CCL_BLOCK | |
| 1345 | |
| 1346 CCL_EOF_CODE := CCL_BLOCK | |
| 1347 | |
| 1348 CCL_BLOCK := STATEMENT | (STATEMENT [STATEMENT ...]) | |
| 1349 | |
| 1350 STATEMENT := | |
| 1351 SET | IF | BRANCH | LOOP | REPEAT | BREAK | READ | WRITE | CALL | |
| 1352 | TRANSLATE | MAP | LOOKUP | END | |
| 1353 | |
| 1354 SET := (REG = EXPRESSION) | |
| 1355 | (REG ASSIGNMENT_OPERATOR EXPRESSION) | |
| 1356 ;; The following form is the same as (r0 = INT-OR-CHAR). | |
| 1357 | INT-OR-CHAR | |
| 1358 | |
| 1359 EXPRESSION := ARG | (EXPRESSION OPERATOR ARG) | |
| 1360 | |
| 1361 ;; Evaluate EXPRESSION. If the result is nonzero, execute | |
| 1362 ;; CCL_BLOCK_0. Otherwise, execute CCL_BLOCK_1. | |
| 1363 IF := (if EXPRESSION CCL_BLOCK_0 [CCL_BLOCK_1]) | |
| 1364 | |
| 1365 ;; Evaluate EXPRESSION. Provided that the result is N, execute | |
| 1366 ;; CCL_BLOCK_N. | |
| 1367 BRANCH := (branch EXPRESSION CCL_BLOCK_0 [CCL_BLOCK_1 ...]) | |
| 1368 | |
| 1369 ;; Execute STATEMENTs until (break) or (end) is executed. | |
| 1370 LOOP := (loop STATEMENT [STATEMENT ...]) | |
| 1371 | |
| 1372 ;; Terminate the innermost loop. | |
| 1373 BREAK := (break) | |
| 1374 | |
| 1375 REPEAT := | |
| 1376 ;; Jump to the head of the innermost loop. | |
| 1377 (repeat) | |
| 1378 ;; Same as: ((write [REG | INT-OR-CHAR | string]) | |
| 1379 ;; (repeat)) | |
| 1380 | (write-repeat [REG | INT-OR-CHAR | string]) | |
| 1381 ;; Same as: ((write REG [ARRAY]) | |
| 1382 ;; (read REG) | |
| 1383 ;; (repeat)) | |
| 1384 | (write-read-repeat REG [ARRAY]) | |
| 1385 ;; Same as: ((write INT-OR-CHAR) | |
| 1386 ;; (read REG) | |
| 1387 ;; (repeat)) | |
| 1388 | (write-read-repeat REG INT-OR-CHAR) | |
| 1389 | |
| 1390 READ := ;; Set REG_0 to a byte read from the input text, set REG_1 | |
| 1391 ;; to the next byte read, and so on. Note that \"byte\" here means | |
| 1392 ;; \"some octet from XEmacs' internal representation\", which may | |
| 1393 ;; not be that useful to you when non-ASCII characters are involved. | |
| 1394 ;; | |
| 1395 ;; Yes, this is exactly the opposite of what (write ...) does. | |
| 1396 (read REG_0 [REG_1 ...]) | |
| 1397 ;; Same as: ((read REG) | |
| 1398 ;; (if (REG OPERATOR ARG) CCL_BLOCK_0 CCL_BLOCK_1)) | |
| 1399 | (read-if (REG OPERATOR ARG) CCL_BLOCK_0 [CCL_BLOCK_1]) | |
| 1400 ;; Same as: ((read REG) | |
| 1401 ;; (branch REG CCL_BLOCK_0 [CCL_BLOCK_1 ...])) | |
| 1402 | (read-branch REG CCL_BLOCK_0 [CCL_BLOCK_1 ...]) | |
| 1403 ;; Read a character from the input text, splitting it into its | |
| 1404 ;; multibyte representation. Set REG_0 to the charset ID of the | |
| 1405 ;; character, and set REG_1 to the code point of the character. If | |
| 1406 ;; the dimension of charset is two, set REG_1 to ((CODE0 << 7) | | |
| 1407 ;; CODE1), where CODE0 is the first code point and CODE1 is the | |
| 1408 ;; second code point. | |
| 1409 | (read-multibyte-character REG_0 REG_1) | |
| 1410 | |
| 1411 WRITE := | |
| 1412 ;; Write REG_0, REG_1, ... to the output buffer. If REG_N is | |
| 1413 ;; a multibyte character, write the corresponding multibyte | |
| 1414 ;; representation. | |
| 1415 (write REG_0 [REG_1 ...]) | |
| 1416 ;; Same as: ((r7 = EXPRESSION) | |
| 1417 ;; (write r7)) | |
| 1418 | (write EXPRESSION) | |
| 1419 ;; Write the value of `INT-OR-CHAR' to the output buffer. If it | |
| 1420 ;; is a multibyte character, write the corresponding multibyte | |
| 1421 ;; representation. | |
| 1422 | (write INT-OR-CHAR) | |
| 1423 ;; Write the byte sequence of `string' as is to the output | |
| 1424 ;; buffer. It is encoded by binary coding system, thus, | |
| 1425 ;; by this operation, you cannot write multibyte string | |
| 1426 ;; as it is. | |
| 1427 | (write string) | |
| 1428 ;; Same as: (write string) | |
| 1429 | string | |
| 1430 ;; Provided that the value of REG is N, write Nth element of | |
| 1431 ;; ARRAY to the output buffer. If it is a multibyte | |
| 1432 ;; character, write the corresponding multibyte | |
| 1433 ;; representation. | |
| 1434 | (write REG ARRAY) | |
| 1435 ;; Write a multibyte representation of a character whose | |
| 1436 ;; charset ID is REG_0 and code point is REG_1. If the | |
| 1437 ;; dimension of the charset is two, REG_1 should be ((CODE0 << | |
| 1438 ;; 7) | CODE1), where CODE0 is the first code point and CODE1 | |
| 1439 ;; is the second code point of the character. | |
| 1440 | (write-multibyte-character REG_0 REG_1) | |
| 1441 | |
| 1442 ;; Call CCL program whose name is ccl-program-name. | |
| 1443 CALL := (call ccl-program-name) | |
| 1444 | |
| 1445 TRANSLATE := ;; Not implemented under XEmacs, except mule-to-unicode and | |
| 1446 ;; unicode-to-mule. | |
| 1447 (translate-character REG(table) REG(charset) REG(codepoint)) | |
| 1448 | (translate-character SYMBOL REG(charset) REG(codepoint)) | |
| 1449 | (mule-to-unicode REG(charset) REG(codepoint)) | |
| 1450 | (unicode-to-mule REG(unicode,code) REG(CHARSET)) | |
| 1451 | |
| 1452 LOOKUP := | |
| 1453 (lookup-character SYMBOL REG(charset) REG(codepoint)) | |
| 1454 | (lookup-integer SYMBOL REG(integer)) | |
| 1455 ;; SYMBOL refers to a table defined by `define-hash-translation-table'. | |
| 1456 | |
| 1457 MAP := | |
| 1458 (iterate-multiple-map REG REG MAP-IDs) | |
| 1459 | (map-multiple REG REG (MAP-SET)) | |
| 1460 | (map-single REG REG MAP-ID) | |
| 1461 MAP-IDs := MAP-ID ... | |
| 1462 MAP-SET := MAP-IDs | (MAP-IDs) MAP-SET | |
| 1463 MAP-ID := INT-OR-CHAR | |
| 1464 | |
| 1465 ;; Terminate the CCL program. | |
| 1466 END := (end) | |
| 1467 | |
| 1468 ;; CCL registers. These can contain any integer value. As r7 is used by the | |
| 1469 ;; CCL interpreter itself, its value can change unexpectedly. | |
| 1470 REG := r0 | r1 | r2 | r3 | r4 | r5 | r6 | r7 | |
| 1471 | |
| 1472 ARG := REG | INT-OR-CHAR | |
| 1473 | |
| 1474 OPERATOR := | |
| 1475 ;; Normal arithmetical operators (same meaning as C code). | |
| 1476 + | - | * | / | % | |
| 1477 | |
| 1478 ;; Bitwise operators (same meaning as C code) | |
| 1479 | & | `|' | ^ | |
| 1480 | |
| 1481 ;; Shifting operators (same meaning as C code) | |
| 1482 | << | >> | |
| 1483 | |
| 1484 ;; (REG = ARG_0 <8 ARG_1) means: | |
| 1485 ;; (REG = ((ARG_0 << 8) | ARG_1)) | |
| 1486 | <8 | |
| 1487 | |
| 1488 ;; (REG = ARG_0 >8 ARG_1) means: | |
| 1489 ;; ((REG = (ARG_0 >> 8)) | |
| 1490 ;; (r7 = (ARG_0 & 255))) | |
| 1491 | >8 | |
| 1492 | |
| 1493 ;; (REG = ARG_0 // ARG_1) means: | |
| 1494 ;; ((REG = (ARG_0 / ARG_1)) | |
| 1495 ;; (r7 = (ARG_0 % ARG_1))) | |
| 1496 | // | |
| 1497 | |
| 1498 ;; Normal comparing operators (same meaning as C code) | |
| 1499 | < | > | == | <= | >= | != | |
| 1500 | |
| 1501 ;; If ARG_0 and ARG_1 are higher and lower byte of Shift-JIS | |
| 1502 ;; code, and CHAR is the corresponding JISX0208 character, | |
| 1503 ;; (REG = ARG_0 de-sjis ARG_1) means: | |
| 1504 ;; ((REG = CODE0) | |
| 1505 ;; (r7 = CODE1)) | |
| 1506 ;; where CODE0 is the first code point of CHAR, CODE1 is the | |
| 1507 ;; second code point of CHAR. | |
| 1508 | de-sjis | |
| 1509 | |
| 1510 ;; If ARG_0 and ARG_1 are the first and second code point of | |
| 1511 ;; JISX0208 character CHAR, and SJIS is the correponding | |
| 1512 ;; Shift-JIS code, | |
| 1513 ;; (REG = ARG_0 en-sjis ARG_1) means: | |
| 1514 ;; ((REG = HIGH) | |
| 1515 ;; (r7 = LOW)) | |
| 1516 ;; where HIGH is the higher byte of SJIS, LOW is the lower | |
| 1517 ;; byte of SJIS. | |
| 1518 | en-sjis | |
| 1519 | |
| 1520 ASSIGNMENT_OPERATOR := | |
| 1521 ;; Same meaning as C code | |
| 1522 += | -= | *= | /= | %= | &= | `|=' | ^= | <<= | >>= | |
| 1523 | |
| 1524 ;; (REG <8= ARG) is the same as: | |
| 1525 ;; ((REG <<= 8) | |
| 1526 ;; (REG |= ARG)) | |
| 1527 | <8= | |
| 1528 | |
| 1529 ;; (REG >8= ARG) is the same as: | |
| 1530 ;; ((r7 = (REG & 255)) | |
| 1531 ;; (REG >>= 8)) | |
| 1532 | |
| 1533 ;; (REG //= ARG) is the same as: | |
| 1534 ;; ((r7 = (REG % ARG)) | |
| 1535 ;; (REG /= ARG)) | |
| 1536 | //= | |
| 1537 | |
| 1538 ARRAY := `[' INT-OR-CHAR ... `]' | |
| 1539 | |
| 1540 INT-OR-CHAR := integer | character | |
| 1541 " | |
| 1542 `(let ((prog ,(ccl-compile (eval ccl-program)))) | |
| 1543 (defconst ,name prog ,doc) | |
| 1544 (put ',name 'ccl-program-idx (register-ccl-program ',name prog)) | |
| 1545 nil)) | |
| 1546 | |
| 1547 ;;;###autoload | |
| 1548 (defmacro check-ccl-program (ccl-program &optional name) | |
| 1549 "Check validity of CCL-PROGRAM. | |
| 1550 If CCL-PROGRAM is a symbol denoting a CCL program, return | |
| 1551 CCL-PROGRAM, else return nil. | |
| 1552 If CCL-PROGRAM is a vector and optional arg NAME (symbol) is supplied, | |
| 1553 register CCL-PROGRAM by name NAME, and return NAME." | |
| 1554 `(if (ccl-program-p ,ccl-program) | |
| 1555 (if (vectorp ,ccl-program) | |
| 1556 (progn | |
| 1557 (register-ccl-program ,name ,ccl-program) | |
| 1558 ,name) | |
| 1559 ,ccl-program))) | |
| 1560 | |
| 1561 (provide 'ccl) | |
| 1562 | |
|
5366
f00192e1cd49
Examining the result of #'length: `eql', not `=', it's better style & cheaper
Aidan Kehoe <kehoea@parhasard.net>
parents:
4080
diff
changeset
|
1563 ;; ccl.el ends here |