mksyntax.c revision 214305
11553Srgrimes/*- 21553Srgrimes * Copyright (c) 1991, 1993 31553Srgrimes * The Regents of the University of California. All rights reserved. 41553Srgrimes * 51553Srgrimes * This code is derived from software contributed to Berkeley by 61553Srgrimes * Kenneth Almquist. 71553Srgrimes * 81553Srgrimes * Redistribution and use in source and binary forms, with or without 91553Srgrimes * modification, are permitted provided that the following conditions 101553Srgrimes * are met: 111553Srgrimes * 1. Redistributions of source code must retain the above copyright 121553Srgrimes * notice, this list of conditions and the following disclaimer. 131553Srgrimes * 2. Redistributions in binary form must reproduce the above copyright 141553Srgrimes * notice, this list of conditions and the following disclaimer in the 151553Srgrimes * documentation and/or other materials provided with the distribution. 161553Srgrimes * 4. Neither the name of the University nor the names of its contributors 171553Srgrimes * may be used to endorse or promote products derived from this software 181553Srgrimes * without specific prior written permission. 191553Srgrimes * 201553Srgrimes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 211553Srgrimes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 221553Srgrimes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 231553Srgrimes * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 241553Srgrimes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 251553Srgrimes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 261553Srgrimes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 271553Srgrimes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 281553Srgrimes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 291553Srgrimes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 301553Srgrimes * SUCH DAMAGE. 311553Srgrimes */ 321553Srgrimes 331553Srgrimes#if 0 341553Srgrimes#ifndef lint 351553Srgrimesstatic char const copyright[] = 361553Srgrimes"@(#) Copyright (c) 1991, 1993\n\ 371553Srgrimes The Regents of the University of California. All rights reserved.\n"; 381553Srgrimes#endif /* not lint */ 391553Srgrimes 401553Srgrimes#ifndef lint 411553Srgrimesstatic char sccsid[] = "@(#)mksyntax.c 8.2 (Berkeley) 5/4/95"; 421553Srgrimes#endif /* not lint */ 431553Srgrimes#endif 441553Srgrimes#include <sys/cdefs.h> 451553Srgrimes__FBSDID("$FreeBSD: head/bin/sh/mksyntax.c 214305 2010-10-24 22:25:38Z jilles $"); 461553Srgrimes 471553Srgrimes/* 481553Srgrimes * This program creates syntax.h and syntax.c. 491553Srgrimes */ 501553Srgrimes 511553Srgrimes#include <stdio.h> 521553Srgrimes#include <stdlib.h> 531553Srgrimes#include <string.h> 541553Srgrimes#include "parser.h" 551553Srgrimes 561553Srgrimes 571553Srgrimesstruct synclass { 581553Srgrimes const char *name; 591553Srgrimes const char *comment; 601553Srgrimes}; 611553Srgrimes 621553Srgrimes/* Syntax classes */ 631553Srgrimesstruct synclass synclass[] = { 641553Srgrimes { "CWORD", "character is nothing special" }, 651553Srgrimes { "CNL", "newline character" }, 661553Srgrimes { "CBACK", "a backslash character" }, 671553Srgrimes { "CSQUOTE", "single quote" }, 681553Srgrimes { "CDQUOTE", "double quote" }, 691553Srgrimes { "CENDQUOTE", "a terminating quote" }, 701553Srgrimes { "CBQUOTE", "backwards single quote" }, 711553Srgrimes { "CVAR", "a dollar sign" }, 721553Srgrimes { "CENDVAR", "a '}' character" }, 731553Srgrimes { "CLP", "a left paren in arithmetic" }, 741553Srgrimes { "CRP", "a right paren in arithmetic" }, 751553Srgrimes { "CEOF", "end of file" }, 761553Srgrimes { "CCTL", "like CWORD, except it must be escaped" }, 771553Srgrimes { "CSPCL", "these terminate a word" }, 781553Srgrimes { "CIGN", "character should be ignored" }, 791553Srgrimes { NULL, NULL } 801553Srgrimes}; 811553Srgrimes 821553Srgrimes 831553Srgrimes/* 841553Srgrimes * Syntax classes for is_ functions. Warning: if you add new classes 851553Srgrimes * you may have to change the definition of the is_in_name macro. 861553Srgrimes */ 871553Srgrimesstruct synclass is_entry[] = { 881553Srgrimes { "ISDIGIT", "a digit" }, 891553Srgrimes { "ISUPPER", "an upper case letter" }, 901553Srgrimes { "ISLOWER", "a lower case letter" }, 911553Srgrimes { "ISUNDER", "an underscore" }, 921553Srgrimes { "ISSPECL", "the name of a special parameter" }, 931553Srgrimes { NULL, NULL } 941553Srgrimes}; 951553Srgrimes 96static char writer[] = "\ 97/*\n\ 98 * This file was generated by the mksyntax program.\n\ 99 */\n\ 100\n"; 101 102 103static FILE *cfile; 104static FILE *hfile; 105static const char *syntax[513]; 106static int base; 107static int size; /* number of values which a char variable can have */ 108static int nbits; /* number of bits in a character */ 109static int digit_contig;/* true if digits are contiguous */ 110 111static void filltable(const char *); 112static void init(void); 113static void add(const char *, const char *); 114static void print(const char *); 115static void output_type_macros(void); 116static void digit_convert(void); 117 118int 119main(int argc __unused, char **argv __unused) 120{ 121 char c; 122 char d; 123 int sign; 124 int i; 125 char buf[80]; 126 int pos; 127 static char digit[] = "0123456789"; 128 129 /* Create output files */ 130 if ((cfile = fopen("syntax.c", "w")) == NULL) { 131 perror("syntax.c"); 132 exit(2); 133 } 134 if ((hfile = fopen("syntax.h", "w")) == NULL) { 135 perror("syntax.h"); 136 exit(2); 137 } 138 fputs(writer, hfile); 139 fputs(writer, cfile); 140 141 /* Determine the characteristics of chars. */ 142 c = -1; 143 sign = (c > 0) ? 0 : 1; 144 for (nbits = 1 ; ; nbits++) { 145 d = (1 << nbits) - 1; 146 if (d == c) 147 break; 148 } 149#if 0 150 printf("%s %d bit chars\n", sign? "signed" : "unsigned", nbits); 151#endif 152 if (nbits > 9) { 153 fputs("Characters can't have more than 9 bits\n", stderr); 154 exit(2); 155 } 156 size = (1 << nbits) + 1; 157 base = 1; 158 if (sign) 159 base += 1 << (nbits - 1); 160 digit_contig = 1; 161 for (i = 0 ; i < 10 ; i++) { 162 if (digit[i] != '0' + i) 163 digit_contig = 0; 164 } 165 166 fputs("#include <sys/cdefs.h>\n", hfile); 167 fputs("#include <ctype.h>\n", hfile); 168 169 /* Generate the #define statements in the header file */ 170 fputs("/* Syntax classes */\n", hfile); 171 for (i = 0 ; synclass[i].name ; i++) { 172 sprintf(buf, "#define %s %d", synclass[i].name, i); 173 fputs(buf, hfile); 174 for (pos = strlen(buf) ; pos < 32 ; pos = (pos + 8) & ~07) 175 putc('\t', hfile); 176 fprintf(hfile, "/* %s */\n", synclass[i].comment); 177 } 178 putc('\n', hfile); 179 fputs("/* Syntax classes for is_ functions */\n", hfile); 180 for (i = 0 ; is_entry[i].name ; i++) { 181 sprintf(buf, "#define %s %#o", is_entry[i].name, 1 << i); 182 fputs(buf, hfile); 183 for (pos = strlen(buf) ; pos < 32 ; pos = (pos + 8) & ~07) 184 putc('\t', hfile); 185 fprintf(hfile, "/* %s */\n", is_entry[i].comment); 186 } 187 putc('\n', hfile); 188 fprintf(hfile, "#define SYNBASE %d\n", base); 189 fprintf(hfile, "#define PEOF %d\n\n", -base); 190 putc('\n', hfile); 191 fputs("#define BASESYNTAX (basesyntax + SYNBASE)\n", hfile); 192 fputs("#define DQSYNTAX (dqsyntax + SYNBASE)\n", hfile); 193 fputs("#define SQSYNTAX (sqsyntax + SYNBASE)\n", hfile); 194 fputs("#define ARISYNTAX (arisyntax + SYNBASE)\n", hfile); 195 putc('\n', hfile); 196 output_type_macros(); /* is_digit, etc. */ 197 putc('\n', hfile); 198 199 /* Generate the syntax tables. */ 200 fputs("#include \"shell.h\"\n", cfile); 201 fputs("#include \"syntax.h\"\n\n", cfile); 202 init(); 203 fputs("/* syntax table used when not in quotes */\n", cfile); 204 add("\n", "CNL"); 205 add("\\", "CBACK"); 206 add("'", "CSQUOTE"); 207 add("\"", "CDQUOTE"); 208 add("`", "CBQUOTE"); 209 add("$", "CVAR"); 210 add("}", "CENDVAR"); 211 add("<>();&| \t", "CSPCL"); 212 print("basesyntax"); 213 init(); 214 fputs("\n/* syntax table used when in double quotes */\n", cfile); 215 add("\n", "CNL"); 216 add("\\", "CBACK"); 217 add("\"", "CENDQUOTE"); 218 add("`", "CBQUOTE"); 219 add("$", "CVAR"); 220 add("}", "CENDVAR"); 221 /* ':/' for tilde expansion, '-' for [a\-x] pattern ranges */ 222 add("!*?[=~:/-", "CCTL"); 223 print("dqsyntax"); 224 init(); 225 fputs("\n/* syntax table used when in single quotes */\n", cfile); 226 add("\n", "CNL"); 227 add("'", "CENDQUOTE"); 228 /* ':/' for tilde expansion, '-' for [a\-x] pattern ranges */ 229 add("!*?[=~:/-", "CCTL"); 230 print("sqsyntax"); 231 init(); 232 fputs("\n/* syntax table used when in arithmetic */\n", cfile); 233 add("\n", "CNL"); 234 add("\\", "CBACK"); 235 add("`", "CBQUOTE"); 236 add("\"", "CIGN"); 237 add("$", "CVAR"); 238 add("}", "CENDVAR"); 239 add("(", "CLP"); 240 add(")", "CRP"); 241 print("arisyntax"); 242 filltable("0"); 243 fputs("\n/* character classification table */\n", cfile); 244 add("0123456789", "ISDIGIT"); 245 add("abcdefghijklmnopqrstuvwxyz", "ISLOWER"); 246 add("ABCDEFGHIJKLMNOPQRSTUVWXYZ", "ISUPPER"); 247 add("_", "ISUNDER"); 248 add("#?$!-*@", "ISSPECL"); 249 print("is_type"); 250 if (! digit_contig) 251 digit_convert(); 252 exit(0); 253} 254 255 256 257/* 258 * Clear the syntax table. 259 */ 260 261static void 262filltable(const char *dftval) 263{ 264 int i; 265 266 for (i = 0 ; i < size ; i++) 267 syntax[i] = dftval; 268} 269 270 271/* 272 * Initialize the syntax table with default values. 273 */ 274 275static void 276init(void) 277{ 278 filltable("CWORD"); 279 syntax[0] = "CEOF"; 280 syntax[base + CTLESC] = "CCTL"; 281 syntax[base + CTLVAR] = "CCTL"; 282 syntax[base + CTLENDVAR] = "CCTL"; 283 syntax[base + CTLBACKQ] = "CCTL"; 284 syntax[base + CTLBACKQ + CTLQUOTE] = "CCTL"; 285 syntax[base + CTLARI] = "CCTL"; 286 syntax[base + CTLENDARI] = "CCTL"; 287 syntax[base + CTLQUOTEMARK] = "CCTL"; 288} 289 290 291/* 292 * Add entries to the syntax table. 293 */ 294 295static void 296add(const char *p, const char *type) 297{ 298 while (*p) 299 syntax[*p++ + base] = type; 300} 301 302 303 304/* 305 * Output the syntax table. 306 */ 307 308static void 309print(const char *name) 310{ 311 int i; 312 int col; 313 314 fprintf(hfile, "extern const char %s[];\n", name); 315 fprintf(cfile, "const char %s[%d] = {\n", name, size); 316 col = 0; 317 for (i = 0 ; i < size ; i++) { 318 if (i == 0) { 319 fputs(" ", cfile); 320 } else if ((i & 03) == 0) { 321 fputs(",\n ", cfile); 322 col = 0; 323 } else { 324 putc(',', cfile); 325 while (++col < 9 * (i & 03)) 326 putc(' ', cfile); 327 } 328 fputs(syntax[i], cfile); 329 col += strlen(syntax[i]); 330 } 331 fputs("\n};\n", cfile); 332} 333 334 335 336/* 337 * Output character classification macros (e.g. is_digit). If digits are 338 * contiguous, we can test for them quickly. 339 */ 340 341static const char *macro[] = { 342 "#define is_digit(c)\t((is_type+SYNBASE)[(int)c] & ISDIGIT)", 343 "#define is_eof(c)\t((c) == PEOF)", 344 "#define is_alpha(c)\t(((c) < CTLESC || (c) > CTLQUOTEMARK) && (is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER))", 345 "#define is_name(c)\t(((c) < CTLESC || (c) > CTLQUOTEMARK) && (is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER|ISUNDER))", 346 "#define is_in_name(c)\t(((c) < CTLESC || (c) > CTLQUOTEMARK) && (is_type+SYNBASE)[(int)c] & (ISUPPER|ISLOWER|ISUNDER|ISDIGIT))", 347 "#define is_special(c)\t((is_type+SYNBASE)[(int)c] & (ISSPECL|ISDIGIT))", 348 NULL 349}; 350 351static void 352output_type_macros(void) 353{ 354 const char **pp; 355 356 if (digit_contig) 357 macro[0] = "#define is_digit(c)\t((unsigned int)((c) - '0') <= 9)"; 358 for (pp = macro ; *pp ; pp++) 359 fprintf(hfile, "%s\n", *pp); 360 if (digit_contig) 361 fputs("#define digit_val(c)\t((c) - '0')\n", hfile); 362 else 363 fputs("#define digit_val(c)\t(digit_value[c])\n", hfile); 364} 365 366 367 368/* 369 * Output digit conversion table (if digits are not contiguous). 370 */ 371 372static void 373digit_convert(void) 374{ 375 int maxdigit; 376 static char digit[] = "0123456789"; 377 char *p; 378 int i; 379 380 maxdigit = 0; 381 for (p = digit ; *p ; p++) 382 if (*p > maxdigit) 383 maxdigit = *p; 384 fputs("extern const char digit_value[];\n", hfile); 385 fputs("\n\nconst char digit_value[] = {\n", cfile); 386 for (i = 0 ; i <= maxdigit ; i++) { 387 for (p = digit ; *p && *p != i ; p++); 388 if (*p == '\0') 389 p = digit; 390 fprintf(cfile, " %d,\n", (int)(p - digit)); 391 } 392 fputs("};\n", cfile); 393} 394