1 /* $NetBSD: db_lex.c,v 1.26 2020/07/29 23:29:42 uwe Exp $ */ 2 3 /* 4 * Mach Operating System 5 * Copyright (c) 1991,1990 Carnegie Mellon University 6 * All Rights Reserved. 7 * 8 * Permission to use, copy, modify and distribute this software and its 9 * documentation is hereby granted, provided that both the copyright 10 * notice and this permission notice appear in all copies of the 11 * software, derivative works or modified versions, and any portions 12 * thereof, and that both notices appear in supporting documentation. 13 * 14 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" 15 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR 16 * ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. 17 * 18 * Carnegie Mellon requests users of this software to return to 19 * 20 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU 21 * School of Computer Science 22 * Carnegie Mellon University 23 * Pittsburgh PA 15213-3890 24 * 25 * any improvements or extensions that they make and grant Carnegie the 26 * rights to redistribute these changes. 27 * 28 * Author: David B. Golub, Carnegie Mellon University 29 * Date: 7/90 30 */ 31 32 /* 33 * Lexical analyzer. 34 */ 35 36 #include <sys/cdefs.h> 37 __KERNEL_RCSID(0, "$NetBSD: db_lex.c,v 1.26 2020/07/29 23:29:42 uwe Exp $"); 38 39 #include <sys/param.h> 40 #include <sys/systm.h> 41 #include <sys/cpu.h> 42 43 #include <ddb/ddb.h> 44 45 db_expr_t db_tok_number; 46 char db_tok_string[TOK_STRING_SIZE]; 47 48 static char db_line[DB_LINE_MAXLEN]; 49 static const char *db_lp; 50 static const char *db_endlp; 51 52 static int db_look_char = 0; 53 static int db_look_token = 0; 54 55 static void db_flush_line(void); 56 static int db_read_char(void); 57 static void db_unread_char(int); 58 static int db_lex(void); 59 60 int 61 db_read_line(void) 62 { 63 int i; 64 65 #ifdef _KERNEL 66 /* 67 * crash(8) prints the prompt using libedit. That's why we used to 68 * print it in db_readline(). But now people are using db_read_line() 69 * for general purpose input, so.. 70 */ 71 #ifdef MULTIPROCESSOR 72 db_printf("db{%ld}> ", (long)cpu_number()); 73 #else 74 db_printf("db> "); 75 #endif 76 #endif 77 i = db_readline(db_line, sizeof(db_line)); 78 if (i == 0) 79 return (0); /* EOI */ 80 db_set_line(db_line, db_line + i); 81 return (i); 82 } 83 84 void 85 db_set_line(const char *sp, const char *ep) 86 { 87 88 db_lp = sp; 89 db_endlp = ep; 90 } 91 92 void 93 db_get_line(const char **psp, const char **pep) 94 { 95 96 if (psp != NULL) 97 *psp = db_lp; 98 if (pep != NULL) 99 *pep = db_endlp; 100 } 101 102 static void 103 db_flush_line(void) 104 { 105 106 db_lp = db_line; 107 db_endlp = db_line; 108 } 109 110 static int 111 db_read_char(void) 112 { 113 int c; 114 115 if (db_look_char != 0) { 116 c = db_look_char; 117 db_look_char = 0; 118 } 119 else if (db_lp >= db_endlp) 120 c = -1; 121 else 122 c = *db_lp++; 123 return (c); 124 } 125 126 static void 127 db_unread_char(int c) 128 { 129 130 db_look_char = c; 131 } 132 133 void 134 db_unread_token(int t) 135 { 136 137 db_look_token = t; 138 } 139 140 int 141 db_read_token(void) 142 { 143 int t; 144 145 if (db_look_token) { 146 t = db_look_token; 147 db_look_token = 0; 148 } 149 else 150 t = db_lex(); 151 return (t); 152 } 153 154 int db_radix = 16; 155 156 /* 157 * Convert the number to a string in the current radix. 158 * This replaces the non-standard %n printf() format. 159 */ 160 161 char * 162 db_num_to_str(db_expr_t val) 163 { 164 165 /* 166 * 2 chars for "0x", 1 for a sign ("-") 167 * up to 21 chars for a 64-bit number: 168 * % echo 2^64 | bc | wc -c 169 * 21 170 * and 1 char for a terminal NUL 171 * 2+1+21+1 => 25 172 */ 173 static char buf[25]; 174 175 if (db_radix == 16) 176 snprintf(buf, sizeof(buf), "%" DDB_EXPR_FMT "x", val); 177 else if (db_radix == 8) 178 snprintf(buf, sizeof(buf), "%" DDB_EXPR_FMT "o", val); 179 else 180 snprintf(buf, sizeof(buf), "%" DDB_EXPR_FMT "u", val); 181 182 return (buf); 183 } 184 185 void 186 db_flush_lex(void) 187 { 188 189 db_flush_line(); 190 db_look_char = 0; 191 db_look_token = 0; 192 } 193 194 static int 195 db_lex(void) 196 { 197 int c; 198 199 c = db_read_char(); 200 while (c <= ' ' || c > '~') { 201 if (c == '\n' || c == -1) 202 return (tEOL); 203 c = db_read_char(); 204 } 205 206 if (c >= '0' && c <= '9') { 207 /* number */ 208 db_expr_t r, digit = 0; 209 210 if (c > '0') 211 r = db_radix; 212 else { 213 c = db_read_char(); 214 if (c == 'O' || c == 'o') 215 r = 8; 216 else if (c == 'T' || c == 't') 217 r = 10; 218 else if (c == 'X' || c == 'x') 219 r = 16; 220 else { 221 r = db_radix; 222 db_unread_char(c); 223 } 224 c = db_read_char(); 225 } 226 db_tok_number = 0; 227 for (;;) { 228 if (c >= '0' && c <= ((r == 8) ? '7' : '9')) 229 digit = c - '0'; 230 else if (r == 16) { 231 if (c >= 'A' && c <= 'F') 232 digit = c - 'A' + 10; 233 else if (c >= 'a' && c <= 'f') 234 digit = c - 'a' + 10; 235 else 236 break; 237 } else 238 break; 239 db_tok_number = db_tok_number * r + digit; 240 c = db_read_char(); 241 } 242 if ((c >= '0' && c <= '9') || 243 (c >= 'A' && c <= 'Z') || 244 (c >= 'a' && c <= 'z') || 245 (c == '_')) { 246 db_error("Bad character in number\n"); 247 /*NOTREACHED*/ 248 } 249 db_unread_char(c); 250 return (tNUMBER); 251 } 252 if ((c >= 'A' && c <= 'Z') || 253 (c >= 'a' && c <= 'z') || 254 c == '_' || c == '\\') { 255 /* string */ 256 char *cp; 257 258 cp = db_tok_string; 259 if (c == '\\') { 260 c = db_read_char(); 261 if (c == '\n' || c == -1) { 262 db_error("Bad escape\n"); 263 /*NOTREACHED*/ 264 } 265 } 266 *cp++ = c; 267 while (1) { 268 c = db_read_char(); 269 if ((c >= 'A' && c <= 'Z') || 270 (c >= 'a' && c <= 'z') || 271 (c >= '0' && c <= '9') || 272 c == '_' || c == '\\' || c == ':') { 273 if (c == '\\') { 274 c = db_read_char(); 275 if (c == '\n' || c == -1) { 276 db_error("Bad escape\n"); 277 /*NOTREACHED*/ 278 } 279 } 280 *cp++ = c; 281 if (cp == db_tok_string+sizeof(db_tok_string)) { 282 db_error("String too long\n"); 283 /*NOTREACHED*/ 284 } 285 continue; 286 } else { 287 *cp = '\0'; 288 break; 289 } 290 } 291 db_unread_char(c); 292 return (tIDENT); 293 } 294 295 switch (c) { 296 case '+': 297 return (tPLUS); 298 case '-': 299 return (tMINUS); 300 case '.': 301 c = db_read_char(); 302 if (c == '.') 303 return (tDOTDOT); 304 db_unread_char(c); 305 return (tDOT); 306 case '*': 307 return (tSTAR); 308 case '/': 309 return (tSLASH); 310 case '=': 311 return (tEQ); 312 case '%': 313 return (tPCT); 314 case '#': 315 return (tHASH); 316 case '(': 317 return (tLPAREN); 318 case ')': 319 return (tRPAREN); 320 case ',': 321 return (tCOMMA); 322 case '"': 323 return (tDITTO); 324 case '$': 325 return (tDOLLAR); 326 case '!': 327 return (tEXCL); 328 case '<': 329 c = db_read_char(); 330 if (c == '<') 331 return (tSHIFT_L); 332 db_unread_char(c); 333 break; 334 case '>': 335 c = db_read_char(); 336 if (c == '>') 337 return (tSHIFT_R); 338 db_unread_char(c); 339 break; 340 case -1: 341 return (tEOF); 342 } 343 db_printf("Bad character\n"); 344 db_flush_lex(); 345 return (tEOF); 346 } 347 348 /* 349 * Utility routine - discard tokens through end-of-line. 350 */ 351 void 352 db_skip_to_eol(void) 353 { 354 int t; 355 356 do { 357 t = db_read_token(); 358 } while (t != tEOL); 359 } 360 361 void 362 db_error(const char *s) 363 { 364 365 if (s) 366 db_printf("%s", s); 367 db_flush_lex(); 368 longjmp(db_recover); 369 } 370