1 /* 2 * keyword-gen.c -- generate keyword scanner finite state machine and 3 * keyword_text array. 4 * 5 * This program is run to generate ntp_keyword.h 6 * After making a change here, two output files should be committed at 7 * the same time as keyword-gen.c: 8 * ntp_keyword.h 9 * keyword-gen-utd 10 * 11 * keyword-gen-utd is a sentinel used by Makefile.am to avoid compiling 12 * keyword_gen.c and generating ntp_keyword.h if the input keyword-gen.c 13 * has not changed. This is not solely an optimization, it also breaks 14 * a dependency chain that otherwise would cause programs to be compiled 15 * when running "make dist" or "make distdir". We want these to package 16 * the existing source without building anything but a tarball. See 17 * [Bug 1470]. 18 */ 19 #include <config.h> 20 #include <stdio.h> 21 #include <stdlib.h> 22 #include <time.h> 23 24 #include <ntp_stdlib.h> 25 #include <ntp_config.h> 26 #include <lib_strbuf.h> 27 #include "ntp_scanner.h" 28 #include "ntp_parser.h" 29 30 31 /* Define a structure to hold a (keyword, token) pair */ 32 struct key_tok { 33 char * key; /* Keyword */ 34 u_short token; /* Associated Token */ 35 follby followedby; /* nonzero indicates the next token(s) 36 forced to be string(s) */ 37 }; 38 39 struct key_tok ntp_keywords[] = { 40 { "...", T_Ellipsis, FOLLBY_TOKEN }, 41 { "allpeers", T_Allpeers, FOLLBY_TOKEN }, 42 { "automax", T_Automax, FOLLBY_TOKEN }, 43 { "broadcast", T_Broadcast, FOLLBY_STRING }, 44 { "broadcastclient", T_Broadcastclient, FOLLBY_TOKEN }, 45 { "broadcastdelay", T_Broadcastdelay, FOLLBY_TOKEN }, 46 { "ctl", T_Ctl, FOLLBY_TOKEN }, 47 { "disable", T_Disable, FOLLBY_TOKEN }, 48 { "driftfile", T_Driftfile, FOLLBY_STRING }, 49 { "enable", T_Enable, FOLLBY_TOKEN }, 50 { "end", T_End, FOLLBY_TOKEN }, 51 { "filegen", T_Filegen, FOLLBY_TOKEN }, 52 { "fudge", T_Fudge, FOLLBY_STRING }, 53 { "io", T_Io, FOLLBY_TOKEN }, 54 { "includefile", T_Includefile, FOLLBY_STRING }, 55 { "leapfile", T_Leapfile, FOLLBY_STRING }, 56 { "logconfig", T_Logconfig, FOLLBY_STRINGS_TO_EOC }, 57 { "logfile", T_Logfile, FOLLBY_STRING }, 58 { "manycastclient", T_Manycastclient, FOLLBY_STRING }, 59 { "manycastserver", T_Manycastserver, FOLLBY_STRINGS_TO_EOC }, 60 { "mem", T_Mem, FOLLBY_TOKEN }, 61 { "multicastclient", T_Multicastclient, FOLLBY_STRINGS_TO_EOC }, 62 { "peer", T_Peer, FOLLBY_STRING }, 63 { "phone", T_Phone, FOLLBY_STRINGS_TO_EOC }, 64 { "pidfile", T_Pidfile, FOLLBY_STRING }, 65 { "pool", T_Pool, FOLLBY_STRING }, 66 { "discard", T_Discard, FOLLBY_TOKEN }, 67 { "reset", T_Reset, FOLLBY_TOKEN }, 68 { "restrict", T_Restrict, FOLLBY_TOKEN }, 69 { "rlimit", T_Rlimit, FOLLBY_TOKEN }, 70 { "server", T_Server, FOLLBY_STRING }, 71 { "setvar", T_Setvar, FOLLBY_STRING }, 72 { "statistics", T_Statistics, FOLLBY_TOKEN }, 73 { "statsdir", T_Statsdir, FOLLBY_STRING }, 74 { "sys", T_Sys, FOLLBY_TOKEN }, 75 { "tick", T_Tick, FOLLBY_TOKEN }, 76 { "timer", T_Timer, FOLLBY_TOKEN }, 77 { "tinker", T_Tinker, FOLLBY_TOKEN }, 78 { "tos", T_Tos, FOLLBY_TOKEN }, 79 { "trap", T_Trap, FOLLBY_STRING }, 80 { "unconfig", T_Unconfig, FOLLBY_STRING }, 81 { "unpeer", T_Unpeer, FOLLBY_STRING }, 82 /* authentication_command */ 83 { "controlkey", T_ControlKey, FOLLBY_TOKEN }, 84 { "crypto", T_Crypto, FOLLBY_TOKEN }, 85 { "keys", T_Keys, FOLLBY_STRING }, 86 { "keysdir", T_Keysdir, FOLLBY_STRING }, 87 { "ntpsigndsocket", T_NtpSignDsocket, FOLLBY_STRING }, 88 { "requestkey", T_Requestkey, FOLLBY_TOKEN }, 89 { "revoke", T_Revoke, FOLLBY_TOKEN }, 90 { "trustedkey", T_Trustedkey, FOLLBY_TOKEN }, 91 /* IPv4/IPv6 protocol override flag */ 92 { "-4", T_Ipv4_flag, FOLLBY_TOKEN }, 93 { "-6", T_Ipv6_flag, FOLLBY_TOKEN }, 94 /* option */ 95 { "autokey", T_Autokey, FOLLBY_TOKEN }, 96 { "burst", T_Burst, FOLLBY_TOKEN }, 97 { "iburst", T_Iburst, FOLLBY_TOKEN }, 98 { "key", T_Key, FOLLBY_TOKEN }, 99 { "maxpoll", T_Maxpoll, FOLLBY_TOKEN }, 100 { "mdnstries", T_Mdnstries, FOLLBY_TOKEN }, 101 { "minpoll", T_Minpoll, FOLLBY_TOKEN }, 102 { "mode", T_Mode, FOLLBY_TOKEN }, 103 { "noselect", T_Noselect, FOLLBY_TOKEN }, 104 { "preempt", T_Preempt, FOLLBY_TOKEN }, 105 { "true", T_True, FOLLBY_TOKEN }, 106 { "prefer", T_Prefer, FOLLBY_TOKEN }, 107 { "ttl", T_Ttl, FOLLBY_TOKEN }, 108 { "version", T_Version, FOLLBY_TOKEN }, 109 { "xleave", T_Xleave, FOLLBY_TOKEN }, 110 /* crypto_command */ 111 { "host", T_Host, FOLLBY_STRING }, 112 { "ident", T_Ident, FOLLBY_STRING }, 113 { "pw", T_Pw, FOLLBY_STRING }, 114 { "randfile", T_Randfile, FOLLBY_STRING }, 115 { "digest", T_Digest, FOLLBY_STRING }, 116 /*** MONITORING COMMANDS ***/ 117 /* stat */ 118 { "clockstats", T_Clockstats, FOLLBY_TOKEN }, 119 { "cryptostats", T_Cryptostats, FOLLBY_TOKEN }, 120 { "loopstats", T_Loopstats, FOLLBY_TOKEN }, 121 { "peerstats", T_Peerstats, FOLLBY_TOKEN }, 122 { "rawstats", T_Rawstats, FOLLBY_TOKEN }, 123 { "sysstats", T_Sysstats, FOLLBY_TOKEN }, 124 { "protostats", T_Protostats, FOLLBY_TOKEN }, 125 { "timingstats", T_Timingstats, FOLLBY_TOKEN }, 126 /* filegen_option */ 127 { "file", T_File, FOLLBY_STRING }, 128 { "link", T_Link, FOLLBY_TOKEN }, 129 { "nolink", T_Nolink, FOLLBY_TOKEN }, 130 { "type", T_Type, FOLLBY_TOKEN }, 131 /* filegen_type */ 132 { "age", T_Age, FOLLBY_TOKEN }, 133 { "day", T_Day, FOLLBY_TOKEN }, 134 { "month", T_Month, FOLLBY_TOKEN }, 135 { "none", T_None, FOLLBY_TOKEN }, 136 { "pid", T_Pid, FOLLBY_TOKEN }, 137 { "week", T_Week, FOLLBY_TOKEN }, 138 { "year", T_Year, FOLLBY_TOKEN }, 139 /*** ORPHAN MODE COMMANDS ***/ 140 /* tos_option */ 141 { "minclock", T_Minclock, FOLLBY_TOKEN }, 142 { "maxclock", T_Maxclock, FOLLBY_TOKEN }, 143 { "minsane", T_Minsane, FOLLBY_TOKEN }, 144 { "floor", T_Floor, FOLLBY_TOKEN }, 145 { "ceiling", T_Ceiling, FOLLBY_TOKEN }, 146 { "cohort", T_Cohort, FOLLBY_TOKEN }, 147 { "mindist", T_Mindist, FOLLBY_TOKEN }, 148 { "maxdist", T_Maxdist, FOLLBY_TOKEN }, 149 { "beacon", T_Beacon, FOLLBY_TOKEN }, 150 { "orphan", T_Orphan, FOLLBY_TOKEN }, 151 { "orphanwait", T_Orphanwait, FOLLBY_TOKEN }, 152 { "nonvolatile", T_Nonvolatile, FOLLBY_TOKEN }, 153 /* access_control_flag */ 154 { "default", T_Default, FOLLBY_TOKEN }, 155 { "source", T_Source, FOLLBY_TOKEN }, 156 { "flake", T_Flake, FOLLBY_TOKEN }, 157 { "ignore", T_Ignore, FOLLBY_TOKEN }, 158 { "limited", T_Limited, FOLLBY_TOKEN }, 159 { "mssntp", T_Mssntp, FOLLBY_TOKEN }, 160 { "kod", T_Kod, FOLLBY_TOKEN }, 161 { "lowpriotrap", T_Lowpriotrap, FOLLBY_TOKEN }, 162 { "mask", T_Mask, FOLLBY_TOKEN }, 163 { "nomodify", T_Nomodify, FOLLBY_TOKEN }, 164 { "nomrulist", T_Nomrulist, FOLLBY_TOKEN }, 165 { "nopeer", T_Nopeer, FOLLBY_TOKEN }, 166 { "noquery", T_Noquery, FOLLBY_TOKEN }, 167 { "noserve", T_Noserve, FOLLBY_TOKEN }, 168 { "notrap", T_Notrap, FOLLBY_TOKEN }, 169 { "notrust", T_Notrust, FOLLBY_TOKEN }, 170 { "ntpport", T_Ntpport, FOLLBY_TOKEN }, 171 /* discard_option */ 172 { "average", T_Average, FOLLBY_TOKEN }, 173 { "minimum", T_Minimum, FOLLBY_TOKEN }, 174 { "monitor", T_Monitor, FOLLBY_TOKEN }, 175 /* mru_option */ 176 { "incalloc", T_Incalloc, FOLLBY_TOKEN }, 177 { "incmem", T_Incmem, FOLLBY_TOKEN }, 178 { "initalloc", T_Initalloc, FOLLBY_TOKEN }, 179 { "initmem", T_Initmem, FOLLBY_TOKEN }, 180 { "mindepth", T_Mindepth, FOLLBY_TOKEN }, 181 { "maxage", T_Maxage, FOLLBY_TOKEN }, 182 { "maxdepth", T_Maxdepth, FOLLBY_TOKEN }, 183 { "maxmem", T_Maxmem, FOLLBY_TOKEN }, 184 { "mru", T_Mru, FOLLBY_TOKEN }, 185 /* fudge_factor */ 186 { "abbrev", T_Abbrev, FOLLBY_STRING }, 187 { "flag1", T_Flag1, FOLLBY_TOKEN }, 188 { "flag2", T_Flag2, FOLLBY_TOKEN }, 189 { "flag3", T_Flag3, FOLLBY_TOKEN }, 190 { "flag4", T_Flag4, FOLLBY_TOKEN }, 191 { "refid", T_Refid, FOLLBY_STRING }, 192 { "stratum", T_Stratum, FOLLBY_TOKEN }, 193 { "time1", T_Time1, FOLLBY_TOKEN }, 194 { "time2", T_Time2, FOLLBY_TOKEN }, 195 /* system_option */ 196 { "auth", T_Auth, FOLLBY_TOKEN }, 197 { "bclient", T_Bclient, FOLLBY_TOKEN }, 198 { "calibrate", T_Calibrate, FOLLBY_TOKEN }, 199 { "kernel", T_Kernel, FOLLBY_TOKEN }, 200 { "ntp", T_Ntp, FOLLBY_TOKEN }, 201 { "mode7", T_Mode7, FOLLBY_TOKEN }, 202 { "stats", T_Stats, FOLLBY_TOKEN }, 203 /* rlimit_option */ 204 { "memlock", T_Memlock, FOLLBY_TOKEN }, 205 { "stacksize", T_Stacksize, FOLLBY_TOKEN }, 206 { "filenum", T_Filenum, FOLLBY_TOKEN }, 207 /* tinker_option */ 208 { "step", T_Step, FOLLBY_TOKEN }, 209 { "stepback", T_Stepback, FOLLBY_TOKEN }, 210 { "stepfwd", T_Stepfwd, FOLLBY_TOKEN }, 211 { "panic", T_Panic, FOLLBY_TOKEN }, 212 { "dispersion", T_Dispersion, FOLLBY_TOKEN }, 213 { "stepout", T_Stepout, FOLLBY_TOKEN }, 214 { "allan", T_Allan, FOLLBY_TOKEN }, 215 { "huffpuff", T_Huffpuff, FOLLBY_TOKEN }, 216 { "freq", T_Freq, FOLLBY_TOKEN }, 217 /* miscellaneous_command */ 218 { "port", T_Port, FOLLBY_TOKEN }, 219 { "interface", T_Interface, FOLLBY_TOKEN }, 220 { "saveconfigdir", T_Saveconfigdir, FOLLBY_STRING }, 221 /* interface_command (ignore and interface already defined) */ 222 { "nic", T_Nic, FOLLBY_TOKEN }, 223 { "all", T_All, FOLLBY_TOKEN }, 224 { "ipv4", T_Ipv4, FOLLBY_TOKEN }, 225 { "ipv6", T_Ipv6, FOLLBY_TOKEN }, 226 { "wildcard", T_Wildcard, FOLLBY_TOKEN }, 227 { "listen", T_Listen, FOLLBY_TOKEN }, 228 { "drop", T_Drop, FOLLBY_TOKEN }, 229 /* simulator commands */ 230 { "simulate", T_Simulate, FOLLBY_TOKEN }, 231 { "simulation_duration",T_Sim_Duration, FOLLBY_TOKEN }, 232 { "beep_delay", T_Beep_Delay, FOLLBY_TOKEN }, 233 { "duration", T_Duration, FOLLBY_TOKEN }, 234 { "server_offset", T_Server_Offset, FOLLBY_TOKEN }, 235 { "freq_offset", T_Freq_Offset, FOLLBY_TOKEN }, 236 { "wander", T_Wander, FOLLBY_TOKEN }, 237 { "jitter", T_Jitter, FOLLBY_TOKEN }, 238 { "prop_delay", T_Prop_Delay, FOLLBY_TOKEN }, 239 { "proc_delay", T_Proc_Delay, FOLLBY_TOKEN }, 240 }; 241 242 typedef struct big_scan_state_tag { 243 char ch; /* Character this state matches on */ 244 char followedby; /* Forces next token(s) to T_String */ 245 u_short finishes_token; /* nonzero ID if last keyword char */ 246 u_short match_next_s; /* next state to check matching ch */ 247 u_short other_next_s; /* next state to check if not ch */ 248 } big_scan_state; 249 250 /* 251 * Note: to increase MAXSTATES beyond 2048, be aware it is currently 252 * crammed into 11 bits in scan_state form. Raising to 4096 would be 253 * relatively easy by storing the followedby value in a separate 254 * array with one entry per token, and shrinking the char value to 255 * 7 bits to free a bit for accepting/non-accepting. More than 4096 256 * states will require expanding scan_state beyond 32 bits each. 257 */ 258 #define MAXSTATES 2048 259 #define MAX_TOK_LEN 63 260 261 const char * current_keyword;/* for error reporting */ 262 big_scan_state sst[MAXSTATES]; /* scanner FSM state entries */ 263 u_short sst_highwater; /* next entry index to consider */ 264 char * symb[1024]; /* map token ID to symbolic name */ 265 266 /* for libntp */ 267 const char * progname = "keyword-gen"; 268 269 int main (int, char **); 270 static void generate_preamble (void); 271 static void generate_fsm (void); 272 static void generate_token_text (void); 273 static u_short create_keyword_scanner (void); 274 static u_short create_scan_states (char *, u_short, follby, u_short); 275 int compare_key_tok_id (const void *, const void *); 276 int compare_key_tok_text (const void *, const void *); 277 void populate_symb (char *); 278 const char * symbname (u_short); 279 280 281 int main(int argc, char **argv) 282 { 283 if (argc < 2) { 284 fprintf(stderr, "Usage:\n%s t_header.h\n", argv[0]); 285 exit(1); 286 } 287 debug = 1; 288 289 populate_symb(argv[1]); 290 291 generate_preamble(); 292 generate_token_text(); 293 generate_fsm(); 294 295 return 0; 296 } 297 298 299 static void 300 generate_preamble(void) 301 { 302 time_t now; 303 char timestamp[128]; 304 char preamble[] = 305 "/*\n" 306 " * ntp_keyword.h\n" 307 " * \n" 308 " * NOTE: edit this file with caution, it is generated by keyword-gen.c\n" 309 " *\t Generated %s UTC diff_ignore_line\n" 310 " *\n" 311 " */\n" 312 "#include \"ntp_scanner.h\"\n" 313 "#include \"ntp_parser.h\"\n" 314 "\n"; 315 316 time(&now); 317 if (!strftime(timestamp, sizeof(timestamp), 318 "%Y-%m-%d %H:%M:%S", gmtime(&now))) 319 timestamp[0] = '\0'; 320 321 printf(preamble, timestamp); 322 } 323 324 325 static void 326 generate_fsm(void) 327 { 328 char rprefix[MAX_TOK_LEN + 1]; 329 char prefix[MAX_TOK_LEN + 1]; 330 char token_id_comment[16 + MAX_TOK_LEN + 1]; 331 size_t prefix_len; 332 char *p; 333 char *r; 334 u_short initial_state; 335 u_short this_state; 336 u_short prev_state; 337 u_short state; 338 u_short i; 339 u_short token; 340 341 /* 342 * Sort ntp_keywords in alphabetical keyword order. This is 343 * not necessary, but minimizes nonfunctional changes in the 344 * generated finite state machine when keywords are modified. 345 */ 346 qsort(ntp_keywords, COUNTOF(ntp_keywords), 347 sizeof(ntp_keywords[0]), compare_key_tok_text); 348 349 /* 350 * To save space, reserve the state array entry matching each 351 * token number for its terminal state, so the token identifier 352 * does not need to be stored in each state, but can be 353 * recovered trivially. To mark the entry reserved, 354 * finishes_token is nonzero. 355 */ 356 357 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 358 token = ntp_keywords[i].token; 359 if (1 > token || token >= COUNTOF(sst)) { 360 fprintf(stderr, 361 "keyword-gen sst[%u] too small " 362 "for keyword '%s' id %d\n", 363 (int)COUNTOF(sst), 364 ntp_keywords[i].key, 365 token); 366 exit(4); 367 } 368 sst[token].finishes_token = token; 369 } 370 371 initial_state = create_keyword_scanner(); 372 373 fprintf(stderr, 374 "%d keywords consumed %d states of %d max.\n", 375 (int)COUNTOF(ntp_keywords), 376 sst_highwater - 1, 377 (int)COUNTOF(sst) - 1); 378 379 printf("#define SCANNER_INIT_S %d\n\n", initial_state); 380 381 printf("const scan_state sst[%d] = {\n" 382 "/*SS_T( ch,\tf-by, match, other ),\t\t\t\t */\n" 383 " 0,\t\t\t\t /* %5d %-17s */\n", 384 sst_highwater, 385 0, ""); 386 387 for (i = 1; i < sst_highwater; i++) { 388 389 /* verify fields will fit */ 390 if (sst[i].followedby & ~0x3) { 391 fprintf(stderr, 392 "keyword-gen internal error " 393 "sst[%d].followedby %d too big\n", 394 i, sst[i].followedby); 395 exit(7); 396 } 397 398 if (sst_highwater <= sst[i].match_next_s 399 || sst[i].match_next_s & ~0x7ff) { 400 fprintf(stderr, 401 "keyword-gen internal error " 402 "sst[%d].match_next_s %d too big\n", 403 i, sst[i].match_next_s); 404 exit(8); 405 } 406 407 if (sst_highwater <= sst[i].other_next_s 408 || sst[i].other_next_s & ~0x7ff) { 409 fprintf(stderr, 410 "keyword-gen internal error " 411 "sst[%d].other_next_s %d too big\n", 412 i, sst[i].other_next_s); 413 exit(9); 414 } 415 416 if (sst[i].finishes_token) { 417 snprintf(token_id_comment, 418 sizeof(token_id_comment), "%5d %-17s", 419 i, symbname(sst[i].finishes_token)); 420 if (i != sst[i].finishes_token) { 421 fprintf(stderr, 422 "keyword-gen internal error " 423 "entry %d finishes token %d\n", 424 i, sst[i].finishes_token); 425 exit(5); 426 } 427 } else { 428 /* 429 * Determine the keyword prefix that leads to this 430 * state. This is expensive but keyword-gen is run 431 * only when it changes. Distributing keyword-gen-utd 432 * achieves that, which is why it must be committed 433 * at the same time as keyword-gen.c and ntp_keyword.h. 434 * 435 * Scan the state array iteratively looking for a state 436 * which leads to the current one, collecting matching 437 * characters along the way. There is only one such 438 * path back to the starting state given the way our 439 * scanner state machine is built and the practice of 440 * using the spelling of the keyword as its T_* token 441 * identifier, which results in never having two 442 * spellings result in the same T_* value. 443 */ 444 prefix_len = 0; 445 prev_state = 0; 446 this_state = i; 447 do { 448 for (state = 1; state < sst_highwater; state++) 449 if (sst[state].other_next_s == this_state) { 450 this_state = state; 451 break; 452 } else if (sst[state].match_next_s == this_state) { 453 this_state = state; 454 rprefix[prefix_len] = sst[state].ch; 455 prefix_len++; 456 break; 457 } 458 } while (this_state != initial_state); 459 460 if (prefix_len) { 461 /* reverse rprefix into prefix */ 462 p = prefix + prefix_len; 463 r = rprefix; 464 while (r < rprefix + prefix_len) 465 *--p = *r++; 466 } 467 prefix[prefix_len] = '\0'; 468 469 snprintf(token_id_comment, 470 sizeof(token_id_comment), "%5d %-17s", 471 i, (initial_state == i) 472 ? "[initial state]" 473 : prefix); 474 } 475 476 printf(" S_ST( '%c',\t%d, %5u, %5u )%s /* %s */\n", 477 sst[i].ch, 478 sst[i].followedby, 479 sst[i].match_next_s, 480 sst[i].other_next_s, 481 (i + 1 < sst_highwater) 482 ? "," 483 : " ", 484 token_id_comment); 485 } 486 487 printf("};\n\n"); 488 } 489 490 491 /* Define a function to create the states of the scanner. This function 492 * is used by the create_keyword_scanner function below. 493 * 494 * This function takes a suffix of a keyword, the token to be returned on 495 * recognizing the complete keyword, and any pre-existing state that exists 496 * for some other keyword that has the same prefix as the current one. 497 */ 498 static u_short 499 create_scan_states( 500 char * text, 501 u_short token, 502 follby followedby, 503 u_short prev_state 504 ) 505 { 506 u_short my_state; 507 u_short return_state; 508 u_short prev_char_s; 509 u_short curr_char_s; 510 511 return_state = prev_state; 512 curr_char_s = prev_state; 513 prev_char_s = 0; 514 515 /* Find the correct position to insert the state. 516 * All states should be in alphabetical order 517 */ 518 while (curr_char_s && (text[0] < sst[curr_char_s].ch)) { 519 prev_char_s = curr_char_s; 520 curr_char_s = sst[curr_char_s].other_next_s; 521 } 522 523 /* 524 * Check if a previously seen keyword has the same prefix as 525 * the current keyword. If so, simply use the state for that 526 * keyword as my_state, otherwise, allocate a new state. 527 */ 528 if (curr_char_s && (text[0] == sst[curr_char_s].ch)) { 529 my_state = curr_char_s; 530 if ('\0' == text[1]) { 531 fprintf(stderr, 532 "Duplicate entries for keyword '%s' in" 533 " keyword_gen.c ntp_keywords[].\n", 534 current_keyword); 535 exit(2); 536 } 537 } else { 538 do 539 my_state = sst_highwater++; 540 while (my_state < COUNTOF(sst) 541 && sst[my_state].finishes_token); 542 if (my_state >= COUNTOF(sst)) { 543 fprintf(stderr, 544 "fatal, keyword scanner state array " 545 "sst[%d] is too small, modify\n" 546 "keyword-gen.c to increase.\n", 547 (int)COUNTOF(sst)); 548 exit(3); 549 } 550 /* Store the next character of the keyword */ 551 sst[my_state].ch = text[0]; 552 sst[my_state].other_next_s = curr_char_s; 553 sst[my_state].followedby = FOLLBY_NON_ACCEPTING; 554 555 if (prev_char_s) 556 sst[prev_char_s].other_next_s = my_state; 557 else 558 return_state = my_state; 559 } 560 561 /* Check if the next character is '\0'. 562 * If yes, we are done with the recognition and this is an accepting 563 * state. 564 * If not, we need to continue scanning 565 */ 566 if ('\0' == text[1]) { 567 sst[my_state].finishes_token = (u_short)token; 568 sst[my_state].followedby = (char)followedby; 569 570 if (sst[token].finishes_token != (u_short)token) { 571 fprintf(stderr, 572 "fatal, sst[%d] not reserved for %s.\n", 573 token, symbname(token)); 574 exit(6); 575 } 576 /* relocate so token id is sst[] index */ 577 if (my_state != token) { 578 sst[token] = sst[my_state]; 579 ZERO(sst[my_state]); 580 do 581 sst_highwater--; 582 while (sst[sst_highwater].finishes_token); 583 my_state = token; 584 if (prev_char_s) 585 sst[prev_char_s].other_next_s = my_state; 586 else 587 return_state = my_state; 588 } 589 } else 590 sst[my_state].match_next_s = 591 create_scan_states( 592 &text[1], 593 token, 594 followedby, 595 sst[my_state].match_next_s); 596 597 return return_state; 598 } 599 600 601 /* Define a function that takes a list of (keyword, token) values and 602 * creates a keywords scanner out of it. 603 */ 604 605 static u_short 606 create_keyword_scanner(void) 607 { 608 u_short scanner; 609 u_short i; 610 611 sst_highwater = 1; /* index 0 invalid, unused */ 612 scanner = 0; 613 614 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 615 current_keyword = ntp_keywords[i].key; 616 scanner = 617 create_scan_states( 618 ntp_keywords[i].key, 619 ntp_keywords[i].token, 620 ntp_keywords[i].followedby, 621 scanner); 622 } 623 624 return scanner; 625 } 626 627 628 static void 629 generate_token_text(void) 630 { 631 u_short lowest_id; 632 u_short highest_id; 633 u_short id_count; 634 u_short id; 635 u_short i; 636 637 /* sort ntp_keywords in token ID order */ 638 qsort(ntp_keywords, COUNTOF(ntp_keywords), 639 sizeof(ntp_keywords[0]), compare_key_tok_id); 640 641 lowest_id = ntp_keywords[0].token; 642 highest_id = ntp_keywords[COUNTOF(ntp_keywords) - 1].token; 643 id_count = highest_id - lowest_id + 1; 644 645 printf("#define LOWEST_KEYWORD_ID %d\n\n", lowest_id); 646 647 printf("const char * const keyword_text[%d] = {", id_count); 648 649 id = lowest_id; 650 i = 0; 651 while (i < COUNTOF(ntp_keywords)) { 652 while (id < ntp_keywords[i].token) { 653 printf(",\n\t/* %-5d %5d %20s */\tNULL", 654 id - lowest_id, id, symbname(id)); 655 id++; 656 } 657 if (i > 0) 658 printf(","); 659 printf("\n\t/* %-5d %5d %20s */\t\"%s\"", 660 id - lowest_id, id, symbname(id), 661 ntp_keywords[i].key); 662 i++; 663 id++; 664 } 665 666 printf("\n};\n\n"); 667 } 668 669 670 int 671 compare_key_tok_id( 672 const void *a1, 673 const void *a2 674 ) 675 { 676 const struct key_tok *p1 = a1; 677 const struct key_tok *p2 = a2; 678 679 if (p1->token == p2->token) 680 return 0; 681 682 if (p1->token < p2->token) 683 return -1; 684 else 685 return 1; 686 } 687 688 689 int 690 compare_key_tok_text( 691 const void *a1, 692 const void *a2 693 ) 694 { 695 const struct key_tok *p1 = a1; 696 const struct key_tok *p2 = a2; 697 698 return strcmp(p1->key, p2->key); 699 } 700 701 702 /* 703 * populate_symb() - populate symb[] lookup array with symbolic token 704 * names such that symb[T_Age] == "T_Age", etc. 705 */ 706 void 707 populate_symb( 708 char *header_file 709 ) 710 { 711 FILE * yh; 712 char line[2 * MAX_TOK_LEN]; 713 char name[2 * MAX_TOK_LEN]; 714 int token; 715 716 yh = fopen(header_file, "r"); 717 if (NULL == yh) { 718 perror("unable to open yacc/bison header file"); 719 exit(4); 720 } 721 722 while (NULL != fgets(line, sizeof(line), yh)) 723 if (2 == sscanf(line, "#define %s %d", name, &token) 724 && 'T' == name[0] && '_' == name[1] && token >= 0 725 && token < COUNTOF(symb)) { 726 727 symb[token] = estrdup(name); 728 if (strlen(name) > MAX_TOK_LEN) { 729 fprintf(stderr, 730 "MAX_TOK_LEN %d too small for '%s'\n" 731 "Edit keyword-gen.c to raise.\n", 732 MAX_TOK_LEN, name); 733 exit(10); 734 } 735 } 736 fclose(yh); 737 } 738 739 740 const char * 741 symbname( 742 u_short token 743 ) 744 { 745 char *name; 746 747 if (token < COUNTOF(symb) && symb[token] != NULL) { 748 name = symb[token]; 749 } else { 750 LIB_GETBUF(name); 751 snprintf(name, LIB_BUFLENGTH, "%d", token); 752 } 753 754 return name; 755 } 756