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 { "dscp", T_Dscp, FOLLBY_TOKEN }, 50 { "enable", T_Enable, FOLLBY_TOKEN }, 51 { "end", T_End, FOLLBY_TOKEN }, 52 { "filegen", T_Filegen, FOLLBY_TOKEN }, 53 { "fudge", T_Fudge, FOLLBY_STRING }, 54 { "io", T_Io, FOLLBY_TOKEN }, 55 { "includefile", T_Includefile, FOLLBY_STRING }, 56 { "leapfile", T_Leapfile, FOLLBY_STRING }, 57 { "leapsmearinterval", T_Leapsmearinterval, FOLLBY_TOKEN }, 58 { "logconfig", T_Logconfig, FOLLBY_STRINGS_TO_EOC }, 59 { "logfile", T_Logfile, FOLLBY_STRING }, 60 { "manycastclient", T_Manycastclient, FOLLBY_STRING }, 61 { "manycastserver", T_Manycastserver, FOLLBY_STRINGS_TO_EOC }, 62 { "mem", T_Mem, FOLLBY_TOKEN }, 63 { "multicastclient", T_Multicastclient, FOLLBY_STRINGS_TO_EOC }, 64 { "peer", T_Peer, FOLLBY_STRING }, 65 { "phone", T_Phone, FOLLBY_STRINGS_TO_EOC }, 66 { "pidfile", T_Pidfile, FOLLBY_STRING }, 67 { "pool", T_Pool, FOLLBY_STRING }, 68 { "discard", T_Discard, FOLLBY_TOKEN }, 69 { "reset", T_Reset, FOLLBY_TOKEN }, 70 { "restrict", T_Restrict, FOLLBY_TOKEN }, 71 { "rlimit", T_Rlimit, FOLLBY_TOKEN }, 72 { "server", T_Server, FOLLBY_STRING }, 73 { "setvar", T_Setvar, FOLLBY_STRING }, 74 { "statistics", T_Statistics, FOLLBY_TOKEN }, 75 { "statsdir", T_Statsdir, FOLLBY_STRING }, 76 { "sys", T_Sys, FOLLBY_TOKEN }, 77 { "tick", T_Tick, FOLLBY_TOKEN }, 78 { "timer", T_Timer, FOLLBY_TOKEN }, 79 { "tinker", T_Tinker, FOLLBY_TOKEN }, 80 { "tos", T_Tos, FOLLBY_TOKEN }, 81 { "trap", T_Trap, FOLLBY_STRING }, 82 { "unconfig", T_Unconfig, FOLLBY_STRING }, 83 { "unpeer", T_Unpeer, FOLLBY_STRING }, 84 /* authentication_command */ 85 { "controlkey", T_ControlKey, FOLLBY_TOKEN }, 86 { "crypto", T_Crypto, FOLLBY_TOKEN }, 87 { "keys", T_Keys, FOLLBY_STRING }, 88 { "keysdir", T_Keysdir, FOLLBY_STRING }, 89 { "ntpsigndsocket", T_NtpSignDsocket, FOLLBY_STRING }, 90 { "requestkey", T_Requestkey, FOLLBY_TOKEN }, 91 { "revoke", T_Revoke, FOLLBY_TOKEN }, 92 { "trustedkey", T_Trustedkey, FOLLBY_TOKEN }, 93 /* IPv4/IPv6 protocol override flag */ 94 { "-4", T_Ipv4_flag, FOLLBY_TOKEN }, 95 { "-6", T_Ipv6_flag, FOLLBY_TOKEN }, 96 /* option */ 97 { "autokey", T_Autokey, FOLLBY_TOKEN }, 98 { "burst", T_Burst, FOLLBY_TOKEN }, 99 { "iburst", T_Iburst, FOLLBY_TOKEN }, 100 { "key", T_Key, FOLLBY_TOKEN }, 101 { "maxpoll", T_Maxpoll, FOLLBY_TOKEN }, 102 { "mdnstries", T_Mdnstries, FOLLBY_TOKEN }, 103 { "minpoll", T_Minpoll, FOLLBY_TOKEN }, 104 { "mode", T_Mode, FOLLBY_TOKEN }, 105 { "noselect", T_Noselect, FOLLBY_TOKEN }, 106 { "preempt", T_Preempt, FOLLBY_TOKEN }, 107 { "true", T_True, FOLLBY_TOKEN }, 108 { "prefer", T_Prefer, FOLLBY_TOKEN }, 109 { "ttl", T_Ttl, FOLLBY_TOKEN }, 110 { "version", T_Version, FOLLBY_TOKEN }, 111 { "xleave", T_Xleave, FOLLBY_TOKEN }, 112 /* crypto_command */ 113 { "host", T_Host, FOLLBY_STRING }, 114 { "ident", T_Ident, FOLLBY_STRING }, 115 { "pw", T_Pw, FOLLBY_STRING }, 116 { "randfile", T_Randfile, FOLLBY_STRING }, 117 { "digest", T_Digest, FOLLBY_STRING }, 118 /*** MONITORING COMMANDS ***/ 119 /* stat */ 120 { "clockstats", T_Clockstats, FOLLBY_TOKEN }, 121 { "cryptostats", T_Cryptostats, FOLLBY_TOKEN }, 122 { "loopstats", T_Loopstats, FOLLBY_TOKEN }, 123 { "peerstats", T_Peerstats, FOLLBY_TOKEN }, 124 { "rawstats", T_Rawstats, FOLLBY_TOKEN }, 125 { "sysstats", T_Sysstats, FOLLBY_TOKEN }, 126 { "protostats", T_Protostats, FOLLBY_TOKEN }, 127 { "timingstats", T_Timingstats, FOLLBY_TOKEN }, 128 /* filegen_option */ 129 { "file", T_File, FOLLBY_STRING }, 130 { "link", T_Link, FOLLBY_TOKEN }, 131 { "nolink", T_Nolink, FOLLBY_TOKEN }, 132 { "type", T_Type, FOLLBY_TOKEN }, 133 /* filegen_type */ 134 { "age", T_Age, FOLLBY_TOKEN }, 135 { "day", T_Day, FOLLBY_TOKEN }, 136 { "month", T_Month, FOLLBY_TOKEN }, 137 { "none", T_None, FOLLBY_TOKEN }, 138 { "pid", T_Pid, FOLLBY_TOKEN }, 139 { "week", T_Week, FOLLBY_TOKEN }, 140 { "year", T_Year, FOLLBY_TOKEN }, 141 /*** ORPHAN MODE COMMANDS ***/ 142 /* tos_option */ 143 { "minclock", T_Minclock, FOLLBY_TOKEN }, 144 { "maxclock", T_Maxclock, FOLLBY_TOKEN }, 145 { "minsane", T_Minsane, FOLLBY_TOKEN }, 146 { "floor", T_Floor, FOLLBY_TOKEN }, 147 { "ceiling", T_Ceiling, FOLLBY_TOKEN }, 148 { "cohort", T_Cohort, FOLLBY_TOKEN }, 149 { "mindist", T_Mindist, FOLLBY_TOKEN }, 150 { "maxdist", T_Maxdist, FOLLBY_TOKEN }, 151 { "bcpollbstep", T_Bcpollbstep, FOLLBY_TOKEN }, 152 { "beacon", T_Beacon, FOLLBY_TOKEN }, 153 { "orphan", T_Orphan, FOLLBY_TOKEN }, 154 { "orphanwait", T_Orphanwait, FOLLBY_TOKEN }, 155 { "nonvolatile", T_Nonvolatile, FOLLBY_TOKEN }, 156 { "basedate", T_Basedate, FOLLBY_STRING }, 157 /* access_control_flag */ 158 { "default", T_Default, FOLLBY_TOKEN }, 159 { "source", T_Source, FOLLBY_TOKEN }, 160 { "epeer", T_Epeer, FOLLBY_TOKEN }, 161 { "noepeer", T_Noepeer, FOLLBY_TOKEN }, 162 { "flake", T_Flake, FOLLBY_TOKEN }, 163 { "ignore", T_Ignore, FOLLBY_TOKEN }, 164 { "ippeerlimit", T_Ippeerlimit, FOLLBY_TOKEN }, 165 { "limited", T_Limited, FOLLBY_TOKEN }, 166 { "mssntp", T_Mssntp, FOLLBY_TOKEN }, 167 { "kod", T_Kod, FOLLBY_TOKEN }, 168 { "lowpriotrap", T_Lowpriotrap, FOLLBY_TOKEN }, 169 { "mask", T_Mask, FOLLBY_TOKEN }, 170 { "nomodify", T_Nomodify, FOLLBY_TOKEN }, 171 { "nomrulist", T_Nomrulist, FOLLBY_TOKEN }, 172 { "nopeer", T_Nopeer, FOLLBY_TOKEN }, 173 { "noquery", T_Noquery, FOLLBY_TOKEN }, 174 { "noserve", T_Noserve, FOLLBY_TOKEN }, 175 { "notrap", T_Notrap, FOLLBY_TOKEN }, 176 { "notrust", T_Notrust, FOLLBY_TOKEN }, 177 { "ntpport", T_Ntpport, FOLLBY_TOKEN }, 178 /* discard_option */ 179 { "average", T_Average, FOLLBY_TOKEN }, 180 { "minimum", T_Minimum, FOLLBY_TOKEN }, 181 { "monitor", T_Monitor, FOLLBY_TOKEN }, 182 /* mru_option */ 183 { "incalloc", T_Incalloc, FOLLBY_TOKEN }, 184 { "incmem", T_Incmem, FOLLBY_TOKEN }, 185 { "initalloc", T_Initalloc, FOLLBY_TOKEN }, 186 { "initmem", T_Initmem, FOLLBY_TOKEN }, 187 { "mindepth", T_Mindepth, FOLLBY_TOKEN }, 188 { "maxage", T_Maxage, FOLLBY_TOKEN }, 189 { "maxdepth", T_Maxdepth, FOLLBY_TOKEN }, 190 { "maxmem", T_Maxmem, FOLLBY_TOKEN }, 191 { "mru", T_Mru, FOLLBY_TOKEN }, 192 /* fudge_factor */ 193 { "abbrev", T_Abbrev, FOLLBY_STRING }, 194 { "flag1", T_Flag1, FOLLBY_TOKEN }, 195 { "flag2", T_Flag2, FOLLBY_TOKEN }, 196 { "flag3", T_Flag3, FOLLBY_TOKEN }, 197 { "flag4", T_Flag4, FOLLBY_TOKEN }, 198 { "refid", T_Refid, FOLLBY_STRING }, 199 { "stratum", T_Stratum, FOLLBY_TOKEN }, 200 { "time1", T_Time1, FOLLBY_TOKEN }, 201 { "time2", T_Time2, FOLLBY_TOKEN }, 202 /* system_option */ 203 { "auth", T_Auth, FOLLBY_TOKEN }, 204 { "bclient", T_Bclient, FOLLBY_TOKEN }, 205 { "calibrate", T_Calibrate, FOLLBY_TOKEN }, 206 { "kernel", T_Kernel, FOLLBY_TOKEN }, 207 { "mode7", T_Mode7, FOLLBY_TOKEN }, 208 { "ntp", T_Ntp, FOLLBY_TOKEN }, 209 { "peer_clear_digest_early", T_PCEdigest, FOLLBY_TOKEN }, 210 { "stats", T_Stats, FOLLBY_TOKEN }, 211 { "unpeer_crypto_early", T_UEcrypto, FOLLBY_TOKEN }, 212 { "unpeer_crypto_nak_early", T_UEcryptonak, FOLLBY_TOKEN }, 213 { "unpeer_digest_early", T_UEdigest, FOLLBY_TOKEN }, 214 /* rlimit_option */ 215 { "memlock", T_Memlock, FOLLBY_TOKEN }, 216 { "stacksize", T_Stacksize, FOLLBY_TOKEN }, 217 { "filenum", T_Filenum, FOLLBY_TOKEN }, 218 /* tinker_option */ 219 { "step", T_Step, FOLLBY_TOKEN }, 220 { "stepback", T_Stepback, FOLLBY_TOKEN }, 221 { "stepfwd", T_Stepfwd, FOLLBY_TOKEN }, 222 { "panic", T_Panic, FOLLBY_TOKEN }, 223 { "dispersion", T_Dispersion, FOLLBY_TOKEN }, 224 { "stepout", T_Stepout, FOLLBY_TOKEN }, 225 { "allan", T_Allan, FOLLBY_TOKEN }, 226 { "huffpuff", T_Huffpuff, FOLLBY_TOKEN }, 227 { "freq", T_Freq, FOLLBY_TOKEN }, 228 /* miscellaneous_command */ 229 { "port", T_Port, FOLLBY_TOKEN }, 230 { "interface", T_Interface, FOLLBY_TOKEN }, 231 { "saveconfigdir", T_Saveconfigdir, FOLLBY_STRING }, 232 /* interface_command (ignore and interface already defined) */ 233 { "nic", T_Nic, FOLLBY_TOKEN }, 234 { "all", T_All, FOLLBY_TOKEN }, 235 { "ipv4", T_Ipv4, FOLLBY_TOKEN }, 236 { "ipv6", T_Ipv6, FOLLBY_TOKEN }, 237 { "wildcard", T_Wildcard, FOLLBY_TOKEN }, 238 { "listen", T_Listen, FOLLBY_TOKEN }, 239 { "drop", T_Drop, FOLLBY_TOKEN }, 240 /* simulator commands */ 241 { "simulate", T_Simulate, FOLLBY_TOKEN }, 242 { "simulation_duration",T_Sim_Duration, FOLLBY_TOKEN }, 243 { "beep_delay", T_Beep_Delay, FOLLBY_TOKEN }, 244 { "duration", T_Duration, FOLLBY_TOKEN }, 245 { "server_offset", T_Server_Offset, FOLLBY_TOKEN }, 246 { "freq_offset", T_Freq_Offset, FOLLBY_TOKEN }, 247 { "wander", T_Wander, FOLLBY_TOKEN }, 248 { "jitter", T_Jitter, FOLLBY_TOKEN }, 249 { "prop_delay", T_Prop_Delay, FOLLBY_TOKEN }, 250 { "proc_delay", T_Proc_Delay, FOLLBY_TOKEN }, 251 }; 252 253 typedef struct big_scan_state_tag { 254 char ch; /* Character this state matches on */ 255 char followedby; /* Forces next token(s) to T_String */ 256 u_short finishes_token; /* nonzero ID if last keyword char */ 257 u_short match_next_s; /* next state to check matching ch */ 258 u_short other_next_s; /* next state to check if not ch */ 259 } big_scan_state; 260 261 /* 262 * Note: to increase MAXSTATES beyond 2048, be aware it is currently 263 * crammed into 11 bits in scan_state form. Raising to 4096 would be 264 * relatively easy by storing the followedby value in a separate 265 * array with one entry per token, and shrinking the char value to 266 * 7 bits to free a bit for accepting/non-accepting. More than 4096 267 * states will require expanding scan_state beyond 32 bits each. 268 */ 269 #define MAXSTATES 2048 270 #define MAX_TOK_LEN 63 271 272 const char * current_keyword;/* for error reporting */ 273 big_scan_state sst[MAXSTATES]; /* scanner FSM state entries */ 274 u_short sst_highwater; /* next entry index to consider */ 275 char * symb[1024]; /* map token ID to symbolic name */ 276 277 /* for libntp */ 278 const char * progname = "keyword-gen"; 279 280 int main (int, char **); 281 static void generate_preamble (void); 282 static void generate_fsm (void); 283 static void generate_token_text (void); 284 static u_short create_keyword_scanner (void); 285 static u_short create_scan_states (char *, u_short, follby, u_short); 286 int compare_key_tok_id (const void *, const void *); 287 int compare_key_tok_text (const void *, const void *); 288 void populate_symb (char *); 289 const char * symbname (u_short); 290 291 292 int main(int argc, char **argv) 293 { 294 if (argc < 2) { 295 fprintf(stderr, "Usage:\n%s t_header.h\n", argv[0]); 296 exit(1); 297 } 298 debug = 1; 299 300 populate_symb(argv[1]); 301 302 generate_preamble(); 303 generate_token_text(); 304 generate_fsm(); 305 306 return 0; 307 } 308 309 310 static void 311 generate_preamble(void) 312 { 313 time_t now; 314 char timestamp[128]; 315 char preamble[] = 316 "/*\n" 317 " * ntp_keyword.h\n" 318 " * \n" 319 " * NOTE: edit this file with caution, it is generated by keyword-gen.c\n" 320 " *\t Generated %s UTC diff_ignore_line\n" 321 " *\n" 322 " */\n" 323 "#include \"ntp_scanner.h\"\n" 324 "#include \"ntp_parser.h\"\n" 325 "\n"; 326 327 time(&now); 328 if (!strftime(timestamp, sizeof(timestamp), 329 "%Y-%m-%d %H:%M:%S", gmtime(&now))) 330 timestamp[0] = '\0'; 331 332 printf(preamble, timestamp); 333 } 334 335 336 static void 337 generate_fsm(void) 338 { 339 char rprefix[MAX_TOK_LEN + 1]; 340 char prefix[MAX_TOK_LEN + 1]; 341 char token_id_comment[16 + MAX_TOK_LEN + 1]; 342 size_t prefix_len; 343 char *p; 344 char *r; 345 u_short initial_state; 346 u_short this_state; 347 u_short state; 348 u_short i; 349 u_short token; 350 351 /* 352 * Sort ntp_keywords in alphabetical keyword order. This is 353 * not necessary, but minimizes nonfunctional changes in the 354 * generated finite state machine when keywords are modified. 355 */ 356 qsort(ntp_keywords, COUNTOF(ntp_keywords), 357 sizeof(ntp_keywords[0]), compare_key_tok_text); 358 359 /* 360 * To save space, reserve the state array entry matching each 361 * token number for its terminal state, so the token identifier 362 * does not need to be stored in each state, but can be 363 * recovered trivially. To mark the entry reserved, 364 * finishes_token is nonzero. 365 */ 366 367 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 368 token = ntp_keywords[i].token; 369 if (1 > token || token >= COUNTOF(sst)) { 370 fprintf(stderr, 371 "keyword-gen sst[%u] too small " 372 "for keyword '%s' id %d\n", 373 (int)COUNTOF(sst), 374 ntp_keywords[i].key, 375 token); 376 exit(4); 377 } 378 sst[token].finishes_token = token; 379 } 380 381 initial_state = create_keyword_scanner(); 382 383 fprintf(stderr, 384 "%d keywords consumed %d states of %d max.\n", 385 (int)COUNTOF(ntp_keywords), 386 sst_highwater - 1, 387 (int)COUNTOF(sst) - 1); 388 389 printf("#define SCANNER_INIT_S %d\n\n", initial_state); 390 391 printf("const scan_state sst[%d] = {\n" 392 "/*SS_T( ch,\tf-by, match, other ),\t\t\t\t */\n" 393 " 0,\t\t\t\t /* %5d %-17s */\n", 394 sst_highwater, 395 0, ""); 396 397 for (i = 1; i < sst_highwater; i++) { 398 399 /* verify fields will fit */ 400 if (sst[i].followedby & ~0x3) { 401 fprintf(stderr, 402 "keyword-gen internal error " 403 "sst[%d].followedby %d too big\n", 404 i, sst[i].followedby); 405 exit(7); 406 } 407 408 if (sst_highwater <= sst[i].match_next_s 409 || sst[i].match_next_s & ~0x7ff) { 410 fprintf(stderr, 411 "keyword-gen internal error " 412 "sst[%d].match_next_s %d too big\n", 413 i, sst[i].match_next_s); 414 exit(8); 415 } 416 417 if (sst_highwater <= sst[i].other_next_s 418 || sst[i].other_next_s & ~0x7ff) { 419 fprintf(stderr, 420 "keyword-gen internal error " 421 "sst[%d].other_next_s %d too big\n", 422 i, sst[i].other_next_s); 423 exit(9); 424 } 425 426 if (sst[i].finishes_token) { 427 snprintf(token_id_comment, 428 sizeof(token_id_comment), "%5d %-17s", 429 i, symbname(sst[i].finishes_token)); 430 if (i != sst[i].finishes_token) { 431 fprintf(stderr, 432 "keyword-gen internal error " 433 "entry %d finishes token %d\n", 434 i, sst[i].finishes_token); 435 exit(5); 436 } 437 } else { 438 /* 439 * Determine the keyword prefix that leads to this 440 * state. This is expensive but keyword-gen is run 441 * only when it changes. Distributing keyword-gen-utd 442 * achieves that, which is why it must be committed 443 * at the same time as keyword-gen.c and ntp_keyword.h. 444 * 445 * Scan the state array iteratively looking for a state 446 * which leads to the current one, collecting matching 447 * characters along the way. There is only one such 448 * path back to the starting state given the way our 449 * scanner state machine is built and the practice of 450 * using the spelling of the keyword as its T_* token 451 * identifier, which results in never having two 452 * spellings result in the same T_* value. 453 */ 454 prefix_len = 0; 455 this_state = i; 456 do { 457 for (state = 1; state < sst_highwater; state++) 458 if (sst[state].other_next_s == this_state) { 459 this_state = state; 460 break; 461 } else if (sst[state].match_next_s == this_state) { 462 this_state = state; 463 rprefix[prefix_len] = sst[state].ch; 464 prefix_len++; 465 break; 466 } 467 } while (this_state != initial_state); 468 469 if (prefix_len) { 470 /* reverse rprefix into prefix */ 471 p = prefix + prefix_len; 472 r = rprefix; 473 while (r < rprefix + prefix_len) 474 *--p = *r++; 475 } 476 prefix[prefix_len] = '\0'; 477 478 snprintf(token_id_comment, 479 sizeof(token_id_comment), "%5d %-17s", 480 i, (initial_state == i) 481 ? "[initial state]" 482 : prefix); 483 } 484 485 printf(" S_ST( '%c',\t%d, %5u, %5u )%s /* %s */\n", 486 sst[i].ch, 487 sst[i].followedby, 488 sst[i].match_next_s, 489 sst[i].other_next_s, 490 (i + 1 < sst_highwater) 491 ? "," 492 : " ", 493 token_id_comment); 494 } 495 496 printf("};\n\n"); 497 } 498 499 500 /* Define a function to create the states of the scanner. This function 501 * is used by the create_keyword_scanner function below. 502 * 503 * This function takes a suffix of a keyword, the token to be returned on 504 * recognizing the complete keyword, and any pre-existing state that exists 505 * for some other keyword that has the same prefix as the current one. 506 */ 507 static u_short 508 create_scan_states( 509 char * text, 510 u_short token, 511 follby followedby, 512 u_short prev_state 513 ) 514 { 515 u_short my_state; 516 u_short return_state; 517 u_short prev_char_s; 518 u_short curr_char_s; 519 520 return_state = prev_state; 521 curr_char_s = prev_state; 522 prev_char_s = 0; 523 524 /* Find the correct position to insert the state. 525 * All states should be in alphabetical order 526 */ 527 while (curr_char_s && (text[0] < sst[curr_char_s].ch)) { 528 prev_char_s = curr_char_s; 529 curr_char_s = sst[curr_char_s].other_next_s; 530 } 531 532 /* 533 * Check if a previously seen keyword has the same prefix as 534 * the current keyword. If so, simply use the state for that 535 * keyword as my_state, otherwise, allocate a new state. 536 */ 537 if (curr_char_s && (text[0] == sst[curr_char_s].ch)) { 538 my_state = curr_char_s; 539 if ('\0' == text[1]) { 540 fprintf(stderr, 541 "Duplicate entries for keyword '%s' in" 542 " keyword_gen.c ntp_keywords[].\n", 543 current_keyword); 544 exit(2); 545 } 546 } else { 547 do 548 my_state = sst_highwater++; 549 while (my_state < COUNTOF(sst) 550 && sst[my_state].finishes_token); 551 if (my_state >= COUNTOF(sst)) { 552 fprintf(stderr, 553 "fatal, keyword scanner state array " 554 "sst[%d] is too small, modify\n" 555 "keyword-gen.c to increase.\n", 556 (int)COUNTOF(sst)); 557 exit(3); 558 } 559 /* Store the next character of the keyword */ 560 sst[my_state].ch = text[0]; 561 sst[my_state].other_next_s = curr_char_s; 562 sst[my_state].followedby = FOLLBY_NON_ACCEPTING; 563 564 if (prev_char_s) 565 sst[prev_char_s].other_next_s = my_state; 566 else 567 return_state = my_state; 568 } 569 570 /* Check if the next character is '\0'. 571 * If yes, we are done with the recognition and this is an accepting 572 * state. 573 * If not, we need to continue scanning 574 */ 575 if ('\0' == text[1]) { 576 sst[my_state].finishes_token = (u_short)token; 577 sst[my_state].followedby = (char)followedby; 578 579 if (sst[token].finishes_token != (u_short)token) { 580 fprintf(stderr, 581 "fatal, sst[%d] not reserved for %s.\n", 582 token, symbname(token)); 583 exit(6); 584 } 585 /* relocate so token id is sst[] index */ 586 if (my_state != token) { 587 sst[token] = sst[my_state]; 588 ZERO(sst[my_state]); 589 do 590 sst_highwater--; 591 while (sst[sst_highwater].finishes_token); 592 my_state = token; 593 if (prev_char_s) 594 sst[prev_char_s].other_next_s = my_state; 595 else 596 return_state = my_state; 597 } 598 } else 599 sst[my_state].match_next_s = 600 create_scan_states( 601 &text[1], 602 token, 603 followedby, 604 sst[my_state].match_next_s); 605 606 return return_state; 607 } 608 609 610 /* Define a function that takes a list of (keyword, token) values and 611 * creates a keywords scanner out of it. 612 */ 613 614 static u_short 615 create_keyword_scanner(void) 616 { 617 u_short scanner; 618 u_short i; 619 620 sst_highwater = 1; /* index 0 invalid, unused */ 621 scanner = 0; 622 623 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 624 current_keyword = ntp_keywords[i].key; 625 scanner = 626 create_scan_states( 627 ntp_keywords[i].key, 628 ntp_keywords[i].token, 629 ntp_keywords[i].followedby, 630 scanner); 631 } 632 633 return scanner; 634 } 635 636 637 static void 638 generate_token_text(void) 639 { 640 u_short lowest_id; 641 u_short highest_id; 642 u_short id_count; 643 u_short id; 644 u_short i; 645 646 /* sort ntp_keywords in token ID order */ 647 qsort(ntp_keywords, COUNTOF(ntp_keywords), 648 sizeof(ntp_keywords[0]), compare_key_tok_id); 649 650 lowest_id = ntp_keywords[0].token; 651 highest_id = ntp_keywords[COUNTOF(ntp_keywords) - 1].token; 652 id_count = highest_id - lowest_id + 1; 653 654 printf("#define LOWEST_KEYWORD_ID %d\n\n", lowest_id); 655 656 printf("const char * const keyword_text[%d] = {", id_count); 657 658 id = lowest_id; 659 i = 0; 660 while (i < COUNTOF(ntp_keywords)) { 661 while (id < ntp_keywords[i].token) { 662 printf(",\n\t/* %-5d %5d %20s */\tNULL", 663 id - lowest_id, id, symbname(id)); 664 id++; 665 } 666 if (i > 0) 667 printf(","); 668 printf("\n\t/* %-5d %5d %20s */\t\"%s\"", 669 id - lowest_id, id, symbname(id), 670 ntp_keywords[i].key); 671 i++; 672 id++; 673 } 674 675 printf("\n};\n\n"); 676 } 677 678 679 int 680 compare_key_tok_id( 681 const void *a1, 682 const void *a2 683 ) 684 { 685 const struct key_tok *p1 = a1; 686 const struct key_tok *p2 = a2; 687 688 if (p1->token == p2->token) 689 return 0; 690 691 if (p1->token < p2->token) 692 return -1; 693 else 694 return 1; 695 } 696 697 698 int 699 compare_key_tok_text( 700 const void *a1, 701 const void *a2 702 ) 703 { 704 const struct key_tok *p1 = a1; 705 const struct key_tok *p2 = a2; 706 707 return strcmp(p1->key, p2->key); 708 } 709 710 711 /* 712 * populate_symb() - populate symb[] lookup array with symbolic token 713 * names such that symb[T_Age] == "T_Age", etc. 714 */ 715 void 716 populate_symb( 717 char *header_file 718 ) 719 { 720 FILE * yh; 721 char line[2 * MAX_TOK_LEN]; 722 char name[2 * MAX_TOK_LEN]; 723 int token; 724 725 yh = fopen(header_file, "r"); 726 if (NULL == yh) { 727 perror("unable to open yacc/bison header file"); 728 exit(4); 729 } 730 731 while (NULL != fgets(line, sizeof(line), yh)) 732 if (2 == sscanf(line, "#define %s %d", name, &token) 733 && 'T' == name[0] && '_' == name[1] && token >= 0 734 && token < COUNTOF(symb)) { 735 736 symb[token] = estrdup(name); 737 if (strlen(name) > MAX_TOK_LEN) { 738 fprintf(stderr, 739 "MAX_TOK_LEN %d too small for '%s'\n" 740 "Edit keyword-gen.c to raise.\n", 741 MAX_TOK_LEN, name); 742 exit(10); 743 } 744 } 745 fclose(yh); 746 } 747 748 749 const char * 750 symbname( 751 u_short token 752 ) 753 { 754 char *name; 755 756 if (token < COUNTOF(symb) && symb[token] != NULL) { 757 name = symb[token]; 758 } else { 759 LIB_GETBUF(name); 760 snprintf(name, LIB_BUFLENGTH, "%d", token); 761 } 762 763 return name; 764 } 765