keyword-gen.c revision 310419
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 */ 32struct 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 39struct 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/* access_control_flag */ 157{ "default", T_Default, FOLLBY_TOKEN }, 158{ "source", T_Source, FOLLBY_TOKEN }, 159{ "flake", T_Flake, FOLLBY_TOKEN }, 160{ "ignore", T_Ignore, FOLLBY_TOKEN }, 161{ "limited", T_Limited, FOLLBY_TOKEN }, 162{ "mssntp", T_Mssntp, FOLLBY_TOKEN }, 163{ "kod", T_Kod, FOLLBY_TOKEN }, 164{ "lowpriotrap", T_Lowpriotrap, FOLLBY_TOKEN }, 165{ "mask", T_Mask, FOLLBY_TOKEN }, 166{ "nomodify", T_Nomodify, FOLLBY_TOKEN }, 167{ "nomrulist", T_Nomrulist, FOLLBY_TOKEN }, 168{ "nopeer", T_Nopeer, FOLLBY_TOKEN }, 169{ "noquery", T_Noquery, FOLLBY_TOKEN }, 170{ "noserve", T_Noserve, FOLLBY_TOKEN }, 171{ "notrap", T_Notrap, FOLLBY_TOKEN }, 172{ "notrust", T_Notrust, FOLLBY_TOKEN }, 173{ "ntpport", T_Ntpport, FOLLBY_TOKEN }, 174/* discard_option */ 175{ "average", T_Average, FOLLBY_TOKEN }, 176{ "minimum", T_Minimum, FOLLBY_TOKEN }, 177{ "monitor", T_Monitor, FOLLBY_TOKEN }, 178/* mru_option */ 179{ "incalloc", T_Incalloc, FOLLBY_TOKEN }, 180{ "incmem", T_Incmem, FOLLBY_TOKEN }, 181{ "initalloc", T_Initalloc, FOLLBY_TOKEN }, 182{ "initmem", T_Initmem, FOLLBY_TOKEN }, 183{ "mindepth", T_Mindepth, FOLLBY_TOKEN }, 184{ "maxage", T_Maxage, FOLLBY_TOKEN }, 185{ "maxdepth", T_Maxdepth, FOLLBY_TOKEN }, 186{ "maxmem", T_Maxmem, FOLLBY_TOKEN }, 187{ "mru", T_Mru, FOLLBY_TOKEN }, 188/* fudge_factor */ 189{ "abbrev", T_Abbrev, FOLLBY_STRING }, 190{ "flag1", T_Flag1, FOLLBY_TOKEN }, 191{ "flag2", T_Flag2, FOLLBY_TOKEN }, 192{ "flag3", T_Flag3, FOLLBY_TOKEN }, 193{ "flag4", T_Flag4, FOLLBY_TOKEN }, 194{ "refid", T_Refid, FOLLBY_STRING }, 195{ "stratum", T_Stratum, FOLLBY_TOKEN }, 196{ "time1", T_Time1, FOLLBY_TOKEN }, 197{ "time2", T_Time2, FOLLBY_TOKEN }, 198/* system_option */ 199{ "auth", T_Auth, FOLLBY_TOKEN }, 200{ "bclient", T_Bclient, FOLLBY_TOKEN }, 201{ "calibrate", T_Calibrate, FOLLBY_TOKEN }, 202{ "kernel", T_Kernel, FOLLBY_TOKEN }, 203{ "mode7", T_Mode7, FOLLBY_TOKEN }, 204{ "ntp", T_Ntp, FOLLBY_TOKEN }, 205{ "peer_clear_digest_early", T_PCEdigest, FOLLBY_TOKEN }, 206{ "stats", T_Stats, FOLLBY_TOKEN }, 207{ "unpeer_crypto_early", T_UEcrypto, FOLLBY_TOKEN }, 208{ "unpeer_crypto_nak_early", T_UEcryptonak, FOLLBY_TOKEN }, 209{ "unpeer_digest_early", T_UEdigest, FOLLBY_TOKEN }, 210/* rlimit_option */ 211{ "memlock", T_Memlock, FOLLBY_TOKEN }, 212{ "stacksize", T_Stacksize, FOLLBY_TOKEN }, 213{ "filenum", T_Filenum, FOLLBY_TOKEN }, 214/* tinker_option */ 215{ "step", T_Step, FOLLBY_TOKEN }, 216{ "stepback", T_Stepback, FOLLBY_TOKEN }, 217{ "stepfwd", T_Stepfwd, FOLLBY_TOKEN }, 218{ "panic", T_Panic, FOLLBY_TOKEN }, 219{ "dispersion", T_Dispersion, FOLLBY_TOKEN }, 220{ "stepout", T_Stepout, FOLLBY_TOKEN }, 221{ "allan", T_Allan, FOLLBY_TOKEN }, 222{ "huffpuff", T_Huffpuff, FOLLBY_TOKEN }, 223{ "freq", T_Freq, FOLLBY_TOKEN }, 224/* miscellaneous_command */ 225{ "port", T_Port, FOLLBY_TOKEN }, 226{ "interface", T_Interface, FOLLBY_TOKEN }, 227{ "saveconfigdir", T_Saveconfigdir, FOLLBY_STRING }, 228/* interface_command (ignore and interface already defined) */ 229{ "nic", T_Nic, FOLLBY_TOKEN }, 230{ "all", T_All, FOLLBY_TOKEN }, 231{ "ipv4", T_Ipv4, FOLLBY_TOKEN }, 232{ "ipv6", T_Ipv6, FOLLBY_TOKEN }, 233{ "wildcard", T_Wildcard, FOLLBY_TOKEN }, 234{ "listen", T_Listen, FOLLBY_TOKEN }, 235{ "drop", T_Drop, FOLLBY_TOKEN }, 236/* simulator commands */ 237{ "simulate", T_Simulate, FOLLBY_TOKEN }, 238{ "simulation_duration",T_Sim_Duration, FOLLBY_TOKEN }, 239{ "beep_delay", T_Beep_Delay, FOLLBY_TOKEN }, 240{ "duration", T_Duration, FOLLBY_TOKEN }, 241{ "server_offset", T_Server_Offset, FOLLBY_TOKEN }, 242{ "freq_offset", T_Freq_Offset, FOLLBY_TOKEN }, 243{ "wander", T_Wander, FOLLBY_TOKEN }, 244{ "jitter", T_Jitter, FOLLBY_TOKEN }, 245{ "prop_delay", T_Prop_Delay, FOLLBY_TOKEN }, 246{ "proc_delay", T_Proc_Delay, FOLLBY_TOKEN }, 247}; 248 249typedef struct big_scan_state_tag { 250 char ch; /* Character this state matches on */ 251 char followedby; /* Forces next token(s) to T_String */ 252 u_short finishes_token; /* nonzero ID if last keyword char */ 253 u_short match_next_s; /* next state to check matching ch */ 254 u_short other_next_s; /* next state to check if not ch */ 255} big_scan_state; 256 257/* 258 * Note: to increase MAXSTATES beyond 2048, be aware it is currently 259 * crammed into 11 bits in scan_state form. Raising to 4096 would be 260 * relatively easy by storing the followedby value in a separate 261 * array with one entry per token, and shrinking the char value to 262 * 7 bits to free a bit for accepting/non-accepting. More than 4096 263 * states will require expanding scan_state beyond 32 bits each. 264 */ 265#define MAXSTATES 2048 266#define MAX_TOK_LEN 63 267 268const char * current_keyword;/* for error reporting */ 269big_scan_state sst[MAXSTATES]; /* scanner FSM state entries */ 270u_short sst_highwater; /* next entry index to consider */ 271char * symb[1024]; /* map token ID to symbolic name */ 272 273/* for libntp */ 274const char * progname = "keyword-gen"; 275 276int main (int, char **); 277static void generate_preamble (void); 278static void generate_fsm (void); 279static void generate_token_text (void); 280static u_short create_keyword_scanner (void); 281static u_short create_scan_states (char *, u_short, follby, u_short); 282int compare_key_tok_id (const void *, const void *); 283int compare_key_tok_text (const void *, const void *); 284void populate_symb (char *); 285const char * symbname (u_short); 286 287 288int main(int argc, char **argv) 289{ 290 if (argc < 2) { 291 fprintf(stderr, "Usage:\n%s t_header.h\n", argv[0]); 292 exit(1); 293 } 294 debug = 1; 295 296 populate_symb(argv[1]); 297 298 generate_preamble(); 299 generate_token_text(); 300 generate_fsm(); 301 302 return 0; 303} 304 305 306static void 307generate_preamble(void) 308{ 309 time_t now; 310 char timestamp[128]; 311 char preamble[] = 312"/*\n" 313" * ntp_keyword.h\n" 314" * \n" 315" * NOTE: edit this file with caution, it is generated by keyword-gen.c\n" 316" *\t Generated %s UTC diff_ignore_line\n" 317" *\n" 318" */\n" 319"#include \"ntp_scanner.h\"\n" 320"#include \"ntp_parser.h\"\n" 321"\n"; 322 323 time(&now); 324 if (!strftime(timestamp, sizeof(timestamp), 325 "%Y-%m-%d %H:%M:%S", gmtime(&now))) 326 timestamp[0] = '\0'; 327 328 printf(preamble, timestamp); 329} 330 331 332static void 333generate_fsm(void) 334{ 335 char rprefix[MAX_TOK_LEN + 1]; 336 char prefix[MAX_TOK_LEN + 1]; 337 char token_id_comment[16 + MAX_TOK_LEN + 1]; 338 size_t prefix_len; 339 char *p; 340 char *r; 341 u_short initial_state; 342 u_short this_state; 343 u_short state; 344 u_short i; 345 u_short token; 346 347 /* 348 * Sort ntp_keywords in alphabetical keyword order. This is 349 * not necessary, but minimizes nonfunctional changes in the 350 * generated finite state machine when keywords are modified. 351 */ 352 qsort(ntp_keywords, COUNTOF(ntp_keywords), 353 sizeof(ntp_keywords[0]), compare_key_tok_text); 354 355 /* 356 * To save space, reserve the state array entry matching each 357 * token number for its terminal state, so the token identifier 358 * does not need to be stored in each state, but can be 359 * recovered trivially. To mark the entry reserved, 360 * finishes_token is nonzero. 361 */ 362 363 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 364 token = ntp_keywords[i].token; 365 if (1 > token || token >= COUNTOF(sst)) { 366 fprintf(stderr, 367 "keyword-gen sst[%u] too small " 368 "for keyword '%s' id %d\n", 369 (int)COUNTOF(sst), 370 ntp_keywords[i].key, 371 token); 372 exit(4); 373 } 374 sst[token].finishes_token = token; 375 } 376 377 initial_state = create_keyword_scanner(); 378 379 fprintf(stderr, 380 "%d keywords consumed %d states of %d max.\n", 381 (int)COUNTOF(ntp_keywords), 382 sst_highwater - 1, 383 (int)COUNTOF(sst) - 1); 384 385 printf("#define SCANNER_INIT_S %d\n\n", initial_state); 386 387 printf("const scan_state sst[%d] = {\n" 388 "/*SS_T( ch,\tf-by, match, other ),\t\t\t\t */\n" 389 " 0,\t\t\t\t /* %5d %-17s */\n", 390 sst_highwater, 391 0, ""); 392 393 for (i = 1; i < sst_highwater; i++) { 394 395 /* verify fields will fit */ 396 if (sst[i].followedby & ~0x3) { 397 fprintf(stderr, 398 "keyword-gen internal error " 399 "sst[%d].followedby %d too big\n", 400 i, sst[i].followedby); 401 exit(7); 402 } 403 404 if (sst_highwater <= sst[i].match_next_s 405 || sst[i].match_next_s & ~0x7ff) { 406 fprintf(stderr, 407 "keyword-gen internal error " 408 "sst[%d].match_next_s %d too big\n", 409 i, sst[i].match_next_s); 410 exit(8); 411 } 412 413 if (sst_highwater <= sst[i].other_next_s 414 || sst[i].other_next_s & ~0x7ff) { 415 fprintf(stderr, 416 "keyword-gen internal error " 417 "sst[%d].other_next_s %d too big\n", 418 i, sst[i].other_next_s); 419 exit(9); 420 } 421 422 if (sst[i].finishes_token) { 423 snprintf(token_id_comment, 424 sizeof(token_id_comment), "%5d %-17s", 425 i, symbname(sst[i].finishes_token)); 426 if (i != sst[i].finishes_token) { 427 fprintf(stderr, 428 "keyword-gen internal error " 429 "entry %d finishes token %d\n", 430 i, sst[i].finishes_token); 431 exit(5); 432 } 433 } else { 434 /* 435 * Determine the keyword prefix that leads to this 436 * state. This is expensive but keyword-gen is run 437 * only when it changes. Distributing keyword-gen-utd 438 * achieves that, which is why it must be committed 439 * at the same time as keyword-gen.c and ntp_keyword.h. 440 * 441 * Scan the state array iteratively looking for a state 442 * which leads to the current one, collecting matching 443 * characters along the way. There is only one such 444 * path back to the starting state given the way our 445 * scanner state machine is built and the practice of 446 * using the spelling of the keyword as its T_* token 447 * identifier, which results in never having two 448 * spellings result in the same T_* value. 449 */ 450 prefix_len = 0; 451 this_state = i; 452 do { 453 for (state = 1; state < sst_highwater; state++) 454 if (sst[state].other_next_s == this_state) { 455 this_state = state; 456 break; 457 } else if (sst[state].match_next_s == this_state) { 458 this_state = state; 459 rprefix[prefix_len] = sst[state].ch; 460 prefix_len++; 461 break; 462 } 463 } while (this_state != initial_state); 464 465 if (prefix_len) { 466 /* reverse rprefix into prefix */ 467 p = prefix + prefix_len; 468 r = rprefix; 469 while (r < rprefix + prefix_len) 470 *--p = *r++; 471 } 472 prefix[prefix_len] = '\0'; 473 474 snprintf(token_id_comment, 475 sizeof(token_id_comment), "%5d %-17s", 476 i, (initial_state == i) 477 ? "[initial state]" 478 : prefix); 479 } 480 481 printf(" S_ST( '%c',\t%d, %5u, %5u )%s /* %s */\n", 482 sst[i].ch, 483 sst[i].followedby, 484 sst[i].match_next_s, 485 sst[i].other_next_s, 486 (i + 1 < sst_highwater) 487 ? "," 488 : " ", 489 token_id_comment); 490 } 491 492 printf("};\n\n"); 493} 494 495 496/* Define a function to create the states of the scanner. This function 497 * is used by the create_keyword_scanner function below. 498 * 499 * This function takes a suffix of a keyword, the token to be returned on 500 * recognizing the complete keyword, and any pre-existing state that exists 501 * for some other keyword that has the same prefix as the current one. 502 */ 503static u_short 504create_scan_states( 505 char * text, 506 u_short token, 507 follby followedby, 508 u_short prev_state 509 ) 510{ 511 u_short my_state; 512 u_short return_state; 513 u_short prev_char_s; 514 u_short curr_char_s; 515 516 return_state = prev_state; 517 curr_char_s = prev_state; 518 prev_char_s = 0; 519 520 /* Find the correct position to insert the state. 521 * All states should be in alphabetical order 522 */ 523 while (curr_char_s && (text[0] < sst[curr_char_s].ch)) { 524 prev_char_s = curr_char_s; 525 curr_char_s = sst[curr_char_s].other_next_s; 526 } 527 528 /* 529 * Check if a previously seen keyword has the same prefix as 530 * the current keyword. If so, simply use the state for that 531 * keyword as my_state, otherwise, allocate a new state. 532 */ 533 if (curr_char_s && (text[0] == sst[curr_char_s].ch)) { 534 my_state = curr_char_s; 535 if ('\0' == text[1]) { 536 fprintf(stderr, 537 "Duplicate entries for keyword '%s' in" 538 " keyword_gen.c ntp_keywords[].\n", 539 current_keyword); 540 exit(2); 541 } 542 } else { 543 do 544 my_state = sst_highwater++; 545 while (my_state < COUNTOF(sst) 546 && sst[my_state].finishes_token); 547 if (my_state >= COUNTOF(sst)) { 548 fprintf(stderr, 549 "fatal, keyword scanner state array " 550 "sst[%d] is too small, modify\n" 551 "keyword-gen.c to increase.\n", 552 (int)COUNTOF(sst)); 553 exit(3); 554 } 555 /* Store the next character of the keyword */ 556 sst[my_state].ch = text[0]; 557 sst[my_state].other_next_s = curr_char_s; 558 sst[my_state].followedby = FOLLBY_NON_ACCEPTING; 559 560 if (prev_char_s) 561 sst[prev_char_s].other_next_s = my_state; 562 else 563 return_state = my_state; 564 } 565 566 /* Check if the next character is '\0'. 567 * If yes, we are done with the recognition and this is an accepting 568 * state. 569 * If not, we need to continue scanning 570 */ 571 if ('\0' == text[1]) { 572 sst[my_state].finishes_token = (u_short)token; 573 sst[my_state].followedby = (char)followedby; 574 575 if (sst[token].finishes_token != (u_short)token) { 576 fprintf(stderr, 577 "fatal, sst[%d] not reserved for %s.\n", 578 token, symbname(token)); 579 exit(6); 580 } 581 /* relocate so token id is sst[] index */ 582 if (my_state != token) { 583 sst[token] = sst[my_state]; 584 ZERO(sst[my_state]); 585 do 586 sst_highwater--; 587 while (sst[sst_highwater].finishes_token); 588 my_state = token; 589 if (prev_char_s) 590 sst[prev_char_s].other_next_s = my_state; 591 else 592 return_state = my_state; 593 } 594 } else 595 sst[my_state].match_next_s = 596 create_scan_states( 597 &text[1], 598 token, 599 followedby, 600 sst[my_state].match_next_s); 601 602 return return_state; 603} 604 605 606/* Define a function that takes a list of (keyword, token) values and 607 * creates a keywords scanner out of it. 608 */ 609 610static u_short 611create_keyword_scanner(void) 612{ 613 u_short scanner; 614 u_short i; 615 616 sst_highwater = 1; /* index 0 invalid, unused */ 617 scanner = 0; 618 619 for (i = 0; i < COUNTOF(ntp_keywords); i++) { 620 current_keyword = ntp_keywords[i].key; 621 scanner = 622 create_scan_states( 623 ntp_keywords[i].key, 624 ntp_keywords[i].token, 625 ntp_keywords[i].followedby, 626 scanner); 627 } 628 629 return scanner; 630} 631 632 633static void 634generate_token_text(void) 635{ 636 u_short lowest_id; 637 u_short highest_id; 638 u_short id_count; 639 u_short id; 640 u_short i; 641 642 /* sort ntp_keywords in token ID order */ 643 qsort(ntp_keywords, COUNTOF(ntp_keywords), 644 sizeof(ntp_keywords[0]), compare_key_tok_id); 645 646 lowest_id = ntp_keywords[0].token; 647 highest_id = ntp_keywords[COUNTOF(ntp_keywords) - 1].token; 648 id_count = highest_id - lowest_id + 1; 649 650 printf("#define LOWEST_KEYWORD_ID %d\n\n", lowest_id); 651 652 printf("const char * const keyword_text[%d] = {", id_count); 653 654 id = lowest_id; 655 i = 0; 656 while (i < COUNTOF(ntp_keywords)) { 657 while (id < ntp_keywords[i].token) { 658 printf(",\n\t/* %-5d %5d %20s */\tNULL", 659 id - lowest_id, id, symbname(id)); 660 id++; 661 } 662 if (i > 0) 663 printf(","); 664 printf("\n\t/* %-5d %5d %20s */\t\"%s\"", 665 id - lowest_id, id, symbname(id), 666 ntp_keywords[i].key); 667 i++; 668 id++; 669 } 670 671 printf("\n};\n\n"); 672} 673 674 675int 676compare_key_tok_id( 677 const void *a1, 678 const void *a2 679 ) 680{ 681 const struct key_tok *p1 = a1; 682 const struct key_tok *p2 = a2; 683 684 if (p1->token == p2->token) 685 return 0; 686 687 if (p1->token < p2->token) 688 return -1; 689 else 690 return 1; 691} 692 693 694int 695compare_key_tok_text( 696 const void *a1, 697 const void *a2 698 ) 699{ 700 const struct key_tok *p1 = a1; 701 const struct key_tok *p2 = a2; 702 703 return strcmp(p1->key, p2->key); 704} 705 706 707/* 708 * populate_symb() - populate symb[] lookup array with symbolic token 709 * names such that symb[T_Age] == "T_Age", etc. 710 */ 711void 712populate_symb( 713 char *header_file 714 ) 715{ 716 FILE * yh; 717 char line[2 * MAX_TOK_LEN]; 718 char name[2 * MAX_TOK_LEN]; 719 int token; 720 721 yh = fopen(header_file, "r"); 722 if (NULL == yh) { 723 perror("unable to open yacc/bison header file"); 724 exit(4); 725 } 726 727 while (NULL != fgets(line, sizeof(line), yh)) 728 if (2 == sscanf(line, "#define %s %d", name, &token) 729 && 'T' == name[0] && '_' == name[1] && token >= 0 730 && token < COUNTOF(symb)) { 731 732 symb[token] = estrdup(name); 733 if (strlen(name) > MAX_TOK_LEN) { 734 fprintf(stderr, 735 "MAX_TOK_LEN %d too small for '%s'\n" 736 "Edit keyword-gen.c to raise.\n", 737 MAX_TOK_LEN, name); 738 exit(10); 739 } 740 } 741 fclose(yh); 742} 743 744 745const char * 746symbname( 747 u_short token 748 ) 749{ 750 char *name; 751 752 if (token < COUNTOF(symb) && symb[token] != NULL) { 753 name = symb[token]; 754 } else { 755 LIB_GETBUF(name); 756 snprintf(name, LIB_BUFLENGTH, "%d", token); 757 } 758 759 return name; 760} 761