1 /* $OpenBSD: man_validate.c,v 1.126 2021/08/10 12:36:42 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2010, 2012-2020 Ingo Schwarze <schwarze@openbsd.org> 4 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 * 18 * Validation module for man(7) syntax trees used by mandoc(1). 19 */ 20 #include <sys/types.h> 21 22 #include <assert.h> 23 #include <ctype.h> 24 #include <errno.h> 25 #include <limits.h> 26 #include <stdarg.h> 27 #include <stdio.h> 28 #include <stdlib.h> 29 #include <string.h> 30 #include <time.h> 31 32 #include "mandoc_aux.h" 33 #include "mandoc.h" 34 #include "roff.h" 35 #include "man.h" 36 #include "libmandoc.h" 37 #include "roff_int.h" 38 #include "libman.h" 39 #include "tag.h" 40 41 #define CHKARGS struct roff_man *man, struct roff_node *n 42 43 typedef void (*v_check)(CHKARGS); 44 45 static void check_abort(CHKARGS) __attribute__((__noreturn__)); 46 static void check_par(CHKARGS); 47 static void check_part(CHKARGS); 48 static void check_root(CHKARGS); 49 static void check_tag(struct roff_node *, struct roff_node *); 50 static void check_text(CHKARGS); 51 52 static void post_AT(CHKARGS); 53 static void post_EE(CHKARGS); 54 static void post_EX(CHKARGS); 55 static void post_IP(CHKARGS); 56 static void post_OP(CHKARGS); 57 static void post_SH(CHKARGS); 58 static void post_TH(CHKARGS); 59 static void post_TP(CHKARGS); 60 static void post_UC(CHKARGS); 61 static void post_UR(CHKARGS); 62 static void post_in(CHKARGS); 63 64 static const v_check man_valids[MAN_MAX - MAN_TH] = { 65 post_TH, /* TH */ 66 post_SH, /* SH */ 67 post_SH, /* SS */ 68 post_TP, /* TP */ 69 post_TP, /* TQ */ 70 check_abort,/* LP */ 71 check_par, /* PP */ 72 check_abort,/* P */ 73 post_IP, /* IP */ 74 NULL, /* HP */ 75 NULL, /* SM */ 76 NULL, /* SB */ 77 NULL, /* BI */ 78 NULL, /* IB */ 79 NULL, /* BR */ 80 NULL, /* RB */ 81 NULL, /* R */ 82 NULL, /* B */ 83 NULL, /* I */ 84 NULL, /* IR */ 85 NULL, /* RI */ 86 NULL, /* RE */ 87 check_part, /* RS */ 88 NULL, /* DT */ 89 post_UC, /* UC */ 90 NULL, /* PD */ 91 post_AT, /* AT */ 92 post_in, /* in */ 93 NULL, /* SY */ 94 NULL, /* YS */ 95 post_OP, /* OP */ 96 post_EX, /* EX */ 97 post_EE, /* EE */ 98 post_UR, /* UR */ 99 NULL, /* UE */ 100 post_UR, /* MT */ 101 NULL, /* ME */ 102 }; 103 104 105 /* Validate the subtree rooted at man->last. */ 106 void 107 man_validate(struct roff_man *man) 108 { 109 struct roff_node *n; 110 const v_check *cp; 111 112 /* 113 * Translate obsolete macros such that later code 114 * does not need to look for them. 115 */ 116 117 n = man->last; 118 switch (n->tok) { 119 case MAN_LP: 120 case MAN_P: 121 n->tok = MAN_PP; 122 break; 123 default: 124 break; 125 } 126 127 /* 128 * Iterate over all children, recursing into each one 129 * in turn, depth-first. 130 */ 131 132 man->last = man->last->child; 133 while (man->last != NULL) { 134 man_validate(man); 135 if (man->last == n) 136 man->last = man->last->child; 137 else 138 man->last = man->last->next; 139 } 140 141 /* Finally validate the macro itself. */ 142 143 man->last = n; 144 man->next = ROFF_NEXT_SIBLING; 145 switch (n->type) { 146 case ROFFT_TEXT: 147 check_text(man, n); 148 break; 149 case ROFFT_ROOT: 150 check_root(man, n); 151 break; 152 case ROFFT_COMMENT: 153 case ROFFT_EQN: 154 case ROFFT_TBL: 155 break; 156 default: 157 if (n->tok < ROFF_MAX) { 158 roff_validate(man); 159 break; 160 } 161 assert(n->tok >= MAN_TH && n->tok < MAN_MAX); 162 cp = man_valids + (n->tok - MAN_TH); 163 if (*cp) 164 (*cp)(man, n); 165 if (man->last == n) 166 n->flags |= NODE_VALID; 167 break; 168 } 169 } 170 171 static void 172 check_root(CHKARGS) 173 { 174 assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0); 175 176 if (n->last == NULL || n->last->type == ROFFT_COMMENT) 177 mandoc_msg(MANDOCERR_DOC_EMPTY, n->line, n->pos, NULL); 178 else 179 man->meta.hasbody = 1; 180 181 if (NULL == man->meta.title) { 182 mandoc_msg(MANDOCERR_TH_NOTITLE, n->line, n->pos, NULL); 183 184 /* 185 * If a title hasn't been set, do so now (by 186 * implication, date and section also aren't set). 187 */ 188 189 man->meta.title = mandoc_strdup(""); 190 man->meta.msec = mandoc_strdup(""); 191 man->meta.date = mandoc_normdate(NULL, NULL); 192 } 193 194 if (man->meta.os_e && 195 (man->meta.rcsids & (1 << man->meta.os_e)) == 0) 196 mandoc_msg(MANDOCERR_RCS_MISSING, 0, 0, 197 man->meta.os_e == MANDOC_OS_OPENBSD ? 198 "(OpenBSD)" : "(NetBSD)"); 199 } 200 201 static void 202 check_abort(CHKARGS) 203 { 204 abort(); 205 } 206 207 /* 208 * Skip leading whitespace, dashes, backslashes, and font escapes, 209 * then create a tag if the first following byte is a letter. 210 * Priority is high unless whitespace is present. 211 */ 212 static void 213 check_tag(struct roff_node *n, struct roff_node *nt) 214 { 215 const char *cp, *arg; 216 int prio, sz; 217 218 if (nt == NULL || nt->type != ROFFT_TEXT) 219 return; 220 221 cp = nt->string; 222 prio = TAG_STRONG; 223 for (;;) { 224 switch (*cp) { 225 case ' ': 226 case '\t': 227 prio = TAG_WEAK; 228 /* FALLTHROUGH */ 229 case '-': 230 cp++; 231 break; 232 case '\\': 233 cp++; 234 switch (mandoc_escape(&cp, &arg, &sz)) { 235 case ESCAPE_FONT: 236 case ESCAPE_FONTBOLD: 237 case ESCAPE_FONTITALIC: 238 case ESCAPE_FONTBI: 239 case ESCAPE_FONTROMAN: 240 case ESCAPE_FONTCR: 241 case ESCAPE_FONTCB: 242 case ESCAPE_FONTCI: 243 case ESCAPE_FONTPREV: 244 case ESCAPE_IGNORE: 245 break; 246 case ESCAPE_SPECIAL: 247 if (sz != 1) 248 return; 249 switch (*arg) { 250 case '-': 251 case 'e': 252 break; 253 default: 254 return; 255 } 256 break; 257 default: 258 return; 259 } 260 break; 261 default: 262 if (isalpha((unsigned char)*cp)) 263 tag_put(cp, prio, n); 264 return; 265 } 266 } 267 } 268 269 static void 270 check_text(CHKARGS) 271 { 272 char *cp, *p; 273 274 if (n->flags & NODE_NOFILL) 275 return; 276 277 cp = n->string; 278 for (p = cp; NULL != (p = strchr(p, '\t')); p++) 279 mandoc_msg(MANDOCERR_FI_TAB, 280 n->line, n->pos + (int)(p - cp), NULL); 281 } 282 283 static void 284 post_EE(CHKARGS) 285 { 286 if ((n->flags & NODE_NOFILL) == 0) 287 mandoc_msg(MANDOCERR_FI_SKIP, n->line, n->pos, "EE"); 288 } 289 290 static void 291 post_EX(CHKARGS) 292 { 293 if (n->flags & NODE_NOFILL) 294 mandoc_msg(MANDOCERR_NF_SKIP, n->line, n->pos, "EX"); 295 } 296 297 static void 298 post_OP(CHKARGS) 299 { 300 301 if (n->child == NULL) 302 mandoc_msg(MANDOCERR_OP_EMPTY, n->line, n->pos, "OP"); 303 else if (n->child->next != NULL && n->child->next->next != NULL) { 304 n = n->child->next->next; 305 mandoc_msg(MANDOCERR_ARG_EXCESS, 306 n->line, n->pos, "OP ... %s", n->string); 307 } 308 } 309 310 static void 311 post_SH(CHKARGS) 312 { 313 struct roff_node *nc; 314 char *cp, *tag; 315 316 nc = n->child; 317 switch (n->type) { 318 case ROFFT_HEAD: 319 tag = NULL; 320 deroff(&tag, n); 321 if (tag != NULL) { 322 for (cp = tag; *cp != '\0'; cp++) 323 if (*cp == ' ') 324 *cp = '_'; 325 if (nc != NULL && nc->type == ROFFT_TEXT && 326 strcmp(nc->string, tag) == 0) 327 tag_put(NULL, TAG_STRONG, n); 328 else 329 tag_put(tag, TAG_FALLBACK, n); 330 free(tag); 331 } 332 return; 333 case ROFFT_BODY: 334 if (nc != NULL) 335 break; 336 return; 337 default: 338 return; 339 } 340 341 if (nc->tok == MAN_PP && nc->body->child != NULL) { 342 while (nc->body->last != NULL) { 343 man->next = ROFF_NEXT_CHILD; 344 roff_node_relink(man, nc->body->last); 345 man->last = n; 346 } 347 } 348 349 if (nc->tok == MAN_PP || nc->tok == ROFF_sp || nc->tok == ROFF_br) { 350 mandoc_msg(MANDOCERR_PAR_SKIP, nc->line, nc->pos, 351 "%s after %s", roff_name[nc->tok], roff_name[n->tok]); 352 roff_node_delete(man, nc); 353 } 354 355 /* 356 * Trailing PP is empty, so it is deleted by check_par(). 357 * Trailing sp is significant. 358 */ 359 360 if ((nc = n->last) != NULL && nc->tok == ROFF_br) { 361 mandoc_msg(MANDOCERR_PAR_SKIP, 362 nc->line, nc->pos, "%s at the end of %s", 363 roff_name[nc->tok], roff_name[n->tok]); 364 roff_node_delete(man, nc); 365 } 366 } 367 368 static void 369 post_UR(CHKARGS) 370 { 371 if (n->type == ROFFT_HEAD && n->child == NULL) 372 mandoc_msg(MANDOCERR_UR_NOHEAD, n->line, n->pos, 373 "%s", roff_name[n->tok]); 374 check_part(man, n); 375 } 376 377 static void 378 check_part(CHKARGS) 379 { 380 381 if (n->type == ROFFT_BODY && n->child == NULL) 382 mandoc_msg(MANDOCERR_BLK_EMPTY, n->line, n->pos, 383 "%s", roff_name[n->tok]); 384 } 385 386 static void 387 check_par(CHKARGS) 388 { 389 390 switch (n->type) { 391 case ROFFT_BLOCK: 392 if (n->body->child == NULL) 393 roff_node_delete(man, n); 394 break; 395 case ROFFT_BODY: 396 if (n->child != NULL && 397 (n->child->tok == ROFF_sp || n->child->tok == ROFF_br)) { 398 mandoc_msg(MANDOCERR_PAR_SKIP, 399 n->child->line, n->child->pos, 400 "%s after %s", roff_name[n->child->tok], 401 roff_name[n->tok]); 402 roff_node_delete(man, n->child); 403 } 404 if (n->child == NULL) 405 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos, 406 "%s empty", roff_name[n->tok]); 407 break; 408 case ROFFT_HEAD: 409 if (n->child != NULL) 410 mandoc_msg(MANDOCERR_ARG_SKIP, 411 n->line, n->pos, "%s %s%s", 412 roff_name[n->tok], n->child->string, 413 n->child->next != NULL ? " ..." : ""); 414 break; 415 default: 416 break; 417 } 418 } 419 420 static void 421 post_IP(CHKARGS) 422 { 423 switch (n->type) { 424 case ROFFT_BLOCK: 425 if (n->head->child == NULL && n->body->child == NULL) 426 roff_node_delete(man, n); 427 break; 428 case ROFFT_HEAD: 429 check_tag(n, n->child); 430 break; 431 case ROFFT_BODY: 432 if (n->parent->head->child == NULL && n->child == NULL) 433 mandoc_msg(MANDOCERR_PAR_SKIP, n->line, n->pos, 434 "%s empty", roff_name[n->tok]); 435 break; 436 default: 437 break; 438 } 439 } 440 441 /* 442 * The first next-line element in the head is the tag. 443 * If that's a font macro, use its first child instead. 444 */ 445 static void 446 post_TP(CHKARGS) 447 { 448 struct roff_node *nt; 449 450 if (n->type != ROFFT_HEAD || (nt = n->child) == NULL) 451 return; 452 453 while ((nt->flags & NODE_LINE) == 0) 454 if ((nt = nt->next) == NULL) 455 return; 456 457 switch (nt->tok) { 458 case MAN_B: 459 case MAN_BI: 460 case MAN_BR: 461 case MAN_I: 462 case MAN_IB: 463 case MAN_IR: 464 nt = nt->child; 465 break; 466 default: 467 break; 468 } 469 check_tag(n, nt); 470 } 471 472 static void 473 post_TH(CHKARGS) 474 { 475 struct roff_node *nb; 476 const char *p; 477 478 free(man->meta.title); 479 free(man->meta.vol); 480 free(man->meta.os); 481 free(man->meta.msec); 482 free(man->meta.date); 483 484 man->meta.title = man->meta.vol = man->meta.date = 485 man->meta.msec = man->meta.os = NULL; 486 487 nb = n; 488 489 /* ->TITLE<- MSEC DATE OS VOL */ 490 491 n = n->child; 492 if (n != NULL && n->string != NULL) { 493 for (p = n->string; *p != '\0'; p++) { 494 /* Only warn about this once... */ 495 if (isalpha((unsigned char)*p) && 496 ! isupper((unsigned char)*p)) { 497 mandoc_msg(MANDOCERR_TITLE_CASE, n->line, 498 n->pos + (int)(p - n->string), 499 "TH %s", n->string); 500 break; 501 } 502 } 503 man->meta.title = mandoc_strdup(n->string); 504 } else { 505 man->meta.title = mandoc_strdup(""); 506 mandoc_msg(MANDOCERR_TH_NOTITLE, nb->line, nb->pos, "TH"); 507 } 508 509 /* TITLE ->MSEC<- DATE OS VOL */ 510 511 if (n != NULL) 512 n = n->next; 513 if (n != NULL && n->string != NULL) { 514 man->meta.msec = mandoc_strdup(n->string); 515 if (man->filesec != '\0' && 516 man->filesec != *n->string && 517 *n->string >= '1' && *n->string <= '9') 518 mandoc_msg(MANDOCERR_MSEC_FILE, n->line, n->pos, 519 "*.%c vs TH ... %c", man->filesec, *n->string); 520 } else { 521 man->meta.msec = mandoc_strdup(""); 522 mandoc_msg(MANDOCERR_MSEC_MISSING, 523 nb->line, nb->pos, "TH %s", man->meta.title); 524 } 525 526 /* TITLE MSEC ->DATE<- OS VOL */ 527 528 if (n != NULL) 529 n = n->next; 530 if (man->quick && n != NULL) 531 man->meta.date = mandoc_strdup(""); 532 else 533 man->meta.date = mandoc_normdate(n, nb); 534 535 /* TITLE MSEC DATE ->OS<- VOL */ 536 537 if (n && (n = n->next)) 538 man->meta.os = mandoc_strdup(n->string); 539 else if (man->os_s != NULL) 540 man->meta.os = mandoc_strdup(man->os_s); 541 if (man->meta.os_e == MANDOC_OS_OTHER && man->meta.os != NULL) { 542 if (strstr(man->meta.os, "OpenBSD") != NULL) 543 man->meta.os_e = MANDOC_OS_OPENBSD; 544 else if (strstr(man->meta.os, "NetBSD") != NULL) 545 man->meta.os_e = MANDOC_OS_NETBSD; 546 } 547 548 /* TITLE MSEC DATE OS ->VOL<- */ 549 /* If missing, use the default VOL name for MSEC. */ 550 551 if (n && (n = n->next)) 552 man->meta.vol = mandoc_strdup(n->string); 553 else if ('\0' != man->meta.msec[0] && 554 (NULL != (p = mandoc_a2msec(man->meta.msec)))) 555 man->meta.vol = mandoc_strdup(p); 556 557 if (n != NULL && (n = n->next) != NULL) 558 mandoc_msg(MANDOCERR_ARG_EXCESS, 559 n->line, n->pos, "TH ... %s", n->string); 560 561 /* 562 * Remove the `TH' node after we've processed it for our 563 * meta-data. 564 */ 565 roff_node_delete(man, man->last); 566 } 567 568 static void 569 post_UC(CHKARGS) 570 { 571 static const char * const bsd_versions[] = { 572 "3rd Berkeley Distribution", 573 "4th Berkeley Distribution", 574 "4.2 Berkeley Distribution", 575 "4.3 Berkeley Distribution", 576 "4.4 Berkeley Distribution", 577 }; 578 579 const char *p, *s; 580 581 n = n->child; 582 583 if (n == NULL || n->type != ROFFT_TEXT) 584 p = bsd_versions[0]; 585 else { 586 s = n->string; 587 if (0 == strcmp(s, "3")) 588 p = bsd_versions[0]; 589 else if (0 == strcmp(s, "4")) 590 p = bsd_versions[1]; 591 else if (0 == strcmp(s, "5")) 592 p = bsd_versions[2]; 593 else if (0 == strcmp(s, "6")) 594 p = bsd_versions[3]; 595 else if (0 == strcmp(s, "7")) 596 p = bsd_versions[4]; 597 else 598 p = bsd_versions[0]; 599 } 600 601 free(man->meta.os); 602 man->meta.os = mandoc_strdup(p); 603 } 604 605 static void 606 post_AT(CHKARGS) 607 { 608 static const char * const unix_versions[] = { 609 "7th Edition", 610 "System III", 611 "System V", 612 "System V Release 2", 613 }; 614 615 struct roff_node *nn; 616 const char *p, *s; 617 618 n = n->child; 619 620 if (n == NULL || n->type != ROFFT_TEXT) 621 p = unix_versions[0]; 622 else { 623 s = n->string; 624 if (0 == strcmp(s, "3")) 625 p = unix_versions[0]; 626 else if (0 == strcmp(s, "4")) 627 p = unix_versions[1]; 628 else if (0 == strcmp(s, "5")) { 629 nn = n->next; 630 if (nn != NULL && 631 nn->type == ROFFT_TEXT && 632 nn->string[0] != '\0') 633 p = unix_versions[3]; 634 else 635 p = unix_versions[2]; 636 } else 637 p = unix_versions[0]; 638 } 639 640 free(man->meta.os); 641 man->meta.os = mandoc_strdup(p); 642 } 643 644 static void 645 post_in(CHKARGS) 646 { 647 char *s; 648 649 if (n->parent->tok != MAN_TP || 650 n->parent->type != ROFFT_HEAD || 651 n->child == NULL || 652 *n->child->string == '+' || 653 *n->child->string == '-') 654 return; 655 mandoc_asprintf(&s, "+%s", n->child->string); 656 free(n->child->string); 657 n->child->string = s; 658 } 659