Annotation of src/usr.bin/mandoc/man_validate.c, Revision 1.98
1.98 ! schwarze 1: /* $OpenBSD: man_validate.c,v 1.97 2017/05/05 13:17:04 schwarze Exp $ */
1.1 kristaps 2: /*
1.38 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.94 schwarze 4: * Copyright (c) 2010, 2012-2017 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.85 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.2 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.85 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.2 schwarze 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.
1.1 kristaps 17: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <ctype.h>
1.6 schwarze 22: #include <errno.h>
23: #include <limits.h>
1.1 kristaps 24: #include <stdarg.h>
25: #include <stdlib.h>
1.28 schwarze 26: #include <string.h>
1.36 schwarze 27: #include <time.h>
1.1 kristaps 28:
1.85 schwarze 29: #include "mandoc_aux.h"
30: #include "mandoc.h"
31: #include "roff.h"
1.44 schwarze 32: #include "man.h"
1.85 schwarze 33: #include "libmandoc.h"
1.89 schwarze 34: #include "roff_int.h"
1.1 kristaps 35: #include "libman.h"
36:
1.88 schwarze 37: #define CHKARGS struct roff_man *man, struct roff_node *n
1.1 kristaps 38:
1.78 schwarze 39: typedef void (*v_check)(CHKARGS);
1.1 kristaps 40:
1.78 schwarze 41: static void check_par(CHKARGS);
42: static void check_part(CHKARGS);
43: static void check_root(CHKARGS);
44: static void check_text(CHKARGS);
45:
46: static void post_AT(CHKARGS);
47: static void post_IP(CHKARGS);
48: static void post_vs(CHKARGS);
1.84 schwarze 49: static void post_OP(CHKARGS);
1.78 schwarze 50: static void post_TH(CHKARGS);
51: static void post_UC(CHKARGS);
52: static void post_UR(CHKARGS);
1.34 schwarze 53:
1.94 schwarze 54: static const v_check __man_valids[MAN_MAX - MAN_TH] = {
1.76 schwarze 55: post_TH, /* TH */
56: NULL, /* SH */
57: NULL, /* SS */
58: NULL, /* TP */
59: check_par, /* LP */
60: check_par, /* PP */
61: check_par, /* P */
62: post_IP, /* IP */
63: NULL, /* HP */
64: NULL, /* SM */
65: NULL, /* SB */
66: NULL, /* BI */
67: NULL, /* IB */
68: NULL, /* BR */
69: NULL, /* RB */
70: NULL, /* R */
71: NULL, /* B */
72: NULL, /* I */
73: NULL, /* IR */
74: NULL, /* RI */
1.92 schwarze 75: NULL, /* nf */
76: NULL, /* fi */
1.76 schwarze 77: NULL, /* RE */
78: check_part, /* RS */
79: NULL, /* DT */
80: post_UC, /* UC */
1.83 schwarze 81: NULL, /* PD */
1.76 schwarze 82: post_AT, /* AT */
83: NULL, /* in */
1.84 schwarze 84: post_OP, /* OP */
1.92 schwarze 85: NULL, /* EX */
86: NULL, /* EE */
1.76 schwarze 87: post_UR, /* UR */
88: NULL, /* UE */
1.1 kristaps 89: };
1.94 schwarze 90: static const v_check *man_valids = __man_valids - MAN_TH;
1.1 kristaps 91:
92:
1.78 schwarze 93: void
1.92 schwarze 94: man_node_validate(struct roff_man *man)
1.1 kristaps 95: {
1.86 schwarze 96: struct roff_node *n;
1.94 schwarze 97: const v_check *cp;
1.1 kristaps 98:
1.76 schwarze 99: n = man->last;
1.92 schwarze 100: man->last = man->last->child;
101: while (man->last != NULL) {
102: man_node_validate(man);
103: if (man->last == n)
104: man->last = man->last->child;
105: else
106: man->last = man->last->next;
107: }
1.1 kristaps 108:
1.92 schwarze 109: man->last = n;
110: man->next = ROFF_NEXT_SIBLING;
1.76 schwarze 111: switch (n->type) {
1.85 schwarze 112: case ROFFT_TEXT:
1.78 schwarze 113: check_text(man, n);
114: break;
1.85 schwarze 115: case ROFFT_ROOT:
1.78 schwarze 116: check_root(man, n);
117: break;
1.85 schwarze 118: case ROFFT_EQN:
119: case ROFFT_TBL:
1.78 schwarze 120: break;
1.1 kristaps 121: default:
1.95 schwarze 122: if (n->tok < ROFF_MAX) {
123: switch (n->tok) {
124: case ROFF_br:
1.98 ! schwarze 125: case ROFF_sp:
1.95 schwarze 126: post_vs(man, n);
127: break;
128: default:
1.96 schwarze 129: roff_validate(man);
130: break;
1.95 schwarze 131: }
132: break;
133: }
134: assert(n->tok >= MAN_TH && n->tok < MAN_MAX);
1.76 schwarze 135: cp = man_valids + n->tok;
1.78 schwarze 136: if (*cp)
137: (*cp)(man, n);
1.92 schwarze 138: if (man->last == n)
139: man_state(man, n);
1.78 schwarze 140: break;
1.1 kristaps 141: }
142: }
143:
1.78 schwarze 144: static void
1.63 schwarze 145: check_root(CHKARGS)
1.4 schwarze 146: {
1.7 schwarze 147:
1.73 schwarze 148: assert((man->flags & (MAN_BLINE | MAN_ELINE)) == 0);
1.7 schwarze 149:
1.65 schwarze 150: if (NULL == man->first->child)
1.75 schwarze 151: mandoc_msg(MANDOCERR_DOC_EMPTY, man->parse,
152: n->line, n->pos, NULL);
1.65 schwarze 153: else
154: man->meta.hasbody = 1;
155:
156: if (NULL == man->meta.title) {
1.77 schwarze 157: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
1.75 schwarze 158: n->line, n->pos, NULL);
1.35 schwarze 159:
1.18 schwarze 160: /*
161: * If a title hasn't been set, do so now (by
162: * implication, date and section also aren't set).
163: */
1.35 schwarze 164:
1.77 schwarze 165: man->meta.title = mandoc_strdup("");
166: man->meta.msec = mandoc_strdup("");
1.60 schwarze 167: man->meta.date = man->quick ? mandoc_strdup("") :
168: mandoc_normdate(man->parse, NULL, n->line, n->pos);
1.17 schwarze 169: }
1.15 schwarze 170: }
171:
1.78 schwarze 172: static void
1.47 schwarze 173: check_text(CHKARGS)
1.4 schwarze 174: {
1.47 schwarze 175: char *cp, *p;
176:
1.57 schwarze 177: if (MAN_LITERAL & man->flags)
1.78 schwarze 178: return;
1.48 schwarze 179:
180: cp = n->string;
181: for (p = cp; NULL != (p = strchr(p, '\t')); p++)
1.71 schwarze 182: mandoc_msg(MANDOCERR_FI_TAB, man->parse,
183: n->line, n->pos + (p - cp), NULL);
1.1 kristaps 184: }
185:
1.84 schwarze 186: static void
187: post_OP(CHKARGS)
188: {
189:
1.93 schwarze 190: if (n->child == NULL)
1.84 schwarze 191: mandoc_msg(MANDOCERR_OP_EMPTY, man->parse,
192: n->line, n->pos, "OP");
1.93 schwarze 193: else if (n->child->next != NULL && n->child->next->next != NULL) {
1.84 schwarze 194: n = n->child->next->next;
195: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
196: n->line, n->pos, "OP ... %s", n->string);
197: }
1.1 kristaps 198: }
199:
1.78 schwarze 200: static void
1.76 schwarze 201: post_UR(CHKARGS)
1.58 schwarze 202: {
203:
1.85 schwarze 204: if (n->type == ROFFT_HEAD && n->child == NULL)
1.84 schwarze 205: mandoc_vmsg(MANDOCERR_UR_NOHEAD, man->parse,
206: n->line, n->pos, "UR");
1.78 schwarze 207: check_part(man, n);
1.32 schwarze 208: }
1.7 schwarze 209:
1.78 schwarze 210: static void
1.8 schwarze 211: check_part(CHKARGS)
212: {
213:
1.85 schwarze 214: if (n->type == ROFFT_BODY && n->child == NULL)
1.81 schwarze 215: mandoc_msg(MANDOCERR_BLK_EMPTY, man->parse,
1.94 schwarze 216: n->line, n->pos, roff_name[n->tok]);
1.8 schwarze 217: }
218:
1.78 schwarze 219: static void
1.36 schwarze 220: check_par(CHKARGS)
221: {
222:
1.39 schwarze 223: switch (n->type) {
1.85 schwarze 224: case ROFFT_BLOCK:
1.93 schwarze 225: if (n->body->child == NULL)
1.89 schwarze 226: roff_node_delete(man, n);
1.39 schwarze 227: break;
1.85 schwarze 228: case ROFFT_BODY:
1.93 schwarze 229: if (n->child == NULL)
1.67 schwarze 230: mandoc_vmsg(MANDOCERR_PAR_SKIP,
231: man->parse, n->line, n->pos,
1.94 schwarze 232: "%s empty", roff_name[n->tok]);
1.39 schwarze 233: break;
1.85 schwarze 234: case ROFFT_HEAD:
1.93 schwarze 235: if (n->child != NULL)
1.69 schwarze 236: mandoc_vmsg(MANDOCERR_ARG_SKIP,
1.94 schwarze 237: man->parse, n->line, n->pos, "%s %s%s",
238: roff_name[n->tok], n->child->string,
1.93 schwarze 239: n->child->next != NULL ? " ..." : "");
1.39 schwarze 240: break;
241: default:
242: break;
243: }
1.36 schwarze 244: }
245:
1.78 schwarze 246: static void
1.55 schwarze 247: post_IP(CHKARGS)
248: {
249:
250: switch (n->type) {
1.85 schwarze 251: case ROFFT_BLOCK:
1.93 schwarze 252: if (n->head->child == NULL && n->body->child == NULL)
1.89 schwarze 253: roff_node_delete(man, n);
1.55 schwarze 254: break;
1.85 schwarze 255: case ROFFT_BODY:
1.93 schwarze 256: if (n->parent->head->child == NULL && n->child == NULL)
1.67 schwarze 257: mandoc_vmsg(MANDOCERR_PAR_SKIP,
258: man->parse, n->line, n->pos,
1.94 schwarze 259: "%s empty", roff_name[n->tok]);
1.55 schwarze 260: break;
261: default:
262: break;
263: }
264: }
1.36 schwarze 265:
1.78 schwarze 266: static void
1.34 schwarze 267: post_TH(CHKARGS)
268: {
1.86 schwarze 269: struct roff_node *nb;
1.40 schwarze 270: const char *p;
1.34 schwarze 271:
1.57 schwarze 272: free(man->meta.title);
273: free(man->meta.vol);
1.87 schwarze 274: free(man->meta.os);
1.57 schwarze 275: free(man->meta.msec);
276: free(man->meta.date);
1.34 schwarze 277:
1.57 schwarze 278: man->meta.title = man->meta.vol = man->meta.date =
1.87 schwarze 279: man->meta.msec = man->meta.os = NULL;
1.34 schwarze 280:
1.64 schwarze 281: nb = n;
282:
1.87 schwarze 283: /* ->TITLE<- MSEC DATE OS VOL */
1.34 schwarze 284:
285: n = n->child;
1.40 schwarze 286: if (n && n->string) {
287: for (p = n->string; '\0' != *p; p++) {
288: /* Only warn about this once... */
1.63 schwarze 289: if (isalpha((unsigned char)*p) &&
290: ! isupper((unsigned char)*p)) {
1.74 schwarze 291: mandoc_vmsg(MANDOCERR_TITLE_CASE,
1.66 schwarze 292: man->parse, n->line,
293: n->pos + (p - n->string),
1.74 schwarze 294: "TH %s", n->string);
1.40 schwarze 295: break;
296: }
297: }
1.57 schwarze 298: man->meta.title = mandoc_strdup(n->string);
1.77 schwarze 299: } else {
1.57 schwarze 300: man->meta.title = mandoc_strdup("");
1.77 schwarze 301: mandoc_msg(MANDOCERR_TH_NOTITLE, man->parse,
302: nb->line, nb->pos, "TH");
303: }
1.34 schwarze 304:
1.87 schwarze 305: /* TITLE ->MSEC<- DATE OS VOL */
1.34 schwarze 306:
1.40 schwarze 307: if (n)
308: n = n->next;
309: if (n && n->string)
1.57 schwarze 310: man->meta.msec = mandoc_strdup(n->string);
1.77 schwarze 311: else {
1.57 schwarze 312: man->meta.msec = mandoc_strdup("");
1.77 schwarze 313: mandoc_vmsg(MANDOCERR_MSEC_MISSING, man->parse,
314: nb->line, nb->pos, "TH %s", man->meta.title);
315: }
1.34 schwarze 316:
1.87 schwarze 317: /* TITLE MSEC ->DATE<- OS VOL */
1.34 schwarze 318:
1.40 schwarze 319: if (n)
320: n = n->next;
1.49 schwarze 321: if (n && n->string && '\0' != n->string[0]) {
1.60 schwarze 322: man->meta.date = man->quick ?
323: mandoc_strdup(n->string) :
324: mandoc_normdate(man->parse, n->string,
325: n->line, n->pos);
1.64 schwarze 326: } else {
1.57 schwarze 327: man->meta.date = mandoc_strdup("");
1.74 schwarze 328: mandoc_msg(MANDOCERR_DATE_MISSING, man->parse,
329: n ? n->line : nb->line,
330: n ? n->pos : nb->pos, "TH");
1.64 schwarze 331: }
1.34 schwarze 332:
1.87 schwarze 333: /* TITLE MSEC DATE ->OS<- VOL */
1.34 schwarze 334:
335: if (n && (n = n->next))
1.87 schwarze 336: man->meta.os = mandoc_strdup(n->string);
1.79 schwarze 337: else if (man->defos != NULL)
1.87 schwarze 338: man->meta.os = mandoc_strdup(man->defos);
1.34 schwarze 339:
1.87 schwarze 340: /* TITLE MSEC DATE OS ->VOL<- */
1.51 schwarze 341: /* If missing, use the default VOL name for MSEC. */
1.34 schwarze 342:
343: if (n && (n = n->next))
1.57 schwarze 344: man->meta.vol = mandoc_strdup(n->string);
345: else if ('\0' != man->meta.msec[0] &&
346: (NULL != (p = mandoc_a2msec(man->meta.msec))))
347: man->meta.vol = mandoc_strdup(p);
1.84 schwarze 348:
349: if (n != NULL && (n = n->next) != NULL)
350: mandoc_vmsg(MANDOCERR_ARG_EXCESS, man->parse,
351: n->line, n->pos, "TH ... %s", n->string);
1.34 schwarze 352:
353: /*
354: * Remove the `TH' node after we've processed it for our
355: * meta-data.
356: */
1.89 schwarze 357: roff_node_delete(man, man->last);
1.34 schwarze 358: }
359:
1.78 schwarze 360: static void
1.34 schwarze 361: post_UC(CHKARGS)
362: {
363: static const char * const bsd_versions[] = {
364: "3rd Berkeley Distribution",
365: "4th Berkeley Distribution",
366: "4.2 Berkeley Distribution",
367: "4.3 Berkeley Distribution",
368: "4.4 Berkeley Distribution",
369: };
370:
371: const char *p, *s;
372:
373: n = n->child;
374:
1.85 schwarze 375: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 376: p = bsd_versions[0];
377: else {
378: s = n->string;
379: if (0 == strcmp(s, "3"))
380: p = bsd_versions[0];
381: else if (0 == strcmp(s, "4"))
382: p = bsd_versions[1];
383: else if (0 == strcmp(s, "5"))
384: p = bsd_versions[2];
385: else if (0 == strcmp(s, "6"))
386: p = bsd_versions[3];
387: else if (0 == strcmp(s, "7"))
388: p = bsd_versions[4];
389: else
390: p = bsd_versions[0];
391: }
392:
1.87 schwarze 393: free(man->meta.os);
394: man->meta.os = mandoc_strdup(p);
1.34 schwarze 395: }
396:
1.78 schwarze 397: static void
1.34 schwarze 398: post_AT(CHKARGS)
399: {
400: static const char * const unix_versions[] = {
401: "7th Edition",
402: "System III",
403: "System V",
404: "System V Release 2",
405: };
406:
1.86 schwarze 407: struct roff_node *nn;
1.34 schwarze 408: const char *p, *s;
409:
410: n = n->child;
411:
1.85 schwarze 412: if (n == NULL || n->type != ROFFT_TEXT)
1.34 schwarze 413: p = unix_versions[0];
414: else {
415: s = n->string;
416: if (0 == strcmp(s, "3"))
417: p = unix_versions[0];
418: else if (0 == strcmp(s, "4"))
419: p = unix_versions[1];
420: else if (0 == strcmp(s, "5")) {
421: nn = n->next;
1.85 schwarze 422: if (nn != NULL &&
423: nn->type == ROFFT_TEXT &&
424: nn->string[0] != '\0')
1.34 schwarze 425: p = unix_versions[3];
426: else
427: p = unix_versions[2];
428: } else
429: p = unix_versions[0];
430: }
431:
1.87 schwarze 432: free(man->meta.os);
433: man->meta.os = mandoc_strdup(p);
1.46 schwarze 434: }
435:
1.78 schwarze 436: static void
1.46 schwarze 437: post_vs(CHKARGS)
438: {
439:
1.54 schwarze 440: if (NULL != n->prev)
1.78 schwarze 441: return;
1.54 schwarze 442:
443: switch (n->parent->tok) {
1.63 schwarze 444: case MAN_SH:
445: case MAN_SS:
1.98 ! schwarze 446: case MAN_PP:
! 447: case MAN_LP:
! 448: case MAN_P:
1.67 schwarze 449: mandoc_vmsg(MANDOCERR_PAR_SKIP, man->parse, n->line, n->pos,
1.94 schwarze 450: "%s after %s", roff_name[n->tok],
451: roff_name[n->parent->tok]);
1.54 schwarze 452: /* FALLTHROUGH */
1.90 schwarze 453: case TOKEN_NONE:
1.63 schwarze 454: /*
1.54 schwarze 455: * Don't warn about this because it occurs in pod2man
456: * and would cause considerable (unfixable) warnage.
457: */
1.89 schwarze 458: roff_node_delete(man, n);
1.54 schwarze 459: break;
460: default:
461: break;
462: }
1.34 schwarze 463: }