Annotation of mandoc/mdoc.c, Revision 1.260
1.260 ! schwarze 1: /* $Id: mdoc.c,v 1.259 2017/01/28 23:30:08 schwarze Exp $ */
1.1 kristaps 2: /*
1.182 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.259 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
1.75 kristaps 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: *
1.241 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.75 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.241 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.75 kristaps 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: */
1.114 kristaps 18: #include "config.h"
19:
1.106 kristaps 20: #include <sys/types.h>
21:
1.1 kristaps 22: #include <assert.h>
1.211 schwarze 23: #include <ctype.h>
1.1 kristaps 24: #include <stdarg.h>
1.73 kristaps 25: #include <stdio.h>
1.1 kristaps 26: #include <stdlib.h>
27: #include <string.h>
1.120 kristaps 28: #include <time.h>
1.1 kristaps 29:
1.239 schwarze 30: #include "mandoc_aux.h"
31: #include "mandoc.h"
32: #include "roff.h"
1.187 kristaps 33: #include "mdoc.h"
1.239 schwarze 34: #include "libmandoc.h"
1.247 schwarze 35: #include "roff_int.h"
1.70 kristaps 36: #include "libmdoc.h"
1.1 kristaps 37:
1.216 schwarze 38: const char *const __mdoc_macronames[MDOC_MAX + 1] = {
1.248 schwarze 39: "text",
1.82 kristaps 40: "Ap", "Dd", "Dt", "Os",
1.1 kristaps 41: "Sh", "Ss", "Pp", "D1",
42: "Dl", "Bd", "Ed", "Bl",
43: "El", "It", "Ad", "An",
44: "Ar", "Cd", "Cm", "Dv",
45: "Er", "Ev", "Ex", "Fa",
46: "Fd", "Fl", "Fn", "Ft",
47: "Ic", "In", "Li", "Nd",
48: "Nm", "Op", "Ot", "Pa",
49: "Rv", "St", "Va", "Vt",
1.114 kristaps 50: "Xr", "%A", "%B", "%D",
51: "%I", "%J", "%N", "%O",
52: "%P", "%R", "%T", "%V",
1.1 kristaps 53: "Ac", "Ao", "Aq", "At",
54: "Bc", "Bf", "Bo", "Bq",
55: "Bsx", "Bx", "Db", "Dc",
56: "Do", "Dq", "Ec", "Ef",
57: "Em", "Eo", "Fx", "Ms",
58: "No", "Ns", "Nx", "Ox",
59: "Pc", "Pf", "Po", "Pq",
60: "Qc", "Ql", "Qo", "Qq",
61: "Re", "Rs", "Sc", "So",
62: "Sq", "Sm", "Sx", "Sy",
63: "Tn", "Ux", "Xc", "Xo",
64: "Fo", "Fc", "Oo", "Oc",
65: "Bk", "Ek", "Bt", "Hf",
1.82 kristaps 66: "Fr", "Ud", "Lb", "Lp",
67: "Lk", "Mt", "Brq", "Bro",
1.114 kristaps 68: "Brc", "%C", "Es", "En",
69: "Dx", "%Q", "br", "sp",
1.248 schwarze 70: "%U", "Ta", "ll",
71: };
1.1 kristaps 72:
1.213 schwarze 73: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 74: "split", "nosplit", "ragged",
1.213 schwarze 75: "unfilled", "literal", "file",
76: "offset", "bullet", "dash",
77: "hyphen", "item", "enum",
78: "tag", "diag", "hang",
79: "ohang", "inset", "column",
80: "width", "compact", "std",
1.52 kristaps 81: "filled", "words", "emphasis",
1.108 kristaps 82: "symbolic", "nested", "centered"
1.1 kristaps 83: };
84:
1.248 schwarze 85: const char * const *mdoc_macronames = __mdoc_macronames + 1;
1.1 kristaps 86: const char * const *mdoc_argnames = __mdoc_argnames;
87:
1.242 schwarze 88: static int mdoc_ptext(struct roff_man *, int, char *, int);
89: static int mdoc_pmacro(struct roff_man *, int, char *, int);
1.88 kristaps 90:
1.181 kristaps 91:
1.50 kristaps 92: /*
1.53 kristaps 93: * Main parse routine. Parses a single line -- really just hands off to
1.123 kristaps 94: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.50 kristaps 95: */
1.20 kristaps 96: int
1.242 schwarze 97: mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 98: {
99:
1.239 schwarze 100: if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
1.228 schwarze 101: mdoc->flags |= MDOC_NEWLINE;
1.153 schwarze 102:
103: /*
104: * Let the roff nS register switch SYNOPSIS mode early,
105: * such that the parser knows at all times
106: * whether this mode is on or off.
107: * Note that this mode is also switched by the Sh macro.
108: */
1.204 schwarze 109: if (roff_getreg(mdoc->roff, "nS"))
110: mdoc->flags |= MDOC_SYNOPSIS;
111: else
112: mdoc->flags &= ~MDOC_SYNOPSIS;
1.153 schwarze 113:
1.253 schwarze 114: return roff_getcontrol(mdoc->roff, buf, &offs) ?
1.213 schwarze 115: mdoc_pmacro(mdoc, ln, buf, offs) :
1.253 schwarze 116: mdoc_ptext(mdoc, ln, buf, offs);
1.1 kristaps 117: }
118:
1.232 schwarze 119: void
1.148 kristaps 120: mdoc_macro(MACRO_PROT_ARGS)
1.88 kristaps 121: {
1.248 schwarze 122: assert(tok > TOKEN_NONE && tok < MDOC_MAX);
1.122 kristaps 123:
1.232 schwarze 124: (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
1.73 kristaps 125: }
126:
1.231 schwarze 127: void
1.242 schwarze 128: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, int tok)
1.17 kristaps 129: {
1.240 schwarze 130: struct roff_node *p;
1.17 kristaps 131:
1.247 schwarze 132: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
133: roff_node_append(mdoc, p);
1.242 schwarze 134: mdoc->next = ROFF_NEXT_CHILD;
1.152 schwarze 135: }
136:
1.240 schwarze 137: struct roff_node *
1.242 schwarze 138: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos, int tok,
1.260 ! schwarze 139: struct roff_node *body)
1.152 schwarze 140: {
1.240 schwarze 141: struct roff_node *p;
1.152 schwarze 142:
1.258 schwarze 143: body->flags |= NODE_ENDED;
144: body->parent->flags |= NODE_ENDED;
1.247 schwarze 145: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.237 schwarze 146: p->body = body;
1.202 schwarze 147: p->norm = body->norm;
1.260 ! schwarze 148: p->end = ENDBODY_SPACE;
1.247 schwarze 149: roff_node_append(mdoc, p);
1.242 schwarze 150: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 151: return p;
1.1 kristaps 152: }
153:
1.240 schwarze 154: struct roff_node *
1.242 schwarze 155: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.240 schwarze 156: int tok, struct mdoc_arg *args)
1.1 kristaps 157: {
1.240 schwarze 158: struct roff_node *p;
1.1 kristaps 159:
1.247 schwarze 160: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.77 kristaps 161: p->args = args;
162: if (p->args)
1.53 kristaps 163: (args->refcnt)++;
1.172 kristaps 164:
165: switch (tok) {
1.213 schwarze 166: case MDOC_Bd:
167: case MDOC_Bf:
168: case MDOC_Bl:
1.217 schwarze 169: case MDOC_En:
1.213 schwarze 170: case MDOC_Rs:
1.172 kristaps 171: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
172: break;
173: default:
174: break;
175: }
1.247 schwarze 176: roff_node_append(mdoc, p);
1.242 schwarze 177: mdoc->next = ROFF_NEXT_CHILD;
1.253 schwarze 178: return p;
1.1 kristaps 179: }
180:
1.231 schwarze 181: void
1.242 schwarze 182: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.240 schwarze 183: int tok, struct mdoc_arg *args)
1.1 kristaps 184: {
1.240 schwarze 185: struct roff_node *p;
1.1 kristaps 186:
1.247 schwarze 187: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.77 kristaps 188: p->args = args;
189: if (p->args)
1.53 kristaps 190: (args->refcnt)++;
1.172 kristaps 191:
192: switch (tok) {
1.213 schwarze 193: case MDOC_An:
1.172 kristaps 194: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
195: break;
196: default:
197: break;
198: }
1.247 schwarze 199: roff_node_append(mdoc, p);
1.242 schwarze 200: mdoc->next = ROFF_NEXT_CHILD;
1.175 kristaps 201: }
1.1 kristaps 202:
1.231 schwarze 203: void
1.242 schwarze 204: mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
1.201 schwarze 205: {
206:
1.247 schwarze 207: roff_node_unlink(mdoc, p);
1.256 schwarze 208: p->prev = p->next = NULL;
1.247 schwarze 209: roff_node_append(mdoc, p);
1.1 kristaps 210: }
211:
1.53 kristaps 212: /*
213: * Parse free-form text, that is, a line that does not begin with the
214: * control character.
215: */
216: static int
1.242 schwarze 217: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 218: {
1.240 schwarze 219: struct roff_node *n;
1.142 kristaps 220: char *c, *ws, *end;
221:
1.203 schwarze 222: n = mdoc->last;
1.142 kristaps 223:
224: /*
1.257 schwarze 225: * If a column list contains plain text, assume an implicit item
226: * macro. This can happen one or more times at the beginning
227: * of such a list, intermixed with non-It mdoc macros and with
228: * nodes generated on the roff level, for example by tbl.
1.142 kristaps 229: */
230:
1.257 schwarze 231: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
232: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
233: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
234: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 235: mdoc->flags |= MDOC_FREECOL;
1.232 schwarze 236: mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
1.253 schwarze 237: return 1;
1.142 kristaps 238: }
1.124 kristaps 239:
1.137 schwarze 240: /*
241: * Search for the beginning of unescaped trailing whitespace (ws)
242: * and for the first character not to be output (end).
243: */
1.139 kristaps 244:
245: /* FIXME: replace with strcspn(). */
1.137 schwarze 246: ws = NULL;
247: for (c = end = buf + offs; *c; c++) {
248: switch (*c) {
249: case ' ':
250: if (NULL == ws)
251: ws = c;
252: continue;
253: case '\t':
254: /*
255: * Always warn about trailing tabs,
256: * even outside literal context,
257: * where they should be put on the next line.
258: */
259: if (NULL == ws)
260: ws = c;
261: /*
262: * Strip trailing tabs in literal context only;
263: * outside, they affect the next line.
264: */
1.203 schwarze 265: if (MDOC_LITERAL & mdoc->flags)
1.137 schwarze 266: continue;
267: break;
268: case '\\':
269: /* Skip the escaped character, too, if any. */
270: if (c[1])
271: c++;
272: /* FALLTHROUGH */
273: default:
274: ws = NULL;
275: break;
276: }
277: end = c + 1;
278: }
279: *end = '\0';
1.91 kristaps 280:
1.137 schwarze 281: if (ws)
1.218 schwarze 282: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
283: line, (int)(ws-buf), NULL);
1.115 kristaps 284:
1.231 schwarze 285: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.218 schwarze 286: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
287: line, (int)(c - buf), NULL);
1.124 kristaps 288:
1.119 kristaps 289: /*
1.165 schwarze 290: * Insert a `sp' in the case of a blank line. Technically,
1.124 kristaps 291: * blank lines aren't allowed, but enough manuals assume this
292: * behaviour that we want to work around it.
1.119 kristaps 293: */
1.250 schwarze 294: roff_elem_alloc(mdoc, line, offs, MDOC_sp);
1.258 schwarze 295: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.242 schwarze 296: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 297: return 1;
1.119 kristaps 298: }
1.68 kristaps 299:
1.249 schwarze 300: roff_word_alloc(mdoc, line, offs, buf+offs);
1.91 kristaps 301:
1.231 schwarze 302: if (mdoc->flags & MDOC_LITERAL)
1.253 schwarze 303: return 1;
1.128 kristaps 304:
305: /*
306: * End-of-sentence check. If the last character is an unescaped
307: * EOS character, then flag the node as being the end of a
308: * sentence. The front-end will know how to interpret this.
309: */
1.132 kristaps 310:
1.137 schwarze 311: assert(buf < end);
312:
1.207 schwarze 313: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.258 schwarze 314: mdoc->last->flags |= NODE_EOS;
1.259 schwarze 315:
316: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
317: if (c - buf < offs + 2)
318: continue;
319: if (end - c < 4)
320: break;
321: if (isalpha((unsigned char)c[-2]) &&
322: isalpha((unsigned char)c[-1]) &&
323: c[1] == ' ' &&
324: isupper((unsigned char)(c[2] == ' ' ? c[3] : c[2])) &&
325: (c[-2] != 'n' || c[-1] != 'c') &&
326: (c[-2] != 'v' || c[-1] != 's'))
327: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
328: line, (int)(c - buf), NULL);
329: }
330:
1.253 schwarze 331: return 1;
1.1 kristaps 332: }
333:
1.53 kristaps 334: /*
335: * Parse a macro line, that is, a line beginning with the control
336: * character.
337: */
1.155 kristaps 338: static int
1.242 schwarze 339: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 340: {
1.240 schwarze 341: struct roff_node *n;
1.229 schwarze 342: const char *cp;
1.240 schwarze 343: int tok;
1.188 kristaps 344: int i, sv;
1.144 kristaps 345: char mac[5];
1.63 kristaps 346:
1.188 kristaps 347: sv = offs;
1.130 kristaps 348:
1.213 schwarze 349: /*
1.162 schwarze 350: * Copy the first word into a nil-terminated buffer.
1.229 schwarze 351: * Stop when a space, tab, escape, or eoln is encountered.
1.160 kristaps 352: */
1.1 kristaps 353:
1.188 kristaps 354: i = 0;
1.229 schwarze 355: while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
1.188 kristaps 356: mac[i++] = buf[offs++];
357:
358: mac[i] = '\0';
359:
1.248 schwarze 360: tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : TOKEN_NONE;
1.1 kristaps 361:
1.248 schwarze 362: if (tok == TOKEN_NONE) {
1.222 schwarze 363: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
364: ln, sv, buf + sv - 1);
1.253 schwarze 365: return 1;
1.53 kristaps 366: }
1.1 kristaps 367:
1.229 schwarze 368: /* Skip a leading escape sequence or tab. */
1.160 kristaps 369:
1.229 schwarze 370: switch (buf[offs]) {
371: case '\\':
372: cp = buf + offs + 1;
373: mandoc_escape(&cp, NULL, NULL);
374: offs = cp - buf;
375: break;
376: case '\t':
1.188 kristaps 377: offs++;
1.229 schwarze 378: break;
379: default:
380: break;
381: }
1.160 kristaps 382:
383: /* Jump to the next non-whitespace word. */
1.1 kristaps 384:
1.188 kristaps 385: while (buf[offs] && ' ' == buf[offs])
386: offs++;
1.1 kristaps 387:
1.213 schwarze 388: /*
1.125 kristaps 389: * Trailing whitespace. Note that tabs are allowed to be passed
390: * into the parser as "text", so we only warn about spaces here.
391: */
1.115 kristaps 392:
1.188 kristaps 393: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.218 schwarze 394: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
395: ln, offs - 1, NULL);
1.115 kristaps 396:
1.144 kristaps 397: /*
398: * If an initial macro or a list invocation, divert directly
399: * into macro processing.
400: */
401:
1.257 schwarze 402: n = mdoc->last;
403: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.232 schwarze 404: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.253 schwarze 405: return 1;
1.232 schwarze 406: }
1.144 kristaps 407:
408: /*
1.257 schwarze 409: * If a column list contains a non-It macro, assume an implicit
410: * item macro. This can happen one or more times at the
411: * beginning of such a list, intermixed with text lines and
412: * with nodes generated on the roff level, for example by tbl.
1.98 kristaps 413: */
1.144 kristaps 414:
1.257 schwarze 415: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
416: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
417: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
418: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 419: mdoc->flags |= MDOC_FREECOL;
1.232 schwarze 420: mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
1.253 schwarze 421: return 1;
1.144 kristaps 422: }
423:
424: /* Normal processing of a macro. */
425:
1.232 schwarze 426: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.208 schwarze 427:
428: /* In quick mode (for mandocdb), abort after the NAME section. */
429:
430: if (mdoc->quick && MDOC_Sh == tok &&
431: SEC_NAME != mdoc->last->sec)
1.253 schwarze 432: return 2;
1.1 kristaps 433:
1.253 schwarze 434: return 1;
1.1 kristaps 435: }
1.100 kristaps 436:
1.186 kristaps 437: enum mdelim
438: mdoc_isdelim(const char *p)
439: {
440:
441: if ('\0' == p[0])
1.253 schwarze 442: return DELIM_NONE;
1.186 kristaps 443:
444: if ('\0' == p[1])
445: switch (p[0]) {
1.213 schwarze 446: case '(':
447: case '[':
1.253 schwarze 448: return DELIM_OPEN;
1.213 schwarze 449: case '|':
1.253 schwarze 450: return DELIM_MIDDLE;
1.213 schwarze 451: case '.':
452: case ',':
453: case ';':
454: case ':':
455: case '?':
456: case '!':
457: case ')':
458: case ']':
1.253 schwarze 459: return DELIM_CLOSE;
1.186 kristaps 460: default:
1.253 schwarze 461: return DELIM_NONE;
1.186 kristaps 462: }
463:
464: if ('\\' != p[0])
1.253 schwarze 465: return DELIM_NONE;
1.100 kristaps 466:
1.186 kristaps 467: if (0 == strcmp(p + 1, "."))
1.253 schwarze 468: return DELIM_CLOSE;
1.200 schwarze 469: if (0 == strcmp(p + 1, "fR|\\fP"))
1.253 schwarze 470: return DELIM_MIDDLE;
1.186 kristaps 471:
1.253 schwarze 472: return DELIM_NONE;
1.255 schwarze 473: }
474:
475: void
476: mdoc_validate(struct roff_man *mdoc)
477: {
478:
479: mdoc->last = mdoc->first;
480: mdoc_node_validate(mdoc);
481: mdoc_state_reset(mdoc);
1.186 kristaps 482: }
CVSweb