Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.156
1.156 ! schwarze 1: /* $OpenBSD: mdoc.c,v 1.155 2017/06/07 20:58:36 schwarze Exp $ */
1.1 kristaps 2: /*
1.77 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.148 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
1.3 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: *
1.130 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.3 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.130 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.3 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: */
1.28 schwarze 18: #include <sys/types.h>
19:
1.1 kristaps 20: #include <assert.h>
1.101 schwarze 21: #include <ctype.h>
1.1 kristaps 22: #include <stdarg.h>
23: #include <stdio.h>
24: #include <stdlib.h>
25: #include <string.h>
1.39 schwarze 26: #include <time.h>
1.1 kristaps 27:
1.128 schwarze 28: #include "mandoc_aux.h"
29: #include "mandoc.h"
30: #include "roff.h"
1.83 schwarze 31: #include "mdoc.h"
1.128 schwarze 32: #include "libmandoc.h"
1.136 schwarze 33: #include "roff_int.h"
1.1 kristaps 34: #include "libmdoc.h"
35:
1.103 schwarze 36: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 37: "split", "nosplit", "ragged",
1.103 schwarze 38: "unfilled", "literal", "file",
39: "offset", "bullet", "dash",
40: "hyphen", "item", "enum",
41: "tag", "diag", "hang",
42: "ohang", "inset", "column",
43: "width", "compact", "std",
1.1 kristaps 44: "filled", "words", "emphasis",
1.30 schwarze 45: "symbolic", "nested", "centered"
1.151 schwarze 46: };
1.1 kristaps 47: const char * const *mdoc_argnames = __mdoc_argnames;
48:
1.131 schwarze 49: static int mdoc_ptext(struct roff_man *, int, char *, int);
50: static int mdoc_pmacro(struct roff_man *, int, char *, int);
1.1 kristaps 51:
1.81 schwarze 52:
1.1 kristaps 53: /*
54: * Main parse routine. Parses a single line -- really just hands off to
1.45 schwarze 55: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.1 kristaps 56: */
57: int
1.131 schwarze 58: mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 59: {
60:
1.128 schwarze 61: if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
1.117 schwarze 62: mdoc->flags |= MDOC_NEWLINE;
1.60 schwarze 63:
64: /*
65: * Let the roff nS register switch SYNOPSIS mode early,
66: * such that the parser knows at all times
67: * whether this mode is on or off.
68: * Note that this mode is also switched by the Sh macro.
69: */
1.94 schwarze 70: if (roff_getreg(mdoc->roff, "nS"))
71: mdoc->flags |= MDOC_SYNOPSIS;
72: else
73: mdoc->flags &= ~MDOC_SYNOPSIS;
1.60 schwarze 74:
1.142 schwarze 75: return roff_getcontrol(mdoc->roff, buf, &offs) ?
1.103 schwarze 76: mdoc_pmacro(mdoc, ln, buf, offs) :
1.142 schwarze 77: mdoc_ptext(mdoc, ln, buf, offs);
1.1 kristaps 78: }
79:
1.121 schwarze 80: void
1.58 schwarze 81: mdoc_macro(MACRO_PROT_ARGS)
1.1 kristaps 82: {
1.151 schwarze 83: assert(tok >= MDOC_Dd && tok < MDOC_MAX);
1.121 schwarze 84: (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
1.1 kristaps 85: }
86:
1.120 schwarze 87: void
1.151 schwarze 88: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
1.1 kristaps 89: {
1.129 schwarze 90: struct roff_node *p;
1.1 kristaps 91:
1.136 schwarze 92: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
93: roff_node_append(mdoc, p);
1.131 schwarze 94: mdoc->next = ROFF_NEXT_CHILD;
1.59 schwarze 95: }
96:
1.129 schwarze 97: struct roff_node *
1.151 schwarze 98: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
99: enum roff_tok tok, struct roff_node *body)
1.59 schwarze 100: {
1.129 schwarze 101: struct roff_node *p;
1.59 schwarze 102:
1.147 schwarze 103: body->flags |= NODE_ENDED;
104: body->parent->flags |= NODE_ENDED;
1.136 schwarze 105: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.126 schwarze 106: p->body = body;
1.92 schwarze 107: p->norm = body->norm;
1.149 schwarze 108: p->end = ENDBODY_SPACE;
1.136 schwarze 109: roff_node_append(mdoc, p);
1.131 schwarze 110: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 111: return p;
1.1 kristaps 112: }
113:
1.129 schwarze 114: struct roff_node *
1.131 schwarze 115: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.151 schwarze 116: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 117: {
1.129 schwarze 118: struct roff_node *p;
1.1 kristaps 119:
1.136 schwarze 120: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.4 schwarze 121: p->args = args;
122: if (p->args)
1.1 kristaps 123: (args->refcnt)++;
1.75 schwarze 124:
125: switch (tok) {
1.103 schwarze 126: case MDOC_Bd:
127: case MDOC_Bf:
128: case MDOC_Bl:
1.107 schwarze 129: case MDOC_En:
1.103 schwarze 130: case MDOC_Rs:
1.75 schwarze 131: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
132: break;
133: default:
134: break;
135: }
1.136 schwarze 136: roff_node_append(mdoc, p);
1.131 schwarze 137: mdoc->next = ROFF_NEXT_CHILD;
1.142 schwarze 138: return p;
1.1 kristaps 139: }
140:
1.120 schwarze 141: void
1.131 schwarze 142: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.151 schwarze 143: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 144: {
1.129 schwarze 145: struct roff_node *p;
1.1 kristaps 146:
1.136 schwarze 147: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.4 schwarze 148: p->args = args;
149: if (p->args)
1.1 kristaps 150: (args->refcnt)++;
1.75 schwarze 151:
152: switch (tok) {
1.103 schwarze 153: case MDOC_An:
1.75 schwarze 154: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
155: break;
156: default:
157: break;
158: }
1.136 schwarze 159: roff_node_append(mdoc, p);
1.131 schwarze 160: mdoc->next = ROFF_NEXT_CHILD;
1.1 kristaps 161: }
162:
1.120 schwarze 163: void
1.131 schwarze 164: mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
1.91 schwarze 165: {
166:
1.136 schwarze 167: roff_node_unlink(mdoc, p);
1.145 schwarze 168: p->prev = p->next = NULL;
1.136 schwarze 169: roff_node_append(mdoc, p);
1.1 kristaps 170: }
171:
172: /*
173: * Parse free-form text, that is, a line that does not begin with the
174: * control character.
175: */
176: static int
1.131 schwarze 177: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 178: {
1.129 schwarze 179: struct roff_node *n;
1.156 ! schwarze 180: const char *cp, *sp;
1.56 schwarze 181: char *c, *ws, *end;
1.46 schwarze 182:
1.93 schwarze 183: n = mdoc->last;
1.56 schwarze 184:
185: /*
1.146 schwarze 186: * If a column list contains plain text, assume an implicit item
187: * macro. This can happen one or more times at the beginning
188: * of such a list, intermixed with non-It mdoc macros and with
189: * nodes generated on the roff level, for example by tbl.
1.56 schwarze 190: */
191:
1.146 schwarze 192: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
193: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
194: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
195: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 196: mdoc->flags |= MDOC_FREECOL;
1.121 schwarze 197: mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
1.142 schwarze 198: return 1;
1.56 schwarze 199: }
200:
1.52 schwarze 201: /*
202: * Search for the beginning of unescaped trailing whitespace (ws)
203: * and for the first character not to be output (end).
204: */
1.56 schwarze 205:
206: /* FIXME: replace with strcspn(). */
1.52 schwarze 207: ws = NULL;
1.53 schwarze 208: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 209: switch (*c) {
210: case ' ':
211: if (NULL == ws)
212: ws = c;
213: continue;
214: case '\t':
215: /*
216: * Always warn about trailing tabs,
217: * even outside literal context,
218: * where they should be put on the next line.
219: */
220: if (NULL == ws)
221: ws = c;
222: /*
223: * Strip trailing tabs in literal context only;
224: * outside, they affect the next line.
225: */
1.93 schwarze 226: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 227: continue;
228: break;
229: case '\\':
230: /* Skip the escaped character, too, if any. */
231: if (c[1])
232: c++;
233: /* FALLTHROUGH */
234: default:
235: ws = NULL;
236: break;
237: }
238: end = c + 1;
239: }
240: *end = '\0';
1.19 schwarze 241:
1.52 schwarze 242: if (ws)
1.108 schwarze 243: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
244: line, (int)(ws-buf), NULL);
1.34 schwarze 245:
1.156 ! schwarze 246: /*
! 247: * Blank lines are allowed in no-fill mode
! 248: * and cancel preceding \c,
! 249: * but add a single vertical space elsewhere.
! 250: */
! 251:
1.120 schwarze 252: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.156 ! schwarze 253: switch (mdoc->last->type) {
! 254: case ROFFT_TEXT:
! 255: sp = mdoc->last->string;
! 256: cp = end = strchr(sp, '\0') - 2;
! 257: if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
! 258: break;
! 259: while (cp > sp && cp[-1] == '\\')
! 260: cp--;
! 261: if ((end - cp) % 2)
! 262: break;
! 263: *end = '\0';
! 264: return 1;
! 265: default:
! 266: break;
! 267: }
1.108 schwarze 268: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
269: line, (int)(c - buf), NULL);
1.153 schwarze 270: roff_elem_alloc(mdoc, line, offs, ROFF_sp);
1.147 schwarze 271: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.131 schwarze 272: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 273: return 1;
1.38 schwarze 274: }
1.1 kristaps 275:
1.138 schwarze 276: roff_word_alloc(mdoc, line, offs, buf+offs);
1.1 kristaps 277:
1.120 schwarze 278: if (mdoc->flags & MDOC_LITERAL)
1.142 schwarze 279: return 1;
1.52 schwarze 280:
1.35 schwarze 281: /*
1.48 schwarze 282: * End-of-sentence check. If the last character is an unescaped
283: * EOS character, then flag the node as being the end of a
284: * sentence. The front-end will know how to interpret this.
1.35 schwarze 285: */
286:
1.52 schwarze 287: assert(buf < end);
1.48 schwarze 288:
1.97 schwarze 289: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.147 schwarze 290: mdoc->last->flags |= NODE_EOS;
1.148 schwarze 291:
292: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
293: if (c - buf < offs + 2)
294: continue;
1.154 schwarze 295: if (end - c < 3)
1.148 schwarze 296: break;
1.155 schwarze 297: if (c[1] != ' ' ||
298: isalpha((unsigned char)c[-2]) == 0 ||
299: isalpha((unsigned char)c[-1]) == 0 ||
300: (c[-2] == 'n' && c[-1] == 'c') ||
301: (c[-2] == 'v' && c[-1] == 's'))
302: continue;
303: c += 2;
304: if (*c == ' ')
305: c++;
306: if (*c == ' ')
307: c++;
308: if (isupper((unsigned char)(*c)))
1.148 schwarze 309: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
310: line, (int)(c - buf), NULL);
311: }
312:
1.142 schwarze 313: return 1;
1.1 kristaps 314: }
1.19 schwarze 315:
1.1 kristaps 316: /*
317: * Parse a macro line, that is, a line beginning with the control
318: * character.
319: */
1.61 schwarze 320: static int
1.131 schwarze 321: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 322: {
1.129 schwarze 323: struct roff_node *n;
1.118 schwarze 324: const char *cp;
1.152 schwarze 325: size_t sz;
1.151 schwarze 326: enum roff_tok tok;
1.152 schwarze 327: int sv;
328:
329: /* Determine the line macro. */
1.1 kristaps 330:
1.83 schwarze 331: sv = offs;
1.152 schwarze 332: tok = TOKEN_NONE;
333: for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
334: offs++;
335: if (sz == 2 || sz == 3)
336: tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
1.137 schwarze 337: if (tok == TOKEN_NONE) {
1.112 schwarze 338: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
339: ln, sv, buf + sv - 1);
1.142 schwarze 340: return 1;
1.1 kristaps 341: }
342:
1.118 schwarze 343: /* Skip a leading escape sequence or tab. */
1.63 schwarze 344:
1.118 schwarze 345: switch (buf[offs]) {
346: case '\\':
347: cp = buf + offs + 1;
348: mandoc_escape(&cp, NULL, NULL);
349: offs = cp - buf;
350: break;
351: case '\t':
1.83 schwarze 352: offs++;
1.118 schwarze 353: break;
354: default:
355: break;
356: }
1.63 schwarze 357:
358: /* Jump to the next non-whitespace word. */
1.1 kristaps 359:
1.150 schwarze 360: while (buf[offs] == ' ')
1.83 schwarze 361: offs++;
1.34 schwarze 362:
1.103 schwarze 363: /*
1.46 schwarze 364: * Trailing whitespace. Note that tabs are allowed to be passed
365: * into the parser as "text", so we only warn about spaces here.
366: */
1.34 schwarze 367:
1.83 schwarze 368: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.108 schwarze 369: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
370: ln, offs - 1, NULL);
1.1 kristaps 371:
1.56 schwarze 372: /*
373: * If an initial macro or a list invocation, divert directly
374: * into macro processing.
375: */
376:
1.146 schwarze 377: n = mdoc->last;
378: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.121 schwarze 379: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.142 schwarze 380: return 1;
1.121 schwarze 381: }
1.56 schwarze 382:
383: /*
1.146 schwarze 384: * If a column list contains a non-It macro, assume an implicit
385: * item macro. This can happen one or more times at the
386: * beginning of such a list, intermixed with text lines and
387: * with nodes generated on the roff level, for example by tbl.
1.24 schwarze 388: */
1.56 schwarze 389:
1.146 schwarze 390: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
391: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
392: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
393: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 394: mdoc->flags |= MDOC_FREECOL;
1.121 schwarze 395: mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
1.142 schwarze 396: return 1;
1.56 schwarze 397: }
398:
399: /* Normal processing of a macro. */
400:
1.121 schwarze 401: mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
1.98 schwarze 402:
403: /* In quick mode (for mandocdb), abort after the NAME section. */
404:
405: if (mdoc->quick && MDOC_Sh == tok &&
406: SEC_NAME != mdoc->last->sec)
1.142 schwarze 407: return 2;
1.1 kristaps 408:
1.142 schwarze 409: return 1;
1.1 kristaps 410: }
1.24 schwarze 411:
1.83 schwarze 412: enum mdelim
413: mdoc_isdelim(const char *p)
414: {
415:
416: if ('\0' == p[0])
1.142 schwarze 417: return DELIM_NONE;
1.83 schwarze 418:
419: if ('\0' == p[1])
420: switch (p[0]) {
1.103 schwarze 421: case '(':
422: case '[':
1.142 schwarze 423: return DELIM_OPEN;
1.103 schwarze 424: case '|':
1.142 schwarze 425: return DELIM_MIDDLE;
1.103 schwarze 426: case '.':
427: case ',':
428: case ';':
429: case ':':
430: case '?':
431: case '!':
432: case ')':
433: case ']':
1.142 schwarze 434: return DELIM_CLOSE;
1.83 schwarze 435: default:
1.142 schwarze 436: return DELIM_NONE;
1.83 schwarze 437: }
438:
439: if ('\\' != p[0])
1.142 schwarze 440: return DELIM_NONE;
1.24 schwarze 441:
1.83 schwarze 442: if (0 == strcmp(p + 1, "."))
1.142 schwarze 443: return DELIM_CLOSE;
1.90 schwarze 444: if (0 == strcmp(p + 1, "fR|\\fP"))
1.142 schwarze 445: return DELIM_MIDDLE;
1.83 schwarze 446:
1.142 schwarze 447: return DELIM_NONE;
1.144 schwarze 448: }
449:
450: void
451: mdoc_validate(struct roff_man *mdoc)
452: {
453:
454: mdoc->last = mdoc->first;
455: mdoc_node_validate(mdoc);
456: mdoc_state_reset(mdoc);
1.83 schwarze 457: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mdoc.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc