Annotation of mandoc/mdoc.c, Revision 1.269
1.269 ! schwarze 1: /* $Id: mdoc.c,v 1.268 2017/08/11 16:56:21 schwarze Exp $ */
1.1 kristaps 2: /*
1.182 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.269 ! schwarze 4: * Copyright (c) 2010, 2012-2018 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:
1.213 schwarze 38: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 39: "split", "nosplit", "ragged",
1.213 schwarze 40: "unfilled", "literal", "file",
41: "offset", "bullet", "dash",
42: "hyphen", "item", "enum",
43: "tag", "diag", "hang",
44: "ohang", "inset", "column",
45: "width", "compact", "std",
1.52 kristaps 46: "filled", "words", "emphasis",
1.108 kristaps 47: "symbolic", "nested", "centered"
1.262 schwarze 48: };
1.1 kristaps 49: const char * const *mdoc_argnames = __mdoc_argnames;
50:
1.242 schwarze 51: static int mdoc_ptext(struct roff_man *, int, char *, int);
52: static int mdoc_pmacro(struct roff_man *, int, char *, int);
1.88 kristaps 53:
1.181 kristaps 54:
1.50 kristaps 55: /*
1.53 kristaps 56: * Main parse routine. Parses a single line -- really just hands off to
1.123 kristaps 57: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.50 kristaps 58: */
1.20 kristaps 59: int
1.242 schwarze 60: mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 61: {
62:
1.239 schwarze 63: if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
1.228 schwarze 64: mdoc->flags |= MDOC_NEWLINE;
1.153 schwarze 65:
66: /*
67: * Let the roff nS register switch SYNOPSIS mode early,
68: * such that the parser knows at all times
69: * whether this mode is on or off.
70: * Note that this mode is also switched by the Sh macro.
71: */
1.204 schwarze 72: if (roff_getreg(mdoc->roff, "nS"))
73: mdoc->flags |= MDOC_SYNOPSIS;
74: else
75: mdoc->flags &= ~MDOC_SYNOPSIS;
1.153 schwarze 76:
1.253 schwarze 77: return roff_getcontrol(mdoc->roff, buf, &offs) ?
1.213 schwarze 78: mdoc_pmacro(mdoc, ln, buf, offs) :
1.253 schwarze 79: mdoc_ptext(mdoc, ln, buf, offs);
1.1 kristaps 80: }
81:
1.232 schwarze 82: void
1.262 schwarze 83: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
1.17 kristaps 84: {
1.240 schwarze 85: struct roff_node *p;
1.17 kristaps 86:
1.247 schwarze 87: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
88: roff_node_append(mdoc, p);
1.242 schwarze 89: mdoc->next = ROFF_NEXT_CHILD;
1.152 schwarze 90: }
91:
1.240 schwarze 92: struct roff_node *
1.262 schwarze 93: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
94: enum roff_tok tok, struct roff_node *body)
1.152 schwarze 95: {
1.240 schwarze 96: struct roff_node *p;
1.152 schwarze 97:
1.258 schwarze 98: body->flags |= NODE_ENDED;
99: body->parent->flags |= NODE_ENDED;
1.247 schwarze 100: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.237 schwarze 101: p->body = body;
1.202 schwarze 102: p->norm = body->norm;
1.260 schwarze 103: p->end = ENDBODY_SPACE;
1.247 schwarze 104: roff_node_append(mdoc, p);
1.242 schwarze 105: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 106: return p;
1.1 kristaps 107: }
108:
1.240 schwarze 109: struct roff_node *
1.242 schwarze 110: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.262 schwarze 111: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 112: {
1.240 schwarze 113: struct roff_node *p;
1.1 kristaps 114:
1.247 schwarze 115: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.77 kristaps 116: p->args = args;
117: if (p->args)
1.53 kristaps 118: (args->refcnt)++;
1.172 kristaps 119:
120: switch (tok) {
1.213 schwarze 121: case MDOC_Bd:
122: case MDOC_Bf:
123: case MDOC_Bl:
1.217 schwarze 124: case MDOC_En:
1.213 schwarze 125: case MDOC_Rs:
1.172 kristaps 126: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
127: break;
128: default:
129: break;
130: }
1.247 schwarze 131: roff_node_append(mdoc, p);
1.242 schwarze 132: mdoc->next = ROFF_NEXT_CHILD;
1.253 schwarze 133: return p;
1.1 kristaps 134: }
135:
1.231 schwarze 136: void
1.242 schwarze 137: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.262 schwarze 138: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 139: {
1.240 schwarze 140: struct roff_node *p;
1.1 kristaps 141:
1.247 schwarze 142: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.77 kristaps 143: p->args = args;
144: if (p->args)
1.53 kristaps 145: (args->refcnt)++;
1.172 kristaps 146:
147: switch (tok) {
1.213 schwarze 148: case MDOC_An:
1.172 kristaps 149: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
150: break;
151: default:
152: break;
153: }
1.247 schwarze 154: roff_node_append(mdoc, p);
1.242 schwarze 155: mdoc->next = ROFF_NEXT_CHILD;
1.175 kristaps 156: }
1.1 kristaps 157:
1.231 schwarze 158: void
1.242 schwarze 159: mdoc_node_relink(struct roff_man *mdoc, struct roff_node *p)
1.201 schwarze 160: {
161:
1.247 schwarze 162: roff_node_unlink(mdoc, p);
1.256 schwarze 163: p->prev = p->next = NULL;
1.247 schwarze 164: roff_node_append(mdoc, p);
1.1 kristaps 165: }
166:
1.53 kristaps 167: /*
168: * Parse free-form text, that is, a line that does not begin with the
169: * control character.
170: */
171: static int
1.242 schwarze 172: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 173: {
1.240 schwarze 174: struct roff_node *n;
1.267 schwarze 175: const char *cp, *sp;
1.142 kristaps 176: char *c, *ws, *end;
177:
1.203 schwarze 178: n = mdoc->last;
1.142 kristaps 179:
180: /*
1.257 schwarze 181: * If a column list contains plain text, assume an implicit item
182: * macro. This can happen one or more times at the beginning
183: * of such a list, intermixed with non-It mdoc macros and with
184: * nodes generated on the roff level, for example by tbl.
1.142 kristaps 185: */
186:
1.257 schwarze 187: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
188: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
189: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
190: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 191: mdoc->flags |= MDOC_FREECOL;
1.269 ! schwarze 192: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
! 193: line, offs, &offs, buf);
1.253 schwarze 194: return 1;
1.142 kristaps 195: }
1.124 kristaps 196:
1.137 schwarze 197: /*
198: * Search for the beginning of unescaped trailing whitespace (ws)
199: * and for the first character not to be output (end).
200: */
1.139 kristaps 201:
202: /* FIXME: replace with strcspn(). */
1.137 schwarze 203: ws = NULL;
204: for (c = end = buf + offs; *c; c++) {
205: switch (*c) {
206: case ' ':
207: if (NULL == ws)
208: ws = c;
209: continue;
210: case '\t':
211: /*
212: * Always warn about trailing tabs,
213: * even outside literal context,
214: * where they should be put on the next line.
215: */
216: if (NULL == ws)
217: ws = c;
218: /*
219: * Strip trailing tabs in literal context only;
220: * outside, they affect the next line.
221: */
1.203 schwarze 222: if (MDOC_LITERAL & mdoc->flags)
1.137 schwarze 223: continue;
224: break;
225: case '\\':
226: /* Skip the escaped character, too, if any. */
227: if (c[1])
228: c++;
229: /* FALLTHROUGH */
230: default:
231: ws = NULL;
232: break;
233: }
234: end = c + 1;
235: }
236: *end = '\0';
1.91 kristaps 237:
1.137 schwarze 238: if (ws)
1.218 schwarze 239: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
240: line, (int)(ws-buf), NULL);
1.115 kristaps 241:
1.267 schwarze 242: /*
243: * Blank lines are allowed in no-fill mode
244: * and cancel preceding \c,
245: * but add a single vertical space elsewhere.
246: */
247:
1.231 schwarze 248: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.267 schwarze 249: switch (mdoc->last->type) {
250: case ROFFT_TEXT:
251: sp = mdoc->last->string;
252: cp = end = strchr(sp, '\0') - 2;
253: if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
254: break;
255: while (cp > sp && cp[-1] == '\\')
256: cp--;
257: if ((end - cp) % 2)
258: break;
259: *end = '\0';
260: return 1;
261: default:
262: break;
263: }
1.218 schwarze 264: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
265: line, (int)(c - buf), NULL);
1.264 schwarze 266: roff_elem_alloc(mdoc, line, offs, ROFF_sp);
1.258 schwarze 267: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.242 schwarze 268: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 269: return 1;
1.119 kristaps 270: }
1.68 kristaps 271:
1.249 schwarze 272: roff_word_alloc(mdoc, line, offs, buf+offs);
1.91 kristaps 273:
1.231 schwarze 274: if (mdoc->flags & MDOC_LITERAL)
1.253 schwarze 275: return 1;
1.128 kristaps 276:
277: /*
278: * End-of-sentence check. If the last character is an unescaped
279: * EOS character, then flag the node as being the end of a
280: * sentence. The front-end will know how to interpret this.
281: */
1.132 kristaps 282:
1.137 schwarze 283: assert(buf < end);
284:
1.207 schwarze 285: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.258 schwarze 286: mdoc->last->flags |= NODE_EOS;
1.259 schwarze 287:
288: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
289: if (c - buf < offs + 2)
290: continue;
1.265 schwarze 291: if (end - c < 3)
1.259 schwarze 292: break;
1.266 schwarze 293: if (c[1] != ' ' ||
1.268 schwarze 294: isalnum((unsigned char)c[-2]) == 0 ||
295: isalnum((unsigned char)c[-1]) == 0 ||
1.266 schwarze 296: (c[-2] == 'n' && c[-1] == 'c') ||
297: (c[-2] == 'v' && c[-1] == 's'))
298: continue;
299: c += 2;
300: if (*c == ' ')
301: c++;
302: if (*c == ' ')
303: c++;
304: if (isupper((unsigned char)(*c)))
1.259 schwarze 305: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
306: line, (int)(c - buf), NULL);
307: }
308:
1.253 schwarze 309: return 1;
1.1 kristaps 310: }
311:
1.53 kristaps 312: /*
313: * Parse a macro line, that is, a line beginning with the control
314: * character.
315: */
1.155 kristaps 316: static int
1.242 schwarze 317: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 318: {
1.240 schwarze 319: struct roff_node *n;
1.229 schwarze 320: const char *cp;
1.263 schwarze 321: size_t sz;
1.262 schwarze 322: enum roff_tok tok;
1.263 schwarze 323: int sv;
324:
325: /* Determine the line macro. */
1.63 kristaps 326:
1.188 kristaps 327: sv = offs;
1.263 schwarze 328: tok = TOKEN_NONE;
329: for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
330: offs++;
331: if (sz == 2 || sz == 3)
332: tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
1.248 schwarze 333: if (tok == TOKEN_NONE) {
1.222 schwarze 334: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
335: ln, sv, buf + sv - 1);
1.253 schwarze 336: return 1;
1.53 kristaps 337: }
1.1 kristaps 338:
1.229 schwarze 339: /* Skip a leading escape sequence or tab. */
1.160 kristaps 340:
1.229 schwarze 341: switch (buf[offs]) {
342: case '\\':
343: cp = buf + offs + 1;
344: mandoc_escape(&cp, NULL, NULL);
345: offs = cp - buf;
346: break;
347: case '\t':
1.188 kristaps 348: offs++;
1.229 schwarze 349: break;
350: default:
351: break;
352: }
1.160 kristaps 353:
354: /* Jump to the next non-whitespace word. */
1.1 kristaps 355:
1.261 schwarze 356: while (buf[offs] == ' ')
1.188 kristaps 357: offs++;
1.1 kristaps 358:
1.213 schwarze 359: /*
1.125 kristaps 360: * Trailing whitespace. Note that tabs are allowed to be passed
361: * into the parser as "text", so we only warn about spaces here.
362: */
1.115 kristaps 363:
1.188 kristaps 364: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.218 schwarze 365: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
366: ln, offs - 1, NULL);
1.115 kristaps 367:
1.144 kristaps 368: /*
369: * If an initial macro or a list invocation, divert directly
370: * into macro processing.
371: */
372:
1.257 schwarze 373: n = mdoc->last;
374: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.269 ! schwarze 375: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.253 schwarze 376: return 1;
1.232 schwarze 377: }
1.144 kristaps 378:
379: /*
1.257 schwarze 380: * If a column list contains a non-It macro, assume an implicit
381: * item macro. This can happen one or more times at the
382: * beginning of such a list, intermixed with text lines and
383: * with nodes generated on the roff level, for example by tbl.
1.98 kristaps 384: */
1.144 kristaps 385:
1.257 schwarze 386: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
387: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
388: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
389: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 390: mdoc->flags |= MDOC_FREECOL;
1.269 ! schwarze 391: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
1.253 schwarze 392: return 1;
1.144 kristaps 393: }
394:
395: /* Normal processing of a macro. */
396:
1.269 ! schwarze 397: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.208 schwarze 398:
399: /* In quick mode (for mandocdb), abort after the NAME section. */
400:
401: if (mdoc->quick && MDOC_Sh == tok &&
402: SEC_NAME != mdoc->last->sec)
1.253 schwarze 403: return 2;
1.1 kristaps 404:
1.253 schwarze 405: return 1;
1.1 kristaps 406: }
1.100 kristaps 407:
1.186 kristaps 408: enum mdelim
409: mdoc_isdelim(const char *p)
410: {
411:
412: if ('\0' == p[0])
1.253 schwarze 413: return DELIM_NONE;
1.186 kristaps 414:
415: if ('\0' == p[1])
416: switch (p[0]) {
1.213 schwarze 417: case '(':
418: case '[':
1.253 schwarze 419: return DELIM_OPEN;
1.213 schwarze 420: case '|':
1.253 schwarze 421: return DELIM_MIDDLE;
1.213 schwarze 422: case '.':
423: case ',':
424: case ';':
425: case ':':
426: case '?':
427: case '!':
428: case ')':
429: case ']':
1.253 schwarze 430: return DELIM_CLOSE;
1.186 kristaps 431: default:
1.253 schwarze 432: return DELIM_NONE;
1.186 kristaps 433: }
434:
435: if ('\\' != p[0])
1.253 schwarze 436: return DELIM_NONE;
1.100 kristaps 437:
1.186 kristaps 438: if (0 == strcmp(p + 1, "."))
1.253 schwarze 439: return DELIM_CLOSE;
1.200 schwarze 440: if (0 == strcmp(p + 1, "fR|\\fP"))
1.253 schwarze 441: return DELIM_MIDDLE;
1.186 kristaps 442:
1.253 schwarze 443: return DELIM_NONE;
1.255 schwarze 444: }
445:
446: void
447: mdoc_validate(struct roff_man *mdoc)
448: {
449:
450: mdoc->last = mdoc->first;
451: mdoc_node_validate(mdoc);
452: mdoc_state_reset(mdoc);
1.186 kristaps 453: }
CVSweb