Annotation of src/usr.bin/mandoc/mdoc.c, Revision 1.159
1.159 ! schwarze 1: /* $OpenBSD: mdoc.c,v 1.158 2018/08/17 20:31:52 schwarze Exp $ */
1.1 kristaps 2: /*
1.77 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.158 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
1.3 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
1.151 schwarze 81: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
1.1 kristaps 82: {
1.129 schwarze 83: struct roff_node *p;
1.1 kristaps 84:
1.136 schwarze 85: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
86: roff_node_append(mdoc, p);
1.131 schwarze 87: mdoc->next = ROFF_NEXT_CHILD;
1.59 schwarze 88: }
89:
1.129 schwarze 90: struct roff_node *
1.151 schwarze 91: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
92: enum roff_tok tok, struct roff_node *body)
1.59 schwarze 93: {
1.129 schwarze 94: struct roff_node *p;
1.59 schwarze 95:
1.147 schwarze 96: body->flags |= NODE_ENDED;
97: body->parent->flags |= NODE_ENDED;
1.136 schwarze 98: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.126 schwarze 99: p->body = body;
1.92 schwarze 100: p->norm = body->norm;
1.149 schwarze 101: p->end = ENDBODY_SPACE;
1.136 schwarze 102: roff_node_append(mdoc, p);
1.131 schwarze 103: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 104: return p;
1.1 kristaps 105: }
106:
1.129 schwarze 107: struct roff_node *
1.131 schwarze 108: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.151 schwarze 109: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 110: {
1.129 schwarze 111: struct roff_node *p;
1.1 kristaps 112:
1.136 schwarze 113: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.4 schwarze 114: p->args = args;
115: if (p->args)
1.1 kristaps 116: (args->refcnt)++;
1.75 schwarze 117:
118: switch (tok) {
1.103 schwarze 119: case MDOC_Bd:
120: case MDOC_Bf:
121: case MDOC_Bl:
1.107 schwarze 122: case MDOC_En:
1.103 schwarze 123: case MDOC_Rs:
1.75 schwarze 124: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
125: break;
126: default:
127: break;
128: }
1.136 schwarze 129: roff_node_append(mdoc, p);
1.131 schwarze 130: mdoc->next = ROFF_NEXT_CHILD;
1.142 schwarze 131: return p;
1.1 kristaps 132: }
133:
1.120 schwarze 134: void
1.131 schwarze 135: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.151 schwarze 136: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 137: {
1.129 schwarze 138: struct roff_node *p;
1.1 kristaps 139:
1.136 schwarze 140: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.4 schwarze 141: p->args = args;
142: if (p->args)
1.1 kristaps 143: (args->refcnt)++;
1.75 schwarze 144:
145: switch (tok) {
1.103 schwarze 146: case MDOC_An:
1.75 schwarze 147: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
148: break;
149: default:
150: break;
151: }
1.136 schwarze 152: roff_node_append(mdoc, p);
1.131 schwarze 153: mdoc->next = ROFF_NEXT_CHILD;
1.1 kristaps 154: }
155:
156: /*
157: * Parse free-form text, that is, a line that does not begin with the
158: * control character.
159: */
160: static int
1.131 schwarze 161: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 162: {
1.129 schwarze 163: struct roff_node *n;
1.156 schwarze 164: const char *cp, *sp;
1.56 schwarze 165: char *c, *ws, *end;
1.46 schwarze 166:
1.93 schwarze 167: n = mdoc->last;
1.56 schwarze 168:
169: /*
1.146 schwarze 170: * If a column list contains plain text, assume an implicit item
171: * macro. This can happen one or more times at the beginning
172: * of such a list, intermixed with non-It mdoc macros and with
173: * nodes generated on the roff level, for example by tbl.
1.56 schwarze 174: */
175:
1.146 schwarze 176: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
177: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
178: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
179: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 180: mdoc->flags |= MDOC_FREECOL;
1.158 schwarze 181: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
182: line, offs, &offs, buf);
1.142 schwarze 183: return 1;
1.56 schwarze 184: }
185:
1.52 schwarze 186: /*
187: * Search for the beginning of unescaped trailing whitespace (ws)
188: * and for the first character not to be output (end).
189: */
1.56 schwarze 190:
191: /* FIXME: replace with strcspn(). */
1.52 schwarze 192: ws = NULL;
1.53 schwarze 193: for (c = end = buf + offs; *c; c++) {
1.52 schwarze 194: switch (*c) {
195: case ' ':
196: if (NULL == ws)
197: ws = c;
198: continue;
199: case '\t':
200: /*
201: * Always warn about trailing tabs,
202: * even outside literal context,
203: * where they should be put on the next line.
204: */
205: if (NULL == ws)
206: ws = c;
207: /*
208: * Strip trailing tabs in literal context only;
209: * outside, they affect the next line.
210: */
1.93 schwarze 211: if (MDOC_LITERAL & mdoc->flags)
1.52 schwarze 212: continue;
213: break;
214: case '\\':
215: /* Skip the escaped character, too, if any. */
216: if (c[1])
217: c++;
218: /* FALLTHROUGH */
219: default:
220: ws = NULL;
221: break;
222: }
223: end = c + 1;
224: }
225: *end = '\0';
1.19 schwarze 226:
1.52 schwarze 227: if (ws)
1.108 schwarze 228: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
229: line, (int)(ws-buf), NULL);
1.34 schwarze 230:
1.156 schwarze 231: /*
232: * Blank lines are allowed in no-fill mode
233: * and cancel preceding \c,
234: * but add a single vertical space elsewhere.
235: */
236:
1.120 schwarze 237: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.156 schwarze 238: switch (mdoc->last->type) {
239: case ROFFT_TEXT:
240: sp = mdoc->last->string;
241: cp = end = strchr(sp, '\0') - 2;
242: if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
243: break;
244: while (cp > sp && cp[-1] == '\\')
245: cp--;
246: if ((end - cp) % 2)
247: break;
248: *end = '\0';
249: return 1;
250: default:
251: break;
252: }
1.108 schwarze 253: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
254: line, (int)(c - buf), NULL);
1.153 schwarze 255: roff_elem_alloc(mdoc, line, offs, ROFF_sp);
1.147 schwarze 256: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.131 schwarze 257: mdoc->next = ROFF_NEXT_SIBLING;
1.142 schwarze 258: return 1;
1.38 schwarze 259: }
1.1 kristaps 260:
1.138 schwarze 261: roff_word_alloc(mdoc, line, offs, buf+offs);
1.1 kristaps 262:
1.120 schwarze 263: if (mdoc->flags & MDOC_LITERAL)
1.142 schwarze 264: return 1;
1.52 schwarze 265:
1.35 schwarze 266: /*
1.48 schwarze 267: * End-of-sentence check. If the last character is an unescaped
268: * EOS character, then flag the node as being the end of a
269: * sentence. The front-end will know how to interpret this.
1.35 schwarze 270: */
271:
1.52 schwarze 272: assert(buf < end);
1.48 schwarze 273:
1.97 schwarze 274: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.147 schwarze 275: mdoc->last->flags |= NODE_EOS;
1.148 schwarze 276:
277: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
278: if (c - buf < offs + 2)
279: continue;
1.154 schwarze 280: if (end - c < 3)
1.148 schwarze 281: break;
1.155 schwarze 282: if (c[1] != ' ' ||
1.157 schwarze 283: isalnum((unsigned char)c[-2]) == 0 ||
284: isalnum((unsigned char)c[-1]) == 0 ||
1.155 schwarze 285: (c[-2] == 'n' && c[-1] == 'c') ||
286: (c[-2] == 'v' && c[-1] == 's'))
287: continue;
288: c += 2;
289: if (*c == ' ')
290: c++;
291: if (*c == ' ')
292: c++;
293: if (isupper((unsigned char)(*c)))
1.148 schwarze 294: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
295: line, (int)(c - buf), NULL);
296: }
297:
1.142 schwarze 298: return 1;
1.1 kristaps 299: }
1.19 schwarze 300:
1.1 kristaps 301: /*
302: * Parse a macro line, that is, a line beginning with the control
303: * character.
304: */
1.61 schwarze 305: static int
1.131 schwarze 306: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 307: {
1.129 schwarze 308: struct roff_node *n;
1.118 schwarze 309: const char *cp;
1.152 schwarze 310: size_t sz;
1.151 schwarze 311: enum roff_tok tok;
1.152 schwarze 312: int sv;
313:
314: /* Determine the line macro. */
1.1 kristaps 315:
1.83 schwarze 316: sv = offs;
1.152 schwarze 317: tok = TOKEN_NONE;
318: for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
319: offs++;
320: if (sz == 2 || sz == 3)
321: tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
1.137 schwarze 322: if (tok == TOKEN_NONE) {
1.112 schwarze 323: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
324: ln, sv, buf + sv - 1);
1.142 schwarze 325: return 1;
1.1 kristaps 326: }
327:
1.118 schwarze 328: /* Skip a leading escape sequence or tab. */
1.63 schwarze 329:
1.118 schwarze 330: switch (buf[offs]) {
331: case '\\':
332: cp = buf + offs + 1;
333: mandoc_escape(&cp, NULL, NULL);
334: offs = cp - buf;
335: break;
336: case '\t':
1.83 schwarze 337: offs++;
1.118 schwarze 338: break;
339: default:
340: break;
341: }
1.63 schwarze 342:
343: /* Jump to the next non-whitespace word. */
1.1 kristaps 344:
1.150 schwarze 345: while (buf[offs] == ' ')
1.83 schwarze 346: offs++;
1.34 schwarze 347:
1.103 schwarze 348: /*
1.46 schwarze 349: * Trailing whitespace. Note that tabs are allowed to be passed
350: * into the parser as "text", so we only warn about spaces here.
351: */
1.34 schwarze 352:
1.83 schwarze 353: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.108 schwarze 354: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
355: ln, offs - 1, NULL);
1.1 kristaps 356:
1.56 schwarze 357: /*
358: * If an initial macro or a list invocation, divert directly
359: * into macro processing.
360: */
361:
1.146 schwarze 362: n = mdoc->last;
363: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.158 schwarze 364: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.142 schwarze 365: return 1;
1.121 schwarze 366: }
1.56 schwarze 367:
368: /*
1.146 schwarze 369: * If a column list contains a non-It macro, assume an implicit
370: * item macro. This can happen one or more times at the
371: * beginning of such a list, intermixed with text lines and
372: * with nodes generated on the roff level, for example by tbl.
1.24 schwarze 373: */
1.56 schwarze 374:
1.146 schwarze 375: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
376: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
377: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
378: n->parent->norm->Bl.type == LIST_column)) {
1.93 schwarze 379: mdoc->flags |= MDOC_FREECOL;
1.158 schwarze 380: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
1.142 schwarze 381: return 1;
1.56 schwarze 382: }
383:
384: /* Normal processing of a macro. */
385:
1.158 schwarze 386: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.98 schwarze 387:
388: /* In quick mode (for mandocdb), abort after the NAME section. */
389:
390: if (mdoc->quick && MDOC_Sh == tok &&
391: SEC_NAME != mdoc->last->sec)
1.142 schwarze 392: return 2;
1.1 kristaps 393:
1.142 schwarze 394: return 1;
1.1 kristaps 395: }
1.24 schwarze 396:
1.83 schwarze 397: enum mdelim
398: mdoc_isdelim(const char *p)
399: {
400:
401: if ('\0' == p[0])
1.142 schwarze 402: return DELIM_NONE;
1.83 schwarze 403:
404: if ('\0' == p[1])
405: switch (p[0]) {
1.103 schwarze 406: case '(':
407: case '[':
1.142 schwarze 408: return DELIM_OPEN;
1.103 schwarze 409: case '|':
1.142 schwarze 410: return DELIM_MIDDLE;
1.103 schwarze 411: case '.':
412: case ',':
413: case ';':
414: case ':':
415: case '?':
416: case '!':
417: case ')':
418: case ']':
1.142 schwarze 419: return DELIM_CLOSE;
1.83 schwarze 420: default:
1.142 schwarze 421: return DELIM_NONE;
1.83 schwarze 422: }
423:
424: if ('\\' != p[0])
1.142 schwarze 425: return DELIM_NONE;
1.24 schwarze 426:
1.83 schwarze 427: if (0 == strcmp(p + 1, "."))
1.142 schwarze 428: return DELIM_CLOSE;
1.90 schwarze 429: if (0 == strcmp(p + 1, "fR|\\fP"))
1.142 schwarze 430: return DELIM_MIDDLE;
1.83 schwarze 431:
1.142 schwarze 432: return DELIM_NONE;
1.144 schwarze 433: }
434:
435: void
436: mdoc_validate(struct roff_man *mdoc)
437: {
438:
439: mdoc->last = mdoc->first;
440: mdoc_node_validate(mdoc);
441: mdoc_state_reset(mdoc);
1.83 schwarze 442: }
![[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