Annotation of mandoc/mdoc.c, Revision 1.275
1.275 ! schwarze 1: /* $Id: mdoc.c,v 1.274 2018/12/31 07:46:07 schwarze Exp $ */
1.1 kristaps 2: /*
1.275 ! schwarze 3: * Copyright (c) 2010, 2012-2018, 2020 Ingo Schwarze <schwarze@openbsd.org>
1.182 schwarze 4: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
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.
1.275 ! schwarze 17: *
! 18: * Top level and utility functions of the mdoc(7) parser for mandoc(1).
1.1 kristaps 19: */
1.114 kristaps 20: #include "config.h"
21:
1.106 kristaps 22: #include <sys/types.h>
23:
1.1 kristaps 24: #include <assert.h>
1.211 schwarze 25: #include <ctype.h>
1.1 kristaps 26: #include <stdarg.h>
1.73 kristaps 27: #include <stdio.h>
1.1 kristaps 28: #include <stdlib.h>
29: #include <string.h>
1.120 kristaps 30: #include <time.h>
1.1 kristaps 31:
1.239 schwarze 32: #include "mandoc_aux.h"
33: #include "mandoc.h"
34: #include "roff.h"
1.187 kristaps 35: #include "mdoc.h"
1.239 schwarze 36: #include "libmandoc.h"
1.247 schwarze 37: #include "roff_int.h"
1.70 kristaps 38: #include "libmdoc.h"
1.1 kristaps 39:
1.213 schwarze 40: const char *const __mdoc_argnames[MDOC_ARG_MAX] = {
1.1 kristaps 41: "split", "nosplit", "ragged",
1.213 schwarze 42: "unfilled", "literal", "file",
43: "offset", "bullet", "dash",
44: "hyphen", "item", "enum",
45: "tag", "diag", "hang",
46: "ohang", "inset", "column",
47: "width", "compact", "std",
1.52 kristaps 48: "filled", "words", "emphasis",
1.108 kristaps 49: "symbolic", "nested", "centered"
1.262 schwarze 50: };
1.1 kristaps 51: const char * const *mdoc_argnames = __mdoc_argnames;
52:
1.242 schwarze 53: static int mdoc_ptext(struct roff_man *, int, char *, int);
54: static int mdoc_pmacro(struct roff_man *, int, char *, int);
1.88 kristaps 55:
1.181 kristaps 56:
1.50 kristaps 57: /*
1.53 kristaps 58: * Main parse routine. Parses a single line -- really just hands off to
1.123 kristaps 59: * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
1.50 kristaps 60: */
1.20 kristaps 61: int
1.242 schwarze 62: mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 63: {
64:
1.239 schwarze 65: if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
1.228 schwarze 66: mdoc->flags |= MDOC_NEWLINE;
1.153 schwarze 67:
68: /*
69: * Let the roff nS register switch SYNOPSIS mode early,
70: * such that the parser knows at all times
71: * whether this mode is on or off.
72: * Note that this mode is also switched by the Sh macro.
73: */
1.204 schwarze 74: if (roff_getreg(mdoc->roff, "nS"))
75: mdoc->flags |= MDOC_SYNOPSIS;
76: else
77: mdoc->flags &= ~MDOC_SYNOPSIS;
1.153 schwarze 78:
1.253 schwarze 79: return roff_getcontrol(mdoc->roff, buf, &offs) ?
1.213 schwarze 80: mdoc_pmacro(mdoc, ln, buf, offs) :
1.253 schwarze 81: mdoc_ptext(mdoc, ln, buf, offs);
1.1 kristaps 82: }
83:
1.232 schwarze 84: void
1.262 schwarze 85: mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
1.17 kristaps 86: {
1.240 schwarze 87: struct roff_node *p;
1.17 kristaps 88:
1.247 schwarze 89: p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
90: roff_node_append(mdoc, p);
1.242 schwarze 91: mdoc->next = ROFF_NEXT_CHILD;
1.152 schwarze 92: }
93:
1.240 schwarze 94: struct roff_node *
1.262 schwarze 95: mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
96: enum roff_tok tok, struct roff_node *body)
1.152 schwarze 97: {
1.240 schwarze 98: struct roff_node *p;
1.152 schwarze 99:
1.258 schwarze 100: body->flags |= NODE_ENDED;
101: body->parent->flags |= NODE_ENDED;
1.247 schwarze 102: p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
1.237 schwarze 103: p->body = body;
1.202 schwarze 104: p->norm = body->norm;
1.260 schwarze 105: p->end = ENDBODY_SPACE;
1.247 schwarze 106: roff_node_append(mdoc, p);
1.242 schwarze 107: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 108: return p;
1.1 kristaps 109: }
110:
1.240 schwarze 111: struct roff_node *
1.242 schwarze 112: mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
1.262 schwarze 113: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 114: {
1.240 schwarze 115: struct roff_node *p;
1.1 kristaps 116:
1.247 schwarze 117: p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
1.77 kristaps 118: p->args = args;
119: if (p->args)
1.53 kristaps 120: (args->refcnt)++;
1.172 kristaps 121:
122: switch (tok) {
1.213 schwarze 123: case MDOC_Bd:
124: case MDOC_Bf:
125: case MDOC_Bl:
1.217 schwarze 126: case MDOC_En:
1.213 schwarze 127: case MDOC_Rs:
1.172 kristaps 128: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
129: break;
130: default:
131: break;
132: }
1.247 schwarze 133: roff_node_append(mdoc, p);
1.242 schwarze 134: mdoc->next = ROFF_NEXT_CHILD;
1.253 schwarze 135: return p;
1.1 kristaps 136: }
137:
1.231 schwarze 138: void
1.242 schwarze 139: mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
1.262 schwarze 140: enum roff_tok tok, struct mdoc_arg *args)
1.1 kristaps 141: {
1.240 schwarze 142: struct roff_node *p;
1.1 kristaps 143:
1.247 schwarze 144: p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
1.77 kristaps 145: p->args = args;
146: if (p->args)
1.53 kristaps 147: (args->refcnt)++;
1.172 kristaps 148:
149: switch (tok) {
1.213 schwarze 150: case MDOC_An:
1.172 kristaps 151: p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
152: break;
153: default:
154: break;
155: }
1.247 schwarze 156: roff_node_append(mdoc, p);
1.242 schwarze 157: mdoc->next = ROFF_NEXT_CHILD;
1.1 kristaps 158: }
159:
1.53 kristaps 160: /*
161: * Parse free-form text, that is, a line that does not begin with the
162: * control character.
163: */
164: static int
1.242 schwarze 165: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 166: {
1.240 schwarze 167: struct roff_node *n;
1.267 schwarze 168: const char *cp, *sp;
1.142 kristaps 169: char *c, *ws, *end;
170:
1.203 schwarze 171: n = mdoc->last;
1.142 kristaps 172:
173: /*
1.257 schwarze 174: * If a column list contains plain text, assume an implicit item
175: * macro. This can happen one or more times at the beginning
176: * of such a list, intermixed with non-It mdoc macros and with
177: * nodes generated on the roff level, for example by tbl.
1.142 kristaps 178: */
179:
1.257 schwarze 180: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
181: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
182: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
183: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 184: mdoc->flags |= MDOC_FREECOL;
1.269 schwarze 185: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
186: line, offs, &offs, buf);
1.253 schwarze 187: return 1;
1.142 kristaps 188: }
1.124 kristaps 189:
1.137 schwarze 190: /*
191: * Search for the beginning of unescaped trailing whitespace (ws)
192: * and for the first character not to be output (end).
193: */
1.139 kristaps 194:
195: /* FIXME: replace with strcspn(). */
1.137 schwarze 196: ws = NULL;
197: for (c = end = buf + offs; *c; c++) {
198: switch (*c) {
199: case ' ':
200: if (NULL == ws)
201: ws = c;
202: continue;
203: case '\t':
204: /*
205: * Always warn about trailing tabs,
206: * even outside literal context,
207: * where they should be put on the next line.
208: */
209: if (NULL == ws)
210: ws = c;
211: /*
212: * Strip trailing tabs in literal context only;
213: * outside, they affect the next line.
214: */
1.274 schwarze 215: if (mdoc->flags & ROFF_NOFILL)
1.137 schwarze 216: continue;
217: break;
218: case '\\':
219: /* Skip the escaped character, too, if any. */
220: if (c[1])
221: c++;
222: /* FALLTHROUGH */
223: default:
224: ws = NULL;
225: break;
226: }
227: end = c + 1;
228: }
229: *end = '\0';
1.91 kristaps 230:
1.137 schwarze 231: if (ws)
1.271 schwarze 232: mandoc_msg(MANDOCERR_SPACE_EOL, line, (int)(ws - buf), NULL);
1.115 kristaps 233:
1.267 schwarze 234: /*
235: * Blank lines are allowed in no-fill mode
236: * and cancel preceding \c,
237: * but add a single vertical space elsewhere.
238: */
239:
1.274 schwarze 240: if (buf[offs] == '\0' && (mdoc->flags & ROFF_NOFILL) == 0) {
1.267 schwarze 241: switch (mdoc->last->type) {
242: case ROFFT_TEXT:
243: sp = mdoc->last->string;
244: cp = end = strchr(sp, '\0') - 2;
245: if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
246: break;
247: while (cp > sp && cp[-1] == '\\')
248: cp--;
249: if ((end - cp) % 2)
250: break;
251: *end = '\0';
252: return 1;
253: default:
254: break;
255: }
1.271 schwarze 256: mandoc_msg(MANDOCERR_FI_BLANK, line, (int)(c - buf), NULL);
1.264 schwarze 257: roff_elem_alloc(mdoc, line, offs, ROFF_sp);
1.258 schwarze 258: mdoc->last->flags |= NODE_VALID | NODE_ENDED;
1.242 schwarze 259: mdoc->next = ROFF_NEXT_SIBLING;
1.253 schwarze 260: return 1;
1.119 kristaps 261: }
1.68 kristaps 262:
1.249 schwarze 263: roff_word_alloc(mdoc, line, offs, buf+offs);
1.91 kristaps 264:
1.274 schwarze 265: if (mdoc->flags & ROFF_NOFILL)
1.253 schwarze 266: return 1;
1.128 kristaps 267:
268: /*
269: * End-of-sentence check. If the last character is an unescaped
270: * EOS character, then flag the node as being the end of a
271: * sentence. The front-end will know how to interpret this.
272: */
1.132 kristaps 273:
1.137 schwarze 274: assert(buf < end);
275:
1.207 schwarze 276: if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
1.258 schwarze 277: mdoc->last->flags |= NODE_EOS;
1.259 schwarze 278:
279: for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
280: if (c - buf < offs + 2)
281: continue;
1.265 schwarze 282: if (end - c < 3)
1.259 schwarze 283: break;
1.266 schwarze 284: if (c[1] != ' ' ||
1.268 schwarze 285: isalnum((unsigned char)c[-2]) == 0 ||
286: isalnum((unsigned char)c[-1]) == 0 ||
1.266 schwarze 287: (c[-2] == 'n' && c[-1] == 'c') ||
288: (c[-2] == 'v' && c[-1] == 's'))
289: continue;
290: c += 2;
291: if (*c == ' ')
292: c++;
293: if (*c == ' ')
294: c++;
295: if (isupper((unsigned char)(*c)))
1.271 schwarze 296: mandoc_msg(MANDOCERR_EOS, line, (int)(c - buf), NULL);
1.259 schwarze 297: }
298:
1.253 schwarze 299: return 1;
1.1 kristaps 300: }
301:
1.53 kristaps 302: /*
303: * Parse a macro line, that is, a line beginning with the control
304: * character.
305: */
1.155 kristaps 306: static int
1.242 schwarze 307: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 308: {
1.240 schwarze 309: struct roff_node *n;
1.229 schwarze 310: const char *cp;
1.263 schwarze 311: size_t sz;
1.262 schwarze 312: enum roff_tok tok;
1.263 schwarze 313: int sv;
314:
315: /* Determine the line macro. */
1.63 kristaps 316:
1.188 kristaps 317: sv = offs;
1.263 schwarze 318: tok = TOKEN_NONE;
319: for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
320: offs++;
321: if (sz == 2 || sz == 3)
322: tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
1.248 schwarze 323: if (tok == TOKEN_NONE) {
1.271 schwarze 324: mandoc_msg(MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
1.253 schwarze 325: return 1;
1.53 kristaps 326: }
1.1 kristaps 327:
1.229 schwarze 328: /* Skip a leading escape sequence or tab. */
1.160 kristaps 329:
1.229 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':
1.188 kristaps 337: offs++;
1.229 schwarze 338: break;
339: default:
340: break;
341: }
1.160 kristaps 342:
343: /* Jump to the next non-whitespace word. */
1.1 kristaps 344:
1.261 schwarze 345: while (buf[offs] == ' ')
1.188 kristaps 346: offs++;
1.1 kristaps 347:
1.213 schwarze 348: /*
1.125 kristaps 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: */
1.115 kristaps 352:
1.188 kristaps 353: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.271 schwarze 354: mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);
1.115 kristaps 355:
1.144 kristaps 356: /*
1.275 ! schwarze 357: * If an initial or transparent macro or a list invocation,
! 358: * divert directly into macro processing.
1.144 kristaps 359: */
360:
1.257 schwarze 361: n = mdoc->last;
1.275 ! schwarze 362: if (n == NULL || tok == MDOC_It || tok == MDOC_El ||
! 363: roff_tok_transparent(tok)) {
1.269 schwarze 364: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.253 schwarze 365: return 1;
1.232 schwarze 366: }
1.144 kristaps 367:
368: /*
1.257 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.
1.98 kristaps 373: */
1.144 kristaps 374:
1.257 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)) {
1.203 schwarze 379: mdoc->flags |= MDOC_FREECOL;
1.269 schwarze 380: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
1.253 schwarze 381: return 1;
1.144 kristaps 382: }
383:
384: /* Normal processing of a macro. */
385:
1.269 schwarze 386: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.208 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)
1.253 schwarze 392: return 2;
1.1 kristaps 393:
1.253 schwarze 394: return 1;
1.1 kristaps 395: }
1.100 kristaps 396:
1.186 kristaps 397: enum mdelim
398: mdoc_isdelim(const char *p)
399: {
400:
401: if ('\0' == p[0])
1.253 schwarze 402: return DELIM_NONE;
1.186 kristaps 403:
404: if ('\0' == p[1])
405: switch (p[0]) {
1.213 schwarze 406: case '(':
407: case '[':
1.253 schwarze 408: return DELIM_OPEN;
1.213 schwarze 409: case '|':
1.253 schwarze 410: return DELIM_MIDDLE;
1.213 schwarze 411: case '.':
412: case ',':
413: case ';':
414: case ':':
415: case '?':
416: case '!':
417: case ')':
418: case ']':
1.253 schwarze 419: return DELIM_CLOSE;
1.186 kristaps 420: default:
1.253 schwarze 421: return DELIM_NONE;
1.186 kristaps 422: }
423:
424: if ('\\' != p[0])
1.253 schwarze 425: return DELIM_NONE;
1.100 kristaps 426:
1.186 kristaps 427: if (0 == strcmp(p + 1, "."))
1.253 schwarze 428: return DELIM_CLOSE;
1.200 schwarze 429: if (0 == strcmp(p + 1, "fR|\\fP"))
1.253 schwarze 430: return DELIM_MIDDLE;
1.186 kristaps 431:
1.253 schwarze 432: return DELIM_NONE;
1.186 kristaps 433: }
CVSweb