Annotation of mandoc/mdoc.c, Revision 1.270
1.270 ! schwarze 1: /* $Id: mdoc.c,v 1.269 2018/08/17 20:33:37 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;
1.1 kristaps 156: }
157:
1.53 kristaps 158: /*
159: * Parse free-form text, that is, a line that does not begin with the
160: * control character.
161: */
162: static int
1.242 schwarze 163: mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
1.1 kristaps 164: {
1.240 schwarze 165: struct roff_node *n;
1.267 schwarze 166: const char *cp, *sp;
1.142 kristaps 167: char *c, *ws, *end;
168:
1.203 schwarze 169: n = mdoc->last;
1.142 kristaps 170:
171: /*
1.257 schwarze 172: * If a column list contains plain text, assume an implicit item
173: * macro. This can happen one or more times at the beginning
174: * of such a list, intermixed with non-It mdoc macros and with
175: * nodes generated on the roff level, for example by tbl.
1.142 kristaps 176: */
177:
1.257 schwarze 178: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
179: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
180: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
181: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 182: mdoc->flags |= MDOC_FREECOL;
1.269 schwarze 183: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
184: line, offs, &offs, buf);
1.253 schwarze 185: return 1;
1.142 kristaps 186: }
1.124 kristaps 187:
1.137 schwarze 188: /*
189: * Search for the beginning of unescaped trailing whitespace (ws)
190: * and for the first character not to be output (end).
191: */
1.139 kristaps 192:
193: /* FIXME: replace with strcspn(). */
1.137 schwarze 194: ws = NULL;
195: for (c = end = buf + offs; *c; c++) {
196: switch (*c) {
197: case ' ':
198: if (NULL == ws)
199: ws = c;
200: continue;
201: case '\t':
202: /*
203: * Always warn about trailing tabs,
204: * even outside literal context,
205: * where they should be put on the next line.
206: */
207: if (NULL == ws)
208: ws = c;
209: /*
210: * Strip trailing tabs in literal context only;
211: * outside, they affect the next line.
212: */
1.203 schwarze 213: if (MDOC_LITERAL & mdoc->flags)
1.137 schwarze 214: continue;
215: break;
216: case '\\':
217: /* Skip the escaped character, too, if any. */
218: if (c[1])
219: c++;
220: /* FALLTHROUGH */
221: default:
222: ws = NULL;
223: break;
224: }
225: end = c + 1;
226: }
227: *end = '\0';
1.91 kristaps 228:
1.137 schwarze 229: if (ws)
1.218 schwarze 230: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
231: line, (int)(ws-buf), NULL);
1.115 kristaps 232:
1.267 schwarze 233: /*
234: * Blank lines are allowed in no-fill mode
235: * and cancel preceding \c,
236: * but add a single vertical space elsewhere.
237: */
238:
1.231 schwarze 239: if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
1.267 schwarze 240: switch (mdoc->last->type) {
241: case ROFFT_TEXT:
242: sp = mdoc->last->string;
243: cp = end = strchr(sp, '\0') - 2;
244: if (cp < sp || cp[0] != '\\' || cp[1] != 'c')
245: break;
246: while (cp > sp && cp[-1] == '\\')
247: cp--;
248: if ((end - cp) % 2)
249: break;
250: *end = '\0';
251: return 1;
252: default:
253: break;
254: }
1.218 schwarze 255: mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
256: 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:
1.231 schwarze 265: if (mdoc->flags & MDOC_LITERAL)
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)))
1.259 schwarze 296: mandoc_msg(MANDOCERR_EOS, mdoc->parse,
297: line, (int)(c - buf), NULL);
298: }
299:
1.253 schwarze 300: return 1;
1.1 kristaps 301: }
302:
1.53 kristaps 303: /*
304: * Parse a macro line, that is, a line beginning with the control
305: * character.
306: */
1.155 kristaps 307: static int
1.242 schwarze 308: mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
1.1 kristaps 309: {
1.240 schwarze 310: struct roff_node *n;
1.229 schwarze 311: const char *cp;
1.263 schwarze 312: size_t sz;
1.262 schwarze 313: enum roff_tok tok;
1.263 schwarze 314: int sv;
315:
316: /* Determine the line macro. */
1.63 kristaps 317:
1.188 kristaps 318: sv = offs;
1.263 schwarze 319: tok = TOKEN_NONE;
320: for (sz = 0; sz < 4 && strchr(" \t\\", buf[offs]) == NULL; sz++)
321: offs++;
322: if (sz == 2 || sz == 3)
323: tok = roffhash_find(mdoc->mdocmac, buf + sv, sz);
1.248 schwarze 324: if (tok == TOKEN_NONE) {
1.222 schwarze 325: mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
326: ln, sv, buf + sv - 1);
1.253 schwarze 327: return 1;
1.53 kristaps 328: }
1.1 kristaps 329:
1.229 schwarze 330: /* Skip a leading escape sequence or tab. */
1.160 kristaps 331:
1.229 schwarze 332: switch (buf[offs]) {
333: case '\\':
334: cp = buf + offs + 1;
335: mandoc_escape(&cp, NULL, NULL);
336: offs = cp - buf;
337: break;
338: case '\t':
1.188 kristaps 339: offs++;
1.229 schwarze 340: break;
341: default:
342: break;
343: }
1.160 kristaps 344:
345: /* Jump to the next non-whitespace word. */
1.1 kristaps 346:
1.261 schwarze 347: while (buf[offs] == ' ')
1.188 kristaps 348: offs++;
1.1 kristaps 349:
1.213 schwarze 350: /*
1.125 kristaps 351: * Trailing whitespace. Note that tabs are allowed to be passed
352: * into the parser as "text", so we only warn about spaces here.
353: */
1.115 kristaps 354:
1.188 kristaps 355: if ('\0' == buf[offs] && ' ' == buf[offs - 1])
1.218 schwarze 356: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
357: ln, offs - 1, NULL);
1.115 kristaps 358:
1.144 kristaps 359: /*
360: * If an initial macro or a list invocation, divert directly
361: * into macro processing.
362: */
363:
1.257 schwarze 364: n = mdoc->last;
365: if (n == NULL || tok == MDOC_It || tok == MDOC_El) {
1.269 schwarze 366: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.253 schwarze 367: return 1;
1.232 schwarze 368: }
1.144 kristaps 369:
370: /*
1.257 schwarze 371: * If a column list contains a non-It macro, assume an implicit
372: * item macro. This can happen one or more times at the
373: * beginning of such a list, intermixed with text lines and
374: * with nodes generated on the roff level, for example by tbl.
1.98 kristaps 375: */
1.144 kristaps 376:
1.257 schwarze 377: if ((n->tok == MDOC_Bl && n->type == ROFFT_BODY &&
378: n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) ||
379: (n->parent != NULL && n->parent->tok == MDOC_Bl &&
380: n->parent->norm->Bl.type == LIST_column)) {
1.203 schwarze 381: mdoc->flags |= MDOC_FREECOL;
1.269 schwarze 382: (*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
1.253 schwarze 383: return 1;
1.144 kristaps 384: }
385:
386: /* Normal processing of a macro. */
387:
1.269 schwarze 388: (*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
1.208 schwarze 389:
390: /* In quick mode (for mandocdb), abort after the NAME section. */
391:
392: if (mdoc->quick && MDOC_Sh == tok &&
393: SEC_NAME != mdoc->last->sec)
1.253 schwarze 394: return 2;
1.1 kristaps 395:
1.253 schwarze 396: return 1;
1.1 kristaps 397: }
1.100 kristaps 398:
1.186 kristaps 399: enum mdelim
400: mdoc_isdelim(const char *p)
401: {
402:
403: if ('\0' == p[0])
1.253 schwarze 404: return DELIM_NONE;
1.186 kristaps 405:
406: if ('\0' == p[1])
407: switch (p[0]) {
1.213 schwarze 408: case '(':
409: case '[':
1.253 schwarze 410: return DELIM_OPEN;
1.213 schwarze 411: case '|':
1.253 schwarze 412: return DELIM_MIDDLE;
1.213 schwarze 413: case '.':
414: case ',':
415: case ';':
416: case ':':
417: case '?':
418: case '!':
419: case ')':
420: case ']':
1.253 schwarze 421: return DELIM_CLOSE;
1.186 kristaps 422: default:
1.253 schwarze 423: return DELIM_NONE;
1.186 kristaps 424: }
425:
426: if ('\\' != p[0])
1.253 schwarze 427: return DELIM_NONE;
1.100 kristaps 428:
1.186 kristaps 429: if (0 == strcmp(p + 1, "."))
1.253 schwarze 430: return DELIM_CLOSE;
1.200 schwarze 431: if (0 == strcmp(p + 1, "fR|\\fP"))
1.253 schwarze 432: return DELIM_MIDDLE;
1.186 kristaps 433:
1.253 schwarze 434: return DELIM_NONE;
1.255 schwarze 435: }
436:
437: void
438: mdoc_validate(struct roff_man *mdoc)
439: {
440:
441: mdoc->last = mdoc->first;
442: mdoc_node_validate(mdoc);
443: mdoc_state_reset(mdoc);
1.186 kristaps 444: }
CVSweb