1/* $Id: mdoc.c,v 1.275 2020/04/06 10:16:17 schwarze Exp $ */
2/*
3 * Copyright (c) 2010, 2012-2018, 2020 Ingo Schwarze <schwarze@openbsd.org>
4 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
5 *
6 * Permission to use, copy, modify, and distribute this software for any
7 * purpose with or without fee is hereby granted, provided that the above
8 * copyright notice and this permission notice appear in all copies.
9 *
10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
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.
17 *
18 * Top level and utility functions of the mdoc(7) parser for mandoc(1).
19 */
20#include "config.h"
21
22#include <sys/types.h>
23
24#include <assert.h>
25#include <ctype.h>
26#include <stdarg.h>
27#include <stdio.h>
28#include <stdlib.h>
29#include <string.h>
30#include <time.h>
31
32#include "mandoc_aux.h"
33#include "mandoc.h"
34#include "roff.h"
35#include "mdoc.h"
36#include "libmandoc.h"
37#include "roff_int.h"
38#include "libmdoc.h"
39
40const	char *const __mdoc_argnames[MDOC_ARG_MAX] = {
41	"split",		"nosplit",		"ragged",
42	"unfilled",		"literal",		"file",
43	"offset",		"bullet",		"dash",
44	"hyphen",		"item",			"enum",
45	"tag",			"diag",			"hang",
46	"ohang",		"inset",		"column",
47	"width",		"compact",		"std",
48	"filled",		"words",		"emphasis",
49	"symbolic",		"nested",		"centered"
50};
51const	char * const *mdoc_argnames = __mdoc_argnames;
52
53static	int		  mdoc_ptext(struct roff_man *, int, char *, int);
54static	int		  mdoc_pmacro(struct roff_man *, int, char *, int);
55
56
57/*
58 * Main parse routine.  Parses a single line -- really just hands off to
59 * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
60 */
61int
62mdoc_parseln(struct roff_man *mdoc, int ln, char *buf, int offs)
63{
64
65	if (mdoc->last->type != ROFFT_EQN || ln > mdoc->last->line)
66		mdoc->flags |= MDOC_NEWLINE;
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	 */
74	if (roff_getreg(mdoc->roff, "nS"))
75		mdoc->flags |= MDOC_SYNOPSIS;
76	else
77		mdoc->flags &= ~MDOC_SYNOPSIS;
78
79	return roff_getcontrol(mdoc->roff, buf, &offs) ?
80	    mdoc_pmacro(mdoc, ln, buf, offs) :
81	    mdoc_ptext(mdoc, ln, buf, offs);
82}
83
84void
85mdoc_tail_alloc(struct roff_man *mdoc, int line, int pos, enum roff_tok tok)
86{
87	struct roff_node *p;
88
89	p = roff_node_alloc(mdoc, line, pos, ROFFT_TAIL, tok);
90	roff_node_append(mdoc, p);
91	mdoc->next = ROFF_NEXT_CHILD;
92}
93
94struct roff_node *
95mdoc_endbody_alloc(struct roff_man *mdoc, int line, int pos,
96    enum roff_tok tok, struct roff_node *body)
97{
98	struct roff_node *p;
99
100	body->flags |= NODE_ENDED;
101	body->parent->flags |= NODE_ENDED;
102	p = roff_node_alloc(mdoc, line, pos, ROFFT_BODY, tok);
103	p->body = body;
104	p->norm = body->norm;
105	p->end = ENDBODY_SPACE;
106	roff_node_append(mdoc, p);
107	mdoc->next = ROFF_NEXT_SIBLING;
108	return p;
109}
110
111struct roff_node *
112mdoc_block_alloc(struct roff_man *mdoc, int line, int pos,
113    enum roff_tok tok, struct mdoc_arg *args)
114{
115	struct roff_node *p;
116
117	p = roff_node_alloc(mdoc, line, pos, ROFFT_BLOCK, tok);
118	p->args = args;
119	if (p->args)
120		(args->refcnt)++;
121
122	switch (tok) {
123	case MDOC_Bd:
124	case MDOC_Bf:
125	case MDOC_Bl:
126	case MDOC_En:
127	case MDOC_Rs:
128		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
129		break;
130	default:
131		break;
132	}
133	roff_node_append(mdoc, p);
134	mdoc->next = ROFF_NEXT_CHILD;
135	return p;
136}
137
138void
139mdoc_elem_alloc(struct roff_man *mdoc, int line, int pos,
140     enum roff_tok tok, struct mdoc_arg *args)
141{
142	struct roff_node *p;
143
144	p = roff_node_alloc(mdoc, line, pos, ROFFT_ELEM, tok);
145	p->args = args;
146	if (p->args)
147		(args->refcnt)++;
148
149	switch (tok) {
150	case MDOC_An:
151		p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
152		break;
153	default:
154		break;
155	}
156	roff_node_append(mdoc, p);
157	mdoc->next = ROFF_NEXT_CHILD;
158}
159
160/*
161 * Parse free-form text, that is, a line that does not begin with the
162 * control character.
163 */
164static int
165mdoc_ptext(struct roff_man *mdoc, int line, char *buf, int offs)
166{
167	struct roff_node *n;
168	const char	 *cp, *sp;
169	char		 *c, *ws, *end;
170
171	n = mdoc->last;
172
173	/*
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.
178	 */
179
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)) {
184		mdoc->flags |= MDOC_FREECOL;
185		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It,
186		    line, offs, &offs, buf);
187		return 1;
188	}
189
190	/*
191	 * Search for the beginning of unescaped trailing whitespace (ws)
192	 * and for the first character not to be output (end).
193	 */
194
195	/* FIXME: replace with strcspn(). */
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			 */
215			if (mdoc->flags & ROFF_NOFILL)
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';
230
231	if (ws)
232		mandoc_msg(MANDOCERR_SPACE_EOL, line, (int)(ws - buf), NULL);
233
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
240	if (buf[offs] == '\0' && (mdoc->flags & ROFF_NOFILL) == 0) {
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		}
256		mandoc_msg(MANDOCERR_FI_BLANK, line, (int)(c - buf), NULL);
257		roff_elem_alloc(mdoc, line, offs, ROFF_sp);
258		mdoc->last->flags |= NODE_VALID | NODE_ENDED;
259		mdoc->next = ROFF_NEXT_SIBLING;
260		return 1;
261	}
262
263	roff_word_alloc(mdoc, line, offs, buf+offs);
264
265	if (mdoc->flags & ROFF_NOFILL)
266		return 1;
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	 */
273
274	assert(buf < end);
275
276	if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
277		mdoc->last->flags |= NODE_EOS;
278
279	for (c = buf + offs; c != NULL; c = strchr(c + 1, '.')) {
280		if (c - buf < offs + 2)
281			continue;
282		if (end - c < 3)
283			break;
284		if (c[1] != ' ' ||
285		    isalnum((unsigned char)c[-2]) == 0 ||
286		    isalnum((unsigned char)c[-1]) == 0 ||
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)))
296			mandoc_msg(MANDOCERR_EOS, line, (int)(c - buf), NULL);
297	}
298
299	return 1;
300}
301
302/*
303 * Parse a macro line, that is, a line beginning with the control
304 * character.
305 */
306static int
307mdoc_pmacro(struct roff_man *mdoc, int ln, char *buf, int offs)
308{
309	struct roff_node *n;
310	const char	 *cp;
311	size_t		  sz;
312	enum roff_tok	  tok;
313	int		  sv;
314
315	/* Determine the line macro. */
316
317	sv = offs;
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);
323	if (tok == TOKEN_NONE) {
324		mandoc_msg(MANDOCERR_MACRO, ln, sv, "%s", buf + sv - 1);
325		return 1;
326	}
327
328	/* Skip a leading escape sequence or tab. */
329
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':
337		offs++;
338		break;
339	default:
340		break;
341	}
342
343	/* Jump to the next non-whitespace word. */
344
345	while (buf[offs] == ' ')
346		offs++;
347
348	/*
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	 */
352
353	if ('\0' == buf[offs] && ' ' == buf[offs - 1])
354		mandoc_msg(MANDOCERR_SPACE_EOL, ln, offs - 1, NULL);
355
356	/*
357	 * If an initial or transparent macro or a list invocation,
358	 * divert directly into macro processing.
359	 */
360
361	n = mdoc->last;
362	if (n == NULL || tok == MDOC_It || tok == MDOC_El ||
363	    roff_tok_transparent(tok)) {
364		(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
365		return 1;
366	}
367
368	/*
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.
373	 */
374
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)) {
379		mdoc->flags |= MDOC_FREECOL;
380		(*mdoc_macro(MDOC_It)->fp)(mdoc, MDOC_It, ln, sv, &sv, buf);
381		return 1;
382	}
383
384	/* Normal processing of a macro. */
385
386	(*mdoc_macro(tok)->fp)(mdoc, tok, ln, sv, &offs, buf);
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)
392		return 2;
393
394	return 1;
395}
396
397enum mdelim
398mdoc_isdelim(const char *p)
399{
400
401	if ('\0' == p[0])
402		return DELIM_NONE;
403
404	if ('\0' == p[1])
405		switch (p[0]) {
406		case '(':
407		case '[':
408			return DELIM_OPEN;
409		case '|':
410			return DELIM_MIDDLE;
411		case '.':
412		case ',':
413		case ';':
414		case ':':
415		case '?':
416		case '!':
417		case ')':
418		case ']':
419			return DELIM_CLOSE;
420		default:
421			return DELIM_NONE;
422		}
423
424	if ('\\' != p[0])
425		return DELIM_NONE;
426
427	if (0 == strcmp(p + 1, "."))
428		return DELIM_CLOSE;
429	if (0 == strcmp(p + 1, "fR|\\fP"))
430		return DELIM_MIDDLE;
431
432	return DELIM_NONE;
433}
434