xref: /openbsd-src/usr.bin/mandoc/tbl_layout.c (revision 60395358bd468881069c44744132008b5b82170f) 
1 /*	$OpenBSD: tbl_layout.c,v 1.38 2025/01/05 18:03:51 schwarze Exp $ */
2 /*
3  * Copyright (c) 2012, 2014, 2015, 2017, 2020, 2021, 2025
4  *               Ingo Schwarze <schwarze@openbsd.org>
5  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
6  *
7  * Permission to use, copy, modify, and distribute this software for any
8  * purpose with or without fee is hereby granted, provided that the above
9  * copyright notice and this permission notice appear in all copies.
10  *
11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18  */
19 #include <sys/types.h>
20 
21 #include <ctype.h>
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <time.h>
27 
28 #include "mandoc_aux.h"
29 #include "mandoc.h"
30 #include "tbl.h"
31 #include "libmandoc.h"
32 #include "tbl_int.h"
33 
34 struct	tbl_phrase {
35 	char		 name;
36 	enum tbl_cellt	 key;
37 };
38 
39 static	const struct tbl_phrase keys[] = {
40 	{ 'c',		 TBL_CELL_CENTRE },
41 	{ 'r',		 TBL_CELL_RIGHT },
42 	{ 'l',		 TBL_CELL_LEFT },
43 	{ 'n',		 TBL_CELL_NUMBER },
44 	{ 's',		 TBL_CELL_SPAN },
45 	{ 'a',		 TBL_CELL_LONG },
46 	{ '^',		 TBL_CELL_DOWN },
47 	{ '-',		 TBL_CELL_HORIZ },
48 	{ '_',		 TBL_CELL_HORIZ },
49 	{ '=',		 TBL_CELL_DHORIZ }
50 };
51 
52 #define KEYS_MAX ((int)(sizeof(keys)/sizeof(keys[0])))
53 
54 static	void		 mods(struct tbl_node *, struct tbl_cell *,
55 				int, const char *, int *);
56 static	void		 cell(struct tbl_node *, struct tbl_row *,
57 				int, const char *, int *);
58 static	struct tbl_cell *cell_alloc(struct tbl_node *, struct tbl_row *,
59 				enum tbl_cellt);
60 
61 
62 static void
63 mods(struct tbl_node *tbl, struct tbl_cell *cp,
64 		int ln, const char *p, int *pos)
65 {
66 	char		*endptr;
67 	unsigned long	 spacing;
68 	int		 isz;
69 	enum mandoc_esc	 fontesc;
70 
71 mod:
72 	while (p[*pos] == ' ' || p[*pos] == '\t')
73 		(*pos)++;
74 
75 	/* Row delimiters and cell specifiers end modifier lists. */
76 
77 	if (strchr(".,-=^_ACLNRSaclnrs", p[*pos]) != NULL)
78 		return;
79 
80 	/* Throw away parenthesised expression. */
81 
82 	if ('(' == p[*pos]) {
83 		(*pos)++;
84 		while (p[*pos] && ')' != p[*pos])
85 			(*pos)++;
86 		if (')' == p[*pos]) {
87 			(*pos)++;
88 			goto mod;
89 		}
90 		mandoc_msg(MANDOCERR_TBLLAYOUT_PAR, ln, *pos, NULL);
91 		return;
92 	}
93 
94 	/* Parse numerical spacing from modifier string. */
95 
96 	if (isdigit((unsigned char)p[*pos])) {
97 		if ((spacing = strtoul(p + *pos, &endptr, 10)) > 9)
98 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPC, ln, *pos,
99 			    "%lu", spacing);
100 		else
101 			cp->spacing = spacing;
102 		*pos = endptr - p;
103 		goto mod;
104 	}
105 
106 	switch (tolower((unsigned char)p[(*pos)++])) {
107 	case 'b':
108 		cp->font = ESCAPE_FONTBOLD;
109 		goto mod;
110 	case 'd':
111 		cp->flags |= TBL_CELL_BALIGN;
112 		goto mod;
113 	case 'e':
114 		cp->flags |= TBL_CELL_EQUAL;
115 		goto mod;
116 	case 'f':
117 		break;
118 	case 'i':
119 		cp->font = ESCAPE_FONTITALIC;
120 		goto mod;
121 	case 'm':
122 		mandoc_msg(MANDOCERR_TBLLAYOUT_MOD, ln, *pos, "m");
123 		goto mod;
124 	case 'p':
125 	case 'v':
126 		if (p[*pos] == '-' || p[*pos] == '+')
127 			(*pos)++;
128 		while (isdigit((unsigned char)p[*pos]))
129 			(*pos)++;
130 		goto mod;
131 	case 't':
132 		cp->flags |= TBL_CELL_TALIGN;
133 		goto mod;
134 	case 'u':
135 		cp->flags |= TBL_CELL_UP;
136 		goto mod;
137 	case 'w':
138 		if (p[*pos] == '(') {
139 			(*pos)++;
140 			isz = 0;
141 			if (roff_evalnum(ln, p, pos, &isz, 'n', 1) == 0 ||
142 			    p[*pos] != ')')
143 				mandoc_msg(MANDOCERR_TBLLAYOUT_WIDTH,
144 				    ln, *pos, "%s", p + *pos);
145 			else {
146 				/* Convert from BU to EN and round. */
147 				cp->width = (isz + 11) /24;
148 				(*pos)++;
149 			}
150 		} else {
151 			cp->width = 0;
152 			while (isdigit((unsigned char)p[*pos])) {
153 				cp->width *= 10;
154 				cp->width += p[(*pos)++] - '0';
155 			}
156 			if (cp->width == 0)
157 				mandoc_msg(MANDOCERR_TBLLAYOUT_WIDTH,
158 				    ln, *pos, "%s", p + *pos);
159 		}
160 		goto mod;
161 	case 'x':
162 		cp->flags |= TBL_CELL_WMAX;
163 		goto mod;
164 	case 'z':
165 		cp->flags |= TBL_CELL_WIGN;
166 		goto mod;
167 	case '|':
168 		if (cp->vert < 2)
169 			cp->vert++;
170 		else
171 			mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
172 			    ln, *pos - 1, NULL);
173 		goto mod;
174 	default:
175 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
176 		    ln, *pos - 1, "%c", p[*pos - 1]);
177 		goto mod;
178 	}
179 
180 	while (p[*pos] == ' ' || p[*pos] == '\t')
181 		(*pos)++;
182 
183 	/* Ignore parenthised font names for now. */
184 
185 	if (p[*pos] == '(')
186 		goto mod;
187 
188 	isz = 0;
189 	if (p[*pos] != '\0')
190 		isz++;
191 	if (strchr(" \t.", p[*pos + isz]) == NULL)
192 		isz++;
193 
194 	fontesc = mandoc_font(p + *pos, isz);
195 
196 	switch (fontesc) {
197 	case ESCAPE_FONTPREV:
198 	case ESCAPE_ERROR:
199 		mandoc_msg(MANDOCERR_FT_BAD,
200 		    ln, *pos, "TS %s", p + *pos - 1);
201 		break;
202 	default:
203 		cp->font = fontesc;
204 		break;
205 	}
206 	*pos += isz;
207 	goto mod;
208 }
209 
210 static void
211 cell(struct tbl_node *tbl, struct tbl_row *rp,
212 		int ln, const char *p, int *pos)
213 {
214 	int		 i;
215 	enum tbl_cellt	 c;
216 
217 	/* Handle leading vertical lines */
218 
219 	while (p[*pos] == ' ' || p[*pos] == '\t' || p[*pos] == '|') {
220 		if (p[*pos] == '|') {
221 			if (rp->vert < 2)
222 				rp->vert++;
223 			else
224 				mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
225 				    ln, *pos, NULL);
226 		}
227 		(*pos)++;
228 	}
229 
230 again:
231 	while (p[*pos] == ' ' || p[*pos] == '\t')
232 		(*pos)++;
233 
234 	if (p[*pos] == '.' || p[*pos] == '\0')
235 		return;
236 
237 	/* Parse the column position (`c', `l', `r', ...). */
238 
239 	for (i = 0; i < KEYS_MAX; i++)
240 		if (tolower((unsigned char)p[*pos]) == keys[i].name)
241 			break;
242 
243 	if (i == KEYS_MAX) {
244 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
245 		    ln, *pos, "%c", p[*pos]);
246 		(*pos)++;
247 		goto again;
248 	}
249 	c = keys[i].key;
250 
251 	/* Special cases of spanners. */
252 
253 	if (c == TBL_CELL_SPAN) {
254 		if (rp->last == NULL)
255 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPAN, ln, *pos, NULL);
256 		else if (rp->last->pos == TBL_CELL_HORIZ ||
257 		    rp->last->pos == TBL_CELL_DHORIZ)
258 			c = rp->last->pos;
259 	} else if (c == TBL_CELL_DOWN && rp == tbl->first_row)
260 		mandoc_msg(MANDOCERR_TBLLAYOUT_DOWN, ln, *pos, NULL);
261 
262 	(*pos)++;
263 
264 	/* Allocate cell then parse its modifiers. */
265 
266 	mods(tbl, cell_alloc(tbl, rp, c), ln, p, pos);
267 }
268 
269 void
270 tbl_layout(struct tbl_node *tbl, int ln, const char *p, int pos)
271 {
272 	struct tbl_row	*rp;
273 
274 	rp = NULL;
275 	for (;;) {
276 		/* Skip whitespace before and after each cell. */
277 
278 		while (p[pos] == ' ' || p[pos] == '\t')
279 			pos++;
280 
281 		switch (p[pos]) {
282 		case ',':  /* Next row on this input line. */
283 			pos++;
284 			rp = NULL;
285 			continue;
286 		case '\0':  /* Next row on next input line. */
287 			return;
288 		case '.':  /* End of layout. */
289 			pos++;
290 			tbl->part = TBL_PART_DATA;
291 
292 			/*
293 			 * When the layout is completely empty,
294 			 * default to one left-justified column.
295 			 */
296 
297 			if (tbl->first_row == NULL) {
298 				tbl->first_row = tbl->last_row =
299 				    mandoc_calloc(1, sizeof(*rp));
300 			}
301 			if (tbl->first_row->first == NULL) {
302 				mandoc_msg(MANDOCERR_TBLLAYOUT_NONE,
303 				    ln, pos, NULL);
304 				cell_alloc(tbl, tbl->first_row,
305 				    TBL_CELL_LEFT);
306 				if (tbl->opts.lvert < tbl->first_row->vert)
307 					tbl->opts.lvert = tbl->first_row->vert;
308 				return;
309 			}
310 
311 			/*
312 			 * Search for the widest line
313 			 * along the left and right margins.
314 			 */
315 
316 			for (rp = tbl->first_row; rp; rp = rp->next) {
317 				if (tbl->opts.lvert < rp->vert)
318 					tbl->opts.lvert = rp->vert;
319 				if (rp->last != NULL &&
320 				    rp->last->col + 1 == tbl->opts.cols &&
321 				    tbl->opts.rvert < rp->last->vert)
322 					tbl->opts.rvert = rp->last->vert;
323 
324 				/* If the last line is empty, drop it. */
325 
326 				if (rp->next != NULL &&
327 				    rp->next->first == NULL) {
328 					free(rp->next);
329 					rp->next = NULL;
330 					tbl->last_row = rp;
331 				}
332 			}
333 			return;
334 		default:  /* Cell. */
335 			break;
336 		}
337 
338 		/*
339 		 * If the last line had at least one cell,
340 		 * start a new one; otherwise, continue it.
341 		 */
342 
343 		if (rp == NULL) {
344 			if (tbl->last_row == NULL ||
345 			    tbl->last_row->first != NULL) {
346 				rp = mandoc_calloc(1, sizeof(*rp));
347 				if (tbl->last_row)
348 					tbl->last_row->next = rp;
349 				else
350 					tbl->first_row = rp;
351 				tbl->last_row = rp;
352 			} else
353 				rp = tbl->last_row;
354 		}
355 		cell(tbl, rp, ln, p, &pos);
356 	}
357 }
358 
359 static struct tbl_cell *
360 cell_alloc(struct tbl_node *tbl, struct tbl_row *rp, enum tbl_cellt pos)
361 {
362 	struct tbl_cell	*p, *pp;
363 
364 	p = mandoc_calloc(1, sizeof(*p));
365 	p->spacing = SIZE_MAX;
366 	p->font = ESCAPE_FONTROMAN;
367 	p->pos = pos;
368 
369 	if ((pp = rp->last) != NULL) {
370 		pp->next = p;
371 		p->col = pp->col + 1;
372 	} else
373 		rp->first = p;
374 	rp->last = p;
375 
376 	if (tbl->opts.cols <= p->col)
377 		tbl->opts.cols = p->col + 1;
378 
379 	return p;
380 }
381