xref: /freebsd/contrib/mandoc/tbl_layout.c (revision b670c9bafc0e31c7609969bf374b2e80bdc00211) 
1 /*	$Id: tbl_layout.c,v 1.51 2025/01/05 18:14:39 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 "config.h"
20 
21 #include <sys/types.h>
22 
23 #include <ctype.h>
24 #include <stdint.h>
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <time.h>
29 
30 #include "mandoc_aux.h"
31 #include "mandoc.h"
32 #include "tbl.h"
33 #include "libmandoc.h"
34 #include "tbl_int.h"
35 
36 struct	tbl_phrase {
37 	char		 name;
38 	enum tbl_cellt	 key;
39 };
40 
41 static	const struct tbl_phrase keys[] = {
42 	{ 'c',		 TBL_CELL_CENTRE },
43 	{ 'r',		 TBL_CELL_RIGHT },
44 	{ 'l',		 TBL_CELL_LEFT },
45 	{ 'n',		 TBL_CELL_NUMBER },
46 	{ 's',		 TBL_CELL_SPAN },
47 	{ 'a',		 TBL_CELL_LONG },
48 	{ '^',		 TBL_CELL_DOWN },
49 	{ '-',		 TBL_CELL_HORIZ },
50 	{ '_',		 TBL_CELL_HORIZ },
51 	{ '=',		 TBL_CELL_DHORIZ }
52 };
53 
54 #define KEYS_MAX ((int)(sizeof(keys)/sizeof(keys[0])))
55 
56 static	void		 mods(struct tbl_node *, struct tbl_cell *,
57 				int, const char *, int *);
58 static	void		 cell(struct tbl_node *, struct tbl_row *,
59 				int, const char *, int *);
60 static	struct tbl_cell *cell_alloc(struct tbl_node *, struct tbl_row *,
61 				enum tbl_cellt);
62 
63 
64 static void
65 mods(struct tbl_node *tbl, struct tbl_cell *cp,
66 		int ln, const char *p, int *pos)
67 {
68 	char		*endptr;
69 	unsigned long	 spacing;
70 	int		 isz;
71 	enum mandoc_esc	 fontesc;
72 
73 mod:
74 	while (p[*pos] == ' ' || p[*pos] == '\t')
75 		(*pos)++;
76 
77 	/* Row delimiters and cell specifiers end modifier lists. */
78 
79 	if (strchr(".,-=^_ACLNRSaclnrs", p[*pos]) != NULL)
80 		return;
81 
82 	/* Throw away parenthesised expression. */
83 
84 	if ('(' == p[*pos]) {
85 		(*pos)++;
86 		while (p[*pos] && ')' != p[*pos])
87 			(*pos)++;
88 		if (')' == p[*pos]) {
89 			(*pos)++;
90 			goto mod;
91 		}
92 		mandoc_msg(MANDOCERR_TBLLAYOUT_PAR, ln, *pos, NULL);
93 		return;
94 	}
95 
96 	/* Parse numerical spacing from modifier string. */
97 
98 	if (isdigit((unsigned char)p[*pos])) {
99 		if ((spacing = strtoul(p + *pos, &endptr, 10)) > 9)
100 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPC, ln, *pos,
101 			    "%lu", spacing);
102 		else
103 			cp->spacing = spacing;
104 		*pos = endptr - p;
105 		goto mod;
106 	}
107 
108 	switch (tolower((unsigned char)p[(*pos)++])) {
109 	case 'b':
110 		cp->font = ESCAPE_FONTBOLD;
111 		goto mod;
112 	case 'd':
113 		cp->flags |= TBL_CELL_BALIGN;
114 		goto mod;
115 	case 'e':
116 		cp->flags |= TBL_CELL_EQUAL;
117 		goto mod;
118 	case 'f':
119 		break;
120 	case 'i':
121 		cp->font = ESCAPE_FONTITALIC;
122 		goto mod;
123 	case 'm':
124 		mandoc_msg(MANDOCERR_TBLLAYOUT_MOD, ln, *pos, "m");
125 		goto mod;
126 	case 'p':
127 	case 'v':
128 		if (p[*pos] == '-' || p[*pos] == '+')
129 			(*pos)++;
130 		while (isdigit((unsigned char)p[*pos]))
131 			(*pos)++;
132 		goto mod;
133 	case 't':
134 		cp->flags |= TBL_CELL_TALIGN;
135 		goto mod;
136 	case 'u':
137 		cp->flags |= TBL_CELL_UP;
138 		goto mod;
139 	case 'w':
140 		if (p[*pos] == '(') {
141 			(*pos)++;
142 			isz = 0;
143 			if (roff_evalnum(ln, p, pos, &isz, 'n', 1) == 0 ||
144 			    p[*pos] != ')')
145 				mandoc_msg(MANDOCERR_TBLLAYOUT_WIDTH,
146 				    ln, *pos, "%s", p + *pos);
147 			else {
148 				/* Convert from BU to EN and round. */
149 				cp->width = (isz + 11) /24;
150 				(*pos)++;
151 			}
152 		} else {
153 			cp->width = 0;
154 			while (isdigit((unsigned char)p[*pos])) {
155 				cp->width *= 10;
156 				cp->width += p[(*pos)++] - '0';
157 			}
158 			if (cp->width == 0)
159 				mandoc_msg(MANDOCERR_TBLLAYOUT_WIDTH,
160 				    ln, *pos, "%s", p + *pos);
161 		}
162 		goto mod;
163 	case 'x':
164 		cp->flags |= TBL_CELL_WMAX;
165 		goto mod;
166 	case 'z':
167 		cp->flags |= TBL_CELL_WIGN;
168 		goto mod;
169 	case '|':
170 		if (cp->vert < 2)
171 			cp->vert++;
172 		else
173 			mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
174 			    ln, *pos - 1, NULL);
175 		goto mod;
176 	default:
177 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
178 		    ln, *pos - 1, "%c", p[*pos - 1]);
179 		goto mod;
180 	}
181 
182 	while (p[*pos] == ' ' || p[*pos] == '\t')
183 		(*pos)++;
184 
185 	/* Ignore parenthised font names for now. */
186 
187 	if (p[*pos] == '(')
188 		goto mod;
189 
190 	isz = 0;
191 	if (p[*pos] != '\0')
192 		isz++;
193 	if (strchr(" \t.", p[*pos + isz]) == NULL)
194 		isz++;
195 
196 	fontesc = mandoc_font(p + *pos, isz);
197 
198 	switch (fontesc) {
199 	case ESCAPE_FONTPREV:
200 	case ESCAPE_ERROR:
201 		mandoc_msg(MANDOCERR_FT_BAD,
202 		    ln, *pos, "TS %s", p + *pos - 1);
203 		break;
204 	default:
205 		cp->font = fontesc;
206 		break;
207 	}
208 	*pos += isz;
209 	goto mod;
210 }
211 
212 static void
213 cell(struct tbl_node *tbl, struct tbl_row *rp,
214 		int ln, const char *p, int *pos)
215 {
216 	int		 i;
217 	enum tbl_cellt	 c;
218 
219 	/* Handle leading vertical lines */
220 
221 	while (p[*pos] == ' ' || p[*pos] == '\t' || p[*pos] == '|') {
222 		if (p[*pos] == '|') {
223 			if (rp->vert < 2)
224 				rp->vert++;
225 			else
226 				mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
227 				    ln, *pos, NULL);
228 		}
229 		(*pos)++;
230 	}
231 
232 again:
233 	while (p[*pos] == ' ' || p[*pos] == '\t')
234 		(*pos)++;
235 
236 	if (p[*pos] == '.' || p[*pos] == '\0')
237 		return;
238 
239 	/* Parse the column position (`c', `l', `r', ...). */
240 
241 	for (i = 0; i < KEYS_MAX; i++)
242 		if (tolower((unsigned char)p[*pos]) == keys[i].name)
243 			break;
244 
245 	if (i == KEYS_MAX) {
246 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
247 		    ln, *pos, "%c", p[*pos]);
248 		(*pos)++;
249 		goto again;
250 	}
251 	c = keys[i].key;
252 
253 	/* Special cases of spanners. */
254 
255 	if (c == TBL_CELL_SPAN) {
256 		if (rp->last == NULL)
257 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPAN, ln, *pos, NULL);
258 		else if (rp->last->pos == TBL_CELL_HORIZ ||
259 		    rp->last->pos == TBL_CELL_DHORIZ)
260 			c = rp->last->pos;
261 	} else if (c == TBL_CELL_DOWN && rp == tbl->first_row)
262 		mandoc_msg(MANDOCERR_TBLLAYOUT_DOWN, ln, *pos, NULL);
263 
264 	(*pos)++;
265 
266 	/* Allocate cell then parse its modifiers. */
267 
268 	mods(tbl, cell_alloc(tbl, rp, c), ln, p, pos);
269 }
270 
271 void
272 tbl_layout(struct tbl_node *tbl, int ln, const char *p, int pos)
273 {
274 	struct tbl_row	*rp;
275 
276 	rp = NULL;
277 	for (;;) {
278 		/* Skip whitespace before and after each cell. */
279 
280 		while (p[pos] == ' ' || p[pos] == '\t')
281 			pos++;
282 
283 		switch (p[pos]) {
284 		case ',':  /* Next row on this input line. */
285 			pos++;
286 			rp = NULL;
287 			continue;
288 		case '\0':  /* Next row on next input line. */
289 			return;
290 		case '.':  /* End of layout. */
291 			pos++;
292 			tbl->part = TBL_PART_DATA;
293 
294 			/*
295 			 * When the layout is completely empty,
296 			 * default to one left-justified column.
297 			 */
298 
299 			if (tbl->first_row == NULL) {
300 				tbl->first_row = tbl->last_row =
301 				    mandoc_calloc(1, sizeof(*rp));
302 			}
303 			if (tbl->first_row->first == NULL) {
304 				mandoc_msg(MANDOCERR_TBLLAYOUT_NONE,
305 				    ln, pos, NULL);
306 				cell_alloc(tbl, tbl->first_row,
307 				    TBL_CELL_LEFT);
308 				if (tbl->opts.lvert < tbl->first_row->vert)
309 					tbl->opts.lvert = tbl->first_row->vert;
310 				return;
311 			}
312 
313 			/*
314 			 * Search for the widest line
315 			 * along the left and right margins.
316 			 */
317 
318 			for (rp = tbl->first_row; rp; rp = rp->next) {
319 				if (tbl->opts.lvert < rp->vert)
320 					tbl->opts.lvert = rp->vert;
321 				if (rp->last != NULL &&
322 				    rp->last->col + 1 == tbl->opts.cols &&
323 				    tbl->opts.rvert < rp->last->vert)
324 					tbl->opts.rvert = rp->last->vert;
325 
326 				/* If the last line is empty, drop it. */
327 
328 				if (rp->next != NULL &&
329 				    rp->next->first == NULL) {
330 					free(rp->next);
331 					rp->next = NULL;
332 					tbl->last_row = rp;
333 				}
334 			}
335 			return;
336 		default:  /* Cell. */
337 			break;
338 		}
339 
340 		/*
341 		 * If the last line had at least one cell,
342 		 * start a new one; otherwise, continue it.
343 		 */
344 
345 		if (rp == NULL) {
346 			if (tbl->last_row == NULL ||
347 			    tbl->last_row->first != NULL) {
348 				rp = mandoc_calloc(1, sizeof(*rp));
349 				if (tbl->last_row)
350 					tbl->last_row->next = rp;
351 				else
352 					tbl->first_row = rp;
353 				tbl->last_row = rp;
354 			} else
355 				rp = tbl->last_row;
356 		}
357 		cell(tbl, rp, ln, p, &pos);
358 	}
359 }
360 
361 static struct tbl_cell *
362 cell_alloc(struct tbl_node *tbl, struct tbl_row *rp, enum tbl_cellt pos)
363 {
364 	struct tbl_cell	*p, *pp;
365 
366 	p = mandoc_calloc(1, sizeof(*p));
367 	p->spacing = SIZE_MAX;
368 	p->font = ESCAPE_FONTROMAN;
369 	p->pos = pos;
370 
371 	if ((pp = rp->last) != NULL) {
372 		pp->next = p;
373 		p->col = pp->col + 1;
374 	} else
375 		rp->first = p;
376 	rp->last = p;
377 
378 	if (tbl->opts.cols <= p->col)
379 		tbl->opts.cols = p->col + 1;
380 
381 	return p;
382 }
383