xref: /illumos-gate/usr/src/cmd/mandoc/tbl_layout.c (revision 67fa3f2c31312dc0caac188f568ab1fdc6b27295) 
1 /*	$Id: tbl_layout.c,v 1.50 2021/08/10 12:55:04 schwarze Exp $ */
2 /*
3  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012, 2014, 2015, 2017, 2020, 2021
5  *               Ingo Schwarze <schwarze@openbsd.org>
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 	size_t		 sz;
71 	int		 isz;
72 	enum mandoc_esc	 fontesc;
73 
74 mod:
75 	while (p[*pos] == ' ' || p[*pos] == '\t')
76 		(*pos)++;
77 
78 	/* Row delimiters and cell specifiers end modifier lists. */
79 
80 	if (strchr(".,-=^_ACLNRSaclnrs", p[*pos]) != NULL)
81 		return;
82 
83 	/* Throw away parenthesised expression. */
84 
85 	if ('(' == p[*pos]) {
86 		(*pos)++;
87 		while (p[*pos] && ')' != p[*pos])
88 			(*pos)++;
89 		if (')' == p[*pos]) {
90 			(*pos)++;
91 			goto mod;
92 		}
93 		mandoc_msg(MANDOCERR_TBLLAYOUT_PAR, ln, *pos, NULL);
94 		return;
95 	}
96 
97 	/* Parse numerical spacing from modifier string. */
98 
99 	if (isdigit((unsigned char)p[*pos])) {
100 		if ((spacing = strtoul(p + *pos, &endptr, 10)) > 9)
101 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPC, ln, *pos,
102 			    "%lu", spacing);
103 		else
104 			cp->spacing = spacing;
105 		*pos = endptr - p;
106 		goto mod;
107 	}
108 
109 	switch (tolower((unsigned char)p[(*pos)++])) {
110 	case 'b':
111 		cp->font = ESCAPE_FONTBOLD;
112 		goto mod;
113 	case 'd':
114 		cp->flags |= TBL_CELL_BALIGN;
115 		goto mod;
116 	case 'e':
117 		cp->flags |= TBL_CELL_EQUAL;
118 		goto mod;
119 	case 'f':
120 		break;
121 	case 'i':
122 		cp->font = ESCAPE_FONTITALIC;
123 		goto mod;
124 	case 'm':
125 		mandoc_msg(MANDOCERR_TBLLAYOUT_MOD, ln, *pos, "m");
126 		goto mod;
127 	case 'p':
128 	case 'v':
129 		if (p[*pos] == '-' || p[*pos] == '+')
130 			(*pos)++;
131 		while (isdigit((unsigned char)p[*pos]))
132 			(*pos)++;
133 		goto mod;
134 	case 't':
135 		cp->flags |= TBL_CELL_TALIGN;
136 		goto mod;
137 	case 'u':
138 		cp->flags |= TBL_CELL_UP;
139 		goto mod;
140 	case 'w':
141 		sz = 0;
142 		if (p[*pos] == '(') {
143 			(*pos)++;
144 			while (p[*pos + sz] != '\0' && p[*pos + sz] != ')')
145 				sz++;
146 		} else
147 			while (isdigit((unsigned char)p[*pos + sz]))
148 				sz++;
149 		if (sz) {
150 			free(cp->wstr);
151 			cp->wstr = mandoc_strndup(p + *pos, sz);
152 			*pos += sz;
153 			if (p[*pos] == ')')
154 				(*pos)++;
155 		}
156 		goto mod;
157 	case 'x':
158 		cp->flags |= TBL_CELL_WMAX;
159 		goto mod;
160 	case 'z':
161 		cp->flags |= TBL_CELL_WIGN;
162 		goto mod;
163 	case '|':
164 		if (cp->vert < 2)
165 			cp->vert++;
166 		else
167 			mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
168 			    ln, *pos - 1, NULL);
169 		goto mod;
170 	default:
171 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
172 		    ln, *pos - 1, "%c", p[*pos - 1]);
173 		goto mod;
174 	}
175 
176 	while (p[*pos] == ' ' || p[*pos] == '\t')
177 		(*pos)++;
178 
179 	/* Ignore parenthised font names for now. */
180 
181 	if (p[*pos] == '(')
182 		goto mod;
183 
184 	isz = 0;
185 	if (p[*pos] != '\0')
186 		isz++;
187 	if (strchr(" \t.", p[*pos + isz]) == NULL)
188 		isz++;
189 
190 	fontesc = mandoc_font(p + *pos, isz);
191 
192 	switch (fontesc) {
193 	case ESCAPE_FONTPREV:
194 	case ESCAPE_ERROR:
195 		mandoc_msg(MANDOCERR_FT_BAD,
196 		    ln, *pos, "TS %s", p + *pos - 1);
197 		break;
198 	default:
199 		cp->font = fontesc;
200 		break;
201 	}
202 	*pos += isz;
203 	goto mod;
204 }
205 
206 static void
207 cell(struct tbl_node *tbl, struct tbl_row *rp,
208 		int ln, const char *p, int *pos)
209 {
210 	int		 i;
211 	enum tbl_cellt	 c;
212 
213 	/* Handle leading vertical lines */
214 
215 	while (p[*pos] == ' ' || p[*pos] == '\t' || p[*pos] == '|') {
216 		if (p[*pos] == '|') {
217 			if (rp->vert < 2)
218 				rp->vert++;
219 			else
220 				mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
221 				    ln, *pos, NULL);
222 		}
223 		(*pos)++;
224 	}
225 
226 again:
227 	while (p[*pos] == ' ' || p[*pos] == '\t')
228 		(*pos)++;
229 
230 	if (p[*pos] == '.' || p[*pos] == '\0')
231 		return;
232 
233 	/* Parse the column position (`c', `l', `r', ...). */
234 
235 	for (i = 0; i < KEYS_MAX; i++)
236 		if (tolower((unsigned char)p[*pos]) == keys[i].name)
237 			break;
238 
239 	if (i == KEYS_MAX) {
240 		mandoc_msg(MANDOCERR_TBLLAYOUT_CHAR,
241 		    ln, *pos, "%c", p[*pos]);
242 		(*pos)++;
243 		goto again;
244 	}
245 	c = keys[i].key;
246 
247 	/* Special cases of spanners. */
248 
249 	if (c == TBL_CELL_SPAN) {
250 		if (rp->last == NULL)
251 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPAN, ln, *pos, NULL);
252 		else if (rp->last->pos == TBL_CELL_HORIZ ||
253 		    rp->last->pos == TBL_CELL_DHORIZ)
254 			c = rp->last->pos;
255 	} else if (c == TBL_CELL_DOWN && rp == tbl->first_row)
256 		mandoc_msg(MANDOCERR_TBLLAYOUT_DOWN, ln, *pos, NULL);
257 
258 	(*pos)++;
259 
260 	/* Allocate cell then parse its modifiers. */
261 
262 	mods(tbl, cell_alloc(tbl, rp, c), ln, p, pos);
263 }
264 
265 void
266 tbl_layout(struct tbl_node *tbl, int ln, const char *p, int pos)
267 {
268 	struct tbl_row	*rp;
269 
270 	rp = NULL;
271 	for (;;) {
272 		/* Skip whitespace before and after each cell. */
273 
274 		while (p[pos] == ' ' || p[pos] == '\t')
275 			pos++;
276 
277 		switch (p[pos]) {
278 		case ',':  /* Next row on this input line. */
279 			pos++;
280 			rp = NULL;
281 			continue;
282 		case '\0':  /* Next row on next input line. */
283 			return;
284 		case '.':  /* End of layout. */
285 			pos++;
286 			tbl->part = TBL_PART_DATA;
287 
288 			/*
289 			 * When the layout is completely empty,
290 			 * default to one left-justified column.
291 			 */
292 
293 			if (tbl->first_row == NULL) {
294 				tbl->first_row = tbl->last_row =
295 				    mandoc_calloc(1, sizeof(*rp));
296 			}
297 			if (tbl->first_row->first == NULL) {
298 				mandoc_msg(MANDOCERR_TBLLAYOUT_NONE,
299 				    ln, pos, NULL);
300 				cell_alloc(tbl, tbl->first_row,
301 				    TBL_CELL_LEFT);
302 				if (tbl->opts.lvert < tbl->first_row->vert)
303 					tbl->opts.lvert = tbl->first_row->vert;
304 				return;
305 			}
306 
307 			/*
308 			 * Search for the widest line
309 			 * along the left and right margins.
310 			 */
311 
312 			for (rp = tbl->first_row; rp; rp = rp->next) {
313 				if (tbl->opts.lvert < rp->vert)
314 					tbl->opts.lvert = rp->vert;
315 				if (rp->last != NULL &&
316 				    rp->last->col + 1 == tbl->opts.cols &&
317 				    tbl->opts.rvert < rp->last->vert)
318 					tbl->opts.rvert = rp->last->vert;
319 
320 				/* If the last line is empty, drop it. */
321 
322 				if (rp->next != NULL &&
323 				    rp->next->first == NULL) {
324 					free(rp->next);
325 					rp->next = NULL;
326 					tbl->last_row = rp;
327 				}
328 			}
329 			return;
330 		default:  /* Cell. */
331 			break;
332 		}
333 
334 		/*
335 		 * If the last line had at least one cell,
336 		 * start a new one; otherwise, continue it.
337 		 */
338 
339 		if (rp == NULL) {
340 			if (tbl->last_row == NULL ||
341 			    tbl->last_row->first != NULL) {
342 				rp = mandoc_calloc(1, sizeof(*rp));
343 				if (tbl->last_row)
344 					tbl->last_row->next = rp;
345 				else
346 					tbl->first_row = rp;
347 				tbl->last_row = rp;
348 			} else
349 				rp = tbl->last_row;
350 		}
351 		cell(tbl, rp, ln, p, &pos);
352 	}
353 }
354 
355 static struct tbl_cell *
356 cell_alloc(struct tbl_node *tbl, struct tbl_row *rp, enum tbl_cellt pos)
357 {
358 	struct tbl_cell	*p, *pp;
359 
360 	p = mandoc_calloc(1, sizeof(*p));
361 	p->spacing = SIZE_MAX;
362 	p->font = ESCAPE_FONTROMAN;
363 	p->pos = pos;
364 
365 	if ((pp = rp->last) != NULL) {
366 		pp->next = p;
367 		p->col = pp->col + 1;
368 	} else
369 		rp->first = p;
370 	rp->last = p;
371 
372 	if (tbl->opts.cols <= p->col)
373 		tbl->opts.cols = p->col + 1;
374 
375 	return p;
376 }
377