xref: /openbsd/usr.bin/mandoc/tbl_layout.c (revision 6f40fd34) 
1 /*	$OpenBSD: tbl_layout.c,v 1.31 2017/06/27 18:23:29 schwarze Exp $ */
2 /*
3  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2012, 2014, 2015, 2017 Ingo Schwarze <schwarze@openbsd.org>
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 AUTHOR DISCLAIMS ALL WARRANTIES
11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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 #include <sys/types.h>
19 
20 #include <ctype.h>
21 #include <stdint.h>
22 #include <stdlib.h>
23 #include <string.h>
24 #include <time.h>
25 
26 #include "mandoc.h"
27 #include "mandoc_aux.h"
28 #include "libmandoc.h"
29 #include "libroff.h"
30 
31 struct	tbl_phrase {
32 	char		 name;
33 	enum tbl_cellt	 key;
34 };
35 
36 static	const struct tbl_phrase keys[] = {
37 	{ 'c',		 TBL_CELL_CENTRE },
38 	{ 'r',		 TBL_CELL_RIGHT },
39 	{ 'l',		 TBL_CELL_LEFT },
40 	{ 'n',		 TBL_CELL_NUMBER },
41 	{ 's',		 TBL_CELL_SPAN },
42 	{ 'a',		 TBL_CELL_LONG },
43 	{ '^',		 TBL_CELL_DOWN },
44 	{ '-',		 TBL_CELL_HORIZ },
45 	{ '_',		 TBL_CELL_HORIZ },
46 	{ '=',		 TBL_CELL_DHORIZ }
47 };
48 
49 #define KEYS_MAX ((int)(sizeof(keys)/sizeof(keys[0])))
50 
51 static	void		 mods(struct tbl_node *, struct tbl_cell *,
52 				int, const char *, int *);
53 static	void		 cell(struct tbl_node *, struct tbl_row *,
54 				int, const char *, int *);
55 static	struct tbl_cell *cell_alloc(struct tbl_node *, struct tbl_row *,
56 				enum tbl_cellt);
57 
58 
59 static void
60 mods(struct tbl_node *tbl, struct tbl_cell *cp,
61 		int ln, const char *p, int *pos)
62 {
63 	char		*endptr;
64 	size_t		 sz;
65 
66 mod:
67 	while (p[*pos] == ' ' || p[*pos] == '\t')
68 		(*pos)++;
69 
70 	/* Row delimiters and cell specifiers end modifier lists. */
71 
72 	if (strchr(".,-=^_ACLNRSaclnrs", p[*pos]) != NULL)
73 		return;
74 
75 	/* Throw away parenthesised expression. */
76 
77 	if ('(' == p[*pos]) {
78 		(*pos)++;
79 		while (p[*pos] && ')' != p[*pos])
80 			(*pos)++;
81 		if (')' == p[*pos]) {
82 			(*pos)++;
83 			goto mod;
84 		}
85 		mandoc_msg(MANDOCERR_TBLLAYOUT_PAR, tbl->parse,
86 		    ln, *pos, NULL);
87 		return;
88 	}
89 
90 	/* Parse numerical spacing from modifier string. */
91 
92 	if (isdigit((unsigned char)p[*pos])) {
93 		cp->spacing = strtoull(p + *pos, &endptr, 10);
94 		*pos = endptr - p;
95 		goto mod;
96 	}
97 
98 	switch (tolower((unsigned char)p[(*pos)++])) {
99 	case 'b':
100 		cp->flags |= TBL_CELL_BOLD;
101 		goto mod;
102 	case 'd':
103 		cp->flags |= TBL_CELL_BALIGN;
104 		goto mod;
105 	case 'e':
106 		cp->flags |= TBL_CELL_EQUAL;
107 		goto mod;
108 	case 'f':
109 		break;
110 	case 'i':
111 		cp->flags |= TBL_CELL_ITALIC;
112 		goto mod;
113 	case 'm':
114 		mandoc_msg(MANDOCERR_TBLLAYOUT_MOD, tbl->parse,
115 		    ln, *pos, "m");
116 		goto mod;
117 	case 'p':
118 	case 'v':
119 		if (p[*pos] == '-' || p[*pos] == '+')
120 			(*pos)++;
121 		while (isdigit((unsigned char)p[*pos]))
122 			(*pos)++;
123 		goto mod;
124 	case 't':
125 		cp->flags |= TBL_CELL_TALIGN;
126 		goto mod;
127 	case 'u':
128 		cp->flags |= TBL_CELL_UP;
129 		goto mod;
130 	case 'w':
131 		sz = 0;
132 		if (p[*pos] == '(') {
133 			(*pos)++;
134 			while (p[*pos + sz] != '\0' && p[*pos + sz] != ')')
135 				sz++;
136 		} else
137 			while (isdigit((unsigned char)p[*pos + sz]))
138 				sz++;
139 		if (sz) {
140 			free(cp->wstr);
141 			cp->wstr = mandoc_strndup(p + *pos, sz);
142 			*pos += sz;
143 			if (p[*pos] == ')')
144 				(*pos)++;
145 		}
146 		goto mod;
147 	case 'x':
148 		cp->flags |= TBL_CELL_WMAX;
149 		goto mod;
150 	case 'z':
151 		cp->flags |= TBL_CELL_WIGN;
152 		goto mod;
153 	case '|':
154 		if (cp->vert < 2)
155 			cp->vert++;
156 		else
157 			mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
158 			    tbl->parse, ln, *pos - 1, NULL);
159 		goto mod;
160 	default:
161 		mandoc_vmsg(MANDOCERR_TBLLAYOUT_CHAR, tbl->parse,
162 		    ln, *pos - 1, "%c", p[*pos - 1]);
163 		goto mod;
164 	}
165 
166 	/* Ignore parenthised font names for now. */
167 
168 	if (p[*pos] == '(')
169 		goto mod;
170 
171 	/* Support only one-character font-names for now. */
172 
173 	if (p[*pos] == '\0' || (p[*pos + 1] != ' ' && p[*pos + 1] != '.')) {
174 		mandoc_vmsg(MANDOCERR_FT_BAD, tbl->parse,
175 		    ln, *pos, "TS %s", p + *pos - 1);
176 		if (p[*pos] != '\0')
177 			(*pos)++;
178 		if (p[*pos] != '\0')
179 			(*pos)++;
180 		goto mod;
181 	}
182 
183 	switch (p[(*pos)++]) {
184 	case '3':
185 	case 'B':
186 		cp->flags |= TBL_CELL_BOLD;
187 		goto mod;
188 	case '2':
189 	case 'I':
190 		cp->flags |= TBL_CELL_ITALIC;
191 		goto mod;
192 	case '1':
193 	case 'R':
194 		goto mod;
195 	default:
196 		mandoc_vmsg(MANDOCERR_FT_BAD, tbl->parse,
197 		    ln, *pos - 1, "TS f%c", p[*pos - 1]);
198 		goto mod;
199 	}
200 }
201 
202 static void
203 cell(struct tbl_node *tbl, struct tbl_row *rp,
204 		int ln, const char *p, int *pos)
205 {
206 	int		 i;
207 	enum tbl_cellt	 c;
208 
209 	/* Handle leading vertical lines */
210 
211 	while (p[*pos] == ' ' || p[*pos] == '\t' || p[*pos] == '|') {
212 		if (p[*pos] == '|') {
213 			if (rp->vert < 2)
214 				rp->vert++;
215 			else
216 				mandoc_msg(MANDOCERR_TBLLAYOUT_VERT,
217 				    tbl->parse, ln, *pos, NULL);
218 		}
219 		(*pos)++;
220 	}
221 
222 again:
223 	while (p[*pos] == ' ' || p[*pos] == '\t')
224 		(*pos)++;
225 
226 	if (p[*pos] == '.' || p[*pos] == '\0')
227 		return;
228 
229 	/* Parse the column position (`c', `l', `r', ...). */
230 
231 	for (i = 0; i < KEYS_MAX; i++)
232 		if (tolower((unsigned char)p[*pos]) == keys[i].name)
233 			break;
234 
235 	if (i == KEYS_MAX) {
236 		mandoc_vmsg(MANDOCERR_TBLLAYOUT_CHAR, tbl->parse,
237 		    ln, *pos, "%c", p[*pos]);
238 		(*pos)++;
239 		goto again;
240 	}
241 	c = keys[i].key;
242 
243 	/* Special cases of spanners. */
244 
245 	if (c == TBL_CELL_SPAN) {
246 		if (rp->last == NULL)
247 			mandoc_msg(MANDOCERR_TBLLAYOUT_SPAN,
248 			    tbl->parse, ln, *pos, NULL);
249 		else if (rp->last->pos == TBL_CELL_HORIZ ||
250 		    rp->last->pos == TBL_CELL_DHORIZ)
251 			c = rp->last->pos;
252 	} else if (c == TBL_CELL_DOWN && rp == tbl->first_row)
253 		mandoc_msg(MANDOCERR_TBLLAYOUT_DOWN,
254 		    tbl->parse, ln, *pos, NULL);
255 
256 	(*pos)++;
257 
258 	/* Allocate cell then parse its modifiers. */
259 
260 	mods(tbl, cell_alloc(tbl, rp, c), ln, p, pos);
261 }
262 
263 void
264 tbl_layout(struct tbl_node *tbl, int ln, const char *p, int pos)
265 {
266 	struct tbl_row	*rp;
267 
268 	rp = NULL;
269 	for (;;) {
270 		/* Skip whitespace before and after each cell. */
271 
272 		while (p[pos] == ' ' || p[pos] == '\t')
273 			pos++;
274 
275 		switch (p[pos]) {
276 		case ',':  /* Next row on this input line. */
277 			pos++;
278 			rp = NULL;
279 			continue;
280 		case '\0':  /* Next row on next input line. */
281 			return;
282 		case '.':  /* End of layout. */
283 			pos++;
284 			tbl->part = TBL_PART_DATA;
285 
286 			/*
287 			 * When the layout is completely empty,
288 			 * default to one left-justified column.
289 			 */
290 
291 			if (tbl->first_row == NULL) {
292 				tbl->first_row = tbl->last_row =
293 				    mandoc_calloc(1, sizeof(*rp));
294 			}
295 			if (tbl->first_row->first == NULL) {
296 				mandoc_msg(MANDOCERR_TBLLAYOUT_NONE,
297 				    tbl->parse, ln, pos, NULL);
298 				cell_alloc(tbl, tbl->first_row,
299 				    TBL_CELL_LEFT);
300 				if (tbl->opts.lvert < tbl->first_row->vert)
301 					tbl->opts.lvert = tbl->first_row->vert;
302 				return;
303 			}
304 
305 			/*
306 			 * Search for the widest line
307 			 * along the left and right margins.
308 			 */
309 
310 			for (rp = tbl->first_row; rp; rp = rp->next) {
311 				if (tbl->opts.lvert < rp->vert)
312 					tbl->opts.lvert = rp->vert;
313 				if (rp->last != NULL &&
314 				    rp->last->col + 1 == tbl->opts.cols &&
315 				    tbl->opts.rvert < rp->last->vert)
316 					tbl->opts.rvert = rp->last->vert;
317 
318 				/* If the last line is empty, drop it. */
319 
320 				if (rp->next != NULL &&
321 				    rp->next->first == NULL) {
322 					free(rp->next);
323 					rp->next = NULL;
324 					tbl->last_row = rp;
325 				}
326 			}
327 			return;
328 		default:  /* Cell. */
329 			break;
330 		}
331 
332 		/*
333 		 * If the last line had at least one cell,
334 		 * start a new one; otherwise, continue it.
335 		 */
336 
337 		if (rp == NULL) {
338 			if (tbl->last_row == NULL ||
339 			    tbl->last_row->first != NULL) {
340 				rp = mandoc_calloc(1, sizeof(*rp));
341 				if (tbl->last_row)
342 					tbl->last_row->next = rp;
343 				else
344 					tbl->first_row = rp;
345 				tbl->last_row = rp;
346 			} else
347 				rp = tbl->last_row;
348 		}
349 		cell(tbl, rp, ln, p, &pos);
350 	}
351 }
352 
353 static struct tbl_cell *
354 cell_alloc(struct tbl_node *tbl, struct tbl_row *rp, enum tbl_cellt pos)
355 {
356 	struct tbl_cell	*p, *pp;
357 
358 	p = mandoc_calloc(1, sizeof(*p));
359 	p->spacing = SIZE_MAX;
360 	p->pos = pos;
361 
362 	if ((pp = rp->last) != NULL) {
363 		pp->next = p;
364 		p->col = pp->col + 1;
365 	} else
366 		rp->first = p;
367 	rp->last = p;
368 
369 	if (tbl->opts.cols <= p->col)
370 		tbl->opts.cols = p->col + 1;
371 
372 	return p;
373 }
374