xref: /openbsd-src/usr.bin/mandoc/tbl_data.c (revision ae3cb403620ab940fbaabb3055fac045a63d56b7)
1 /*	$OpenBSD: tbl_data.c,v 1.32 2017/07/08 17:52:42 schwarze Exp $ */
2 /*
3  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4  * Copyright (c) 2011, 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 <assert.h>
21 #include <ctype.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 static	void		 getdata(struct tbl_node *, struct tbl_span *,
32 				int, const char *, int *);
33 static	struct tbl_span	*newspan(struct tbl_node *, int,
34 				struct tbl_row *);
35 
36 
37 static void
38 getdata(struct tbl_node *tbl, struct tbl_span *dp,
39 		int ln, const char *p, int *pos)
40 {
41 	struct tbl_dat	*dat;
42 	struct tbl_cell	*cp;
43 	int		 sv;
44 
45 	/* Advance to the next layout cell, skipping spanners. */
46 
47 	cp = dp->last == NULL ? dp->layout->first : dp->last->layout->next;
48 	while (cp != NULL && cp->pos == TBL_CELL_SPAN)
49 		cp = cp->next;
50 
51 	/*
52 	 * If the current layout row is out of cells, allocate
53 	 * a new cell if another row of the table has at least
54 	 * this number of columns, or discard the input if we
55 	 * are beyond the last column of the table as a whole.
56 	 */
57 
58 	if (cp == NULL) {
59 		if (dp->layout->last->col + 1 < dp->opts->cols) {
60 			cp = mandoc_calloc(1, sizeof(*cp));
61 			cp->pos = TBL_CELL_LEFT;
62 			dp->layout->last->next = cp;
63 			cp->col = dp->layout->last->col + 1;
64 			dp->layout->last = cp;
65 		} else {
66 			mandoc_msg(MANDOCERR_TBLDATA_EXTRA, tbl->parse,
67 			    ln, *pos, p + *pos);
68 			while (p[*pos])
69 				(*pos)++;
70 			return;
71 		}
72 	}
73 
74 	dat = mandoc_calloc(1, sizeof(*dat));
75 	dat->layout = cp;
76 	dat->pos = TBL_DATA_NONE;
77 	dat->spans = 0;
78 	for (cp = cp->next; cp != NULL; cp = cp->next)
79 		if (cp->pos == TBL_CELL_SPAN)
80 			dat->spans++;
81 		else
82 			break;
83 
84 	if (dp->last == NULL)
85 		dp->first = dat;
86 	else
87 		dp->last->next = dat;
88 	dp->last = dat;
89 
90 	sv = *pos;
91 	while (p[*pos] && p[*pos] != tbl->opts.tab)
92 		(*pos)++;
93 
94 	/*
95 	 * Check for a continued-data scope opening.  This consists of a
96 	 * trailing `T{' at the end of the line.  Subsequent lines,
97 	 * until a standalone `T}', are included in our cell.
98 	 */
99 
100 	if (*pos - sv == 2 && p[sv] == 'T' && p[sv + 1] == '{') {
101 		tbl->part = TBL_PART_CDATA;
102 		return;
103 	}
104 
105 	dat->string = mandoc_strndup(p + sv, *pos - sv);
106 
107 	if (p[*pos])
108 		(*pos)++;
109 
110 	if ( ! strcmp(dat->string, "_"))
111 		dat->pos = TBL_DATA_HORIZ;
112 	else if ( ! strcmp(dat->string, "="))
113 		dat->pos = TBL_DATA_DHORIZ;
114 	else if ( ! strcmp(dat->string, "\\_"))
115 		dat->pos = TBL_DATA_NHORIZ;
116 	else if ( ! strcmp(dat->string, "\\="))
117 		dat->pos = TBL_DATA_NDHORIZ;
118 	else
119 		dat->pos = TBL_DATA_DATA;
120 
121 	if ((dat->layout->pos == TBL_CELL_HORIZ ||
122 	    dat->layout->pos == TBL_CELL_DHORIZ ||
123 	    dat->layout->pos == TBL_CELL_DOWN) &&
124 	    dat->pos == TBL_DATA_DATA && *dat->string != '\0')
125 		mandoc_msg(MANDOCERR_TBLDATA_SPAN,
126 		    tbl->parse, ln, sv, dat->string);
127 }
128 
129 void
130 tbl_cdata(struct tbl_node *tbl, int ln, const char *p, int pos)
131 {
132 	struct tbl_dat	*dat;
133 	size_t		 sz;
134 
135 	dat = tbl->last_span->last;
136 
137 	if (p[pos] == 'T' && p[pos + 1] == '}') {
138 		pos += 2;
139 		if (p[pos] == tbl->opts.tab) {
140 			tbl->part = TBL_PART_DATA;
141 			pos++;
142 			while (p[pos] != '\0')
143 				getdata(tbl, tbl->last_span, ln, p, &pos);
144 			return;
145 		} else if (p[pos] == '\0') {
146 			tbl->part = TBL_PART_DATA;
147 			return;
148 		}
149 
150 		/* Fallthrough: T} is part of a word. */
151 	}
152 
153 	dat->pos = TBL_DATA_DATA;
154 	dat->block = 1;
155 
156 	if (dat->string != NULL) {
157 		sz = strlen(p + pos) + strlen(dat->string) + 2;
158 		dat->string = mandoc_realloc(dat->string, sz);
159 		(void)strlcat(dat->string, " ", sz);
160 		(void)strlcat(dat->string, p + pos, sz);
161 	} else
162 		dat->string = mandoc_strdup(p + pos);
163 
164 	if (dat->layout->pos == TBL_CELL_DOWN)
165 		mandoc_msg(MANDOCERR_TBLDATA_SPAN, tbl->parse,
166 		    ln, pos, dat->string);
167 }
168 
169 static struct tbl_span *
170 newspan(struct tbl_node *tbl, int line, struct tbl_row *rp)
171 {
172 	struct tbl_span	*dp;
173 
174 	dp = mandoc_calloc(1, sizeof(*dp));
175 	dp->line = line;
176 	dp->opts = &tbl->opts;
177 	dp->layout = rp;
178 	dp->prev = tbl->last_span;
179 
180 	if (dp->prev == NULL) {
181 		tbl->first_span = dp;
182 		tbl->current_span = NULL;
183 	} else
184 		dp->prev->next = dp;
185 	tbl->last_span = dp;
186 
187 	return dp;
188 }
189 
190 void
191 tbl_data(struct tbl_node *tbl, int ln, const char *p, int pos)
192 {
193 	struct tbl_row	*rp;
194 	struct tbl_cell	*cp;
195 	struct tbl_span	*sp;
196 
197 	rp = (sp = tbl->last_span) == NULL ? tbl->first_row :
198 	    sp->pos == TBL_SPAN_DATA && sp->layout->next != NULL ?
199 	    sp->layout->next : sp->layout;
200 
201 	assert(rp != NULL);
202 
203 	if ( ! strcmp(p, "_")) {
204 		sp = newspan(tbl, ln, rp);
205 		sp->pos = TBL_SPAN_HORIZ;
206 		return;
207 	} else if ( ! strcmp(p, "=")) {
208 		sp = newspan(tbl, ln, rp);
209 		sp->pos = TBL_SPAN_DHORIZ;
210 		return;
211 	}
212 
213 	/*
214 	 * If the layout row contains nothing but horizontal lines,
215 	 * allocate an empty span for it and assign the current span
216 	 * to the next layout row accepting data.
217 	 */
218 
219 	while (rp->next != NULL) {
220 		if (rp->last->col + 1 < tbl->opts.cols)
221 			break;
222 		for (cp = rp->first; cp != NULL; cp = cp->next)
223 			if (cp->pos != TBL_CELL_HORIZ &&
224 			    cp->pos != TBL_CELL_DHORIZ)
225 				break;
226 		if (cp != NULL)
227 			break;
228 		sp = newspan(tbl, ln, rp);
229 		sp->pos = TBL_SPAN_DATA;
230 		rp = rp->next;
231 	}
232 
233 	/* Process a real data row. */
234 
235 	sp = newspan(tbl, ln, rp);
236 	sp->pos = TBL_SPAN_DATA;
237 	while (p[pos] != '\0')
238 		getdata(tbl, sp, ln, p, &pos);
239 }
240