GCC Code Coverage Report
Directory: ./ Exec Total Coverage
File: usr.bin/mandoc/dba.c Lines: 193 203 95.1 %
Date: 2017-11-07 Branches: 77 86 89.5 %

Line Branch Exec Source
1
/*	$OpenBSD: dba.c,v 1.7 2017/02/09 18:26:17 schwarze Exp $ */
2
/*
3
 * Copyright (c) 2016, 2017 Ingo Schwarze <schwarze@openbsd.org>
4
 *
5
 * Permission to use, copy, modify, and distribute this software for any
6
 * purpose with or without fee is hereby granted, provided that the above
7
 * copyright notice and this permission notice appear in all copies.
8
 *
9
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16
 *
17
 * Allocation-based version of the mandoc database, for read-write access.
18
 * The interface is defined in "dba.h".
19
 */
20
#include <sys/types.h>
21
#include <endian.h>
22
#include <errno.h>
23
#include <stddef.h>
24
#include <stdint.h>
25
#include <stdlib.h>
26
#include <string.h>
27
#include <unistd.h>
28
29
#include "mandoc_aux.h"
30
#include "mandoc_ohash.h"
31
#include "mansearch.h"
32
#include "dba_write.h"
33
#include "dba_array.h"
34
#include "dba.h"
35
36
struct macro_entry {
37
	struct dba_array	*pages;
38
	char			 value[];
39
};
40
41
static void	*prepend(const char *, char);
42
static void	 dba_pages_write(struct dba_array *);
43
static int	 compare_names(const void *, const void *);
44
static int	 compare_strings(const void *, const void *);
45
46
static struct macro_entry
47
		*get_macro_entry(struct ohash *, const char *, int32_t);
48
static void	 dba_macros_write(struct dba_array *);
49
static void	 dba_macro_write(struct ohash *);
50
static int	 compare_entries(const void *, const void *);
51
52
53
/*** top-level functions **********************************************/
54
55
struct dba *
56
dba_new(int32_t npages)
57
{
58
	struct dba	*dba;
59
	struct ohash	*macro;
60
	int32_t		 im;
61
62
192
	dba = mandoc_malloc(sizeof(*dba));
63
96
	dba->pages = dba_array_new(npages, DBA_GROW);
64
96
	dba->macros = dba_array_new(MACRO_MAX, 0);
65
7104
	for (im = 0; im < MACRO_MAX; im++) {
66
3456
		macro = mandoc_malloc(sizeof(*macro));
67
3456
		mandoc_ohash_init(macro, 4,
68
		    offsetof(struct macro_entry, value));
69
3456
		dba_array_set(dba->macros, im, macro);
70
	}
71
96
	return dba;
72
}
73
74
void
75
dba_free(struct dba *dba)
76
{
77
	struct dba_array	*page;
78
	struct ohash		*macro;
79
	struct macro_entry	*entry;
80
192
	unsigned int		 slot;
81
82
7104
	dba_array_FOREACH(dba->macros, macro) {
83
272626
		for (entry = ohash_first(macro, &slot); entry != NULL;
84
132857
		     entry = ohash_next(macro, &slot)) {
85
132857
			dba_array_free(entry->pages);
86
132857
			free(entry);
87
		}
88
3456
		ohash_delete(macro);
89
3456
		free(macro);
90
	}
91
96
	dba_array_free(dba->macros);
92
93
96
	dba_array_undel(dba->pages);
94
30426
	dba_array_FOREACH(dba->pages, page) {
95
15117
		dba_array_free(dba_array_get(page, DBP_NAME));
96
15117
		dba_array_free(dba_array_get(page, DBP_SECT));
97
15117
		dba_array_free(dba_array_get(page, DBP_ARCH));
98
15117
		free(dba_array_get(page, DBP_DESC));
99
15117
		dba_array_free(dba_array_get(page, DBP_FILE));
100
15117
		dba_array_free(page);
101
	}
102
96
	dba_array_free(dba->pages);
103
104
96
	free(dba);
105
96
}
106
107
/*
108
 * Write the complete mandoc database to disk; the format is:
109
 * - One integer each for magic and version.
110
 * - One pointer each to the macros table and to the final magic.
111
 * - The pages table.
112
 * - The macros table.
113
 * - And at the very end, the magic integer again.
114
 */
115
int
116
dba_write(const char *fname, struct dba *dba)
117
{
118
	int	 save_errno;
119
	int32_t	 pos_end, pos_macros, pos_macros_ptr;
120
121
182
	if (dba_open(fname) == -1)
122
		return -1;
123
91
	dba_int_write(MANDOCDB_MAGIC);
124
91
	dba_int_write(MANDOCDB_VERSION);
125
91
	pos_macros_ptr = dba_skip(1, 2);
126
91
	dba_pages_write(dba->pages);
127
91
	pos_macros = dba_tell();
128
91
	dba_macros_write(dba->macros);
129
91
	pos_end = dba_tell();
130
91
	dba_int_write(MANDOCDB_MAGIC);
131
91
	dba_seek(pos_macros_ptr);
132
91
	dba_int_write(pos_macros);
133
91
	dba_int_write(pos_end);
134
91
	if (dba_close() == -1) {
135
		save_errno = errno;
136
		unlink(fname);
137
		errno = save_errno;
138
		return -1;
139
	}
140
91
	return 0;
141
91
}
142
143
144
/*** functions for handling pages *************************************/
145
146
/*
147
 * Create a new page and append it to the pages table.
148
 */
149
struct dba_array *
150
dba_page_new(struct dba_array *pages, const char *arch,
151
    const char *desc, const char *file, enum form form)
152
{
153
	struct dba_array *page, *entry;
154
155
30234
	page = dba_array_new(DBP_MAX, 0);
156
15117
	entry = dba_array_new(1, DBA_STR | DBA_GROW);
157
15117
	dba_array_add(page, entry);
158
15117
	entry = dba_array_new(1, DBA_STR | DBA_GROW);
159
15117
	dba_array_add(page, entry);
160

26895
	if (arch != NULL && *arch != '\0') {
161
1247
		entry = dba_array_new(1, DBA_STR | DBA_GROW);
162
1247
		dba_array_add(entry, (void *)arch);
163
1247
	} else
164
		entry = NULL;
165
15117
	dba_array_add(page, entry);
166
15117
	dba_array_add(page, mandoc_strdup(desc));
167
15117
	entry = dba_array_new(1, DBA_STR | DBA_GROW);
168
15117
	dba_array_add(entry, prepend(file, form));
169
15117
	dba_array_add(page, entry);
170
15117
	dba_array_add(pages, page);
171
15117
	return page;
172
}
173
174
/*
175
 * Add a section, architecture, or file name to an existing page.
176
 * Passing the NULL pointer for the architecture makes the page MI.
177
 * In that case, any earlier or later architectures are ignored.
178
 */
179
void
180
dba_page_add(struct dba_array *page, int32_t ie, const char *str)
181
{
182
	struct dba_array	*entries;
183
	char			*entry;
184
185
139206
	entries = dba_array_get(page, ie);
186
69603
	if (ie == DBP_ARCH) {
187
12128
		if (entries == NULL)
188
10876
			return;
189

2504
		if (str == NULL || *str == '\0') {
190
			dba_array_free(entries);
191
			dba_array_set(page, DBP_ARCH, NULL);
192
			return;
193
		}
194
	}
195
58727
	if (*str == '\0')
196
		return;
197
131279
	dba_array_FOREACH(entries, entry) {
198

66019
		if (ie == DBP_FILE && *entry < ' ')
199
12709
			entry++;
200
50895
		if (strcmp(entry, str) == 0)
201
37070
			return;
202
	}
203
21657
	dba_array_add(entries, (void *)str);
204
91260
}
205
206
/*
207
 * Add an additional name to an existing page.
208
 */
209
void
210
dba_page_alias(struct dba_array *page, const char *name, uint64_t mask)
211
{
212
	struct dba_array	*entries;
213
	char			*entry;
214
	char			 maskbyte;
215
216
80972
	if (*name == '\0')
217
		return;
218
40486
	maskbyte = mask & NAME_MASK;
219
40486
	entries = dba_array_get(page, DBP_NAME);
220
264833
	dba_array_FOREACH(entries, entry) {
221
195979
		if (strcmp(entry + 1, name) == 0) {
222
12118
			*entry |= maskbyte;
223
12118
			return;
224
		}
225
	}
226
28368
	dba_array_add(entries, prepend(name, maskbyte));
227
68854
}
228
229
/*
230
 * Return a pointer to a temporary copy of instr with inbyte prepended.
231
 */
232
static void *
233
prepend(const char *instr, char inbyte)
234
{
235
	static char	*outstr = NULL;
236
	static size_t	 outlen = 0;
237
	size_t		 newlen;
238
239
86970
	newlen = strlen(instr) + 1;
240
43485
	if (newlen > outlen) {
241
182
		outstr = mandoc_realloc(outstr, newlen + 1);
242
182
		outlen = newlen;
243
182
	}
244
43485
	*outstr = inbyte;
245
43485
	memcpy(outstr + 1, instr, newlen);
246
43485
	return outstr;
247
}
248
249
/*
250
 * Write the pages table to disk; the format is:
251
 * - One integer containing the number of pages.
252
 * - For each page, five pointers to the names, sections,
253
 *   architectures, description, and file names of the page.
254
 *   MI pages write 0 instead of the architecture pointer.
255
 * - One list each for names, sections, architectures, descriptions and
256
 *   file names.  The description for each page ends with a NUL byte.
257
 *   For all the other lists, each string ends with a NUL byte,
258
 *   and the last string for a page ends with two NUL bytes.
259
 * - To assure alignment of following integers,
260
 *   the end is padded with NUL bytes up to a multiple of four bytes.
261
 */
262
static void
263
dba_pages_write(struct dba_array *pages)
264
{
265
	struct dba_array	*page, *entry;
266
	int32_t			 pos_pages, pos_end;
267
268
182
	pos_pages = dba_array_writelen(pages, 5);
269
30416
	dba_array_FOREACH(pages, page) {
270
15117
		dba_array_setpos(page, DBP_NAME, dba_tell());
271
15117
		entry = dba_array_get(page, DBP_NAME);
272
15117
		dba_array_sort(entry, compare_names);
273
15117
		dba_array_writelst(entry);
274
	}
275
30416
	dba_array_FOREACH(pages, page) {
276
15117
		dba_array_setpos(page, DBP_SECT, dba_tell());
277
15117
		entry = dba_array_get(page, DBP_SECT);
278
15117
		dba_array_sort(entry, compare_strings);
279
15117
		dba_array_writelst(entry);
280
	}
281
15299
	dba_array_FOREACH(pages, page) {
282
15117
		if ((entry = dba_array_get(page, DBP_ARCH)) != NULL) {
283
1247
			dba_array_setpos(page, DBP_ARCH, dba_tell());
284
1247
			dba_array_sort(entry, compare_strings);
285
1247
			dba_array_writelst(entry);
286
1247
		} else
287
13870
			dba_array_setpos(page, DBP_ARCH, 0);
288
	}
289
30416
	dba_array_FOREACH(pages, page) {
290
15117
		dba_array_setpos(page, DBP_DESC, dba_tell());
291
15117
		dba_str_write(dba_array_get(page, DBP_DESC));
292
	}
293
30416
	dba_array_FOREACH(pages, page) {
294
15117
		dba_array_setpos(page, DBP_FILE, dba_tell());
295
15117
		dba_array_writelst(dba_array_get(page, DBP_FILE));
296
	}
297
91
	pos_end = dba_align();
298
91
	dba_seek(pos_pages);
299
30416
	dba_array_FOREACH(pages, page)
300
15117
		dba_array_writepos(page);
301
91
	dba_seek(pos_end);
302
91
}
303
304
static int
305
compare_names(const void *vp1, const void *vp2)
306
{
307
	const char	*cp1, *cp2;
308
	int		 diff;
309
310
115100
	cp1 = *(const char * const *)vp1;
311
57550
	cp2 = *(const char * const *)vp2;
312
172650
	return (diff = *cp2 - *cp1) ? diff :
313
46088
	    strcasecmp(cp1 + 1, cp2 + 1);
314
}
315
316
static int
317
compare_strings(const void *vp1, const void *vp2)
318
{
319
	const char	*cp1, *cp2;
320
321
1212
	cp1 = *(const char * const *)vp1;
322
606
	cp2 = *(const char * const *)vp2;
323
606
	return strcmp(cp1, cp2);
324
}
325
326
/*** functions for handling macros ************************************/
327
328
/*
329
 * In the hash table for a single macro, look up an entry by
330
 * the macro value or add an empty one if it doesn't exist yet.
331
 */
332
static struct macro_entry *
333
get_macro_entry(struct ohash *macro, const char *value, int32_t np)
334
{
335
	struct macro_entry	*entry;
336
	size_t			 len;
337
	unsigned int		 slot;
338
339
557982
	slot = ohash_qlookup(macro, value);
340
278991
	if ((entry = ohash_find(macro, slot)) == NULL) {
341
132857
		len = strlen(value) + 1;
342
132857
		entry = mandoc_malloc(sizeof(*entry) + len);
343
132857
		memcpy(&entry->value, value, len);
344
132857
		entry->pages = dba_array_new(np, DBA_GROW);
345
132857
		ohash_insert(macro, slot, entry);
346
132857
	}
347
278991
	return entry;
348
}
349
350
/*
351
 * In addition to get_macro_entry(), add multiple page references,
352
 * converting them from the on-disk format (byte offsets in the file)
353
 * to page pointers in memory.
354
 */
355
void
356
dba_macro_new(struct dba *dba, int32_t im, const char *value,
357
    const int32_t *pp)
358
{
359
	struct macro_entry	*entry;
360
	const int32_t		*ip;
361
	int32_t			 np;
362
363
	np = 0;
364
29017
	for (ip = pp; *ip; ip++)
365
7457
		np++;
366
367
4701
	entry = get_macro_entry(dba_array_get(dba->macros, im), value, np);
368
24316
	for (ip = pp; *ip; ip++)
369
14914
		dba_array_add(entry->pages, dba_array_get(dba->pages,
370
7457
		    be32toh(*ip) / 5 / sizeof(*ip) - 1));
371
4701
}
372
373
/*
374
 * In addition to get_macro_entry(), add one page reference,
375
 * directly taking the in-memory page pointer as an argument.
376
 */
377
void
378
dba_macro_add(struct dba_array *macros, int32_t im, const char *value,
379
    struct dba_array *page)
380
{
381
	struct macro_entry	*entry;
382
383
548604
	if (*value == '\0')
384
12
		return;
385
274290
	entry = get_macro_entry(dba_array_get(macros, im), value, 1);
386
274290
	dba_array_add(entry->pages, page);
387
548592
}
388
389
/*
390
 * Write the macros table to disk; the format is:
391
 * - The number of macro tables (actually, MACRO_MAX).
392
 * - That number of pointers to the individual macro tables.
393
 * - The individual macro tables.
394
 */
395
static void
396
dba_macros_write(struct dba_array *macros)
397
{
398
	struct ohash		*macro;
399
	int32_t			 im, pos_macros, pos_end;
400
401
182
	pos_macros = dba_array_writelen(macros, 1);
402
	im = 0;
403
6734
	dba_array_FOREACH(macros, macro) {
404
3276
		dba_array_setpos(macros, im++, dba_tell());
405
3276
		dba_macro_write(macro);
406
	}
407
91
	pos_end = dba_tell();
408
91
	dba_seek(pos_macros);
409
91
	dba_array_writepos(macros);
410
91
	dba_seek(pos_end);
411
91
}
412
413
/*
414
 * Write one individual macro table to disk; the format is:
415
 * - The number of entries in the table.
416
 * - For each entry, two pointers, the first one to the value
417
 *   and the second one to the list of pages.
418
 * - A list of values, each ending in a NUL byte.
419
 * - To assure alignment of following integers,
420
 *   padding with NUL bytes up to a multiple of four bytes.
421
 * - A list of pointers to pages, each list ending in a 0 integer.
422
 */
423
static void
424
dba_macro_write(struct ohash *macro)
425
{
426
	struct macro_entry	**entries, *entry;
427
	struct dba_array	 *page;
428
	int32_t			 *kpos, *dpos;
429
6552
	unsigned int		  ie, ne, slot;
430
	int			  use;
431
	int32_t			  addr, pos_macro, pos_end;
432
433
	/* Temporary storage for filtering and sorting. */
434
435
3276
	ne = ohash_entries(macro);
436
3276
	entries = mandoc_reallocarray(NULL, ne, sizeof(*entries));
437
3276
	kpos = mandoc_reallocarray(NULL, ne, sizeof(*kpos));
438
3276
	dpos = mandoc_reallocarray(NULL, ne, sizeof(*dpos));
439
440
	/* Build a list of non-empty entries and sort it. */
441
442
	ne = 0;
443
272266
	for (entry = ohash_first(macro, &slot); entry != NULL;
444
132857
	     entry = ohash_next(macro, &slot)) {
445
		use = 0;
446
547461
		dba_array_FOREACH(entry->pages, page)
447
281747
			if (dba_array_getpos(page))
448
281747
				use = 1;
449
132857
		if (use)
450
132857
			entries[ne++] = entry;
451
	}
452
3276
	qsort(entries, ne, sizeof(*entries), compare_entries);
453
454
	/* Number of entries, and space for the pointer pairs. */
455
456
3276
	dba_int_write(ne);
457
3276
	pos_macro = dba_skip(2, ne);
458
459
	/* String table. */
460
461
272266
	for (ie = 0; ie < ne; ie++) {
462
132857
		kpos[ie] = dba_tell();
463
132857
		dba_str_write(entries[ie]->value);
464
	}
465
3276
	dba_align();
466
467
	/* Pages table. */
468
469
408399
	for (ie = 0; ie < ne; ie++) {
470
268990
		dpos[ie] = dba_tell();
471
547461
		dba_array_FOREACH(entries[ie]->pages, page)
472
281747
			if ((addr = dba_array_getpos(page)))
473
281747
				dba_int_write(addr);
474
132857
		dba_int_write(0);
475
	}
476
	pos_end = dba_tell();
477
478
	/* Fill in the pointer pairs. */
479
480
3276
	dba_seek(pos_macro);
481
272266
	for (ie = 0; ie < ne; ie++) {
482
132857
		dba_int_write(kpos[ie]);
483
132857
		dba_int_write(dpos[ie]);
484
	}
485
3276
	dba_seek(pos_end);
486
487
3276
	free(entries);
488
3276
	free(kpos);
489
3276
	free(dpos);
490
3276
}
491
492
static int
493
compare_entries(const void *vp1, const void *vp2)
494
{
495
	const struct macro_entry *ep1, *ep2;
496
497
2955122
	ep1 = *(const struct macro_entry * const *)vp1;
498
1477561
	ep2 = *(const struct macro_entry * const *)vp2;
499
1477561
	return strcmp(ep1->value, ep2->value);
500
}