xref: /illumos-gate/usr/src/cmd/mandoc/dba.c (revision a40ea1a7d80eee1b409e9dcc2e48c730988147ea)
1*a40ea1a7SYuri Pankov /*	$Id: dba.c,v 1.10 2017/02/17 14:43:54 schwarze Exp $ */
2*a40ea1a7SYuri Pankov /*
3*a40ea1a7SYuri Pankov  * Copyright (c) 2016, 2017 Ingo Schwarze <schwarze@openbsd.org>
4*a40ea1a7SYuri Pankov  *
5*a40ea1a7SYuri Pankov  * Permission to use, copy, modify, and distribute this software for any
6*a40ea1a7SYuri Pankov  * purpose with or without fee is hereby granted, provided that the above
7*a40ea1a7SYuri Pankov  * copyright notice and this permission notice appear in all copies.
8*a40ea1a7SYuri Pankov  *
9*a40ea1a7SYuri Pankov  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10*a40ea1a7SYuri Pankov  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11*a40ea1a7SYuri Pankov  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12*a40ea1a7SYuri Pankov  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13*a40ea1a7SYuri Pankov  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14*a40ea1a7SYuri Pankov  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15*a40ea1a7SYuri Pankov  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16*a40ea1a7SYuri Pankov  *
17*a40ea1a7SYuri Pankov  * Allocation-based version of the mandoc database, for read-write access.
18*a40ea1a7SYuri Pankov  * The interface is defined in "dba.h".
19*a40ea1a7SYuri Pankov  */
20*a40ea1a7SYuri Pankov #include "config.h"
21*a40ea1a7SYuri Pankov 
22*a40ea1a7SYuri Pankov #include <sys/types.h>
23*a40ea1a7SYuri Pankov #if HAVE_ENDIAN
24*a40ea1a7SYuri Pankov #include <endian.h>
25*a40ea1a7SYuri Pankov #elif HAVE_SYS_ENDIAN
26*a40ea1a7SYuri Pankov #include <sys/endian.h>
27*a40ea1a7SYuri Pankov #elif HAVE_NTOHL
28*a40ea1a7SYuri Pankov #include <arpa/inet.h>
29*a40ea1a7SYuri Pankov #endif
30*a40ea1a7SYuri Pankov #include <errno.h>
31*a40ea1a7SYuri Pankov #include <stddef.h>
32*a40ea1a7SYuri Pankov #include <stdint.h>
33*a40ea1a7SYuri Pankov #include <stdlib.h>
34*a40ea1a7SYuri Pankov #include <string.h>
35*a40ea1a7SYuri Pankov #include <unistd.h>
36*a40ea1a7SYuri Pankov 
37*a40ea1a7SYuri Pankov #include "mandoc_aux.h"
38*a40ea1a7SYuri Pankov #include "mandoc_ohash.h"
39*a40ea1a7SYuri Pankov #include "mansearch.h"
40*a40ea1a7SYuri Pankov #include "dba_write.h"
41*a40ea1a7SYuri Pankov #include "dba_array.h"
42*a40ea1a7SYuri Pankov #include "dba.h"
43*a40ea1a7SYuri Pankov 
44*a40ea1a7SYuri Pankov struct macro_entry {
45*a40ea1a7SYuri Pankov 	struct dba_array	*pages;
46*a40ea1a7SYuri Pankov 	char			 value[];
47*a40ea1a7SYuri Pankov };
48*a40ea1a7SYuri Pankov 
49*a40ea1a7SYuri Pankov static void	*prepend(const char *, char);
50*a40ea1a7SYuri Pankov static void	 dba_pages_write(struct dba_array *);
51*a40ea1a7SYuri Pankov static int	 compare_names(const void *, const void *);
52*a40ea1a7SYuri Pankov static int	 compare_strings(const void *, const void *);
53*a40ea1a7SYuri Pankov 
54*a40ea1a7SYuri Pankov static struct macro_entry
55*a40ea1a7SYuri Pankov 		*get_macro_entry(struct ohash *, const char *, int32_t);
56*a40ea1a7SYuri Pankov static void	 dba_macros_write(struct dba_array *);
57*a40ea1a7SYuri Pankov static void	 dba_macro_write(struct ohash *);
58*a40ea1a7SYuri Pankov static int	 compare_entries(const void *, const void *);
59*a40ea1a7SYuri Pankov 
60*a40ea1a7SYuri Pankov 
61*a40ea1a7SYuri Pankov /*** top-level functions **********************************************/
62*a40ea1a7SYuri Pankov 
63*a40ea1a7SYuri Pankov struct dba *
dba_new(int32_t npages)64*a40ea1a7SYuri Pankov dba_new(int32_t npages)
65*a40ea1a7SYuri Pankov {
66*a40ea1a7SYuri Pankov 	struct dba	*dba;
67*a40ea1a7SYuri Pankov 	struct ohash	*macro;
68*a40ea1a7SYuri Pankov 	int32_t		 im;
69*a40ea1a7SYuri Pankov 
70*a40ea1a7SYuri Pankov 	dba = mandoc_malloc(sizeof(*dba));
71*a40ea1a7SYuri Pankov 	dba->pages = dba_array_new(npages, DBA_GROW);
72*a40ea1a7SYuri Pankov 	dba->macros = dba_array_new(MACRO_MAX, 0);
73*a40ea1a7SYuri Pankov 	for (im = 0; im < MACRO_MAX; im++) {
74*a40ea1a7SYuri Pankov 		macro = mandoc_malloc(sizeof(*macro));
75*a40ea1a7SYuri Pankov 		mandoc_ohash_init(macro, 4,
76*a40ea1a7SYuri Pankov 		    offsetof(struct macro_entry, value));
77*a40ea1a7SYuri Pankov 		dba_array_set(dba->macros, im, macro);
78*a40ea1a7SYuri Pankov 	}
79*a40ea1a7SYuri Pankov 	return dba;
80*a40ea1a7SYuri Pankov }
81*a40ea1a7SYuri Pankov 
82*a40ea1a7SYuri Pankov void
dba_free(struct dba * dba)83*a40ea1a7SYuri Pankov dba_free(struct dba *dba)
84*a40ea1a7SYuri Pankov {
85*a40ea1a7SYuri Pankov 	struct dba_array	*page;
86*a40ea1a7SYuri Pankov 	struct ohash		*macro;
87*a40ea1a7SYuri Pankov 	struct macro_entry	*entry;
88*a40ea1a7SYuri Pankov 	unsigned int		 slot;
89*a40ea1a7SYuri Pankov 
90*a40ea1a7SYuri Pankov 	dba_array_FOREACH(dba->macros, macro) {
91*a40ea1a7SYuri Pankov 		for (entry = ohash_first(macro, &slot); entry != NULL;
92*a40ea1a7SYuri Pankov 		     entry = ohash_next(macro, &slot)) {
93*a40ea1a7SYuri Pankov 			dba_array_free(entry->pages);
94*a40ea1a7SYuri Pankov 			free(entry);
95*a40ea1a7SYuri Pankov 		}
96*a40ea1a7SYuri Pankov 		ohash_delete(macro);
97*a40ea1a7SYuri Pankov 		free(macro);
98*a40ea1a7SYuri Pankov 	}
99*a40ea1a7SYuri Pankov 	dba_array_free(dba->macros);
100*a40ea1a7SYuri Pankov 
101*a40ea1a7SYuri Pankov 	dba_array_undel(dba->pages);
102*a40ea1a7SYuri Pankov 	dba_array_FOREACH(dba->pages, page) {
103*a40ea1a7SYuri Pankov 		dba_array_free(dba_array_get(page, DBP_NAME));
104*a40ea1a7SYuri Pankov 		dba_array_free(dba_array_get(page, DBP_SECT));
105*a40ea1a7SYuri Pankov 		dba_array_free(dba_array_get(page, DBP_ARCH));
106*a40ea1a7SYuri Pankov 		free(dba_array_get(page, DBP_DESC));
107*a40ea1a7SYuri Pankov 		dba_array_free(dba_array_get(page, DBP_FILE));
108*a40ea1a7SYuri Pankov 		dba_array_free(page);
109*a40ea1a7SYuri Pankov 	}
110*a40ea1a7SYuri Pankov 	dba_array_free(dba->pages);
111*a40ea1a7SYuri Pankov 
112*a40ea1a7SYuri Pankov 	free(dba);
113*a40ea1a7SYuri Pankov }
114*a40ea1a7SYuri Pankov 
115*a40ea1a7SYuri Pankov /*
116*a40ea1a7SYuri Pankov  * Write the complete mandoc database to disk; the format is:
117*a40ea1a7SYuri Pankov  * - One integer each for magic and version.
118*a40ea1a7SYuri Pankov  * - One pointer each to the macros table and to the final magic.
119*a40ea1a7SYuri Pankov  * - The pages table.
120*a40ea1a7SYuri Pankov  * - The macros table.
121*a40ea1a7SYuri Pankov  * - And at the very end, the magic integer again.
122*a40ea1a7SYuri Pankov  */
123*a40ea1a7SYuri Pankov int
dba_write(const char * fname,struct dba * dba)124*a40ea1a7SYuri Pankov dba_write(const char *fname, struct dba *dba)
125*a40ea1a7SYuri Pankov {
126*a40ea1a7SYuri Pankov 	int	 save_errno;
127*a40ea1a7SYuri Pankov 	int32_t	 pos_end, pos_macros, pos_macros_ptr;
128*a40ea1a7SYuri Pankov 
129*a40ea1a7SYuri Pankov 	if (dba_open(fname) == -1)
130*a40ea1a7SYuri Pankov 		return -1;
131*a40ea1a7SYuri Pankov 	dba_int_write(MANDOCDB_MAGIC);
132*a40ea1a7SYuri Pankov 	dba_int_write(MANDOCDB_VERSION);
133*a40ea1a7SYuri Pankov 	pos_macros_ptr = dba_skip(1, 2);
134*a40ea1a7SYuri Pankov 	dba_pages_write(dba->pages);
135*a40ea1a7SYuri Pankov 	pos_macros = dba_tell();
136*a40ea1a7SYuri Pankov 	dba_macros_write(dba->macros);
137*a40ea1a7SYuri Pankov 	pos_end = dba_tell();
138*a40ea1a7SYuri Pankov 	dba_int_write(MANDOCDB_MAGIC);
139*a40ea1a7SYuri Pankov 	dba_seek(pos_macros_ptr);
140*a40ea1a7SYuri Pankov 	dba_int_write(pos_macros);
141*a40ea1a7SYuri Pankov 	dba_int_write(pos_end);
142*a40ea1a7SYuri Pankov 	if (dba_close() == -1) {
143*a40ea1a7SYuri Pankov 		save_errno = errno;
144*a40ea1a7SYuri Pankov 		unlink(fname);
145*a40ea1a7SYuri Pankov 		errno = save_errno;
146*a40ea1a7SYuri Pankov 		return -1;
147*a40ea1a7SYuri Pankov 	}
148*a40ea1a7SYuri Pankov 	return 0;
149*a40ea1a7SYuri Pankov }
150*a40ea1a7SYuri Pankov 
151*a40ea1a7SYuri Pankov 
152*a40ea1a7SYuri Pankov /*** functions for handling pages *************************************/
153*a40ea1a7SYuri Pankov 
154*a40ea1a7SYuri Pankov /*
155*a40ea1a7SYuri Pankov  * Create a new page and append it to the pages table.
156*a40ea1a7SYuri Pankov  */
157*a40ea1a7SYuri Pankov struct dba_array *
dba_page_new(struct dba_array * pages,const char * arch,const char * desc,const char * file,enum form form)158*a40ea1a7SYuri Pankov dba_page_new(struct dba_array *pages, const char *arch,
159*a40ea1a7SYuri Pankov     const char *desc, const char *file, enum form form)
160*a40ea1a7SYuri Pankov {
161*a40ea1a7SYuri Pankov 	struct dba_array *page, *entry;
162*a40ea1a7SYuri Pankov 
163*a40ea1a7SYuri Pankov 	page = dba_array_new(DBP_MAX, 0);
164*a40ea1a7SYuri Pankov 	entry = dba_array_new(1, DBA_STR | DBA_GROW);
165*a40ea1a7SYuri Pankov 	dba_array_add(page, entry);
166*a40ea1a7SYuri Pankov 	entry = dba_array_new(1, DBA_STR | DBA_GROW);
167*a40ea1a7SYuri Pankov 	dba_array_add(page, entry);
168*a40ea1a7SYuri Pankov 	if (arch != NULL && *arch != '\0') {
169*a40ea1a7SYuri Pankov 		entry = dba_array_new(1, DBA_STR | DBA_GROW);
170*a40ea1a7SYuri Pankov 		dba_array_add(entry, (void *)arch);
171*a40ea1a7SYuri Pankov 	} else
172*a40ea1a7SYuri Pankov 		entry = NULL;
173*a40ea1a7SYuri Pankov 	dba_array_add(page, entry);
174*a40ea1a7SYuri Pankov 	dba_array_add(page, mandoc_strdup(desc));
175*a40ea1a7SYuri Pankov 	entry = dba_array_new(1, DBA_STR | DBA_GROW);
176*a40ea1a7SYuri Pankov 	dba_array_add(entry, prepend(file, form));
177*a40ea1a7SYuri Pankov 	dba_array_add(page, entry);
178*a40ea1a7SYuri Pankov 	dba_array_add(pages, page);
179*a40ea1a7SYuri Pankov 	return page;
180*a40ea1a7SYuri Pankov }
181*a40ea1a7SYuri Pankov 
182*a40ea1a7SYuri Pankov /*
183*a40ea1a7SYuri Pankov  * Add a section, architecture, or file name to an existing page.
184*a40ea1a7SYuri Pankov  * Passing the NULL pointer for the architecture makes the page MI.
185*a40ea1a7SYuri Pankov  * In that case, any earlier or later architectures are ignored.
186*a40ea1a7SYuri Pankov  */
187*a40ea1a7SYuri Pankov void
dba_page_add(struct dba_array * page,int32_t ie,const char * str)188*a40ea1a7SYuri Pankov dba_page_add(struct dba_array *page, int32_t ie, const char *str)
189*a40ea1a7SYuri Pankov {
190*a40ea1a7SYuri Pankov 	struct dba_array	*entries;
191*a40ea1a7SYuri Pankov 	char			*entry;
192*a40ea1a7SYuri Pankov 
193*a40ea1a7SYuri Pankov 	entries = dba_array_get(page, ie);
194*a40ea1a7SYuri Pankov 	if (ie == DBP_ARCH) {
195*a40ea1a7SYuri Pankov 		if (entries == NULL)
196*a40ea1a7SYuri Pankov 			return;
197*a40ea1a7SYuri Pankov 		if (str == NULL || *str == '\0') {
198*a40ea1a7SYuri Pankov 			dba_array_free(entries);
199*a40ea1a7SYuri Pankov 			dba_array_set(page, DBP_ARCH, NULL);
200*a40ea1a7SYuri Pankov 			return;
201*a40ea1a7SYuri Pankov 		}
202*a40ea1a7SYuri Pankov 	}
203*a40ea1a7SYuri Pankov 	if (*str == '\0')
204*a40ea1a7SYuri Pankov 		return;
205*a40ea1a7SYuri Pankov 	dba_array_FOREACH(entries, entry) {
206*a40ea1a7SYuri Pankov 		if (ie == DBP_FILE && *entry < ' ')
207*a40ea1a7SYuri Pankov 			entry++;
208*a40ea1a7SYuri Pankov 		if (strcmp(entry, str) == 0)
209*a40ea1a7SYuri Pankov 			return;
210*a40ea1a7SYuri Pankov 	}
211*a40ea1a7SYuri Pankov 	dba_array_add(entries, (void *)str);
212*a40ea1a7SYuri Pankov }
213*a40ea1a7SYuri Pankov 
214*a40ea1a7SYuri Pankov /*
215*a40ea1a7SYuri Pankov  * Add an additional name to an existing page.
216*a40ea1a7SYuri Pankov  */
217*a40ea1a7SYuri Pankov void
dba_page_alias(struct dba_array * page,const char * name,uint64_t mask)218*a40ea1a7SYuri Pankov dba_page_alias(struct dba_array *page, const char *name, uint64_t mask)
219*a40ea1a7SYuri Pankov {
220*a40ea1a7SYuri Pankov 	struct dba_array	*entries;
221*a40ea1a7SYuri Pankov 	char			*entry;
222*a40ea1a7SYuri Pankov 	char			 maskbyte;
223*a40ea1a7SYuri Pankov 
224*a40ea1a7SYuri Pankov 	if (*name == '\0')
225*a40ea1a7SYuri Pankov 		return;
226*a40ea1a7SYuri Pankov 	maskbyte = mask & NAME_MASK;
227*a40ea1a7SYuri Pankov 	entries = dba_array_get(page, DBP_NAME);
228*a40ea1a7SYuri Pankov 	dba_array_FOREACH(entries, entry) {
229*a40ea1a7SYuri Pankov 		if (strcmp(entry + 1, name) == 0) {
230*a40ea1a7SYuri Pankov 			*entry |= maskbyte;
231*a40ea1a7SYuri Pankov 			return;
232*a40ea1a7SYuri Pankov 		}
233*a40ea1a7SYuri Pankov 	}
234*a40ea1a7SYuri Pankov 	dba_array_add(entries, prepend(name, maskbyte));
235*a40ea1a7SYuri Pankov }
236*a40ea1a7SYuri Pankov 
237*a40ea1a7SYuri Pankov /*
238*a40ea1a7SYuri Pankov  * Return a pointer to a temporary copy of instr with inbyte prepended.
239*a40ea1a7SYuri Pankov  */
240*a40ea1a7SYuri Pankov static void *
prepend(const char * instr,char inbyte)241*a40ea1a7SYuri Pankov prepend(const char *instr, char inbyte)
242*a40ea1a7SYuri Pankov {
243*a40ea1a7SYuri Pankov 	static char	*outstr = NULL;
244*a40ea1a7SYuri Pankov 	static size_t	 outlen = 0;
245*a40ea1a7SYuri Pankov 	size_t		 newlen;
246*a40ea1a7SYuri Pankov 
247*a40ea1a7SYuri Pankov 	newlen = strlen(instr) + 1;
248*a40ea1a7SYuri Pankov 	if (newlen > outlen) {
249*a40ea1a7SYuri Pankov 		outstr = mandoc_realloc(outstr, newlen + 1);
250*a40ea1a7SYuri Pankov 		outlen = newlen;
251*a40ea1a7SYuri Pankov 	}
252*a40ea1a7SYuri Pankov 	*outstr = inbyte;
253*a40ea1a7SYuri Pankov 	memcpy(outstr + 1, instr, newlen);
254*a40ea1a7SYuri Pankov 	return outstr;
255*a40ea1a7SYuri Pankov }
256*a40ea1a7SYuri Pankov 
257*a40ea1a7SYuri Pankov /*
258*a40ea1a7SYuri Pankov  * Write the pages table to disk; the format is:
259*a40ea1a7SYuri Pankov  * - One integer containing the number of pages.
260*a40ea1a7SYuri Pankov  * - For each page, five pointers to the names, sections,
261*a40ea1a7SYuri Pankov  *   architectures, description, and file names of the page.
262*a40ea1a7SYuri Pankov  *   MI pages write 0 instead of the architecture pointer.
263*a40ea1a7SYuri Pankov  * - One list each for names, sections, architectures, descriptions and
264*a40ea1a7SYuri Pankov  *   file names.  The description for each page ends with a NUL byte.
265*a40ea1a7SYuri Pankov  *   For all the other lists, each string ends with a NUL byte,
266*a40ea1a7SYuri Pankov  *   and the last string for a page ends with two NUL bytes.
267*a40ea1a7SYuri Pankov  * - To assure alignment of following integers,
268*a40ea1a7SYuri Pankov  *   the end is padded with NUL bytes up to a multiple of four bytes.
269*a40ea1a7SYuri Pankov  */
270*a40ea1a7SYuri Pankov static void
dba_pages_write(struct dba_array * pages)271*a40ea1a7SYuri Pankov dba_pages_write(struct dba_array *pages)
272*a40ea1a7SYuri Pankov {
273*a40ea1a7SYuri Pankov 	struct dba_array	*page, *entry;
274*a40ea1a7SYuri Pankov 	int32_t			 pos_pages, pos_end;
275*a40ea1a7SYuri Pankov 
276*a40ea1a7SYuri Pankov 	pos_pages = dba_array_writelen(pages, 5);
277*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page) {
278*a40ea1a7SYuri Pankov 		dba_array_setpos(page, DBP_NAME, dba_tell());
279*a40ea1a7SYuri Pankov 		entry = dba_array_get(page, DBP_NAME);
280*a40ea1a7SYuri Pankov 		dba_array_sort(entry, compare_names);
281*a40ea1a7SYuri Pankov 		dba_array_writelst(entry);
282*a40ea1a7SYuri Pankov 	}
283*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page) {
284*a40ea1a7SYuri Pankov 		dba_array_setpos(page, DBP_SECT, dba_tell());
285*a40ea1a7SYuri Pankov 		entry = dba_array_get(page, DBP_SECT);
286*a40ea1a7SYuri Pankov 		dba_array_sort(entry, compare_strings);
287*a40ea1a7SYuri Pankov 		dba_array_writelst(entry);
288*a40ea1a7SYuri Pankov 	}
289*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page) {
290*a40ea1a7SYuri Pankov 		if ((entry = dba_array_get(page, DBP_ARCH)) != NULL) {
291*a40ea1a7SYuri Pankov 			dba_array_setpos(page, DBP_ARCH, dba_tell());
292*a40ea1a7SYuri Pankov 			dba_array_sort(entry, compare_strings);
293*a40ea1a7SYuri Pankov 			dba_array_writelst(entry);
294*a40ea1a7SYuri Pankov 		} else
295*a40ea1a7SYuri Pankov 			dba_array_setpos(page, DBP_ARCH, 0);
296*a40ea1a7SYuri Pankov 	}
297*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page) {
298*a40ea1a7SYuri Pankov 		dba_array_setpos(page, DBP_DESC, dba_tell());
299*a40ea1a7SYuri Pankov 		dba_str_write(dba_array_get(page, DBP_DESC));
300*a40ea1a7SYuri Pankov 	}
301*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page) {
302*a40ea1a7SYuri Pankov 		dba_array_setpos(page, DBP_FILE, dba_tell());
303*a40ea1a7SYuri Pankov 		dba_array_writelst(dba_array_get(page, DBP_FILE));
304*a40ea1a7SYuri Pankov 	}
305*a40ea1a7SYuri Pankov 	pos_end = dba_align();
306*a40ea1a7SYuri Pankov 	dba_seek(pos_pages);
307*a40ea1a7SYuri Pankov 	dba_array_FOREACH(pages, page)
308*a40ea1a7SYuri Pankov 		dba_array_writepos(page);
309*a40ea1a7SYuri Pankov 	dba_seek(pos_end);
310*a40ea1a7SYuri Pankov }
311*a40ea1a7SYuri Pankov 
312*a40ea1a7SYuri Pankov static int
compare_names(const void * vp1,const void * vp2)313*a40ea1a7SYuri Pankov compare_names(const void *vp1, const void *vp2)
314*a40ea1a7SYuri Pankov {
315*a40ea1a7SYuri Pankov 	const char	*cp1, *cp2;
316*a40ea1a7SYuri Pankov 	int		 diff;
317*a40ea1a7SYuri Pankov 
318*a40ea1a7SYuri Pankov 	cp1 = *(const char * const *)vp1;
319*a40ea1a7SYuri Pankov 	cp2 = *(const char * const *)vp2;
320*a40ea1a7SYuri Pankov 	return (diff = *cp2 - *cp1) ? diff :
321*a40ea1a7SYuri Pankov 	    strcasecmp(cp1 + 1, cp2 + 1);
322*a40ea1a7SYuri Pankov }
323*a40ea1a7SYuri Pankov 
324*a40ea1a7SYuri Pankov static int
compare_strings(const void * vp1,const void * vp2)325*a40ea1a7SYuri Pankov compare_strings(const void *vp1, const void *vp2)
326*a40ea1a7SYuri Pankov {
327*a40ea1a7SYuri Pankov 	const char	*cp1, *cp2;
328*a40ea1a7SYuri Pankov 
329*a40ea1a7SYuri Pankov 	cp1 = *(const char * const *)vp1;
330*a40ea1a7SYuri Pankov 	cp2 = *(const char * const *)vp2;
331*a40ea1a7SYuri Pankov 	return strcmp(cp1, cp2);
332*a40ea1a7SYuri Pankov }
333*a40ea1a7SYuri Pankov 
334*a40ea1a7SYuri Pankov /*** functions for handling macros ************************************/
335*a40ea1a7SYuri Pankov 
336*a40ea1a7SYuri Pankov /*
337*a40ea1a7SYuri Pankov  * In the hash table for a single macro, look up an entry by
338*a40ea1a7SYuri Pankov  * the macro value or add an empty one if it doesn't exist yet.
339*a40ea1a7SYuri Pankov  */
340*a40ea1a7SYuri Pankov static struct macro_entry *
get_macro_entry(struct ohash * macro,const char * value,int32_t np)341*a40ea1a7SYuri Pankov get_macro_entry(struct ohash *macro, const char *value, int32_t np)
342*a40ea1a7SYuri Pankov {
343*a40ea1a7SYuri Pankov 	struct macro_entry	*entry;
344*a40ea1a7SYuri Pankov 	size_t			 len;
345*a40ea1a7SYuri Pankov 	unsigned int		 slot;
346*a40ea1a7SYuri Pankov 
347*a40ea1a7SYuri Pankov 	slot = ohash_qlookup(macro, value);
348*a40ea1a7SYuri Pankov 	if ((entry = ohash_find(macro, slot)) == NULL) {
349*a40ea1a7SYuri Pankov 		len = strlen(value) + 1;
350*a40ea1a7SYuri Pankov 		entry = mandoc_malloc(sizeof(*entry) + len);
351*a40ea1a7SYuri Pankov 		memcpy(&entry->value, value, len);
352*a40ea1a7SYuri Pankov 		entry->pages = dba_array_new(np, DBA_GROW);
353*a40ea1a7SYuri Pankov 		ohash_insert(macro, slot, entry);
354*a40ea1a7SYuri Pankov 	}
355*a40ea1a7SYuri Pankov 	return entry;
356*a40ea1a7SYuri Pankov }
357*a40ea1a7SYuri Pankov 
358*a40ea1a7SYuri Pankov /*
359*a40ea1a7SYuri Pankov  * In addition to get_macro_entry(), add multiple page references,
360*a40ea1a7SYuri Pankov  * converting them from the on-disk format (byte offsets in the file)
361*a40ea1a7SYuri Pankov  * to page pointers in memory.
362*a40ea1a7SYuri Pankov  */
363*a40ea1a7SYuri Pankov void
dba_macro_new(struct dba * dba,int32_t im,const char * value,const int32_t * pp)364*a40ea1a7SYuri Pankov dba_macro_new(struct dba *dba, int32_t im, const char *value,
365*a40ea1a7SYuri Pankov     const int32_t *pp)
366*a40ea1a7SYuri Pankov {
367*a40ea1a7SYuri Pankov 	struct macro_entry	*entry;
368*a40ea1a7SYuri Pankov 	const int32_t		*ip;
369*a40ea1a7SYuri Pankov 	int32_t			 np;
370*a40ea1a7SYuri Pankov 
371*a40ea1a7SYuri Pankov 	np = 0;
372*a40ea1a7SYuri Pankov 	for (ip = pp; *ip; ip++)
373*a40ea1a7SYuri Pankov 		np++;
374*a40ea1a7SYuri Pankov 
375*a40ea1a7SYuri Pankov 	entry = get_macro_entry(dba_array_get(dba->macros, im), value, np);
376*a40ea1a7SYuri Pankov 	for (ip = pp; *ip; ip++)
377*a40ea1a7SYuri Pankov 		dba_array_add(entry->pages, dba_array_get(dba->pages,
378*a40ea1a7SYuri Pankov 		    be32toh(*ip) / 5 / sizeof(*ip) - 1));
379*a40ea1a7SYuri Pankov }
380*a40ea1a7SYuri Pankov 
381*a40ea1a7SYuri Pankov /*
382*a40ea1a7SYuri Pankov  * In addition to get_macro_entry(), add one page reference,
383*a40ea1a7SYuri Pankov  * directly taking the in-memory page pointer as an argument.
384*a40ea1a7SYuri Pankov  */
385*a40ea1a7SYuri Pankov void
dba_macro_add(struct dba_array * macros,int32_t im,const char * value,struct dba_array * page)386*a40ea1a7SYuri Pankov dba_macro_add(struct dba_array *macros, int32_t im, const char *value,
387*a40ea1a7SYuri Pankov     struct dba_array *page)
388*a40ea1a7SYuri Pankov {
389*a40ea1a7SYuri Pankov 	struct macro_entry	*entry;
390*a40ea1a7SYuri Pankov 
391*a40ea1a7SYuri Pankov 	if (*value == '\0')
392*a40ea1a7SYuri Pankov 		return;
393*a40ea1a7SYuri Pankov 	entry = get_macro_entry(dba_array_get(macros, im), value, 1);
394*a40ea1a7SYuri Pankov 	dba_array_add(entry->pages, page);
395*a40ea1a7SYuri Pankov }
396*a40ea1a7SYuri Pankov 
397*a40ea1a7SYuri Pankov /*
398*a40ea1a7SYuri Pankov  * Write the macros table to disk; the format is:
399*a40ea1a7SYuri Pankov  * - The number of macro tables (actually, MACRO_MAX).
400*a40ea1a7SYuri Pankov  * - That number of pointers to the individual macro tables.
401*a40ea1a7SYuri Pankov  * - The individual macro tables.
402*a40ea1a7SYuri Pankov  */
403*a40ea1a7SYuri Pankov static void
dba_macros_write(struct dba_array * macros)404*a40ea1a7SYuri Pankov dba_macros_write(struct dba_array *macros)
405*a40ea1a7SYuri Pankov {
406*a40ea1a7SYuri Pankov 	struct ohash		*macro;
407*a40ea1a7SYuri Pankov 	int32_t			 im, pos_macros, pos_end;
408*a40ea1a7SYuri Pankov 
409*a40ea1a7SYuri Pankov 	pos_macros = dba_array_writelen(macros, 1);
410*a40ea1a7SYuri Pankov 	im = 0;
411*a40ea1a7SYuri Pankov 	dba_array_FOREACH(macros, macro) {
412*a40ea1a7SYuri Pankov 		dba_array_setpos(macros, im++, dba_tell());
413*a40ea1a7SYuri Pankov 		dba_macro_write(macro);
414*a40ea1a7SYuri Pankov 	}
415*a40ea1a7SYuri Pankov 	pos_end = dba_tell();
416*a40ea1a7SYuri Pankov 	dba_seek(pos_macros);
417*a40ea1a7SYuri Pankov 	dba_array_writepos(macros);
418*a40ea1a7SYuri Pankov 	dba_seek(pos_end);
419*a40ea1a7SYuri Pankov }
420*a40ea1a7SYuri Pankov 
421*a40ea1a7SYuri Pankov /*
422*a40ea1a7SYuri Pankov  * Write one individual macro table to disk; the format is:
423*a40ea1a7SYuri Pankov  * - The number of entries in the table.
424*a40ea1a7SYuri Pankov  * - For each entry, two pointers, the first one to the value
425*a40ea1a7SYuri Pankov  *   and the second one to the list of pages.
426*a40ea1a7SYuri Pankov  * - A list of values, each ending in a NUL byte.
427*a40ea1a7SYuri Pankov  * - To assure alignment of following integers,
428*a40ea1a7SYuri Pankov  *   padding with NUL bytes up to a multiple of four bytes.
429*a40ea1a7SYuri Pankov  * - A list of pointers to pages, each list ending in a 0 integer.
430*a40ea1a7SYuri Pankov  */
431*a40ea1a7SYuri Pankov static void
dba_macro_write(struct ohash * macro)432*a40ea1a7SYuri Pankov dba_macro_write(struct ohash *macro)
433*a40ea1a7SYuri Pankov {
434*a40ea1a7SYuri Pankov 	struct macro_entry	**entries, *entry;
435*a40ea1a7SYuri Pankov 	struct dba_array	 *page;
436*a40ea1a7SYuri Pankov 	int32_t			 *kpos, *dpos;
437*a40ea1a7SYuri Pankov 	unsigned int		  ie, ne, slot;
438*a40ea1a7SYuri Pankov 	int			  use;
439*a40ea1a7SYuri Pankov 	int32_t			  addr, pos_macro, pos_end;
440*a40ea1a7SYuri Pankov 
441*a40ea1a7SYuri Pankov 	/* Temporary storage for filtering and sorting. */
442*a40ea1a7SYuri Pankov 
443*a40ea1a7SYuri Pankov 	ne = ohash_entries(macro);
444*a40ea1a7SYuri Pankov 	entries = mandoc_reallocarray(NULL, ne, sizeof(*entries));
445*a40ea1a7SYuri Pankov 	kpos = mandoc_reallocarray(NULL, ne, sizeof(*kpos));
446*a40ea1a7SYuri Pankov 	dpos = mandoc_reallocarray(NULL, ne, sizeof(*dpos));
447*a40ea1a7SYuri Pankov 
448*a40ea1a7SYuri Pankov 	/* Build a list of non-empty entries and sort it. */
449*a40ea1a7SYuri Pankov 
450*a40ea1a7SYuri Pankov 	ne = 0;
451*a40ea1a7SYuri Pankov 	for (entry = ohash_first(macro, &slot); entry != NULL;
452*a40ea1a7SYuri Pankov 	     entry = ohash_next(macro, &slot)) {
453*a40ea1a7SYuri Pankov 		use = 0;
454*a40ea1a7SYuri Pankov 		dba_array_FOREACH(entry->pages, page)
455*a40ea1a7SYuri Pankov 			if (dba_array_getpos(page))
456*a40ea1a7SYuri Pankov 				use = 1;
457*a40ea1a7SYuri Pankov 		if (use)
458*a40ea1a7SYuri Pankov 			entries[ne++] = entry;
459*a40ea1a7SYuri Pankov 	}
460*a40ea1a7SYuri Pankov 	qsort(entries, ne, sizeof(*entries), compare_entries);
461*a40ea1a7SYuri Pankov 
462*a40ea1a7SYuri Pankov 	/* Number of entries, and space for the pointer pairs. */
463*a40ea1a7SYuri Pankov 
464*a40ea1a7SYuri Pankov 	dba_int_write(ne);
465*a40ea1a7SYuri Pankov 	pos_macro = dba_skip(2, ne);
466*a40ea1a7SYuri Pankov 
467*a40ea1a7SYuri Pankov 	/* String table. */
468*a40ea1a7SYuri Pankov 
469*a40ea1a7SYuri Pankov 	for (ie = 0; ie < ne; ie++) {
470*a40ea1a7SYuri Pankov 		kpos[ie] = dba_tell();
471*a40ea1a7SYuri Pankov 		dba_str_write(entries[ie]->value);
472*a40ea1a7SYuri Pankov 	}
473*a40ea1a7SYuri Pankov 	dba_align();
474*a40ea1a7SYuri Pankov 
475*a40ea1a7SYuri Pankov 	/* Pages table. */
476*a40ea1a7SYuri Pankov 
477*a40ea1a7SYuri Pankov 	for (ie = 0; ie < ne; ie++) {
478*a40ea1a7SYuri Pankov 		dpos[ie] = dba_tell();
479*a40ea1a7SYuri Pankov 		dba_array_FOREACH(entries[ie]->pages, page)
480*a40ea1a7SYuri Pankov 			if ((addr = dba_array_getpos(page)))
481*a40ea1a7SYuri Pankov 				dba_int_write(addr);
482*a40ea1a7SYuri Pankov 		dba_int_write(0);
483*a40ea1a7SYuri Pankov 	}
484*a40ea1a7SYuri Pankov 	pos_end = dba_tell();
485*a40ea1a7SYuri Pankov 
486*a40ea1a7SYuri Pankov 	/* Fill in the pointer pairs. */
487*a40ea1a7SYuri Pankov 
488*a40ea1a7SYuri Pankov 	dba_seek(pos_macro);
489*a40ea1a7SYuri Pankov 	for (ie = 0; ie < ne; ie++) {
490*a40ea1a7SYuri Pankov 		dba_int_write(kpos[ie]);
491*a40ea1a7SYuri Pankov 		dba_int_write(dpos[ie]);
492*a40ea1a7SYuri Pankov 	}
493*a40ea1a7SYuri Pankov 	dba_seek(pos_end);
494*a40ea1a7SYuri Pankov 
495*a40ea1a7SYuri Pankov 	free(entries);
496*a40ea1a7SYuri Pankov 	free(kpos);
497*a40ea1a7SYuri Pankov 	free(dpos);
498*a40ea1a7SYuri Pankov }
499*a40ea1a7SYuri Pankov 
500*a40ea1a7SYuri Pankov static int
compare_entries(const void * vp1,const void * vp2)501*a40ea1a7SYuri Pankov compare_entries(const void *vp1, const void *vp2)
502*a40ea1a7SYuri Pankov {
503*a40ea1a7SYuri Pankov 	const struct macro_entry *ep1, *ep2;
504*a40ea1a7SYuri Pankov 
505*a40ea1a7SYuri Pankov 	ep1 = *(const struct macro_entry * const *)vp1;
506*a40ea1a7SYuri Pankov 	ep2 = *(const struct macro_entry * const *)vp2;
507*a40ea1a7SYuri Pankov 	return strcmp(ep1->value, ep2->value);
508*a40ea1a7SYuri Pankov }
509