xref: /freebsd/contrib/mandoc/dbm.c (revision dc318a4ffabcbfa23bb56a33403aad36e6de30af)
1 /*	$Id: dbm.c,v 1.7 2019/07/01 22:56:24 schwarze Exp $ */
2 /*
3  * Copyright (c) 2016 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  * Map-based version of the mandoc database, for read-only access.
18  * The interface is defined in "dbm.h".
19  */
20 #include "config.h"
21 
22 #include <assert.h>
23 #if HAVE_ENDIAN
24 #include <endian.h>
25 #elif HAVE_SYS_ENDIAN
26 #include <sys/endian.h>
27 #elif HAVE_NTOHL
28 #include <arpa/inet.h>
29 #endif
30 #if HAVE_ERR
31 #include <err.h>
32 #endif
33 #include <errno.h>
34 #include <regex.h>
35 #include <stdint.h>
36 #include <stdio.h>
37 #include <stdlib.h>
38 #include <string.h>
39 
40 #include "mansearch.h"
41 #include "dbm_map.h"
42 #include "dbm.h"
43 
44 struct macro {
45 	int32_t	value;
46 	int32_t	pages;
47 };
48 
49 struct page {
50 	int32_t	name;
51 	int32_t	sect;
52 	int32_t	arch;
53 	int32_t	desc;
54 	int32_t	file;
55 };
56 
57 enum iter {
58 	ITER_NONE = 0,
59 	ITER_NAME,
60 	ITER_SECT,
61 	ITER_ARCH,
62 	ITER_DESC,
63 	ITER_MACRO
64 };
65 
66 static struct macro	*macros[MACRO_MAX];
67 static int32_t		 nvals[MACRO_MAX];
68 static struct page	*pages;
69 static int32_t		 npages;
70 static enum iter	 iteration;
71 
72 static struct dbm_res	 page_bytitle(enum iter, const struct dbm_match *);
73 static struct dbm_res	 page_byarch(const struct dbm_match *);
74 static struct dbm_res	 page_bymacro(int32_t, const struct dbm_match *);
75 static char		*macro_bypage(int32_t, int32_t);
76 
77 
78 /*** top level functions **********************************************/
79 
80 /*
81  * Open a disk-based mandoc database for read-only access.
82  * Map the pages and macros[] arrays.
83  * Return 0 on success.  Return -1 and set errno on failure.
84  */
85 int
86 dbm_open(const char *fname)
87 {
88 	const int32_t	*mp, *ep;
89 	int32_t		 im;
90 
91 	if (dbm_map(fname) == -1)
92 		return -1;
93 
94 	if ((npages = be32toh(*dbm_getint(4))) < 0) {
95 		warnx("dbm_open(%s): Invalid number of pages: %d",
96 		    fname, npages);
97 		goto fail;
98 	}
99 	pages = (struct page *)dbm_getint(5);
100 
101 	if ((mp = dbm_get(*dbm_getint(2))) == NULL) {
102 		warnx("dbm_open(%s): Invalid offset of macros array", fname);
103 		goto fail;
104 	}
105 	if (be32toh(*mp) != MACRO_MAX) {
106 		warnx("dbm_open(%s): Invalid number of macros: %d",
107 		    fname, be32toh(*mp));
108 		goto fail;
109 	}
110 	for (im = 0; im < MACRO_MAX; im++) {
111 		if ((ep = dbm_get(*++mp)) == NULL) {
112 			warnx("dbm_open(%s): Invalid offset of macro %d",
113 			    fname, im);
114 			goto fail;
115 		}
116 		nvals[im] = be32toh(*ep);
117 		macros[im] = (struct macro *)++ep;
118 	}
119 	return 0;
120 
121 fail:
122 	dbm_unmap();
123 	errno = EFTYPE;
124 	return -1;
125 }
126 
127 void
128 dbm_close(void)
129 {
130 	dbm_unmap();
131 }
132 
133 
134 /*** functions for handling pages *************************************/
135 
136 int32_t
137 dbm_page_count(void)
138 {
139 	return npages;
140 }
141 
142 /*
143  * Give the caller pointers to the data for one manual page.
144  */
145 struct dbm_page *
146 dbm_page_get(int32_t ip)
147 {
148 	static struct dbm_page	 res;
149 
150 	assert(ip >= 0);
151 	assert(ip < npages);
152 	res.name = dbm_get(pages[ip].name);
153 	if (res.name == NULL)
154 		res.name = "(NULL)\0";
155 	res.sect = dbm_get(pages[ip].sect);
156 	if (res.sect == NULL)
157 		res.sect = "(NULL)\0";
158 	res.arch = pages[ip].arch ? dbm_get(pages[ip].arch) : NULL;
159 	res.desc = dbm_get(pages[ip].desc);
160 	if (res.desc == NULL)
161 		res.desc = "(NULL)";
162 	res.file = dbm_get(pages[ip].file);
163 	if (res.file == NULL)
164 		res.file = " (NULL)\0";
165 	res.addr = dbm_addr(pages + ip);
166 	return &res;
167 }
168 
169 /*
170  * Functions to start filtered iterations over manual pages.
171  */
172 void
173 dbm_page_byname(const struct dbm_match *match)
174 {
175 	assert(match != NULL);
176 	page_bytitle(ITER_NAME, match);
177 }
178 
179 void
180 dbm_page_bysect(const struct dbm_match *match)
181 {
182 	assert(match != NULL);
183 	page_bytitle(ITER_SECT, match);
184 }
185 
186 void
187 dbm_page_byarch(const struct dbm_match *match)
188 {
189 	assert(match != NULL);
190 	page_byarch(match);
191 }
192 
193 void
194 dbm_page_bydesc(const struct dbm_match *match)
195 {
196 	assert(match != NULL);
197 	page_bytitle(ITER_DESC, match);
198 }
199 
200 void
201 dbm_page_bymacro(int32_t im, const struct dbm_match *match)
202 {
203 	assert(im >= 0);
204 	assert(im < MACRO_MAX);
205 	assert(match != NULL);
206 	page_bymacro(im, match);
207 }
208 
209 /*
210  * Return the number of the next manual page in the current iteration.
211  */
212 struct dbm_res
213 dbm_page_next(void)
214 {
215 	struct dbm_res			 res = {-1, 0};
216 
217 	switch(iteration) {
218 	case ITER_NONE:
219 		return res;
220 	case ITER_ARCH:
221 		return page_byarch(NULL);
222 	case ITER_MACRO:
223 		return page_bymacro(0, NULL);
224 	default:
225 		return page_bytitle(iteration, NULL);
226 	}
227 }
228 
229 /*
230  * Functions implementing the iteration over manual pages.
231  */
232 static struct dbm_res
233 page_bytitle(enum iter arg_iter, const struct dbm_match *arg_match)
234 {
235 	static const struct dbm_match	*match;
236 	static const char		*cp;
237 	static int32_t			 ip;
238 	struct dbm_res			 res = {-1, 0};
239 
240 	assert(arg_iter == ITER_NAME || arg_iter == ITER_DESC ||
241 	    arg_iter == ITER_SECT);
242 
243 	/* Initialize for a new iteration. */
244 
245 	if (arg_match != NULL) {
246 		iteration = arg_iter;
247 		match = arg_match;
248 		switch (iteration) {
249 		case ITER_NAME:
250 			cp = dbm_get(pages[0].name);
251 			break;
252 		case ITER_SECT:
253 			cp = dbm_get(pages[0].sect);
254 			break;
255 		case ITER_DESC:
256 			cp = dbm_get(pages[0].desc);
257 			break;
258 		default:
259 			abort();
260 		}
261 		if (cp == NULL) {
262 			iteration = ITER_NONE;
263 			match = NULL;
264 			cp = NULL;
265 			ip = npages;
266 		} else
267 			ip = 0;
268 		return res;
269 	}
270 
271 	/* Search for a name. */
272 
273 	while (ip < npages) {
274 		if (iteration == ITER_NAME)
275 			cp++;
276 		if (dbm_match(match, cp))
277 			break;
278 		cp = strchr(cp, '\0') + 1;
279 		if (iteration == ITER_DESC)
280 			ip++;
281 		else if (*cp == '\0') {
282 			cp++;
283 			ip++;
284 		}
285 	}
286 
287 	/* Reached the end without a match. */
288 
289 	if (ip == npages) {
290 		iteration = ITER_NONE;
291 		match = NULL;
292 		cp = NULL;
293 		return res;
294 	}
295 
296 	/* Found a match; save the quality for later retrieval. */
297 
298 	res.page = ip;
299 	res.bits = iteration == ITER_NAME ? cp[-1] : 0;
300 
301 	/* Skip the remaining names of this page. */
302 
303 	if (++ip < npages) {
304 		do {
305 			cp++;
306 		} while (cp[-1] != '\0' ||
307 		    (iteration != ITER_DESC && cp[-2] != '\0'));
308 	}
309 	return res;
310 }
311 
312 static struct dbm_res
313 page_byarch(const struct dbm_match *arg_match)
314 {
315 	static const struct dbm_match	*match;
316 	struct dbm_res			 res = {-1, 0};
317 	static int32_t			 ip;
318 	const char			*cp;
319 
320 	/* Initialize for a new iteration. */
321 
322 	if (arg_match != NULL) {
323 		iteration = ITER_ARCH;
324 		match = arg_match;
325 		ip = 0;
326 		return res;
327 	}
328 
329 	/* Search for an architecture. */
330 
331 	for ( ; ip < npages; ip++)
332 		if (pages[ip].arch)
333 			for (cp = dbm_get(pages[ip].arch);
334 			    *cp != '\0';
335 			    cp = strchr(cp, '\0') + 1)
336 				if (dbm_match(match, cp)) {
337 					res.page = ip++;
338 					return res;
339 				}
340 
341 	/* Reached the end without a match. */
342 
343 	iteration = ITER_NONE;
344 	match = NULL;
345 	return res;
346 }
347 
348 static struct dbm_res
349 page_bymacro(int32_t arg_im, const struct dbm_match *arg_match)
350 {
351 	static const struct dbm_match	*match;
352 	static const int32_t		*pp;
353 	static const char		*cp;
354 	static int32_t			 im, iv;
355 	struct dbm_res			 res = {-1, 0};
356 
357 	assert(im >= 0);
358 	assert(im < MACRO_MAX);
359 
360 	/* Initialize for a new iteration. */
361 
362 	if (arg_match != NULL) {
363 		iteration = ITER_MACRO;
364 		match = arg_match;
365 		im = arg_im;
366 		cp = nvals[im] ? dbm_get(macros[im]->value) : NULL;
367 		pp = NULL;
368 		iv = -1;
369 		return res;
370 	}
371 	if (iteration != ITER_MACRO)
372 		return res;
373 
374 	/* Find the next matching macro value. */
375 
376 	while (pp == NULL || *pp == 0) {
377 		if (++iv == nvals[im]) {
378 			iteration = ITER_NONE;
379 			return res;
380 		}
381 		if (iv)
382 			cp = strchr(cp, '\0') + 1;
383 		if (dbm_match(match, cp))
384 			pp = dbm_get(macros[im][iv].pages);
385 	}
386 
387 	/* Found a matching page. */
388 
389 	res.page = (struct page *)dbm_get(*pp++) - pages;
390 	return res;
391 }
392 
393 
394 /*** functions for handling macros ************************************/
395 
396 int32_t
397 dbm_macro_count(int32_t im)
398 {
399 	assert(im >= 0);
400 	assert(im < MACRO_MAX);
401 	return nvals[im];
402 }
403 
404 struct dbm_macro *
405 dbm_macro_get(int32_t im, int32_t iv)
406 {
407 	static struct dbm_macro macro;
408 
409 	assert(im >= 0);
410 	assert(im < MACRO_MAX);
411 	assert(iv >= 0);
412 	assert(iv < nvals[im]);
413 	macro.value = dbm_get(macros[im][iv].value);
414 	macro.pp = dbm_get(macros[im][iv].pages);
415 	return &macro;
416 }
417 
418 /*
419  * Filtered iteration over macro entries.
420  */
421 void
422 dbm_macro_bypage(int32_t im, int32_t ip)
423 {
424 	assert(im >= 0);
425 	assert(im < MACRO_MAX);
426 	assert(ip != 0);
427 	macro_bypage(im, ip);
428 }
429 
430 char *
431 dbm_macro_next(void)
432 {
433 	return macro_bypage(MACRO_MAX, 0);
434 }
435 
436 static char *
437 macro_bypage(int32_t arg_im, int32_t arg_ip)
438 {
439 	static const int32_t	*pp;
440 	static int32_t		 im, ip, iv;
441 
442 	/* Initialize for a new iteration. */
443 
444 	if (arg_im < MACRO_MAX && arg_ip != 0) {
445 		im = arg_im;
446 		ip = arg_ip;
447 		pp = dbm_get(macros[im]->pages);
448 		iv = 0;
449 		return NULL;
450 	}
451 	if (im >= MACRO_MAX)
452 		return NULL;
453 
454 	/* Search for the next value. */
455 
456 	while (iv < nvals[im]) {
457 		if (*pp == ip)
458 			break;
459 		if (*pp == 0)
460 			iv++;
461 		pp++;
462 	}
463 
464 	/* Reached the end without a match. */
465 
466 	if (iv == nvals[im]) {
467 		im = MACRO_MAX;
468 		ip = 0;
469 		pp = NULL;
470 		return NULL;
471 	}
472 
473 	/* Found a match; skip the remaining pages of this entry. */
474 
475 	if (++iv < nvals[im])
476 		while (*pp++ != 0)
477 			continue;
478 
479 	return dbm_get(macros[im][iv - 1].value);
480 }
481