1219019Sgabor/* $FreeBSD$ */
2219019Sgabor/* $NetBSD: citrus_csmapper.c,v 1.10 2009/01/11 02:46:24 christos Exp $ */
3219019Sgabor
4219019Sgabor/*-
5219019Sgabor * Copyright (c)2003 Citrus Project,
6219019Sgabor * All rights reserved.
7219019Sgabor *
8219019Sgabor * Redistribution and use in source and binary forms, with or without
9219019Sgabor * modification, are permitted provided that the following conditions
10219019Sgabor * are met:
11219019Sgabor * 1. Redistributions of source code must retain the above copyright
12219019Sgabor *    notice, this list of conditions and the following disclaimer.
13219019Sgabor * 2. Redistributions in binary form must reproduce the above copyright
14219019Sgabor *    notice, this list of conditions and the following disclaimer in the
15219019Sgabor *    documentation and/or other materials provided with the distribution.
16219019Sgabor *
17219019Sgabor * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
18219019Sgabor * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19219019Sgabor * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20219019Sgabor * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
21219019Sgabor * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22219019Sgabor * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23219019Sgabor * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24219019Sgabor * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25219019Sgabor * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26219019Sgabor * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27219019Sgabor * SUCH DAMAGE.
28219019Sgabor */
29219019Sgabor
30219019Sgabor#include <sys/cdefs.h>
31219019Sgabor#include <sys/endian.h>
32219019Sgabor#include <sys/types.h>
33219019Sgabor#include <sys/queue.h>
34219019Sgabor
35219019Sgabor#include <assert.h>
36219019Sgabor#include <errno.h>
37219019Sgabor#include <limits.h>
38219019Sgabor#include <paths.h>
39219019Sgabor#include <stdio.h>
40219019Sgabor#include <stdlib.h>
41219019Sgabor#include <string.h>
42219019Sgabor
43219019Sgabor#include "citrus_namespace.h"
44219019Sgabor#include "citrus_types.h"
45219019Sgabor#include "citrus_bcs.h"
46219019Sgabor#include "citrus_region.h"
47219019Sgabor#include "citrus_lock.h"
48219019Sgabor#include "citrus_memstream.h"
49219019Sgabor#include "citrus_mmap.h"
50219019Sgabor#include "citrus_module.h"
51219019Sgabor#include "citrus_hash.h"
52219019Sgabor#include "citrus_mapper.h"
53219019Sgabor#include "citrus_csmapper.h"
54219019Sgabor#include "citrus_pivot_file.h"
55219019Sgabor#include "citrus_db.h"
56219019Sgabor#include "citrus_db_hash.h"
57219019Sgabor#include "citrus_lookup.h"
58219019Sgabor
59219019Sgaborstatic struct _citrus_mapper_area	*maparea = NULL;
60219019Sgabor
61252584Speterstatic pthread_rwlock_t			ma_lock = PTHREAD_RWLOCK_INITIALIZER;
62252584Speter
63219019Sgabor#define CS_ALIAS	_PATH_CSMAPPER "/charset.alias"
64219019Sgabor#define CS_PIVOT	_PATH_CSMAPPER "/charset.pivot"
65219019Sgabor
66219019Sgabor
67219019Sgabor/* ---------------------------------------------------------------------- */
68219019Sgabor
69219019Sgaborstatic int
70219019Sgaborget32(struct _region *r, uint32_t *rval)
71219019Sgabor{
72219019Sgabor
73219019Sgabor	if (_region_size(r) != 4)
74219019Sgabor		return (EFTYPE);
75219019Sgabor
76219019Sgabor	memcpy(rval, _region_head(r), (size_t)4);
77219019Sgabor	*rval = be32toh(*rval);
78219019Sgabor
79219019Sgabor	return (0);
80219019Sgabor}
81219019Sgabor
82219019Sgaborstatic int
83219019Sgaboropen_subdb(struct _citrus_db **subdb, struct _citrus_db *db, const char *src)
84219019Sgabor{
85219019Sgabor	struct _region r;
86219019Sgabor	int ret;
87219019Sgabor
88219019Sgabor	ret = _db_lookup_by_s(db, src, &r, NULL);
89219019Sgabor	if (ret)
90219019Sgabor		return (ret);
91219019Sgabor	ret = _db_open(subdb, &r, _CITRUS_PIVOT_SUB_MAGIC, _db_hash_std, NULL);
92219019Sgabor	if (ret)
93219019Sgabor		return (ret);
94219019Sgabor
95219019Sgabor	return (0);
96219019Sgabor}
97219019Sgabor
98219019Sgabor
99219019Sgabor#define NO_SUCH_FILE	EOPNOTSUPP
100219019Sgaborstatic int
101219019Sgaborfind_best_pivot_pvdb(const char *src, const char *dst, char *pivot,
102219019Sgabor    size_t pvlen, unsigned long *rnorm)
103219019Sgabor{
104219019Sgabor	struct _citrus_db *db1, *db2, *db3;
105219019Sgabor	struct _region fr, r1, r2;
106219019Sgabor	char buf[LINE_MAX];
107219019Sgabor	uint32_t val32;
108219019Sgabor	unsigned long norm;
109219019Sgabor	int i, num, ret;
110219019Sgabor
111219019Sgabor	ret = _map_file(&fr, CS_PIVOT ".pvdb");
112219019Sgabor	if (ret) {
113219019Sgabor		if (ret == ENOENT)
114219019Sgabor			ret = NO_SUCH_FILE;
115219019Sgabor		return (ret);
116219019Sgabor	}
117219019Sgabor	ret = _db_open(&db1, &fr, _CITRUS_PIVOT_MAGIC, _db_hash_std, NULL);
118219019Sgabor	if (ret)
119219019Sgabor		goto quit1;
120219019Sgabor	ret = open_subdb(&db2, db1, src);
121219019Sgabor	if (ret)
122219019Sgabor		goto quit2;
123219019Sgabor
124219019Sgabor	num = _db_get_num_entries(db2);
125219019Sgabor	*rnorm = ULONG_MAX;
126219019Sgabor	for (i = 0; i < num; i++) {
127219019Sgabor		/* iterate each pivot */
128219019Sgabor		ret = _db_get_entry(db2, i, &r1, &r2);
129219019Sgabor		if (ret)
130219019Sgabor			goto quit3;
131219019Sgabor		/* r1:pivot name, r2:norm among src and pivot */
132219019Sgabor		ret = get32(&r2, &val32);
133219019Sgabor		if (ret)
134219019Sgabor			goto quit3;
135219019Sgabor		norm = val32;
136219019Sgabor		snprintf(buf, sizeof(buf), "%.*s",
137219019Sgabor			 (int)_region_size(&r1), (char *)_region_head(&r1));
138219019Sgabor		/* buf: pivot name */
139219019Sgabor		ret = open_subdb(&db3, db1, buf);
140219019Sgabor		if (ret)
141219019Sgabor			goto quit3;
142219019Sgabor		if (_db_lookup_by_s(db3, dst, &r2, NULL) != 0)
143219019Sgabor			goto quit4;
144219019Sgabor		/* r2: norm among pivot and dst */
145219019Sgabor		ret = get32(&r2, &val32);
146219019Sgabor		if (ret)
147219019Sgabor			goto quit4;
148219019Sgabor		norm += val32;
149219019Sgabor		/* judge minimum norm */
150219019Sgabor		if (norm < *rnorm) {
151219019Sgabor			*rnorm = norm;
152219019Sgabor			strlcpy(pivot, buf, pvlen);
153219019Sgabor		}
154219019Sgaborquit4:
155219019Sgabor		_db_close(db3);
156219019Sgabor		if (ret)
157219019Sgabor			goto quit3;
158219019Sgabor	}
159219019Sgaborquit3:
160219019Sgabor	_db_close(db2);
161219019Sgaborquit2:
162219019Sgabor	_db_close(db1);
163219019Sgaborquit1:
164219019Sgabor	_unmap_file(&fr);
165219019Sgabor	if (ret)
166219019Sgabor		return (ret);
167219019Sgabor
168219019Sgabor	if (*rnorm == ULONG_MAX)
169219019Sgabor		return (ENOENT);
170219019Sgabor
171219019Sgabor	return (0);
172219019Sgabor}
173219019Sgabor
174219019Sgabor/* ---------------------------------------------------------------------- */
175219019Sgabor
176219019Sgaborstruct zone {
177219019Sgabor	const char *begin, *end;
178219019Sgabor};
179219019Sgabor
180219019Sgaborstruct parse_arg {
181219019Sgabor	char dst[PATH_MAX];
182219019Sgabor	unsigned long norm;
183219019Sgabor};
184219019Sgabor
185219019Sgaborstatic int
186219019Sgaborparse_line(struct parse_arg *pa, struct _region *r)
187219019Sgabor{
188219019Sgabor	struct zone z1, z2;
189219019Sgabor	char buf[20];
190219019Sgabor	size_t len;
191219019Sgabor
192219019Sgabor	len = _region_size(r);
193219019Sgabor	z1.begin = _bcs_skip_ws_len(_region_head(r), &len);
194219019Sgabor	if (len == 0)
195219019Sgabor		return (EFTYPE);
196219019Sgabor	z1.end = _bcs_skip_nonws_len(z1.begin, &len);
197219019Sgabor	if (len == 0)
198219019Sgabor		return (EFTYPE);
199219019Sgabor	z2.begin = _bcs_skip_ws_len(z1.end, &len);
200219019Sgabor	if (len == 0)
201219019Sgabor		return (EFTYPE);
202219019Sgabor	z2.end = _bcs_skip_nonws_len(z2.begin, &len);
203219019Sgabor
204219019Sgabor	/* z1 : dst name, z2 : norm */
205219019Sgabor	snprintf(pa->dst, sizeof(pa->dst),
206219019Sgabor	    "%.*s", (int)(z1.end-z1.begin), z1.begin);
207219019Sgabor	snprintf(buf, sizeof(buf),
208219019Sgabor	    "%.*s", (int)(z2.end-z2.begin), z2.begin);
209219019Sgabor	pa->norm = _bcs_strtoul(buf, NULL, 0);
210219019Sgabor
211219019Sgabor	return (0);
212219019Sgabor}
213219019Sgabor
214219019Sgaborstatic int
215219019Sgaborfind_dst(struct parse_arg *pasrc, const char *dst)
216219019Sgabor{
217219019Sgabor	struct _lookup *cl;
218219019Sgabor	struct parse_arg padst;
219219019Sgabor	struct _region data;
220219019Sgabor	int ret;
221219019Sgabor
222219019Sgabor	ret = _lookup_seq_open(&cl, CS_PIVOT, _LOOKUP_CASE_IGNORE);
223219019Sgabor	if (ret)
224219019Sgabor		return (ret);
225219019Sgabor
226219019Sgabor	ret = _lookup_seq_lookup(cl, pasrc->dst, &data);
227219019Sgabor	while (ret == 0) {
228219019Sgabor		ret = parse_line(&padst, &data);
229219019Sgabor		if (ret)
230219019Sgabor			break;
231219019Sgabor		if (strcmp(dst, padst.dst) == 0) {
232219019Sgabor			pasrc->norm += padst.norm;
233219019Sgabor			break;
234219019Sgabor		}
235219019Sgabor		ret = _lookup_seq_next(cl, NULL, &data);
236219019Sgabor	}
237219019Sgabor	_lookup_seq_close(cl);
238219019Sgabor
239219019Sgabor	return (ret);
240219019Sgabor}
241219019Sgabor
242219019Sgaborstatic int
243219019Sgaborfind_best_pivot_lookup(const char *src, const char *dst, char *pivot,
244219019Sgabor    size_t pvlen, unsigned long *rnorm)
245219019Sgabor{
246219019Sgabor	struct _lookup *cl;
247219019Sgabor	struct _region data;
248219019Sgabor	struct parse_arg pa;
249219019Sgabor	char pivot_min[PATH_MAX];
250219019Sgabor	unsigned long norm_min;
251219019Sgabor	int ret;
252219019Sgabor
253219019Sgabor	ret = _lookup_seq_open(&cl, CS_PIVOT, _LOOKUP_CASE_IGNORE);
254219019Sgabor	if (ret)
255219019Sgabor		return (ret);
256219019Sgabor
257219019Sgabor	norm_min = ULONG_MAX;
258219019Sgabor
259219019Sgabor	/* find pivot code */
260219019Sgabor	ret = _lookup_seq_lookup(cl, src, &data);
261219019Sgabor	while (ret == 0) {
262219019Sgabor		ret = parse_line(&pa, &data);
263219019Sgabor		if (ret)
264219019Sgabor			break;
265219019Sgabor		ret = find_dst(&pa, dst);
266219019Sgabor		if (ret)
267219019Sgabor			break;
268219019Sgabor		if (pa.norm < norm_min) {
269219019Sgabor			norm_min = pa.norm;
270219019Sgabor			strlcpy(pivot_min, pa.dst, sizeof(pivot_min));
271219019Sgabor		}
272219019Sgabor		ret = _lookup_seq_next(cl, NULL, &data);
273219019Sgabor	}
274219019Sgabor	_lookup_seq_close(cl);
275219019Sgabor
276219019Sgabor	if (ret != ENOENT)
277219019Sgabor		return (ret);
278219019Sgabor	if (norm_min == ULONG_MAX)
279219019Sgabor		return (ENOENT);
280219019Sgabor	strlcpy(pivot, pivot_min, pvlen);
281219019Sgabor	if (rnorm)
282219019Sgabor		*rnorm = norm_min;
283219019Sgabor
284219019Sgabor	return (0);
285219019Sgabor}
286219019Sgabor
287219019Sgaborstatic int
288219019Sgaborfind_best_pivot(const char *src, const char *dst, char *pivot, size_t pvlen,
289219019Sgabor    unsigned long *rnorm)
290219019Sgabor{
291219019Sgabor	int ret;
292219019Sgabor
293219019Sgabor	ret = find_best_pivot_pvdb(src, dst, pivot, pvlen, rnorm);
294219019Sgabor	if (ret == NO_SUCH_FILE)
295219019Sgabor		ret = find_best_pivot_lookup(src, dst, pivot, pvlen, rnorm);
296219019Sgabor
297219019Sgabor	return (ret);
298219019Sgabor}
299219019Sgabor
300219019Sgaborstatic __inline int
301219019Sgaboropen_serial_mapper(struct _citrus_mapper_area *__restrict ma,
302219019Sgabor    struct _citrus_mapper * __restrict * __restrict rcm,
303219019Sgabor    const char *src, const char *pivot, const char *dst)
304219019Sgabor{
305219019Sgabor	char buf[PATH_MAX];
306219019Sgabor
307219019Sgabor	snprintf(buf, sizeof(buf), "%s/%s,%s/%s", src, pivot, pivot, dst);
308219019Sgabor
309219019Sgabor	return (_mapper_open_direct(ma, rcm, "mapper_serial", buf));
310219019Sgabor}
311219019Sgabor
312219019Sgaborstatic struct _citrus_csmapper *csm_none = NULL;
313219019Sgaborstatic int
314219019Sgaborget_none(struct _citrus_mapper_area *__restrict ma,
315219019Sgabor    struct _citrus_csmapper *__restrict *__restrict rcsm)
316219019Sgabor{
317219019Sgabor	int ret;
318219019Sgabor
319252584Speter	WLOCK(&ma_lock);
320219019Sgabor	if (csm_none) {
321219019Sgabor		*rcsm = csm_none;
322219019Sgabor		ret = 0;
323219019Sgabor		goto quit;
324219019Sgabor	}
325219019Sgabor
326219019Sgabor	ret = _mapper_open_direct(ma, &csm_none, "mapper_none", "");
327219019Sgabor	if (ret)
328219019Sgabor		goto quit;
329219019Sgabor	_mapper_set_persistent(csm_none);
330219019Sgabor
331219019Sgabor	*rcsm = csm_none;
332219019Sgabor	ret = 0;
333219019Sgaborquit:
334252584Speter	UNLOCK(&ma_lock);
335219019Sgabor	return (ret);
336219019Sgabor}
337219019Sgabor
338219019Sgaborint
339219019Sgabor_citrus_csmapper_open(struct _citrus_csmapper * __restrict * __restrict rcsm,
340219019Sgabor    const char * __restrict src, const char * __restrict dst, uint32_t flags,
341219019Sgabor    unsigned long *rnorm)
342219019Sgabor{
343219019Sgabor	const char *realsrc, *realdst;
344219019Sgabor	char buf1[PATH_MAX], buf2[PATH_MAX], key[PATH_MAX], pivot[PATH_MAX];
345219019Sgabor	unsigned long norm;
346219019Sgabor	int ret;
347219019Sgabor
348219019Sgabor	norm = 0;
349219019Sgabor
350219019Sgabor	ret = _citrus_mapper_create_area(&maparea, _PATH_CSMAPPER);
351219019Sgabor	if (ret)
352219019Sgabor		return (ret);
353219019Sgabor
354219019Sgabor	realsrc = _lookup_alias(CS_ALIAS, src, buf1, sizeof(buf1),
355219019Sgabor	    _LOOKUP_CASE_IGNORE);
356219019Sgabor	realdst = _lookup_alias(CS_ALIAS, dst, buf2, sizeof(buf2),
357219019Sgabor	    _LOOKUP_CASE_IGNORE);
358219019Sgabor	if (!strcmp(realsrc, realdst)) {
359219019Sgabor		ret = get_none(maparea, rcsm);
360219019Sgabor		if (ret == 0 && rnorm != NULL)
361219019Sgabor			*rnorm = 0;
362219019Sgabor		return (ret);
363219019Sgabor	}
364219019Sgabor
365219019Sgabor	snprintf(key, sizeof(key), "%s/%s", realsrc, realdst);
366219019Sgabor
367219019Sgabor	ret = _mapper_open(maparea, rcsm, key);
368219019Sgabor	if (ret == 0) {
369219019Sgabor		if (rnorm != NULL)
370219019Sgabor			*rnorm = 0;
371219019Sgabor		return (0);
372219019Sgabor	}
373219019Sgabor	if (ret != ENOENT || (flags & _CSMAPPER_F_PREVENT_PIVOT)!=0)
374219019Sgabor		return (ret);
375219019Sgabor
376219019Sgabor	ret = find_best_pivot(realsrc, realdst, pivot, sizeof(pivot), &norm);
377219019Sgabor	if (ret)
378219019Sgabor		return (ret);
379219019Sgabor
380219019Sgabor	ret = open_serial_mapper(maparea, rcsm, realsrc, pivot, realdst);
381219019Sgabor	if (ret == 0 && rnorm != NULL)
382219019Sgabor		*rnorm = norm;
383219019Sgabor
384219019Sgabor	return (ret);
385219019Sgabor}
386