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
61219019Sgabor#define CS_ALIAS	_PATH_CSMAPPER "/charset.alias"
62219019Sgabor#define CS_PIVOT	_PATH_CSMAPPER "/charset.pivot"
63219019Sgabor
64219019Sgabor
65219019Sgabor/* ---------------------------------------------------------------------- */
66219019Sgabor
67219019Sgaborstatic int
68219019Sgaborget32(struct _region *r, uint32_t *rval)
69219019Sgabor{
70219019Sgabor
71219019Sgabor	if (_region_size(r) != 4)
72219019Sgabor		return (EFTYPE);
73219019Sgabor
74219019Sgabor	memcpy(rval, _region_head(r), (size_t)4);
75219019Sgabor	*rval = be32toh(*rval);
76219019Sgabor
77219019Sgabor	return (0);
78219019Sgabor}
79219019Sgabor
80219019Sgaborstatic int
81219019Sgaboropen_subdb(struct _citrus_db **subdb, struct _citrus_db *db, const char *src)
82219019Sgabor{
83219019Sgabor	struct _region r;
84219019Sgabor	int ret;
85219019Sgabor
86219019Sgabor	ret = _db_lookup_by_s(db, src, &r, NULL);
87219019Sgabor	if (ret)
88219019Sgabor		return (ret);
89219019Sgabor	ret = _db_open(subdb, &r, _CITRUS_PIVOT_SUB_MAGIC, _db_hash_std, NULL);
90219019Sgabor	if (ret)
91219019Sgabor		return (ret);
92219019Sgabor
93219019Sgabor	return (0);
94219019Sgabor}
95219019Sgabor
96219019Sgabor
97219019Sgabor#define NO_SUCH_FILE	EOPNOTSUPP
98219019Sgaborstatic int
99219019Sgaborfind_best_pivot_pvdb(const char *src, const char *dst, char *pivot,
100219019Sgabor    size_t pvlen, unsigned long *rnorm)
101219019Sgabor{
102219019Sgabor	struct _citrus_db *db1, *db2, *db3;
103219019Sgabor	struct _region fr, r1, r2;
104219019Sgabor	char buf[LINE_MAX];
105219019Sgabor	uint32_t val32;
106219019Sgabor	unsigned long norm;
107219019Sgabor	int i, num, ret;
108219019Sgabor
109219019Sgabor	ret = _map_file(&fr, CS_PIVOT ".pvdb");
110219019Sgabor	if (ret) {
111219019Sgabor		if (ret == ENOENT)
112219019Sgabor			ret = NO_SUCH_FILE;
113219019Sgabor		return (ret);
114219019Sgabor	}
115219019Sgabor	ret = _db_open(&db1, &fr, _CITRUS_PIVOT_MAGIC, _db_hash_std, NULL);
116219019Sgabor	if (ret)
117219019Sgabor		goto quit1;
118219019Sgabor	ret = open_subdb(&db2, db1, src);
119219019Sgabor	if (ret)
120219019Sgabor		goto quit2;
121219019Sgabor
122219019Sgabor	num = _db_get_num_entries(db2);
123219019Sgabor	*rnorm = ULONG_MAX;
124219019Sgabor	for (i = 0; i < num; i++) {
125219019Sgabor		/* iterate each pivot */
126219019Sgabor		ret = _db_get_entry(db2, i, &r1, &r2);
127219019Sgabor		if (ret)
128219019Sgabor			goto quit3;
129219019Sgabor		/* r1:pivot name, r2:norm among src and pivot */
130219019Sgabor		ret = get32(&r2, &val32);
131219019Sgabor		if (ret)
132219019Sgabor			goto quit3;
133219019Sgabor		norm = val32;
134219019Sgabor		snprintf(buf, sizeof(buf), "%.*s",
135219019Sgabor			 (int)_region_size(&r1), (char *)_region_head(&r1));
136219019Sgabor		/* buf: pivot name */
137219019Sgabor		ret = open_subdb(&db3, db1, buf);
138219019Sgabor		if (ret)
139219019Sgabor			goto quit3;
140219019Sgabor		if (_db_lookup_by_s(db3, dst, &r2, NULL) != 0)
141219019Sgabor			goto quit4;
142219019Sgabor		/* r2: norm among pivot and dst */
143219019Sgabor		ret = get32(&r2, &val32);
144219019Sgabor		if (ret)
145219019Sgabor			goto quit4;
146219019Sgabor		norm += val32;
147219019Sgabor		/* judge minimum norm */
148219019Sgabor		if (norm < *rnorm) {
149219019Sgabor			*rnorm = norm;
150219019Sgabor			strlcpy(pivot, buf, pvlen);
151219019Sgabor		}
152219019Sgaborquit4:
153219019Sgabor		_db_close(db3);
154219019Sgabor		if (ret)
155219019Sgabor			goto quit3;
156219019Sgabor	}
157219019Sgaborquit3:
158219019Sgabor	_db_close(db2);
159219019Sgaborquit2:
160219019Sgabor	_db_close(db1);
161219019Sgaborquit1:
162219019Sgabor	_unmap_file(&fr);
163219019Sgabor	if (ret)
164219019Sgabor		return (ret);
165219019Sgabor
166219019Sgabor	if (*rnorm == ULONG_MAX)
167219019Sgabor		return (ENOENT);
168219019Sgabor
169219019Sgabor	return (0);
170219019Sgabor}
171219019Sgabor
172219019Sgabor/* ---------------------------------------------------------------------- */
173219019Sgabor
174219019Sgaborstruct zone {
175219019Sgabor	const char *begin, *end;
176219019Sgabor};
177219019Sgabor
178219019Sgaborstruct parse_arg {
179219019Sgabor	char dst[PATH_MAX];
180219019Sgabor	unsigned long norm;
181219019Sgabor};
182219019Sgabor
183219019Sgaborstatic int
184219019Sgaborparse_line(struct parse_arg *pa, struct _region *r)
185219019Sgabor{
186219019Sgabor	struct zone z1, z2;
187219019Sgabor	char buf[20];
188219019Sgabor	size_t len;
189219019Sgabor
190219019Sgabor	len = _region_size(r);
191219019Sgabor	z1.begin = _bcs_skip_ws_len(_region_head(r), &len);
192219019Sgabor	if (len == 0)
193219019Sgabor		return (EFTYPE);
194219019Sgabor	z1.end = _bcs_skip_nonws_len(z1.begin, &len);
195219019Sgabor	if (len == 0)
196219019Sgabor		return (EFTYPE);
197219019Sgabor	z2.begin = _bcs_skip_ws_len(z1.end, &len);
198219019Sgabor	if (len == 0)
199219019Sgabor		return (EFTYPE);
200219019Sgabor	z2.end = _bcs_skip_nonws_len(z2.begin, &len);
201219019Sgabor
202219019Sgabor	/* z1 : dst name, z2 : norm */
203219019Sgabor	snprintf(pa->dst, sizeof(pa->dst),
204219019Sgabor	    "%.*s", (int)(z1.end-z1.begin), z1.begin);
205219019Sgabor	snprintf(buf, sizeof(buf),
206219019Sgabor	    "%.*s", (int)(z2.end-z2.begin), z2.begin);
207219019Sgabor	pa->norm = _bcs_strtoul(buf, NULL, 0);
208219019Sgabor
209219019Sgabor	return (0);
210219019Sgabor}
211219019Sgabor
212219019Sgaborstatic int
213219019Sgaborfind_dst(struct parse_arg *pasrc, const char *dst)
214219019Sgabor{
215219019Sgabor	struct _lookup *cl;
216219019Sgabor	struct parse_arg padst;
217219019Sgabor	struct _region data;
218219019Sgabor	int ret;
219219019Sgabor
220219019Sgabor	ret = _lookup_seq_open(&cl, CS_PIVOT, _LOOKUP_CASE_IGNORE);
221219019Sgabor	if (ret)
222219019Sgabor		return (ret);
223219019Sgabor
224219019Sgabor	ret = _lookup_seq_lookup(cl, pasrc->dst, &data);
225219019Sgabor	while (ret == 0) {
226219019Sgabor		ret = parse_line(&padst, &data);
227219019Sgabor		if (ret)
228219019Sgabor			break;
229219019Sgabor		if (strcmp(dst, padst.dst) == 0) {
230219019Sgabor			pasrc->norm += padst.norm;
231219019Sgabor			break;
232219019Sgabor		}
233219019Sgabor		ret = _lookup_seq_next(cl, NULL, &data);
234219019Sgabor	}
235219019Sgabor	_lookup_seq_close(cl);
236219019Sgabor
237219019Sgabor	return (ret);
238219019Sgabor}
239219019Sgabor
240219019Sgaborstatic int
241219019Sgaborfind_best_pivot_lookup(const char *src, const char *dst, char *pivot,
242219019Sgabor    size_t pvlen, unsigned long *rnorm)
243219019Sgabor{
244219019Sgabor	struct _lookup *cl;
245219019Sgabor	struct _region data;
246219019Sgabor	struct parse_arg pa;
247219019Sgabor	char pivot_min[PATH_MAX];
248219019Sgabor	unsigned long norm_min;
249219019Sgabor	int ret;
250219019Sgabor
251219019Sgabor	ret = _lookup_seq_open(&cl, CS_PIVOT, _LOOKUP_CASE_IGNORE);
252219019Sgabor	if (ret)
253219019Sgabor		return (ret);
254219019Sgabor
255219019Sgabor	norm_min = ULONG_MAX;
256219019Sgabor
257219019Sgabor	/* find pivot code */
258219019Sgabor	ret = _lookup_seq_lookup(cl, src, &data);
259219019Sgabor	while (ret == 0) {
260219019Sgabor		ret = parse_line(&pa, &data);
261219019Sgabor		if (ret)
262219019Sgabor			break;
263219019Sgabor		ret = find_dst(&pa, dst);
264219019Sgabor		if (ret)
265219019Sgabor			break;
266219019Sgabor		if (pa.norm < norm_min) {
267219019Sgabor			norm_min = pa.norm;
268219019Sgabor			strlcpy(pivot_min, pa.dst, sizeof(pivot_min));
269219019Sgabor		}
270219019Sgabor		ret = _lookup_seq_next(cl, NULL, &data);
271219019Sgabor	}
272219019Sgabor	_lookup_seq_close(cl);
273219019Sgabor
274219019Sgabor	if (ret != ENOENT)
275219019Sgabor		return (ret);
276219019Sgabor	if (norm_min == ULONG_MAX)
277219019Sgabor		return (ENOENT);
278219019Sgabor	strlcpy(pivot, pivot_min, pvlen);
279219019Sgabor	if (rnorm)
280219019Sgabor		*rnorm = norm_min;
281219019Sgabor
282219019Sgabor	return (0);
283219019Sgabor}
284219019Sgabor
285219019Sgaborstatic int
286219019Sgaborfind_best_pivot(const char *src, const char *dst, char *pivot, size_t pvlen,
287219019Sgabor    unsigned long *rnorm)
288219019Sgabor{
289219019Sgabor	int ret;
290219019Sgabor
291219019Sgabor	ret = find_best_pivot_pvdb(src, dst, pivot, pvlen, rnorm);
292219019Sgabor	if (ret == NO_SUCH_FILE)
293219019Sgabor		ret = find_best_pivot_lookup(src, dst, pivot, pvlen, rnorm);
294219019Sgabor
295219019Sgabor	return (ret);
296219019Sgabor}
297219019Sgabor
298219019Sgaborstatic __inline int
299219019Sgaboropen_serial_mapper(struct _citrus_mapper_area *__restrict ma,
300219019Sgabor    struct _citrus_mapper * __restrict * __restrict rcm,
301219019Sgabor    const char *src, const char *pivot, const char *dst)
302219019Sgabor{
303219019Sgabor	char buf[PATH_MAX];
304219019Sgabor
305219019Sgabor	snprintf(buf, sizeof(buf), "%s/%s,%s/%s", src, pivot, pivot, dst);
306219019Sgabor
307219019Sgabor	return (_mapper_open_direct(ma, rcm, "mapper_serial", buf));
308219019Sgabor}
309219019Sgabor
310219019Sgaborstatic struct _citrus_csmapper *csm_none = NULL;
311219019Sgaborstatic int
312219019Sgaborget_none(struct _citrus_mapper_area *__restrict ma,
313219019Sgabor    struct _citrus_csmapper *__restrict *__restrict rcsm)
314219019Sgabor{
315219019Sgabor	int ret;
316219019Sgabor
317219019Sgabor	WLOCK;
318219019Sgabor	if (csm_none) {
319219019Sgabor		*rcsm = csm_none;
320219019Sgabor		ret = 0;
321219019Sgabor		goto quit;
322219019Sgabor	}
323219019Sgabor
324219019Sgabor	ret = _mapper_open_direct(ma, &csm_none, "mapper_none", "");
325219019Sgabor	if (ret)
326219019Sgabor		goto quit;
327219019Sgabor	_mapper_set_persistent(csm_none);
328219019Sgabor
329219019Sgabor	*rcsm = csm_none;
330219019Sgabor	ret = 0;
331219019Sgaborquit:
332219019Sgabor	UNLOCK;
333219019Sgabor	return (ret);
334219019Sgabor}
335219019Sgabor
336219019Sgaborint
337219019Sgabor_citrus_csmapper_open(struct _citrus_csmapper * __restrict * __restrict rcsm,
338219019Sgabor    const char * __restrict src, const char * __restrict dst, uint32_t flags,
339219019Sgabor    unsigned long *rnorm)
340219019Sgabor{
341219019Sgabor	const char *realsrc, *realdst;
342219019Sgabor	char buf1[PATH_MAX], buf2[PATH_MAX], key[PATH_MAX], pivot[PATH_MAX];
343219019Sgabor	unsigned long norm;
344219019Sgabor	int ret;
345219019Sgabor
346219019Sgabor	norm = 0;
347219019Sgabor
348219019Sgabor	ret = _citrus_mapper_create_area(&maparea, _PATH_CSMAPPER);
349219019Sgabor	if (ret)
350219019Sgabor		return (ret);
351219019Sgabor
352219019Sgabor	realsrc = _lookup_alias(CS_ALIAS, src, buf1, sizeof(buf1),
353219019Sgabor	    _LOOKUP_CASE_IGNORE);
354219019Sgabor	realdst = _lookup_alias(CS_ALIAS, dst, buf2, sizeof(buf2),
355219019Sgabor	    _LOOKUP_CASE_IGNORE);
356219019Sgabor	if (!strcmp(realsrc, realdst)) {
357219019Sgabor		ret = get_none(maparea, rcsm);
358219019Sgabor		if (ret == 0 && rnorm != NULL)
359219019Sgabor			*rnorm = 0;
360219019Sgabor		return (ret);
361219019Sgabor	}
362219019Sgabor
363219019Sgabor	snprintf(key, sizeof(key), "%s/%s", realsrc, realdst);
364219019Sgabor
365219019Sgabor	ret = _mapper_open(maparea, rcsm, key);
366219019Sgabor	if (ret == 0) {
367219019Sgabor		if (rnorm != NULL)
368219019Sgabor			*rnorm = 0;
369219019Sgabor		return (0);
370219019Sgabor	}
371219019Sgabor	if (ret != ENOENT || (flags & _CSMAPPER_F_PREVENT_PIVOT)!=0)
372219019Sgabor		return (ret);
373219019Sgabor
374219019Sgabor	ret = find_best_pivot(realsrc, realdst, pivot, sizeof(pivot), &norm);
375219019Sgabor	if (ret)
376219019Sgabor		return (ret);
377219019Sgabor
378219019Sgabor	ret = open_serial_mapper(maparea, rcsm, realsrc, pivot, realdst);
379219019Sgabor	if (ret == 0 && rnorm != NULL)
380219019Sgabor		*rnorm = norm;
381219019Sgabor
382219019Sgabor	return (ret);
383219019Sgabor}
384