]>
git.ipfire.org Git - thirdparty/git.git/blob - xdiff/xprepare.c
2 * LibXDiff by Davide Libenzi ( File Differential Library )
3 * Copyright (C) 2003 Davide Libenzi
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2.1 of the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
19 * Davide Libenzi <davidel@xmailserver.org>
26 #define XDL_KPDIS_RUN 4
27 #define XDL_MAX_EQLIMIT 1024
28 #define XDL_SIMSCAN_WINDOW 100
31 typedef struct s_xdlclass
{
32 struct s_xdlclass
*next
;
39 typedef struct s_xdlclassifier
{
51 static int xdl_init_classifier(xdlclassifier_t
*cf
, long size
, long flags
);
52 static void xdl_free_classifier(xdlclassifier_t
*cf
);
53 static int xdl_classify_record(xdlclassifier_t
*cf
, xrecord_t
**rhash
, unsigned int hbits
,
55 static int xdl_prepare_ctx(mmfile_t
*mf
, long narec
, xpparam_t
const *xpp
,
56 xdlclassifier_t
*cf
, xdfile_t
*xdf
);
57 static void xdl_free_ctx(xdfile_t
*xdf
);
58 static int xdl_clean_mmatch(char const *dis
, long i
, long s
, long e
);
59 static int xdl_cleanup_records(xdfile_t
*xdf1
, xdfile_t
*xdf2
);
60 static int xdl_trim_ends(xdfile_t
*xdf1
, xdfile_t
*xdf2
);
61 static int xdl_optimize_ctxs(xdfile_t
*xdf1
, xdfile_t
*xdf2
);
66 static int xdl_init_classifier(xdlclassifier_t
*cf
, long size
, long flags
) {
69 cf
->hbits
= xdl_hashbits((unsigned int) size
);
70 cf
->hsize
= 1 << cf
->hbits
;
72 if (xdl_cha_init(&cf
->ncha
, sizeof(xdlclass_t
), size
/ 4 + 1) < 0) {
76 if (!(cf
->rchash
= (xdlclass_t
**) xdl_malloc(cf
->hsize
* sizeof(xdlclass_t
*)))) {
78 xdl_cha_free(&cf
->ncha
);
81 memset(cf
->rchash
, 0, cf
->hsize
* sizeof(xdlclass_t
*));
89 static void xdl_free_classifier(xdlclassifier_t
*cf
) {
92 xdl_cha_free(&cf
->ncha
);
96 static int xdl_classify_record(xdlclassifier_t
*cf
, xrecord_t
**rhash
, unsigned int hbits
,
103 hi
= (long) XDL_HASHLONG(rec
->ha
, cf
->hbits
);
104 for (rcrec
= cf
->rchash
[hi
]; rcrec
; rcrec
= rcrec
->next
)
105 if (rcrec
->ha
== rec
->ha
&&
106 xdl_recmatch(rcrec
->line
, rcrec
->size
,
107 rec
->ptr
, rec
->size
, cf
->flags
))
111 if (!(rcrec
= xdl_cha_alloc(&cf
->ncha
))) {
115 rcrec
->idx
= cf
->count
++;
117 rcrec
->size
= rec
->size
;
119 rcrec
->next
= cf
->rchash
[hi
];
120 cf
->rchash
[hi
] = rcrec
;
123 rec
->ha
= (unsigned long) rcrec
->idx
;
125 hi
= (long) XDL_HASHLONG(rec
->ha
, hbits
);
126 rec
->next
= rhash
[hi
];
133 static int xdl_prepare_ctx(mmfile_t
*mf
, long narec
, xpparam_t
const *xpp
,
134 xdlclassifier_t
*cf
, xdfile_t
*xdf
) {
136 long nrec
, hsize
, bsize
;
138 char const *blk
, *cur
, *top
, *prev
;
140 xrecord_t
**recs
, **rrecs
;
152 if (xdl_cha_init(&xdf
->rcha
, sizeof(xrecord_t
), narec
/ 4 + 1) < 0)
154 if (!(recs
= (xrecord_t
**) xdl_malloc(narec
* sizeof(xrecord_t
*))))
157 if (xpp
->flags
& XDF_HISTOGRAM_DIFF
)
160 hbits
= xdl_hashbits((unsigned int) narec
);
162 if (!(rhash
= (xrecord_t
**) xdl_malloc(hsize
* sizeof(xrecord_t
*))))
164 memset(rhash
, 0, hsize
* sizeof(xrecord_t
*));
168 if ((cur
= blk
= xdl_mmfile_first(mf
, &bsize
)) != NULL
) {
169 for (top
= blk
+ bsize
;;) {
171 if (!(cur
= blk
= xdl_mmfile_next(mf
, &bsize
)))
176 hav
= xdl_hash_record(&cur
, top
, xpp
->flags
);
179 if (!(rrecs
= (xrecord_t
**) xdl_realloc(recs
, narec
* sizeof(xrecord_t
*))))
183 if (!(crec
= xdl_cha_alloc(&xdf
->rcha
)))
186 crec
->size
= (long) (cur
- prev
);
190 if (!(xpp
->flags
& XDF_HISTOGRAM_DIFF
) &&
191 xdl_classify_record(cf
, rhash
, hbits
, crec
) < 0)
196 if (!(rchg
= (char *) xdl_malloc((nrec
+ 2) * sizeof(char))))
198 memset(rchg
, 0, (nrec
+ 2) * sizeof(char));
200 if (!(rindex
= (long *) xdl_malloc((nrec
+ 1) * sizeof(long))))
202 if (!(ha
= (unsigned long *) xdl_malloc((nrec
+ 1) * sizeof(unsigned long))))
209 xdf
->rchg
= rchg
+ 1;
210 xdf
->rindex
= rindex
;
214 xdf
->dend
= nrec
- 1;
224 xdl_cha_free(&xdf
->rcha
);
229 static void xdl_free_ctx(xdfile_t
*xdf
) {
231 xdl_free(xdf
->rhash
);
232 xdl_free(xdf
->rindex
);
233 xdl_free(xdf
->rchg
- 1);
236 xdl_cha_free(&xdf
->rcha
);
240 int xdl_prepare_env(mmfile_t
*mf1
, mmfile_t
*mf2
, xpparam_t
const *xpp
,
245 enl1
= xdl_guess_lines(mf1
) + 1;
246 enl2
= xdl_guess_lines(mf2
) + 1;
248 if (!(xpp
->flags
& XDF_HISTOGRAM_DIFF
) &&
249 xdl_init_classifier(&cf
, enl1
+ enl2
+ 1, xpp
->flags
) < 0) {
254 if (xdl_prepare_ctx(mf1
, enl1
, xpp
, &cf
, &xe
->xdf1
) < 0) {
256 xdl_free_classifier(&cf
);
259 if (xdl_prepare_ctx(mf2
, enl2
, xpp
, &cf
, &xe
->xdf2
) < 0) {
261 xdl_free_ctx(&xe
->xdf1
);
262 xdl_free_classifier(&cf
);
266 if (!(xpp
->flags
& XDF_HISTOGRAM_DIFF
))
267 xdl_free_classifier(&cf
);
269 if (!(xpp
->flags
& XDF_PATIENCE_DIFF
) &&
270 !(xpp
->flags
& XDF_HISTOGRAM_DIFF
) &&
271 xdl_optimize_ctxs(&xe
->xdf1
, &xe
->xdf2
) < 0) {
273 xdl_free_ctx(&xe
->xdf2
);
274 xdl_free_ctx(&xe
->xdf1
);
282 void xdl_free_env(xdfenv_t
*xe
) {
284 xdl_free_ctx(&xe
->xdf2
);
285 xdl_free_ctx(&xe
->xdf1
);
289 static int xdl_clean_mmatch(char const *dis
, long i
, long s
, long e
) {
290 long r
, rdis0
, rpdis0
, rdis1
, rpdis1
;
293 * Limits the window the is examined during the similar-lines
294 * scan. The loops below stops when dis[i - r] == 1 (line that
295 * has no match), but there are corner cases where the loop
296 * proceed all the way to the extremities by causing huge
297 * performance penalties in case of big files.
299 if (i
- s
> XDL_SIMSCAN_WINDOW
)
300 s
= i
- XDL_SIMSCAN_WINDOW
;
301 if (e
- i
> XDL_SIMSCAN_WINDOW
)
302 e
= i
+ XDL_SIMSCAN_WINDOW
;
305 * Scans the lines before 'i' to find a run of lines that either
306 * have no match (dis[j] == 0) or have multiple matches (dis[j] > 1).
307 * Note that we always call this function with dis[i] > 1, so the
308 * current line (i) is already a multimatch line.
310 for (r
= 1, rdis0
= 0, rpdis0
= 1; (i
- r
) >= s
; r
++) {
313 else if (dis
[i
- r
] == 2)
319 * If the run before the line 'i' found only multimatch lines, we
320 * return 0 and hence we don't make the current line (i) discarded.
321 * We want to discard multimatch lines only when they appear in the
322 * middle of runs with nomatch lines (dis[j] == 0).
326 for (r
= 1, rdis1
= 0, rpdis1
= 1; (i
+ r
) <= e
; r
++) {
329 else if (dis
[i
+ r
] == 2)
335 * If the run after the line 'i' found only multimatch lines, we
336 * return 0 and hence we don't make the current line (i) discarded.
343 return rpdis1
* XDL_KPDIS_RUN
< (rpdis1
+ rdis1
);
348 * Try to reduce the problem complexity, discard records that have no
349 * matches on the other file. Also, lines that have multiple matches
350 * might be potentially discarded if they happear in a run of discardable.
352 static int xdl_cleanup_records(xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
353 long i
, nm
, rhi
, nreff
, mlim
;
357 char *dis
, *dis1
, *dis2
;
359 if (!(dis
= (char *) xdl_malloc(xdf1
->nrec
+ xdf2
->nrec
+ 2))) {
363 memset(dis
, 0, xdf1
->nrec
+ xdf2
->nrec
+ 2);
365 dis2
= dis1
+ xdf1
->nrec
+ 1;
367 if ((mlim
= xdl_bogosqrt(xdf1
->nrec
)) > XDL_MAX_EQLIMIT
)
368 mlim
= XDL_MAX_EQLIMIT
;
369 for (i
= xdf1
->dstart
, recs
= &xdf1
->recs
[xdf1
->dstart
]; i
<= xdf1
->dend
; i
++, recs
++) {
371 rhi
= (long) XDL_HASHLONG(hav
, xdf2
->hbits
);
372 for (nm
= 0, rec
= xdf2
->rhash
[rhi
]; rec
; rec
= rec
->next
)
373 if (rec
->ha
== hav
&& ++nm
== mlim
)
375 dis1
[i
] = (nm
== 0) ? 0: (nm
>= mlim
) ? 2: 1;
378 if ((mlim
= xdl_bogosqrt(xdf2
->nrec
)) > XDL_MAX_EQLIMIT
)
379 mlim
= XDL_MAX_EQLIMIT
;
380 for (i
= xdf2
->dstart
, recs
= &xdf2
->recs
[xdf2
->dstart
]; i
<= xdf2
->dend
; i
++, recs
++) {
382 rhi
= (long) XDL_HASHLONG(hav
, xdf1
->hbits
);
383 for (nm
= 0, rec
= xdf1
->rhash
[rhi
]; rec
; rec
= rec
->next
)
384 if (rec
->ha
== hav
&& ++nm
== mlim
)
386 dis2
[i
] = (nm
== 0) ? 0: (nm
>= mlim
) ? 2: 1;
389 for (nreff
= 0, i
= xdf1
->dstart
, recs
= &xdf1
->recs
[xdf1
->dstart
];
390 i
<= xdf1
->dend
; i
++, recs
++) {
392 (dis1
[i
] == 2 && !xdl_clean_mmatch(dis1
, i
, xdf1
->dstart
, xdf1
->dend
))) {
393 xdf1
->rindex
[nreff
] = i
;
394 xdf1
->ha
[nreff
] = (*recs
)->ha
;
401 for (nreff
= 0, i
= xdf2
->dstart
, recs
= &xdf2
->recs
[xdf2
->dstart
];
402 i
<= xdf2
->dend
; i
++, recs
++) {
404 (dis2
[i
] == 2 && !xdl_clean_mmatch(dis2
, i
, xdf2
->dstart
, xdf2
->dend
))) {
405 xdf2
->rindex
[nreff
] = i
;
406 xdf2
->ha
[nreff
] = (*recs
)->ha
;
420 * Early trim initial and terminal matching records.
422 static int xdl_trim_ends(xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
424 xrecord_t
**recs1
, **recs2
;
428 for (i
= 0, lim
= XDL_MIN(xdf1
->nrec
, xdf2
->nrec
); i
< lim
;
429 i
++, recs1
++, recs2
++)
430 if ((*recs1
)->ha
!= (*recs2
)->ha
)
433 xdf1
->dstart
= xdf2
->dstart
= i
;
435 recs1
= xdf1
->recs
+ xdf1
->nrec
- 1;
436 recs2
= xdf2
->recs
+ xdf2
->nrec
- 1;
437 for (lim
-= i
, i
= 0; i
< lim
; i
++, recs1
--, recs2
--)
438 if ((*recs1
)->ha
!= (*recs2
)->ha
)
441 xdf1
->dend
= xdf1
->nrec
- i
- 1;
442 xdf2
->dend
= xdf2
->nrec
- i
- 1;
448 static int xdl_optimize_ctxs(xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
450 if (xdl_trim_ends(xdf1
, xdf2
) < 0 ||
451 xdl_cleanup_records(xdf1
, xdf2
) < 0) {