]>
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, see
17 * <http://www.gnu.org/licenses/>.
19 * Davide Libenzi <davidel@xmailserver.org>
26 #define XDL_KPDIS_RUN 4
27 #define XDL_MAX_EQLIMIT 1024
28 #define XDL_SIMSCAN_WINDOW 100
29 #define XDL_GUESS_NLINES1 256
30 #define XDL_GUESS_NLINES2 20
33 typedef struct s_xdlclass
{
34 struct s_xdlclass
*next
;
42 typedef struct s_xdlclassifier
{
56 static int xdl_init_classifier(xdlclassifier_t
*cf
, long size
, long flags
);
57 static void xdl_free_classifier(xdlclassifier_t
*cf
);
58 static int xdl_classify_record(unsigned int pass
, xdlclassifier_t
*cf
, xrecord_t
**rhash
,
59 unsigned int hbits
, xrecord_t
*rec
);
60 static int xdl_prepare_ctx(unsigned int pass
, mmfile_t
*mf
, long narec
, xpparam_t
const *xpp
,
61 xdlclassifier_t
*cf
, xdfile_t
*xdf
);
62 static void xdl_free_ctx(xdfile_t
*xdf
);
63 static int xdl_clean_mmatch(char const *dis
, long i
, long s
, long e
);
64 static int xdl_cleanup_records(xdlclassifier_t
*cf
, xdfile_t
*xdf1
, xdfile_t
*xdf2
);
65 static int xdl_trim_ends(xdfile_t
*xdf1
, xdfile_t
*xdf2
);
66 static int xdl_optimize_ctxs(xdlclassifier_t
*cf
, xdfile_t
*xdf1
, xdfile_t
*xdf2
);
71 static int xdl_init_classifier(xdlclassifier_t
*cf
, long size
, long flags
) {
74 cf
->hbits
= xdl_hashbits((unsigned int) size
);
75 cf
->hsize
= 1 << cf
->hbits
;
77 if (xdl_cha_init(&cf
->ncha
, sizeof(xdlclass_t
), size
/ 4 + 1) < 0) {
81 if (!(cf
->rchash
= (xdlclass_t
**) xdl_malloc(cf
->hsize
* sizeof(xdlclass_t
*)))) {
83 xdl_cha_free(&cf
->ncha
);
86 memset(cf
->rchash
, 0, cf
->hsize
* sizeof(xdlclass_t
*));
89 if (!(cf
->rcrecs
= (xdlclass_t
**) xdl_malloc(cf
->alloc
* sizeof(xdlclass_t
*)))) {
92 xdl_cha_free(&cf
->ncha
);
102 static void xdl_free_classifier(xdlclassifier_t
*cf
) {
104 xdl_free(cf
->rcrecs
);
105 xdl_free(cf
->rchash
);
106 xdl_cha_free(&cf
->ncha
);
110 static int xdl_classify_record(unsigned int pass
, xdlclassifier_t
*cf
, xrecord_t
**rhash
,
111 unsigned int hbits
, xrecord_t
*rec
) {
118 hi
= (long) XDL_HASHLONG(rec
->ha
, cf
->hbits
);
119 for (rcrec
= cf
->rchash
[hi
]; rcrec
; rcrec
= rcrec
->next
)
120 if (rcrec
->ha
== rec
->ha
&&
121 xdl_recmatch(rcrec
->line
, rcrec
->size
,
122 rec
->ptr
, rec
->size
, cf
->flags
))
126 if (!(rcrec
= xdl_cha_alloc(&cf
->ncha
))) {
130 rcrec
->idx
= cf
->count
++;
131 if (cf
->count
> cf
->alloc
) {
133 if (!(rcrecs
= (xdlclass_t
**) xdl_realloc(cf
->rcrecs
, cf
->alloc
* sizeof(xdlclass_t
*)))) {
139 cf
->rcrecs
[rcrec
->idx
] = rcrec
;
141 rcrec
->size
= rec
->size
;
143 rcrec
->len1
= rcrec
->len2
= 0;
144 rcrec
->next
= cf
->rchash
[hi
];
145 cf
->rchash
[hi
] = rcrec
;
148 (pass
== 1) ? rcrec
->len1
++ : rcrec
->len2
++;
150 rec
->ha
= (unsigned long) rcrec
->idx
;
152 hi
= (long) XDL_HASHLONG(rec
->ha
, hbits
);
153 rec
->next
= rhash
[hi
];
160 static int xdl_prepare_ctx(unsigned int pass
, mmfile_t
*mf
, long narec
, xpparam_t
const *xpp
,
161 xdlclassifier_t
*cf
, xdfile_t
*xdf
) {
163 long nrec
, hsize
, bsize
;
165 char const *blk
, *cur
, *top
, *prev
;
167 xrecord_t
**recs
, **rrecs
;
179 if (xdl_cha_init(&xdf
->rcha
, sizeof(xrecord_t
), narec
/ 4 + 1) < 0)
181 if (!(recs
= (xrecord_t
**) xdl_malloc(narec
* sizeof(xrecord_t
*))))
184 hbits
= xdl_hashbits((unsigned int) narec
);
186 if (!(rhash
= (xrecord_t
**) xdl_malloc(hsize
* sizeof(xrecord_t
*))))
188 memset(rhash
, 0, hsize
* sizeof(xrecord_t
*));
191 if ((cur
= blk
= xdl_mmfile_first(mf
, &bsize
))) {
192 for (top
= blk
+ bsize
; cur
< top
; ) {
194 hav
= xdl_hash_record(&cur
, top
, xpp
->flags
);
197 if (!(rrecs
= (xrecord_t
**) xdl_realloc(recs
, narec
* sizeof(xrecord_t
*))))
201 if (!(crec
= xdl_cha_alloc(&xdf
->rcha
)))
204 crec
->size
= (long) (cur
- prev
);
207 if (xdl_classify_record(pass
, cf
, rhash
, hbits
, crec
) < 0)
212 if (!(rchg
= (char *) xdl_malloc((nrec
+ 2) * sizeof(char))))
214 memset(rchg
, 0, (nrec
+ 2) * sizeof(char));
216 if ((XDF_DIFF_ALG(xpp
->flags
) != XDF_PATIENCE_DIFF
) &&
217 (XDF_DIFF_ALG(xpp
->flags
) != XDF_HISTOGRAM_DIFF
)) {
218 if (!(rindex
= xdl_malloc((nrec
+ 1) * sizeof(*rindex
))))
220 if (!(ha
= xdl_malloc((nrec
+ 1) * sizeof(*ha
))))
228 xdf
->rchg
= rchg
+ 1;
229 xdf
->rindex
= rindex
;
233 xdf
->dend
= nrec
- 1;
243 xdl_cha_free(&xdf
->rcha
);
248 static void xdl_free_ctx(xdfile_t
*xdf
) {
250 xdl_free(xdf
->rhash
);
251 xdl_free(xdf
->rindex
);
252 xdl_free(xdf
->rchg
- 1);
255 xdl_cha_free(&xdf
->rcha
);
259 int xdl_prepare_env(mmfile_t
*mf1
, mmfile_t
*mf2
, xpparam_t
const *xpp
,
261 long enl1
, enl2
, sample
;
264 memset(&cf
, 0, sizeof(cf
));
267 * For histogram diff, we can afford a smaller sample size and
268 * thus a poorer estimate of the number of lines, as the hash
269 * table (rhash) won't be filled up/grown. The number of lines
270 * (nrecs) will be updated correctly anyway by
273 sample
= (XDF_DIFF_ALG(xpp
->flags
) == XDF_HISTOGRAM_DIFF
274 ? XDL_GUESS_NLINES2
: XDL_GUESS_NLINES1
);
276 enl1
= xdl_guess_lines(mf1
, sample
) + 1;
277 enl2
= xdl_guess_lines(mf2
, sample
) + 1;
279 if (xdl_init_classifier(&cf
, enl1
+ enl2
+ 1, xpp
->flags
) < 0)
282 if (xdl_prepare_ctx(1, mf1
, enl1
, xpp
, &cf
, &xe
->xdf1
) < 0) {
284 xdl_free_classifier(&cf
);
287 if (xdl_prepare_ctx(2, mf2
, enl2
, xpp
, &cf
, &xe
->xdf2
) < 0) {
289 xdl_free_ctx(&xe
->xdf1
);
290 xdl_free_classifier(&cf
);
294 if ((XDF_DIFF_ALG(xpp
->flags
) != XDF_PATIENCE_DIFF
) &&
295 (XDF_DIFF_ALG(xpp
->flags
) != XDF_HISTOGRAM_DIFF
) &&
296 xdl_optimize_ctxs(&cf
, &xe
->xdf1
, &xe
->xdf2
) < 0) {
298 xdl_free_ctx(&xe
->xdf2
);
299 xdl_free_ctx(&xe
->xdf1
);
300 xdl_free_classifier(&cf
);
304 xdl_free_classifier(&cf
);
310 void xdl_free_env(xdfenv_t
*xe
) {
312 xdl_free_ctx(&xe
->xdf2
);
313 xdl_free_ctx(&xe
->xdf1
);
317 static int xdl_clean_mmatch(char const *dis
, long i
, long s
, long e
) {
318 long r
, rdis0
, rpdis0
, rdis1
, rpdis1
;
321 * Limits the window the is examined during the similar-lines
322 * scan. The loops below stops when dis[i - r] == 1 (line that
323 * has no match), but there are corner cases where the loop
324 * proceed all the way to the extremities by causing huge
325 * performance penalties in case of big files.
327 if (i
- s
> XDL_SIMSCAN_WINDOW
)
328 s
= i
- XDL_SIMSCAN_WINDOW
;
329 if (e
- i
> XDL_SIMSCAN_WINDOW
)
330 e
= i
+ XDL_SIMSCAN_WINDOW
;
333 * Scans the lines before 'i' to find a run of lines that either
334 * have no match (dis[j] == 0) or have multiple matches (dis[j] > 1).
335 * Note that we always call this function with dis[i] > 1, so the
336 * current line (i) is already a multimatch line.
338 for (r
= 1, rdis0
= 0, rpdis0
= 1; (i
- r
) >= s
; r
++) {
341 else if (dis
[i
- r
] == 2)
347 * If the run before the line 'i' found only multimatch lines, we
348 * return 0 and hence we don't make the current line (i) discarded.
349 * We want to discard multimatch lines only when they appear in the
350 * middle of runs with nomatch lines (dis[j] == 0).
354 for (r
= 1, rdis1
= 0, rpdis1
= 1; (i
+ r
) <= e
; r
++) {
357 else if (dis
[i
+ r
] == 2)
363 * If the run after the line 'i' found only multimatch lines, we
364 * return 0 and hence we don't make the current line (i) discarded.
371 return rpdis1
* XDL_KPDIS_RUN
< (rpdis1
+ rdis1
);
376 * Try to reduce the problem complexity, discard records that have no
377 * matches on the other file. Also, lines that have multiple matches
378 * might be potentially discarded if they happear in a run of discardable.
380 static int xdl_cleanup_records(xdlclassifier_t
*cf
, xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
381 long i
, nm
, nreff
, mlim
;
384 char *dis
, *dis1
, *dis2
;
386 if (!(dis
= (char *) xdl_malloc(xdf1
->nrec
+ xdf2
->nrec
+ 2))) {
390 memset(dis
, 0, xdf1
->nrec
+ xdf2
->nrec
+ 2);
392 dis2
= dis1
+ xdf1
->nrec
+ 1;
394 if ((mlim
= xdl_bogosqrt(xdf1
->nrec
)) > XDL_MAX_EQLIMIT
)
395 mlim
= XDL_MAX_EQLIMIT
;
396 for (i
= xdf1
->dstart
, recs
= &xdf1
->recs
[xdf1
->dstart
]; i
<= xdf1
->dend
; i
++, recs
++) {
397 rcrec
= cf
->rcrecs
[(*recs
)->ha
];
398 nm
= rcrec
? rcrec
->len2
: 0;
399 dis1
[i
] = (nm
== 0) ? 0: (nm
>= mlim
) ? 2: 1;
402 if ((mlim
= xdl_bogosqrt(xdf2
->nrec
)) > XDL_MAX_EQLIMIT
)
403 mlim
= XDL_MAX_EQLIMIT
;
404 for (i
= xdf2
->dstart
, recs
= &xdf2
->recs
[xdf2
->dstart
]; i
<= xdf2
->dend
; i
++, recs
++) {
405 rcrec
= cf
->rcrecs
[(*recs
)->ha
];
406 nm
= rcrec
? rcrec
->len1
: 0;
407 dis2
[i
] = (nm
== 0) ? 0: (nm
>= mlim
) ? 2: 1;
410 for (nreff
= 0, i
= xdf1
->dstart
, recs
= &xdf1
->recs
[xdf1
->dstart
];
411 i
<= xdf1
->dend
; i
++, recs
++) {
413 (dis1
[i
] == 2 && !xdl_clean_mmatch(dis1
, i
, xdf1
->dstart
, xdf1
->dend
))) {
414 xdf1
->rindex
[nreff
] = i
;
415 xdf1
->ha
[nreff
] = (*recs
)->ha
;
422 for (nreff
= 0, i
= xdf2
->dstart
, recs
= &xdf2
->recs
[xdf2
->dstart
];
423 i
<= xdf2
->dend
; i
++, recs
++) {
425 (dis2
[i
] == 2 && !xdl_clean_mmatch(dis2
, i
, xdf2
->dstart
, xdf2
->dend
))) {
426 xdf2
->rindex
[nreff
] = i
;
427 xdf2
->ha
[nreff
] = (*recs
)->ha
;
441 * Early trim initial and terminal matching records.
443 static int xdl_trim_ends(xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
445 xrecord_t
**recs1
, **recs2
;
449 for (i
= 0, lim
= XDL_MIN(xdf1
->nrec
, xdf2
->nrec
); i
< lim
;
450 i
++, recs1
++, recs2
++)
451 if ((*recs1
)->ha
!= (*recs2
)->ha
)
454 xdf1
->dstart
= xdf2
->dstart
= i
;
456 recs1
= xdf1
->recs
+ xdf1
->nrec
- 1;
457 recs2
= xdf2
->recs
+ xdf2
->nrec
- 1;
458 for (lim
-= i
, i
= 0; i
< lim
; i
++, recs1
--, recs2
--)
459 if ((*recs1
)->ha
!= (*recs2
)->ha
)
462 xdf1
->dend
= xdf1
->nrec
- i
- 1;
463 xdf2
->dend
= xdf2
->nrec
- i
- 1;
469 static int xdl_optimize_ctxs(xdlclassifier_t
*cf
, xdfile_t
*xdf1
, xdfile_t
*xdf2
) {
471 if (xdl_trim_ends(xdf1
, xdf2
) < 0 ||
472 xdl_cleanup_records(cf
, xdf1
, xdf2
) < 0) {