Mercurial > vim

annotate src/xdiff/xprepare.c @ 22567:2245779ade24

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added tag v8.2.1831 for changeset 12b4e321a625a409d762da1a4cc9a5fd4a1ebcdb
author Bram Moolenaar <Bram@vim.org>
date Sun, 11 Oct 2020 14:30:04 +0200
parents 195e8b1fcbbf
children f84e5db372ea
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
14696
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
1 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
2 * LibXDiff by Davide Libenzi ( File Differential Library )
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
3 * Copyright (C) 2003 Davide Libenzi
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
4 *
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
5 * This library is free software; you can redistribute it and/or
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
6 * modify it under the terms of the GNU Lesser General Public
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
7 * License as published by the Free Software Foundation; either
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
8 * version 2.1 of the License, or (at your option) any later version.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
9 *
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
10 * This library is distributed in the hope that it will be useful,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
13 * Lesser General Public License for more details.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
14 *
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
15 * You should have received a copy of the GNU Lesser General Public
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
16 * License along with this library; if not, see
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
17 * <http://www.gnu.org/licenses/>.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
18 *
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
19 * Davide Libenzi <davidel@xmailserver.org>
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
20 *
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
21 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
22
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
23 #include "xinclude.h"
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
24
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
25
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
26 #define XDL_KPDIS_RUN 4
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
27 #define XDL_MAX_EQLIMIT 1024
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
28 #define XDL_SIMSCAN_WINDOW 100
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
29 #define XDL_GUESS_NLINES1 256
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
30 #define XDL_GUESS_NLINES2 20
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
31
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
32
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
33 typedef struct s_xdlclass {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
34 struct s_xdlclass *next;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
35 unsigned long ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
36 char const *line;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
37 long size;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
38 long idx;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
39 long len1, len2;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
40 } xdlclass_t;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
41
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
42 typedef struct s_xdlclassifier {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
43 unsigned int hbits;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
44 long hsize;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
45 xdlclass_t **rchash;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
46 chastore_t ncha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
47 xdlclass_t **rcrecs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
48 long alloc;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
49 long count;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
50 long flags;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
51 } xdlclassifier_t;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
52
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
53
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
54
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
55
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
56 static int xdl_init_classifier(xdlclassifier_t *cf, long size, long flags);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
57 static void xdl_free_classifier(xdlclassifier_t *cf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
58 static int xdl_classify_record(unsigned int pass, xdlclassifier_t *cf, xrecord_t **rhash,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
59 unsigned int hbits, xrecord_t *rec);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
60 static int xdl_prepare_ctx(unsigned int pass, mmfile_t *mf, long narec, xpparam_t const *xpp,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
61 xdlclassifier_t *cf, xdfile_t *xdf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
62 static void xdl_free_ctx(xdfile_t *xdf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
63 static int xdl_clean_mmatch(char const *dis, long i, long s, long e);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
64 static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
65 static int xdl_trim_ends(xdfile_t *xdf1, xdfile_t *xdf2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
66 static int xdl_optimize_ctxs(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
67
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
68
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
69
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
70
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
71 static int xdl_init_classifier(xdlclassifier_t *cf, long size, long flags) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
72 cf->flags = flags;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
73
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
74 cf->hbits = xdl_hashbits((unsigned int) size);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
75 cf->hsize = 1 << cf->hbits;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
76
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
77 if (xdl_cha_init(&cf->ncha, sizeof(xdlclass_t), size / 4 + 1) < 0) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
78
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
79 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
80 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
81 if (!(cf->rchash = (xdlclass_t **) xdl_malloc(cf->hsize * sizeof(xdlclass_t *)))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
82
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
83 xdl_cha_free(&cf->ncha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
84 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
85 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
86 memset(cf->rchash, 0, cf->hsize * sizeof(xdlclass_t *));
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
87
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
88 cf->alloc = size;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
89 if (!(cf->rcrecs = (xdlclass_t **) xdl_malloc(cf->alloc * sizeof(xdlclass_t *)))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
90
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
91 xdl_free(cf->rchash);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
92 xdl_cha_free(&cf->ncha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
93 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
94 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
95
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
96 cf->count = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
97
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
98 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
99 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
100
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
101
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
102 static void xdl_free_classifier(xdlclassifier_t *cf) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
103
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
104 xdl_free(cf->rcrecs);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
105 xdl_free(cf->rchash);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
106 xdl_cha_free(&cf->ncha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
107 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
108
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
109
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
110 static int xdl_classify_record(unsigned int pass, xdlclassifier_t *cf, xrecord_t **rhash,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
111 unsigned int hbits, xrecord_t *rec) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
112 long hi;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
113 char const *line;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
114 xdlclass_t *rcrec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
115 xdlclass_t **rcrecs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
116
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
117 line = rec->ptr;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
118 hi = (long) XDL_HASHLONG(rec->ha, cf->hbits);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
119 for (rcrec = cf->rchash[hi]; rcrec; rcrec = rcrec->next)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
120 if (rcrec->ha == rec->ha &&
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
121 xdl_recmatch(rcrec->line, rcrec->size,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
122 rec->ptr, rec->size, cf->flags))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
123 break;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
124
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
125 if (!rcrec) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
126 if (!(rcrec = xdl_cha_alloc(&cf->ncha))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
127
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
128 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
129 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
130 rcrec->idx = cf->count++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
131 if (cf->count > cf->alloc) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
132 cf->alloc *= 2;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
133 if (!(rcrecs = (xdlclass_t **) xdl_realloc(cf->rcrecs, cf->alloc * sizeof(xdlclass_t *)))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
134
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
135 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
136 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
137 cf->rcrecs = rcrecs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
138 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
139 cf->rcrecs[rcrec->idx] = rcrec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
140 rcrec->line = line;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
141 rcrec->size = rec->size;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
142 rcrec->ha = rec->ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
143 rcrec->len1 = rcrec->len2 = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
144 rcrec->next = cf->rchash[hi];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
145 cf->rchash[hi] = rcrec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
146 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
147
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
148 (pass == 1) ? rcrec->len1++ : rcrec->len2++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
149
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
150 rec->ha = (unsigned long) rcrec->idx;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
151
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
152 hi = (long) XDL_HASHLONG(rec->ha, hbits);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
153 rec->next = rhash[hi];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
154 rhash[hi] = rec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
155
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
156 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
157 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
158
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
159
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
160 static int xdl_prepare_ctx(unsigned int pass, mmfile_t *mf, long narec, xpparam_t const *xpp,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
161 xdlclassifier_t *cf, xdfile_t *xdf) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
162 unsigned int hbits;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
163 long nrec, hsize, bsize;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
164 unsigned long hav;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
165 char const *blk, *cur, *top, *prev;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
166 xrecord_t *crec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
167 xrecord_t **recs, **rrecs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
168 xrecord_t **rhash;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
169 unsigned long *ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
170 char *rchg;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
171 long *rindex;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
172
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
173 ha = NULL;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
174 rindex = NULL;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
175 rchg = NULL;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
176 rhash = NULL;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
177 recs = NULL;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
178
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
179 if (xdl_cha_init(&xdf->rcha, sizeof(xrecord_t), narec / 4 + 1) < 0)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
180 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
181 if (!(recs = (xrecord_t **) xdl_malloc(narec * sizeof(xrecord_t *))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
182 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
183
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
184 if (XDF_DIFF_ALG(xpp->flags) == XDF_HISTOGRAM_DIFF)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
185 hbits = hsize = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
186 else {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
187 hbits = xdl_hashbits((unsigned int) narec);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
188 hsize = 1 << hbits;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
189 if (!(rhash = (xrecord_t **) xdl_malloc(hsize * sizeof(xrecord_t *))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
190 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
191 memset(rhash, 0, hsize * sizeof(xrecord_t *));
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
192 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
193
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
194 nrec = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
195 if ((cur = blk = xdl_mmfile_first(mf, &bsize)) != NULL) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
196 for (top = blk + bsize; cur < top; ) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
197 prev = cur;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
198 hav = xdl_hash_record(&cur, top, xpp->flags);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
199 if (nrec >= narec) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
200 narec *= 2;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
201 if (!(rrecs = (xrecord_t **) xdl_realloc(recs, narec * sizeof(xrecord_t *))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
202 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
203 recs = rrecs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
204 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
205 if (!(crec = xdl_cha_alloc(&xdf->rcha)))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
206 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
207 crec->ptr = prev;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
208 crec->size = (long) (cur - prev);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
209 crec->ha = hav;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
210 recs[nrec++] = crec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
211
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
212 if ((XDF_DIFF_ALG(xpp->flags) != XDF_HISTOGRAM_DIFF) &&
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
213 xdl_classify_record(pass, cf, rhash, hbits, crec) < 0)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
214 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
215 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
216 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
217
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
218 if (!(rchg = (char *) xdl_malloc((nrec + 2) * sizeof(char))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
219 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
220 memset(rchg, 0, (nrec + 2) * sizeof(char));
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
221
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
222 if (!(rindex = (long *) xdl_malloc((nrec + 1) * sizeof(long))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
223 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
224 if (!(ha = (unsigned long *) xdl_malloc((nrec + 1) * sizeof(unsigned long))))
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
225 goto abort;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
226
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
227 xdf->nrec = nrec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
228 xdf->recs = recs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
229 xdf->hbits = hbits;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
230 xdf->rhash = rhash;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
231 xdf->rchg = rchg + 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
232 xdf->rindex = rindex;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
233 xdf->nreff = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
234 xdf->ha = ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
235 xdf->dstart = 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
236 xdf->dend = nrec - 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
237
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
238 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
239
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
240 abort:
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
241 xdl_free(ha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
242 xdl_free(rindex);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
243 xdl_free(rchg);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
244 xdl_free(rhash);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
245 xdl_free(recs);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
246 xdl_cha_free(&xdf->rcha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
247 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
248 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
249
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
250
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
251 static void xdl_free_ctx(xdfile_t *xdf) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
252
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
253 xdl_free(xdf->rhash);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
254 xdl_free(xdf->rindex);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
255 xdl_free(xdf->rchg - 1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
256 xdl_free(xdf->ha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
257 xdl_free(xdf->recs);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
258 xdl_cha_free(&xdf->rcha);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
259 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
260
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
261
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
262 int xdl_prepare_env(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
263 xdfenv_t *xe) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
264 long enl1, enl2, sample;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
265 xdlclassifier_t cf;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
266
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
267 memset(&cf, 0, sizeof(cf));
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
268
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
269 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
270 * For histogram diff, we can afford a smaller sample size and
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
271 * thus a poorer estimate of the number of lines, as the hash
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
272 * table (rhash) won't be filled up/grown. The number of lines
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
273 * (nrecs) will be updated correctly anyway by
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
274 * xdl_prepare_ctx().
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
275 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
276 sample = (XDF_DIFF_ALG(xpp->flags) == XDF_HISTOGRAM_DIFF
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
277 ? XDL_GUESS_NLINES2 : XDL_GUESS_NLINES1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
278
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
279 enl1 = xdl_guess_lines(mf1, sample) + 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
280 enl2 = xdl_guess_lines(mf2, sample) + 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
281
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
282 if (XDF_DIFF_ALG(xpp->flags) != XDF_HISTOGRAM_DIFF &&
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
283 xdl_init_classifier(&cf, enl1 + enl2 + 1, xpp->flags) < 0)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
284 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
285
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
286 if (xdl_prepare_ctx(1, mf1, enl1, xpp, &cf, &xe->xdf1) < 0) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
287
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
288 xdl_free_classifier(&cf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
289 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
290 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
291 if (xdl_prepare_ctx(2, mf2, enl2, xpp, &cf, &xe->xdf2) < 0) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
292
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
293 xdl_free_ctx(&xe->xdf1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
294 xdl_free_classifier(&cf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
295 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
296 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
297
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
298 if ((XDF_DIFF_ALG(xpp->flags) != XDF_PATIENCE_DIFF) &&
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
299 (XDF_DIFF_ALG(xpp->flags) != XDF_HISTOGRAM_DIFF) &&
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
300 xdl_optimize_ctxs(&cf, &xe->xdf1, &xe->xdf2) < 0) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
301
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
302 xdl_free_ctx(&xe->xdf2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
303 xdl_free_ctx(&xe->xdf1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
304 xdl_free_classifier(&cf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
305 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
306 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
307
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
308 if (XDF_DIFF_ALG(xpp->flags) != XDF_HISTOGRAM_DIFF)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
309 xdl_free_classifier(&cf);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
310
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
311 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
312 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
313
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
314
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
315 void xdl_free_env(xdfenv_t *xe) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
316
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
317 xdl_free_ctx(&xe->xdf2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
318 xdl_free_ctx(&xe->xdf1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
319 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
320
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
321
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
322 static int xdl_clean_mmatch(char const *dis, long i, long s, long e) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
323 long r, rdis0, rpdis0, rdis1, rpdis1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
324
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
325 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
326 * Limits the window the is examined during the similar-lines
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
327 * scan. The loops below stops when dis[i - r] == 1 (line that
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
328 * has no match), but there are corner cases where the loop
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
329 * proceed all the way to the extremities by causing huge
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
330 * performance penalties in case of big files.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
331 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
332 if (i - s > XDL_SIMSCAN_WINDOW)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
333 s = i - XDL_SIMSCAN_WINDOW;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
334 if (e - i > XDL_SIMSCAN_WINDOW)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
335 e = i + XDL_SIMSCAN_WINDOW;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
336
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
337 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
338 * Scans the lines before 'i' to find a run of lines that either
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
339 * have no match (dis[j] == 0) or have multiple matches (dis[j] > 1).
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
340 * Note that we always call this function with dis[i] > 1, so the
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
341 * current line (i) is already a multimatch line.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
342 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
343 for (r = 1, rdis0 = 0, rpdis0 = 1; (i - r) >= s; r++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
344 if (!dis[i - r])
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
345 rdis0++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
346 else if (dis[i - r] == 2)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
347 rpdis0++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
348 else
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
349 break;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
350 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
351 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
352 * If the run before the line 'i' found only multimatch lines, we
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
353 * return 0 and hence we don't make the current line (i) discarded.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
354 * We want to discard multimatch lines only when they appear in the
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
355 * middle of runs with nomatch lines (dis[j] == 0).
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
356 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
357 if (rdis0 == 0)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
358 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
359 for (r = 1, rdis1 = 0, rpdis1 = 1; (i + r) <= e; r++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
360 if (!dis[i + r])
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
361 rdis1++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
362 else if (dis[i + r] == 2)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
363 rpdis1++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
364 else
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
365 break;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
366 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
367 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
368 * If the run after the line 'i' found only multimatch lines, we
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
369 * return 0 and hence we don't make the current line (i) discarded.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
370 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
371 if (rdis1 == 0)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
372 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
373 rdis1 += rdis0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
374 rpdis1 += rpdis0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
375
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
376 return rpdis1 * XDL_KPDIS_RUN < (rpdis1 + rdis1);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
377 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
378
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
379
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
380 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
381 * Try to reduce the problem complexity, discard records that have no
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
382 * matches on the other file. Also, lines that have multiple matches
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
383 * might be potentially discarded if they happear in a run of discardable.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
384 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
385 static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
386 long i, nm, nreff, mlim;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
387 xrecord_t **recs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
388 xdlclass_t *rcrec;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
389 char *dis, *dis1, *dis2;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
390
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
391 if (!(dis = (char *) xdl_malloc(xdf1->nrec + xdf2->nrec + 2))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
392
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
393 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
394 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
395 memset(dis, 0, xdf1->nrec + xdf2->nrec + 2);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
396 dis1 = dis;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
397 dis2 = dis1 + xdf1->nrec + 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
398
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
399 if ((mlim = xdl_bogosqrt(xdf1->nrec)) > XDL_MAX_EQLIMIT)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
400 mlim = XDL_MAX_EQLIMIT;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
401 for (i = xdf1->dstart, recs = &xdf1->recs[xdf1->dstart]; i <= xdf1->dend; i++, recs++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
402 rcrec = cf->rcrecs[(*recs)->ha];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
403 nm = rcrec ? rcrec->len2 : 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
404 dis1[i] = (nm == 0) ? 0: (nm >= mlim) ? 2: 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
405 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
406
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
407 if ((mlim = xdl_bogosqrt(xdf2->nrec)) > XDL_MAX_EQLIMIT)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
408 mlim = XDL_MAX_EQLIMIT;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
409 for (i = xdf2->dstart, recs = &xdf2->recs[xdf2->dstart]; i <= xdf2->dend; i++, recs++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
410 rcrec = cf->rcrecs[(*recs)->ha];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
411 nm = rcrec ? rcrec->len1 : 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
412 dis2[i] = (nm == 0) ? 0: (nm >= mlim) ? 2: 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
413 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
414
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
415 for (nreff = 0, i = xdf1->dstart, recs = &xdf1->recs[xdf1->dstart];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
416 i <= xdf1->dend; i++, recs++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
417 if (dis1[i] == 1 ||
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
418 (dis1[i] == 2 && !xdl_clean_mmatch(dis1, i, xdf1->dstart, xdf1->dend))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
419 xdf1->rindex[nreff] = i;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
420 xdf1->ha[nreff] = (*recs)->ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
421 nreff++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
422 } else
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
423 xdf1->rchg[i] = 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
424 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
425 xdf1->nreff = nreff;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
426
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
427 for (nreff = 0, i = xdf2->dstart, recs = &xdf2->recs[xdf2->dstart];
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
428 i <= xdf2->dend; i++, recs++) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
429 if (dis2[i] == 1 ||
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
430 (dis2[i] == 2 && !xdl_clean_mmatch(dis2, i, xdf2->dstart, xdf2->dend))) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
431 xdf2->rindex[nreff] = i;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
432 xdf2->ha[nreff] = (*recs)->ha;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
433 nreff++;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
434 } else
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
435 xdf2->rchg[i] = 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
436 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
437 xdf2->nreff = nreff;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
438
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
439 xdl_free(dis);
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
440
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
441 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
442 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
443
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
444
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
445 /*
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
446 * Early trim initial and terminal matching records.
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
447 */
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
448 static int xdl_trim_ends(xdfile_t *xdf1, xdfile_t *xdf2) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
449 long i, lim;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
450 xrecord_t **recs1, **recs2;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
451
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
452 recs1 = xdf1->recs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
453 recs2 = xdf2->recs;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
454 for (i = 0, lim = XDL_MIN(xdf1->nrec, xdf2->nrec); i < lim;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
455 i++, recs1++, recs2++)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
456 if ((*recs1)->ha != (*recs2)->ha)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
457 break;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
458
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
459 xdf1->dstart = xdf2->dstart = i;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
460
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
461 recs1 = xdf1->recs + xdf1->nrec - 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
462 recs2 = xdf2->recs + xdf2->nrec - 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
463 for (lim -= i, i = 0; i < lim; i++, recs1--, recs2--)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
464 if ((*recs1)->ha != (*recs2)->ha)
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
465 break;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
466
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
467 xdf1->dend = xdf1->nrec - i - 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
468 xdf2->dend = xdf2->nrec - i - 1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
469
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
470 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
471 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
472
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
473
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
474 static int xdl_optimize_ctxs(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
475
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
476 if (xdl_trim_ends(xdf1, xdf2) < 0 ||
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
477 xdl_cleanup_records(cf, xdf1, xdf2) < 0) {
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
478
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
479 return -1;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
480 }
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
481
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
482 return 0;
195e8b1fcbbf patch 8.1.0360: using an external diff program is slow and inflexible
Christian Brabandt <cb@256bit.org>
parents:
diff changeset
483 }