annotate mercurial/thirdparty/xdiff/xprepare.c @ 47890:3853e6ee160d

dirstatemap: replace `removefile` by an explicit `entry.set_untracked()` All the other caller goes through `reset_state`, so we can safely have an explicit method on `DirstateItem` object. This means that all the logic to preserve the previous state (from p2, merged, etc) is now properly encapsulated within the DirstateItem. This pave the way to using different storage for these information. Differential Revision: https://phab.mercurial-scm.org/D11315
author Pierre-Yves David <pierre-yves.david@octobus.net>
date Fri, 20 Aug 2021 11:27:01 +0200
parents d40b9e29c114
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
1 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
2 * LibXDiff by Davide Libenzi ( File Differential Library )
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
3 * Copyright (C) 2003 Davide Libenzi
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
4 *
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
5 * This library is free software; you can redistribute it and/or
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
6 * modify it under the terms of the GNU Lesser General Public
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
7 * License as published by the Free Software Foundation; either
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
8 * version 2.1 of the License, or (at your option) any later version.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
9 *
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
10 * This library is distributed in the hope that it will be useful,
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
13 * Lesser General Public License for more details.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
14 *
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
15 * You should have received a copy of the GNU Lesser General Public
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
16 * License along with this library; if not, see
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
17 * <http://www.gnu.org/licenses/>.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
18 *
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
19 * Davide Libenzi <davidel@xmailserver.org>
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
20 *
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
21 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
22
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
23 #include "xinclude.h"
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
24
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
25
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
26 #define XDL_KPDIS_RUN 4
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
27 #define XDL_MAX_EQLIMIT 1024
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
28 #define XDL_SIMSCAN_WINDOW 100
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
29 #define XDL_GUESS_NLINES1 256
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
30
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
31
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
32 typedef struct s_xdlclass {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
33 struct s_xdlclass *next;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
34 uint64_t ha;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
35 char const *line;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
36 int64_t size;
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
37 int64_t idx;
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
38 int64_t len1, len2;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
39 } xdlclass_t;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
40
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
41 typedef struct s_xdlclassifier {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
42 unsigned int hbits;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
43 int64_t hsize;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
44 xdlclass_t **rchash;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
45 chastore_t ncha;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
46 xdlclass_t **rcrecs;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
47 int64_t alloc;
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
48 int64_t count;
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
49 int64_t flags;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
50 } xdlclassifier_t;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
51
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
52
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
53
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
54
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
55 static int xdl_init_classifier(xdlclassifier_t *cf, int64_t size, int64_t flags);
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
56 static void xdl_free_classifier(xdlclassifier_t *cf);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
57 static int xdl_classify_record(unsigned int pass, xdlclassifier_t *cf, xrecord_t **rhash,
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
58 unsigned int hbits, xrecord_t *rec);
36824
f0d9811dda8e xdiff: remove unused xpp and xecfg parameters
Jun Wu <quark@fb.com>
parents: 36823
diff changeset
59 static int xdl_prepare_ctx(unsigned int pass, mmfile_t *mf, int64_t narec,
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
60 xdlclassifier_t *cf, xdfile_t *xdf);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
61 static void xdl_free_ctx(xdfile_t *xdf);
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
62 static int xdl_clean_mmatch(char const *dis, int64_t i, int64_t s, int64_t e);
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
63 static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
64 static int xdl_trim_ends(xdfile_t *xdf1, xdfile_t *xdf2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
65 static int xdl_optimize_ctxs(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
66
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
67
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
68
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
69
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
70 static int xdl_init_classifier(xdlclassifier_t *cf, int64_t size, int64_t flags) {
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
71 cf->flags = flags;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
72
36825
f1ef0e53e628 xdiff: use int64 for hash table size
Jun Wu <quark@fb.com>
parents: 36824
diff changeset
73 cf->hbits = xdl_hashbits(size);
36915
651c80720eed xdiff: silence a 32-bit shift warning on Windows
Matt Harbison <matt_harbison@yahoo.com>
parents: 36825
diff changeset
74 cf->hsize = ((uint64_t)1) << cf->hbits;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
75
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
76 if (xdl_cha_init(&cf->ncha, sizeof(xdlclass_t), size / 4 + 1) < 0) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
77
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
78 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
79 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
80 if (!(cf->rchash = (xdlclass_t **) xdl_malloc(cf->hsize * sizeof(xdlclass_t *)))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
81
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
82 xdl_cha_free(&cf->ncha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
83 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
84 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
85 memset(cf->rchash, 0, cf->hsize * sizeof(xdlclass_t *));
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
86
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
87 cf->alloc = size;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
88 if (!(cf->rcrecs = (xdlclass_t **) xdl_malloc(cf->alloc * sizeof(xdlclass_t *)))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
89
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
90 xdl_free(cf->rchash);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
91 xdl_cha_free(&cf->ncha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
92 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
93 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
94
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
95 cf->count = 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
96
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
97 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
98 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
99
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
100
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
101 static void xdl_free_classifier(xdlclassifier_t *cf) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
102
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
103 xdl_free(cf->rcrecs);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
104 xdl_free(cf->rchash);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
105 xdl_cha_free(&cf->ncha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
106 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
107
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
108
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
109 static int xdl_classify_record(unsigned int pass, xdlclassifier_t *cf, xrecord_t **rhash,
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
110 unsigned int hbits, xrecord_t *rec) {
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
111 int64_t hi;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
112 char const *line;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
113 xdlclass_t *rcrec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
114 xdlclass_t **rcrecs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
115
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
116 line = rec->ptr;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
117 hi = (long) XDL_HASHLONG(rec->ha, cf->hbits);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
118 for (rcrec = cf->rchash[hi]; rcrec; rcrec = rcrec->next)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
119 if (rcrec->ha == rec->ha &&
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
120 xdl_recmatch(rcrec->line, rcrec->size,
36823
49fe6249937a xdiff: remove unused flags parameter
Jun Wu <quark@fb.com>
parents: 36822
diff changeset
121 rec->ptr, rec->size))
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
122 break;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
123
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
124 if (!rcrec) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
125 if (!(rcrec = xdl_cha_alloc(&cf->ncha))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
126
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
127 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
128 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
129 rcrec->idx = cf->count++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
130 if (cf->count > cf->alloc) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
131 cf->alloc *= 2;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
132 if (!(rcrecs = (xdlclass_t **) xdl_realloc(cf->rcrecs, cf->alloc * sizeof(xdlclass_t *)))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
133
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
134 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
135 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
136 cf->rcrecs = rcrecs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
137 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
138 cf->rcrecs[rcrec->idx] = rcrec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
139 rcrec->line = line;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
140 rcrec->size = rec->size;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
141 rcrec->ha = rec->ha;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
142 rcrec->len1 = rcrec->len2 = 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
143 rcrec->next = cf->rchash[hi];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
144 cf->rchash[hi] = rcrec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
145 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
146
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
147 (pass == 1) ? rcrec->len1++ : rcrec->len2++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
148
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
149 rec->ha = (unsigned long) rcrec->idx;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
150
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
151 hi = (long) XDL_HASHLONG(rec->ha, hbits);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
152 rec->next = rhash[hi];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
153 rhash[hi] = rec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
154
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
155 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
156 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
157
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
158
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
159 /*
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
160 * Trim common prefix from files.
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
161 *
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
162 * Note: trimming could affect hunk shifting. But the performance benefit
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
163 * outweighs the shift change. A diff result with suboptimal shifting is still
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
164 * valid.
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
165 */
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
166 static void xdl_trim_files(mmfile_t *mf1, mmfile_t *mf2, int64_t reserved,
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
167 xdfenv_t *xe, mmfile_t *out_mf1, mmfile_t *out_mf2) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
168 mmfile_t msmall, mlarge;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
169 /* prefix lines, prefix bytes, suffix lines, suffix bytes */
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
170 int64_t plines = 0, pbytes = 0, slines = 0, sbytes = 0, i;
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
171 /* prefix char pointer for msmall and mlarge */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
172 const char *pp1, *pp2;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
173 /* suffix char pointer for msmall and mlarge */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
174 const char *ps1, *ps2;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
175
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
176 /* reserved must >= 0 for the line boundary adjustment to work */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
177 if (reserved < 0)
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
178 reserved = 0;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
179
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
180 if (mf1->size < mf2->size) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
181 memcpy(&msmall, mf1, sizeof(mmfile_t));
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
182 memcpy(&mlarge, mf2, sizeof(mmfile_t));
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
183 } else {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
184 memcpy(&msmall, mf2, sizeof(mmfile_t));
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
185 memcpy(&mlarge, mf1, sizeof(mmfile_t));
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
186 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
187
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
188 pp1 = msmall.ptr, pp2 = mlarge.ptr;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
189 for (i = 0; i < msmall.size && *pp1 == *pp2; ++i) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
190 plines += (*pp1 == '\n');
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
191 pp1++, pp2++;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
192 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
193
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
194 ps1 = msmall.ptr + msmall.size - 1, ps2 = mlarge.ptr + mlarge.size - 1;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
195 while (ps1 > pp1 && *ps1 == *ps2) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
196 slines += (*ps1 == '\n');
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
197 ps1--, ps2--;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
198 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
199
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
200 /* Retract common prefix and suffix boundaries for reserved lines */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
201 if (plines <= reserved + 1) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
202 plines = 0;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
203 } else {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
204 i = 0;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
205 while (i <= reserved) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
206 pp1--;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
207 i += (*pp1 == '\n');
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
208 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
209 /* The new mmfile starts at the next char just after '\n' */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
210 pbytes = pp1 - msmall.ptr + 1;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
211 plines -= reserved;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
212 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
213
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
214 if (slines <= reserved + 1) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
215 slines = 0;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
216 } else {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
217 /* Note: with compiler SIMD support (ex. -O3 -mavx2), this
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
218 * might perform better than memchr. */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
219 i = 0;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
220 while (i <= reserved) {
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
221 ps1++;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
222 i += (*ps1 == '\n');
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
223 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
224 /* The new mmfile includes this '\n' */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
225 sbytes = msmall.ptr + msmall.size - ps1 - 1;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
226 slines -= reserved;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
227 if (msmall.ptr[msmall.size - 1] == '\n')
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
228 slines -= 1;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
229 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
230
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
231 xe->nprefix = plines;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
232 xe->nsuffix = slines;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
233 out_mf1->ptr = mf1->ptr + pbytes;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
234 out_mf1->size = mf1->size - pbytes - sbytes;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
235 out_mf2->ptr = mf2->ptr + pbytes;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
236 out_mf2->size = mf2->size - pbytes - sbytes;
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
237 }
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
238
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
239
36824
f0d9811dda8e xdiff: remove unused xpp and xecfg parameters
Jun Wu <quark@fb.com>
parents: 36823
diff changeset
240 static int xdl_prepare_ctx(unsigned int pass, mmfile_t *mf, int64_t narec,
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
241 xdlclassifier_t *cf, xdfile_t *xdf) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
242 unsigned int hbits;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
243 int64_t nrec, hsize, bsize;
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
244 uint64_t hav;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
245 char const *blk, *cur, *top, *prev;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
246 xrecord_t *crec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
247 xrecord_t **recs, **rrecs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
248 xrecord_t **rhash;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
249 uint64_t *ha;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
250 char *rchg;
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
251 int64_t *rindex;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
252
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
253 ha = NULL;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
254 rindex = NULL;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
255 rchg = NULL;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
256 rhash = NULL;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
257 recs = NULL;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
258
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
259 if (xdl_cha_init(&xdf->rcha, sizeof(xrecord_t), narec / 4 + 1) < 0)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
260 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
261 if (!(recs = (xrecord_t **) xdl_malloc(narec * sizeof(xrecord_t *))))
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
262 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
263
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
264 {
36825
f1ef0e53e628 xdiff: use int64 for hash table size
Jun Wu <quark@fb.com>
parents: 36824
diff changeset
265 hbits = xdl_hashbits(narec);
36915
651c80720eed xdiff: silence a 32-bit shift warning on Windows
Matt Harbison <matt_harbison@yahoo.com>
parents: 36825
diff changeset
266 hsize = ((uint64_t)1) << hbits;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
267 if (!(rhash = (xrecord_t **) xdl_malloc(hsize * sizeof(xrecord_t *))))
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
268 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
269 memset(rhash, 0, hsize * sizeof(xrecord_t *));
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
270 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
271
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
272 nrec = 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
273 if ((cur = blk = xdl_mmfile_first(mf, &bsize)) != NULL) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
274 for (top = blk + bsize; cur < top; ) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
275 prev = cur;
36823
49fe6249937a xdiff: remove unused flags parameter
Jun Wu <quark@fb.com>
parents: 36822
diff changeset
276 hav = xdl_hash_record(&cur, top);
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
277 if (nrec >= narec) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
278 narec *= 2;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
279 if (!(rrecs = (xrecord_t **) xdl_realloc(recs, narec * sizeof(xrecord_t *))))
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
280 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
281 recs = rrecs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
282 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
283 if (!(crec = xdl_cha_alloc(&xdf->rcha)))
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
284 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
285 crec->ptr = prev;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
286 crec->size = (long) (cur - prev);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
287 crec->ha = hav;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
288 recs[nrec++] = crec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
289
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
290 if (xdl_classify_record(pass, cf, rhash, hbits, crec) < 0)
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
291 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
292 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
293 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
294
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
295 if (!(rchg = (char *) xdl_malloc((nrec + 2) * sizeof(char))))
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
296 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
297 memset(rchg, 0, (nrec + 2) * sizeof(char));
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
298
36923
d40b9e29c114 xdiff: fix a hard crash on Windows
Matt Harbison <matt_harbison@yahoo.com>
parents: 36915
diff changeset
299 if (!(rindex = (int64_t *) xdl_malloc((nrec + 1) * sizeof(int64_t))))
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
300 goto abort;
36923
d40b9e29c114 xdiff: fix a hard crash on Windows
Matt Harbison <matt_harbison@yahoo.com>
parents: 36915
diff changeset
301 if (!(ha = (uint64_t *) xdl_malloc((nrec + 1) * sizeof(uint64_t))))
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
302 goto abort;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
303
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
304 xdf->nrec = nrec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
305 xdf->recs = recs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
306 xdf->hbits = hbits;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
307 xdf->rhash = rhash;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
308 xdf->rchg = rchg + 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
309 xdf->rindex = rindex;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
310 xdf->nreff = 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
311 xdf->ha = ha;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
312 xdf->dstart = 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
313 xdf->dend = nrec - 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
314
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
315 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
316
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
317 abort:
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
318 xdl_free(ha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
319 xdl_free(rindex);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
320 xdl_free(rchg);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
321 xdl_free(rhash);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
322 xdl_free(recs);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
323 xdl_cha_free(&xdf->rcha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
324 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
325 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
326
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
327
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
328 static void xdl_free_ctx(xdfile_t *xdf) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
329
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
330 xdl_free(xdf->rhash);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
331 xdl_free(xdf->rindex);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
332 xdl_free(xdf->rchg - 1);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
333 xdl_free(xdf->ha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
334 xdl_free(xdf->recs);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
335 xdl_cha_free(&xdf->rcha);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
336 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
337
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
338 /* Reserved lines for trimming, to leave room for shifting */
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
339 #define TRIM_RESERVED_LINES 100
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
340
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
341 int xdl_prepare_env(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
342 xdfenv_t *xe) {
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
343 int64_t enl1, enl2, sample;
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
344 mmfile_t tmf1, tmf2;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
345 xdlclassifier_t cf;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
346
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
347 memset(&cf, 0, sizeof(cf));
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
348
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
349 sample = XDL_GUESS_NLINES1;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
350
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
351 enl1 = xdl_guess_lines(mf1, sample) + 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
352 enl2 = xdl_guess_lines(mf2, sample) + 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
353
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
354 if (xdl_init_classifier(&cf, enl1 + enl2 + 1, xpp->flags) < 0)
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
355 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
356
36820
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
357 xdl_trim_files(mf1, mf2, TRIM_RESERVED_LINES, xe, &tmf1, &tmf2);
f33a87cf60cc xdiff: add a preprocessing step that trims files
Jun Wu <quark@fb.com>
parents: 36672
diff changeset
358
36824
f0d9811dda8e xdiff: remove unused xpp and xecfg parameters
Jun Wu <quark@fb.com>
parents: 36823
diff changeset
359 if (xdl_prepare_ctx(1, &tmf1, enl1, &cf, &xe->xdf1) < 0) {
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
360
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
361 xdl_free_classifier(&cf);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
362 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
363 }
36824
f0d9811dda8e xdiff: remove unused xpp and xecfg parameters
Jun Wu <quark@fb.com>
parents: 36823
diff changeset
364 if (xdl_prepare_ctx(2, &tmf2, enl2, &cf, &xe->xdf2) < 0) {
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
365
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
366 xdl_free_ctx(&xe->xdf1);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
367 xdl_free_classifier(&cf);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
368 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
369 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
370
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
371 if (xdl_optimize_ctxs(&cf, &xe->xdf1, &xe->xdf2) < 0) {
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
372 xdl_free_ctx(&xe->xdf2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
373 xdl_free_ctx(&xe->xdf1);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
374 xdl_free_classifier(&cf);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
375 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
376 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
377
36672
9e7b14caf67f xdiff: remove patience and histogram diff algorithms
Jun Wu <quark@fb.com>
parents: 36671
diff changeset
378 xdl_free_classifier(&cf);
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
379
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
380 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
381 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
382
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
383
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
384 void xdl_free_env(xdfenv_t *xe) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
385
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
386 xdl_free_ctx(&xe->xdf2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
387 xdl_free_ctx(&xe->xdf1);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
388 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
389
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
390
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
391 static int xdl_clean_mmatch(char const *dis, int64_t i, int64_t s, int64_t e) {
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
392 int64_t r, rdis0, rpdis0, rdis1, rpdis1;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
393
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
394 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
395 * Limits the window the is examined during the similar-lines
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
396 * scan. The loops below stops when dis[i - r] == 1 (line that
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
397 * has no match), but there are corner cases where the loop
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
398 * proceed all the way to the extremities by causing huge
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
399 * performance penalties in case of big files.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
400 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
401 if (i - s > XDL_SIMSCAN_WINDOW)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
402 s = i - XDL_SIMSCAN_WINDOW;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
403 if (e - i > XDL_SIMSCAN_WINDOW)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
404 e = i + XDL_SIMSCAN_WINDOW;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
405
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
406 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
407 * Scans the lines before 'i' to find a run of lines that either
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
408 * have no match (dis[j] == 0) or have multiple matches (dis[j] > 1).
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
409 * Note that we always call this function with dis[i] > 1, so the
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
410 * current line (i) is already a multimatch line.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
411 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
412 for (r = 1, rdis0 = 0, rpdis0 = 1; (i - r) >= s; r++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
413 if (!dis[i - r])
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
414 rdis0++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
415 else if (dis[i - r] == 2)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
416 rpdis0++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
417 else
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
418 break;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
419 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
420 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
421 * If the run before the line 'i' found only multimatch lines, we
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
422 * return 0 and hence we don't make the current line (i) discarded.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
423 * We want to discard multimatch lines only when they appear in the
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
424 * middle of runs with nomatch lines (dis[j] == 0).
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
425 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
426 if (rdis0 == 0)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
427 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
428 for (r = 1, rdis1 = 0, rpdis1 = 1; (i + r) <= e; r++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
429 if (!dis[i + r])
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
430 rdis1++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
431 else if (dis[i + r] == 2)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
432 rpdis1++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
433 else
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
434 break;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
435 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
436 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
437 * If the run after the line 'i' found only multimatch lines, we
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
438 * return 0 and hence we don't make the current line (i) discarded.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
439 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
440 if (rdis1 == 0)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
441 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
442 rdis1 += rdis0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
443 rpdis1 += rpdis0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
444
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
445 return rpdis1 * XDL_KPDIS_RUN < (rpdis1 + rdis1);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
446 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
447
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
448
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
449 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
450 * Try to reduce the problem complexity, discard records that have no
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
451 * matches on the other file. Also, lines that have multiple matches
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
452 * might be potentially discarded if they happear in a run of discardable.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
453 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
454 static int xdl_cleanup_records(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2) {
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
455 int64_t i, nm, nreff, mlim;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
456 xrecord_t **recs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
457 xdlclass_t *rcrec;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
458 char *dis, *dis1, *dis2;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
459
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
460 if (!(dis = (char *) xdl_malloc(xdf1->nrec + xdf2->nrec + 2))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
461
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
462 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
463 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
464 memset(dis, 0, xdf1->nrec + xdf2->nrec + 2);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
465 dis1 = dis;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
466 dis2 = dis1 + xdf1->nrec + 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
467
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
468 if ((mlim = xdl_bogosqrt(xdf1->nrec)) > XDL_MAX_EQLIMIT)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
469 mlim = XDL_MAX_EQLIMIT;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
470 for (i = xdf1->dstart, recs = &xdf1->recs[xdf1->dstart]; i <= xdf1->dend; i++, recs++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
471 rcrec = cf->rcrecs[(*recs)->ha];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
472 nm = rcrec ? rcrec->len2 : 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
473 dis1[i] = (nm == 0) ? 0: (nm >= mlim) ? 2: 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
474 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
475
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
476 if ((mlim = xdl_bogosqrt(xdf2->nrec)) > XDL_MAX_EQLIMIT)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
477 mlim = XDL_MAX_EQLIMIT;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
478 for (i = xdf2->dstart, recs = &xdf2->recs[xdf2->dstart]; i <= xdf2->dend; i++, recs++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
479 rcrec = cf->rcrecs[(*recs)->ha];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
480 nm = rcrec ? rcrec->len1 : 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
481 dis2[i] = (nm == 0) ? 0: (nm >= mlim) ? 2: 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
482 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
483
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
484 for (nreff = 0, i = xdf1->dstart, recs = &xdf1->recs[xdf1->dstart];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
485 i <= xdf1->dend; i++, recs++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
486 if (dis1[i] == 1 ||
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
487 (dis1[i] == 2 && !xdl_clean_mmatch(dis1, i, xdf1->dstart, xdf1->dend))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
488 xdf1->rindex[nreff] = i;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
489 xdf1->ha[nreff] = (*recs)->ha;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
490 nreff++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
491 } else
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
492 xdf1->rchg[i] = 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
493 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
494 xdf1->nreff = nreff;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
495
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
496 for (nreff = 0, i = xdf2->dstart, recs = &xdf2->recs[xdf2->dstart];
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
497 i <= xdf2->dend; i++, recs++) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
498 if (dis2[i] == 1 ||
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
499 (dis2[i] == 2 && !xdl_clean_mmatch(dis2, i, xdf2->dstart, xdf2->dend))) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
500 xdf2->rindex[nreff] = i;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
501 xdf2->ha[nreff] = (*recs)->ha;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
502 nreff++;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
503 } else
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
504 xdf2->rchg[i] = 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
505 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
506 xdf2->nreff = nreff;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
507
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
508 xdl_free(dis);
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
509
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
510 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
511 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
512
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
513
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
514 /*
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
515 * Early trim initial and terminal matching records.
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
516 */
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
517 static int xdl_trim_ends(xdfile_t *xdf1, xdfile_t *xdf2) {
36822
882657a9f768 xdiff: replace {unsigned ,}long with {u,}int64_t
Jun Wu <quark@fb.com>
parents: 36820
diff changeset
518 int64_t i, lim;
36671
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
519 xrecord_t **recs1, **recs2;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
520
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
521 recs1 = xdf1->recs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
522 recs2 = xdf2->recs;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
523 for (i = 0, lim = XDL_MIN(xdf1->nrec, xdf2->nrec); i < lim;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
524 i++, recs1++, recs2++)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
525 if ((*recs1)->ha != (*recs2)->ha)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
526 break;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
527
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
528 xdf1->dstart = xdf2->dstart = i;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
529
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
530 recs1 = xdf1->recs + xdf1->nrec - 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
531 recs2 = xdf2->recs + xdf2->nrec - 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
532 for (lim -= i, i = 0; i < lim; i++, recs1--, recs2--)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
533 if ((*recs1)->ha != (*recs2)->ha)
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
534 break;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
535
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
536 xdf1->dend = xdf1->nrec - i - 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
537 xdf2->dend = xdf2->nrec - i - 1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
538
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
539 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
540 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
541
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
542
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
543 static int xdl_optimize_ctxs(xdlclassifier_t *cf, xdfile_t *xdf1, xdfile_t *xdf2) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
544
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
545 if (xdl_trim_ends(xdf1, xdf2) < 0 ||
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
546 xdl_cleanup_records(cf, xdf1, xdf2) < 0) {
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
547
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
548 return -1;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
549 }
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
550
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
551 return 0;
34e2ff1f9cd8 xdiff: vendor xdiff library from git
Jun Wu <quark@fb.com>
parents:
diff changeset
552 }