]>
Commit | Line | Data |
---|---|---|
a64215b6 | 1 | #include "git-compat-util.h" |
65416758 | 2 | #include "diffcore.h" |
ba23bbc8 JH |
3 | |
4 | /* | |
5 | * Idea here is very simple. | |
6 | * | |
af3abef9 JH |
7 | * Almost all data we are interested in are text, but sometimes we have |
8 | * to deal with binary data. So we cut them into chunks delimited by | |
9 | * LF byte, or 64-byte sequence, whichever comes first, and hash them. | |
ba23bbc8 | 10 | * |
af3abef9 JH |
11 | * For those chunks, if the source buffer has more instances of it |
12 | * than the destination buffer, that means the difference are the | |
13 | * number of bytes not copied from source to destination. If the | |
14 | * counts are the same, everything was copied from source to | |
15 | * destination. If the destination has more, everything was copied, | |
16 | * and destination added more. | |
ba23bbc8 JH |
17 | * |
18 | * We are doing an approximation so we do not really have to waste | |
19 | * memory by actually storing the sequence. We just hash them into | |
20 | * somewhere around 2^16 hashbuckets and count the occurrences. | |
ba23bbc8 JH |
21 | */ |
22 | ||
c06c7966 | 23 | /* Wild guess at the initial hash size */ |
2821104d | 24 | #define INITIAL_HASH_SIZE 9 |
fc66d213 | 25 | |
2821104d JH |
26 | /* We leave more room in smaller hash but do not let it |
27 | * grow to have unused hole too much. | |
28 | */ | |
29 | #define INITIAL_FREE(sz_log2) ((1<<(sz_log2))*(sz_log2-3)/(sz_log2)) | |
ba23bbc8 | 30 | |
fc66d213 JH |
31 | /* A prime rather carefully chosen between 2^16..2^17, so that |
32 | * HASHBASE < INITIAL_FREE(17). We want to keep the maximum hashtable | |
33 | * size under the current 2<<17 maximum, which can hold this many | |
34 | * different values before overflowing to hashtable of size 2<<18. | |
35 | */ | |
36 | #define HASHBASE 107927 | |
37 | ||
c06c7966 | 38 | struct spanhash { |
e31c9f24 LT |
39 | unsigned int hashval; |
40 | unsigned int cnt; | |
c06c7966 JH |
41 | }; |
42 | struct spanhash_top { | |
43 | int alloc_log2; | |
44 | int free; | |
45 | struct spanhash data[FLEX_ARRAY]; | |
46 | }; | |
47 | ||
c06c7966 JH |
48 | static struct spanhash_top *spanhash_rehash(struct spanhash_top *orig) |
49 | { | |
94a5c5d5 | 50 | struct spanhash_top *new_spanhash; |
c06c7966 JH |
51 | int i; |
52 | int osz = 1 << orig->alloc_log2; | |
53 | int sz = osz << 1; | |
54 | ||
94a5c5d5 | 55 | new_spanhash = xmalloc(st_add(sizeof(*orig), |
50a6c8ef | 56 | st_mult(sizeof(struct spanhash), sz))); |
94a5c5d5 BW |
57 | new_spanhash->alloc_log2 = orig->alloc_log2 + 1; |
58 | new_spanhash->free = INITIAL_FREE(new_spanhash->alloc_log2); | |
59 | memset(new_spanhash->data, 0, sizeof(struct spanhash) * sz); | |
c06c7966 JH |
60 | for (i = 0; i < osz; i++) { |
61 | struct spanhash *o = &(orig->data[i]); | |
62 | int bucket; | |
63 | if (!o->cnt) | |
64 | continue; | |
65 | bucket = o->hashval & (sz - 1); | |
66 | while (1) { | |
94a5c5d5 | 67 | struct spanhash *h = &(new_spanhash->data[bucket++]); |
c06c7966 JH |
68 | if (!h->cnt) { |
69 | h->hashval = o->hashval; | |
70 | h->cnt = o->cnt; | |
94a5c5d5 | 71 | new_spanhash->free--; |
c06c7966 JH |
72 | break; |
73 | } | |
74 | if (sz <= bucket) | |
75 | bucket = 0; | |
76 | } | |
77 | } | |
78 | free(orig); | |
94a5c5d5 | 79 | return new_spanhash; |
c06c7966 JH |
80 | } |
81 | ||
82 | static struct spanhash_top *add_spanhash(struct spanhash_top *top, | |
e31c9f24 | 83 | unsigned int hashval, int cnt) |
c06c7966 JH |
84 | { |
85 | int bucket, lim; | |
86 | struct spanhash *h; | |
87 | ||
88 | lim = (1 << top->alloc_log2); | |
89 | bucket = hashval & (lim - 1); | |
90 | while (1) { | |
91 | h = &(top->data[bucket++]); | |
92 | if (!h->cnt) { | |
93 | h->hashval = hashval; | |
3c7ceba4 | 94 | h->cnt = cnt; |
c06c7966 JH |
95 | top->free--; |
96 | if (top->free < 0) | |
97 | return spanhash_rehash(top); | |
98 | return top; | |
99 | } | |
100 | if (h->hashval == hashval) { | |
3c7ceba4 | 101 | h->cnt += cnt; |
c06c7966 JH |
102 | return top; |
103 | } | |
104 | if (lim <= bucket) | |
105 | bucket = 0; | |
106 | } | |
107 | } | |
108 | ||
eb4d0e3f LT |
109 | static int spanhash_cmp(const void *a_, const void *b_) |
110 | { | |
111 | const struct spanhash *a = a_; | |
112 | const struct spanhash *b = b_; | |
113 | ||
114 | /* A count of zero compares at the end.. */ | |
115 | if (!a->cnt) | |
116 | return !b->cnt ? 0 : 1; | |
117 | if (!b->cnt) | |
118 | return -1; | |
119 | return a->hashval < b->hashval ? -1 : | |
120 | a->hashval > b->hashval ? 1 : 0; | |
121 | } | |
122 | ||
b78ea5fc NTND |
123 | static struct spanhash_top *hash_chars(struct repository *r, |
124 | struct diff_filespec *one) | |
65416758 | 125 | { |
3c7ceba4 | 126 | int i, n; |
e31c9f24 | 127 | unsigned int accum1, accum2, hashval; |
c06c7966 | 128 | struct spanhash_top *hash; |
b9905fed JH |
129 | unsigned char *buf = one->data; |
130 | unsigned int sz = one->size; | |
b78ea5fc | 131 | int is_text = !diff_filespec_is_binary(r, one); |
c06c7966 JH |
132 | |
133 | i = INITIAL_HASH_SIZE; | |
50a6c8ef | 134 | hash = xmalloc(st_add(sizeof(*hash), |
62e8452c | 135 | st_mult(sizeof(struct spanhash), (size_t)1 << i))); |
c06c7966 | 136 | hash->alloc_log2 = i; |
2821104d | 137 | hash->free = INITIAL_FREE(i); |
62e8452c | 138 | memset(hash->data, 0, sizeof(struct spanhash) * ((size_t)1 << i)); |
65416758 | 139 | |
3c7ceba4 LT |
140 | n = 0; |
141 | accum1 = accum2 = 0; | |
ba23bbc8 | 142 | while (sz) { |
e31c9f24 LT |
143 | unsigned int c = *buf++; |
144 | unsigned int old_1 = accum1; | |
ba23bbc8 | 145 | sz--; |
b9905fed JH |
146 | |
147 | /* Ignore CR in CRLF sequence if text */ | |
148 | if (is_text && c == '\r' && sz && *buf == '\n') | |
149 | continue; | |
150 | ||
e31c9f24 LT |
151 | accum1 = (accum1 << 7) ^ (accum2 >> 25); |
152 | accum2 = (accum2 << 7) ^ (old_1 >> 25); | |
3c7ceba4 LT |
153 | accum1 += c; |
154 | if (++n < 64 && c != '\n') | |
155 | continue; | |
156 | hashval = (accum1 + accum2 * 0x61) % HASHBASE; | |
157 | hash = add_spanhash(hash, hashval, n); | |
158 | n = 0; | |
159 | accum1 = accum2 = 0; | |
65416758 | 160 | } |
62e8452c | 161 | QSORT(hash->data, (size_t)1ul << hash->alloc_log2, spanhash_cmp); |
c06c7966 | 162 | return hash; |
65416758 JH |
163 | } |
164 | ||
b78ea5fc NTND |
165 | int diffcore_count_changes(struct repository *r, |
166 | struct diff_filespec *src, | |
d8c3d03a | 167 | struct diff_filespec *dst, |
c06c7966 JH |
168 | void **src_count_p, |
169 | void **dst_count_p, | |
65416758 JH |
170 | unsigned long *src_copied, |
171 | unsigned long *literal_added) | |
172 | { | |
eb4d0e3f | 173 | struct spanhash *s, *d; |
c06c7966 | 174 | struct spanhash_top *src_count, *dst_count; |
ba23bbc8 JH |
175 | unsigned long sc, la; |
176 | ||
c06c7966 JH |
177 | src_count = dst_count = NULL; |
178 | if (src_count_p) | |
179 | src_count = *src_count_p; | |
180 | if (!src_count) { | |
b78ea5fc | 181 | src_count = hash_chars(r, src); |
c06c7966 JH |
182 | if (src_count_p) |
183 | *src_count_p = src_count; | |
184 | } | |
185 | if (dst_count_p) | |
186 | dst_count = *dst_count_p; | |
187 | if (!dst_count) { | |
b78ea5fc | 188 | dst_count = hash_chars(r, dst); |
c06c7966 JH |
189 | if (dst_count_p) |
190 | *dst_count_p = dst_count; | |
191 | } | |
ba23bbc8 | 192 | sc = la = 0; |
c06c7966 | 193 | |
eb4d0e3f LT |
194 | s = src_count->data; |
195 | d = dst_count->data; | |
196 | for (;;) { | |
c06c7966 JH |
197 | unsigned dst_cnt, src_cnt; |
198 | if (!s->cnt) | |
eb4d0e3f LT |
199 | break; /* we checked all in src */ |
200 | while (d->cnt) { | |
201 | if (d->hashval >= s->hashval) | |
202 | break; | |
77cd6ab6 | 203 | la += d->cnt; |
eb4d0e3f LT |
204 | d++; |
205 | } | |
c06c7966 | 206 | src_cnt = s->cnt; |
77cd6ab6 LT |
207 | dst_cnt = 0; |
208 | if (d->cnt && d->hashval == s->hashval) { | |
209 | dst_cnt = d->cnt; | |
210 | d++; | |
211 | } | |
c06c7966 JH |
212 | if (src_cnt < dst_cnt) { |
213 | la += dst_cnt - src_cnt; | |
214 | sc += src_cnt; | |
ba23bbc8 | 215 | } |
c06c7966 JH |
216 | else |
217 | sc += dst_cnt; | |
eb4d0e3f | 218 | s++; |
ba23bbc8 | 219 | } |
77cd6ab6 LT |
220 | while (d->cnt) { |
221 | la += d->cnt; | |
222 | d++; | |
223 | } | |
c06c7966 JH |
224 | |
225 | if (!src_count_p) | |
226 | free(src_count); | |
227 | if (!dst_count_p) | |
228 | free(dst_count); | |
ba23bbc8 JH |
229 | *src_copied = sc; |
230 | *literal_added = la; | |
ba23bbc8 | 231 | return 0; |
65416758 | 232 | } |