hashmap: factor out getting a hash code from a SHA1
Copying the first bytes of a SHA1 is duplicated in six places, however, the implications (the actual value would depend on the endianness of the platform) is documented only once. Add a properly documented API for this. Signed-off-by: Karsten Blees <blees@dcon.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Karsten Blees committed
Jul 3, 2014 at 00:20 UTC
039dc71a7cb824300e242f8abc0fcb19dac93641
8 files changed
+29
-40
Documentation/technical/api-hashmap.txt
+9
@@ -58,6 +58,15 @@ Functions
58
+
59
`strihash` and `memihash` are case insensitive versions.
60
61
+`unsigned int sha1hash(const unsigned char *sha1)`::
62
+
63
+ Converts a cryptographic hash (e.g. SHA-1) into an int-sized hash code
64
+ for use in hash tables. Cryptographic hashes are supposed to have
65
+ uniform distribution, so in contrast to `memhash()`, this just copies
66
+ the first `sizeof(int)` bytes without shuffling any bits. Note that
67
+ the results will be different on big-endian and little-endian
68
+ platforms, so they should not be stored or transferred over the net.
69
+
70
`void hashmap_init(struct hashmap *map, hashmap_cmp_fn equals_function, size_t initial_size)`::
71
72
Initializes a hashmap structure.
builtin/describe.c
+2
-9
@@ -56,17 +56,10 @@ static int commit_name_cmp(const struct commit_name *cn1,
56
return hashcmp(cn1->peeled, peeled ? peeled : cn2->peeled);
57
}
58
59
-static inline unsigned int hash_sha1(const unsigned char *sha1)
60
-{
61
- unsigned int hash;
62
- memcpy(&hash, sha1, sizeof(hash));
63
- return hash;
64
-}
65
-
59
static inline struct commit_name *find_commit_name(const unsigned char *peeled)
60
{
61
struct commit_name key;
69
- hashmap_entry_init(&key, hash_sha1(peeled));
62
+ hashmap_entry_init(&key, sha1hash(peeled));
63
return hashmap_get(&names, &key, peeled);
64
}
65
@@ -114,7 +107,7 @@ static void add_to_known_names(const char *path,
107
if (!e) {
108
e = xmalloc(sizeof(struct commit_name));
109
hashcpy(e->peeled, peeled);
117
- hashmap_entry_init(e, hash_sha1(peeled));
110
+ hashmap_entry_init(e, sha1hash(peeled));
111
hashmap_add(&names, e);
112
e->path = NULL;
113
}
decorate.c
+1
-4
@@ -8,10 +8,7 @@
8
9
static unsigned int hash_obj(const struct object *obj, unsigned int n)
10
{
11
- unsigned int hash;
12
-
13
- memcpy(&hash, obj->sha1, sizeof(unsigned int));
14
- return hash % n;
11
+ return sha1hash(obj->sha1) % n;
12
}
13
14
static void *insert_decoration(struct decoration *n, const struct object *base, void *decoration)
diffcore-rename.c
+1
-3
@@ -242,14 +242,12 @@ struct file_similarity {
242
243
static unsigned int hash_filespec(struct diff_filespec *filespec)
244
{
245
- unsigned int hash;
245
if (!filespec->sha1_valid) {
246
if (diff_populate_filespec(filespec, 0))
247
return 0;
248
hash_sha1_file(filespec->data, filespec->size, "blob", filespec->sha1);
249
}
251
- memcpy(&hash, filespec->sha1, sizeof(hash));
252
- return hash;
250
+ return sha1hash(filespec->sha1);
251
}
252
253
static int find_identical_files(struct hashmap *srcs,
hashmap.h
+11
@@ -13,6 +13,17 @@ extern unsigned int strihash(const char *buf);
13
extern unsigned int memhash(const void *buf, size_t len);
14
extern unsigned int memihash(const void *buf, size_t len);
15
16
+static inline unsigned int sha1hash(const unsigned char *sha1)
17
+{
18
+ /*
19
+ * Equivalent to 'return *(unsigned int *)sha1;', but safe on
20
+ * platforms that don't support unaligned reads.
21
+ */
22
+ unsigned int hash;
23
+ memcpy(&hash, sha1, sizeof(hash));
24
+ return hash;
25
+}
26
+
27
/* data structures */
28
29
struct hashmap_entry {
khash.h
+2
-9
@@ -320,19 +320,12 @@ static const double __ac_HASH_UPPER = 0.77;
320
code; \
321
} }
322
323
-static inline khint_t __kh_oid_hash(const unsigned char *oid)
324
-{
325
- khint_t hash;
326
- memcpy(&hash, oid, sizeof(hash));
327
- return hash;
328
-}
329
-
323
#define __kh_oid_cmp(a, b) (hashcmp(a, b) == 0)
324
332
-KHASH_INIT(sha1, const unsigned char *, void *, 1, __kh_oid_hash, __kh_oid_cmp)
325
+KHASH_INIT(sha1, const unsigned char *, void *, 1, sha1hash, __kh_oid_cmp)
326
typedef kh_sha1_t khash_sha1;
327
335
-KHASH_INIT(sha1_pos, const unsigned char *, int, 1, __kh_oid_hash, __kh_oid_cmp)
328
+KHASH_INIT(sha1_pos, const unsigned char *, int, 1, sha1hash, __kh_oid_cmp)
329
typedef kh_sha1_pos_t khash_sha1_pos;
330
331
#endif /* __AC_KHASH_H */
object.c
+1
-12
@@ -50,18 +50,7 @@ int type_from_string(const char *str)
50
*/
51
static unsigned int hash_obj(const unsigned char *sha1, unsigned int n)
52
{
53
- unsigned int hash;
54
-
55
- /*
56
- * Since the sha1 is essentially random, we just take the
57
- * required number of bits directly from the first
58
- * sizeof(unsigned int) bytes of sha1. First we have to copy
59
- * the bytes into a properly aligned integer. If we cared
60
- * about getting consistent results across architectures, we
61
- * would have to call ntohl() here, too.
62
- */
63
- memcpy(&hash, sha1, sizeof(unsigned int));
64
- return hash & (n - 1);
53
+ return sha1hash(sha1) & (n - 1);
54
}
55
56
/*
pack-objects.c
+2
-3
@@ -7,10 +7,9 @@ static uint32_t locate_object_entry_hash(struct packing_data *pdata,
7
const unsigned char *sha1,
8
int *found)
9
{
10
- uint32_t i, hash, mask = (pdata->index_size - 1);
10
+ uint32_t i, mask = (pdata->index_size - 1);
11
12
- memcpy(&hash, sha1, sizeof(uint32_t));
13
- i = hash & mask;
12
+ i = sha1hash(sha1) & mask;
13
14
while (pdata->index[i] > 0) {
15
uint32_t pos = pdata->index[i] - 1;