| … | |
… | |
| 6 | #include <cstdlib> |
6 | #include <cstdlib> |
| 7 | |
7 | |
| 8 | #include <tr1/unordered_set> |
8 | #include <tr1/unordered_set> |
| 9 | |
9 | |
| 10 | #include "shstr.h" |
10 | #include "shstr.h" |
|
|
11 | #include "util.h" |
| 11 | |
12 | |
| 12 | // NOTE: even with lots of stuff loaded, we do not usually have >>20000 strings. |
|
|
| 13 | // maybe refcounting is just overhead? |
|
|
| 14 | |
|
|
| 15 | struct hash |
|
|
| 16 | { |
|
|
| 17 | std::size_t operator ()(const char *s) const |
|
|
| 18 | { |
|
|
| 19 | unsigned long hash = 0; |
|
|
| 20 | unsigned int i = 0; |
|
|
| 21 | |
|
|
| 22 | /* use the one-at-a-time hash function, which supposedly is |
|
|
| 23 | * better than the djb2-like one used by perl5.005, but |
|
|
| 24 | * certainly is better then the bug used here before. |
|
|
| 25 | * see http://burtleburtle.net/bob/hash/doobs.html |
|
|
| 26 | */ |
|
|
| 27 | while (*s) |
|
|
| 28 | { |
|
|
| 29 | hash += *s++; |
|
|
| 30 | hash += hash << 10; |
|
|
| 31 | hash ^= hash >> 6; |
|
|
| 32 | } |
|
|
| 33 | |
|
|
| 34 | hash += hash << 3; |
|
|
| 35 | hash ^= hash >> 11; |
|
|
| 36 | hash += hash << 15; |
|
|
| 37 | |
|
|
| 38 | return hash; |
|
|
| 39 | } |
|
|
| 40 | }; |
|
|
| 41 | |
|
|
| 42 | struct equal |
|
|
| 43 | { |
|
|
| 44 | bool operator ()(const char *a, const char *b) const |
|
|
| 45 | { |
|
|
| 46 | return !strcmp (a, b); |
|
|
| 47 | } |
|
|
| 48 | }; |
|
|
| 49 | |
|
|
| 50 | typedef std::tr1::unordered_set<const char *, hash, equal> HT; |
13 | typedef std::tr1::unordered_set<const char *, str_hash, str_equal> HT; |
| 51 | |
14 | |
| 52 | static HT ht; |
15 | static HT ht; |
|
|
16 | |
|
|
17 | static const char *makevec (const char *s) |
|
|
18 | { |
|
|
19 | int len = strlen (s); |
|
|
20 | |
|
|
21 | const char *v = (const char *)(2 + (int *)malloc (sizeof (int) * 2 + len + 1)); |
|
|
22 | |
|
|
23 | shstr::length (v) = len; |
|
|
24 | shstr::refcnt (v) = 1; |
|
|
25 | |
|
|
26 | memcpy ((char *)v, s, len + 1); |
|
|
27 | |
|
|
28 | return v; |
|
|
29 | } |
|
|
30 | |
|
|
31 | const char *shstr::null = makevec ("<nil>"); |
|
|
32 | |
|
|
33 | // what weird misoptimisation is this again? |
|
|
34 | const shstr undead_name ("undead"); |
| 53 | |
35 | |
| 54 | const char * |
36 | const char * |
| 55 | shstr::find (const char *s) |
37 | shstr::find (const char *s) |
| 56 | { |
38 | { |
| 57 | if (!s) |
39 | if (!s) |
| … | |
… | |
| 66 | |
48 | |
| 67 | const char * |
49 | const char * |
| 68 | shstr::intern (const char *s) |
50 | shstr::intern (const char *s) |
| 69 | { |
51 | { |
| 70 | if (!s) |
52 | if (!s) |
| 71 | return s; |
53 | return null; |
| 72 | |
54 | |
| 73 | if (const char *found = find (s)) |
55 | if (const char *found = find (s)) |
| 74 | { |
56 | { |
| 75 | ++refcnt (found); |
57 | ++refcnt (found); |
| 76 | return found; |
58 | return found; |
| 77 | } |
59 | } |
| 78 | |
60 | |
| 79 | int len = strlen (s); |
61 | s = makevec (s); |
| 80 | |
|
|
| 81 | const char *v = (const char *)(2 + (int *)malloc (sizeof (int) * 2 + len + 1)); |
|
|
| 82 | |
|
|
| 83 | length (v) = len; |
|
|
| 84 | refcnt (v) = 1; |
|
|
| 85 | |
|
|
| 86 | memcpy ((char *)v, s, len + 1); |
|
|
| 87 | |
|
|
| 88 | ht.insert (v); |
62 | ht.insert (s); |
| 89 | |
|
|
| 90 | return v; |
63 | return s; |
| 91 | } |
64 | } |
| 92 | |
65 | |
| 93 | // periodically test refcounts == 0 for a few strings |
66 | // periodically test refcounts == 0 for a few strings |
| 94 | // this is the ONLY thing that erases stuff from ht. keep it that way. |
67 | // this is the ONLY thing that erases stuff from ht. keep it that way. |
| 95 | void |
68 | void |
| … | |
… | |
| 101 | |
74 | |
| 102 | if (i == ht.end ()) |
75 | if (i == ht.end ()) |
| 103 | i = ht.begin (); |
76 | i = ht.begin (); |
| 104 | |
77 | |
| 105 | // go through all strings roughly once every 4 minutes |
78 | // go through all strings roughly once every 4 minutes |
| 106 | for (int n = ht.size () / 256 + 16; --n; ) |
79 | int n = ht.size () / 256 + 16; |
|
|
80 | |
|
|
81 | for (;;) |
| 107 | { |
82 | { |
| 108 | if (i == ht.end ()) |
83 | if (i == ht.end ()) |
| 109 | { |
84 | { |
| 110 | curpos = 0; |
85 | curpos = 0; |
| 111 | return; |
86 | return; |
| 112 | } |
87 | } |
| 113 | |
88 | else if (!--n) |
|
|
89 | break; |
| 114 | if (!refcnt (*i)) |
90 | else if (!refcnt (*i)) |
| 115 | { |
91 | { |
| 116 | HT::iterator o = i++; |
92 | HT::iterator o = i++; |
| 117 | const char *s = *o; |
93 | const char *s = *o; |
| 118 | ht.erase (o); |
94 | ht.erase (o); |
| 119 | |
95 | |
