| … | |
… | |
| 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 | |
|
|
| 12 | // NOTE: even with lots of stuff loaded, we do not usually have >>20000 strings. |
|
|
| 13 | // maybe refcounting is just overhead? |
|
|
| 14 | |
11 | |
| 15 | struct hash |
12 | struct hash |
| 16 | { |
13 | { |
| 17 | std::size_t operator ()(const char *s) const |
14 | std::size_t operator ()(const char *s) const |
| 18 | { |
15 | { |
| … | |
… | |
| 49 | |
46 | |
| 50 | typedef std::tr1::unordered_set<const char *, hash, equal> HT; |
47 | typedef std::tr1::unordered_set<const char *, hash, equal> HT; |
| 51 | |
48 | |
| 52 | static HT ht; |
49 | static HT ht; |
| 53 | |
50 | |
|
|
51 | static const char *makevec (const char *s) |
|
|
52 | { |
|
|
53 | int len = strlen (s); |
|
|
54 | |
|
|
55 | const char *v = (const char *)(2 + (int *)malloc (sizeof (int) * 2 + len + 1)); |
|
|
56 | |
|
|
57 | shstr::length (v) = len; |
|
|
58 | shstr::refcnt (v) = 1; |
|
|
59 | |
|
|
60 | memcpy ((char *)v, s, len + 1); |
|
|
61 | |
|
|
62 | return v; |
|
|
63 | } |
|
|
64 | |
|
|
65 | const char *shstr::null = makevec ("<nil>"); |
|
|
66 | |
|
|
67 | // what weird misoptimisation is this again? |
|
|
68 | const shstr undead_name ("undead"); |
|
|
69 | |
| 54 | const char * |
70 | const char * |
| 55 | shstr::find (const char *s) |
71 | shstr::find (const char *s) |
| 56 | { |
72 | { |
| 57 | if (!s) |
73 | if (!s) |
| 58 | return s; |
74 | return s; |
| … | |
… | |
| 66 | |
82 | |
| 67 | const char * |
83 | const char * |
| 68 | shstr::intern (const char *s) |
84 | shstr::intern (const char *s) |
| 69 | { |
85 | { |
| 70 | if (!s) |
86 | if (!s) |
| 71 | return s; |
87 | return null; |
| 72 | |
88 | |
| 73 | if (const char *found = find (s)) |
89 | if (const char *found = find (s)) |
| 74 | { |
90 | { |
| 75 | ++refcnt (found); |
91 | ++refcnt (found); |
| 76 | return found; |
92 | return found; |
| 77 | } |
93 | } |
| 78 | |
94 | |
| 79 | int len = strlen (s); |
95 | 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); |
96 | ht.insert (s); |
| 89 | |
|
|
| 90 | return v; |
97 | return s; |
| 91 | } |
98 | } |
| 92 | |
99 | |
| 93 | // periodically test refcounts == 0 for a few strings |
100 | // periodically test refcounts == 0 for a few strings |
| 94 | // this is the ONLY thing that erases stuff from ht. keep it that way. |
101 | // this is the ONLY thing that erases stuff from ht. keep it that way. |
| 95 | void |
102 | void |
| … | |
… | |
| 101 | |
108 | |
| 102 | if (i == ht.end ()) |
109 | if (i == ht.end ()) |
| 103 | i = ht.begin (); |
110 | i = ht.begin (); |
| 104 | |
111 | |
| 105 | // go through all strings roughly once every 4 minutes |
112 | // go through all strings roughly once every 4 minutes |
| 106 | for (int n = ht.size () / 256 + 16; --n; ) |
113 | int n = ht.size () / 256 + 16; |
|
|
114 | |
|
|
115 | for (;;) |
| 107 | { |
116 | { |
| 108 | if (i == ht.end ()) |
117 | if (i == ht.end ()) |
| 109 | { |
118 | { |
| 110 | curpos = 0; |
119 | curpos = 0; |
| 111 | return; |
120 | return; |
| 112 | } |
121 | } |
| 113 | |
122 | else if (!--n) |
|
|
123 | break; |
| 114 | if (!refcnt (*i)) |
124 | else if (!refcnt (*i)) |
| 115 | { |
125 | { |
| 116 | HT::iterator o = i++; |
126 | HT::iterator o = i++; |
| 117 | const char *s = *o; |
127 | const char *s = *o; |
| 118 | ht.erase (o); |
128 | ht.erase (o); |
| 119 | |
129 | |
