|
|
1 | /* |
|
|
2 | * This file is part of Deliantra, the Roguelike Realtime MMORPG. |
|
|
3 | * |
|
|
4 | * Copyright (©) 2005,2006,2007,2008,2009 Marc Alexander Lehmann / Robin Redeker / the Deliantra team |
|
|
5 | * |
|
|
6 | * Deliantra is free software: you can redistribute it and/or modify it under |
|
|
7 | * the terms of the Affero GNU General Public License as published by the |
|
|
8 | * Free Software Foundation, either version 3 of the License, or (at your |
|
|
9 | * option) any later version. |
|
|
10 | * |
|
|
11 | * This program is distributed in the hope that it will be useful, |
|
|
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
|
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
|
14 | * GNU General Public License for more details. |
|
|
15 | * |
|
|
16 | * You should have received a copy of the Affero GNU General Public License |
|
|
17 | * and the GNU General Public License along with this program. If not, see |
|
|
18 | * <http://www.gnu.org/licenses/>. |
|
|
19 | * |
|
|
20 | * The authors can be reached via e-mail to <support@deliantra.net> |
|
|
21 | */ |
|
|
22 | |
1 | #ifndef SHSTR_H__ |
23 | #ifndef SHSTR_H__ |
2 | #define SHSTR_H__ |
24 | #define SHSTR_H__ |
3 | |
25 | |
4 | //// OLD STUFF |
26 | #include <cstring> |
|
|
27 | #include <sstream> |
5 | |
28 | |
6 | /* The size of the shared strings hashtable. This must be smaller than |
29 | #include "traits.h" |
7 | * 32767, but 947 ought to be plenty enough. |
|
|
8 | */ |
|
|
9 | #define TABLESIZE 4133 |
|
|
10 | |
30 | |
11 | /* This specifies how many characters the hashing routine should look at. |
31 | extern size_t shstr_alloc; |
12 | * You may actually save CPU by increasing this number if the typical string |
32 | |
13 | * is large. |
33 | extern int buf_overflow (const char *buf1, const char *buf2, int bufsize); |
14 | */ |
34 | |
15 | #ifndef MAXSTRING |
35 | template<int size> |
16 | #define MAXSTRING 20 |
36 | struct shstr_vec |
|
|
37 | { |
|
|
38 | uint32_t hash; |
|
|
39 | uint32_t len; |
|
|
40 | uint32_t refcnt; |
|
|
41 | // pointer points here |
|
|
42 | char string [size]; |
|
|
43 | }; |
|
|
44 | |
|
|
45 | // this class is a non-refcounted shared string |
|
|
46 | // it cannot be used to create or store shared strings, but |
|
|
47 | // it can be used to apss shared strings around, i.e. as function arguments |
|
|
48 | // or return values. their lifetime must not span a gc () call, i.e. |
|
|
49 | // they are only valid as temporary values within the same server tick. |
|
|
50 | struct shstr_tmp |
|
|
51 | { |
|
|
52 | static shstr_vec<sizeof ("(null)")> nullvec; |
|
|
53 | static const char *null () { return nullvec.string; } // this is the null pointer value |
|
|
54 | |
|
|
55 | const char *s; |
|
|
56 | |
|
|
57 | static unsigned int &hash (const char *s) |
|
|
58 | { |
|
|
59 | return *((unsigned int *)s - 3); |
|
|
60 | } |
|
|
61 | |
|
|
62 | int hash () const |
|
|
63 | { |
|
|
64 | return hash (s); |
|
|
65 | } |
|
|
66 | |
|
|
67 | static unsigned int &length (const char *s) |
|
|
68 | { |
|
|
69 | return *((unsigned int *)s - 2); |
|
|
70 | } |
|
|
71 | |
|
|
72 | int length () const |
|
|
73 | { |
|
|
74 | return length (s); |
|
|
75 | } |
|
|
76 | |
|
|
77 | // returns whether this shared string begins with the given prefix, |
|
|
78 | // used mainly for searches when users give only the start of a name. |
|
|
79 | bool starts_with (const char *prefix) const |
|
|
80 | { |
|
|
81 | int plen = strlen (prefix); |
|
|
82 | |
|
|
83 | return length () >= plen && !strncasecmp (s, prefix, plen); |
|
|
84 | } |
|
|
85 | |
|
|
86 | // returns true if the substring is contained in the shstr |
|
|
87 | // if the shstr is 0, then this always returns false. |
|
|
88 | // the shstr is (theoretically) treated as a comma/colon/space etc. separated list. |
|
|
89 | bool contains (const char *substring) const |
|
|
90 | { |
|
|
91 | return s != null () && strstr (s, substring); |
|
|
92 | } |
|
|
93 | |
|
|
94 | //TODO: case sensitive should be eradicated |
|
|
95 | bool eq_nc (const char *otherstring) const |
|
|
96 | { |
|
|
97 | return !strcasecmp (s, otherstring); |
|
|
98 | } |
|
|
99 | |
|
|
100 | shstr_tmp () |
|
|
101 | : s (null ()) |
|
|
102 | { |
|
|
103 | } |
|
|
104 | |
|
|
105 | shstr_tmp (const shstr_tmp &sh) |
|
|
106 | : s (sh.s) |
|
|
107 | { |
|
|
108 | } |
|
|
109 | |
|
|
110 | shstr_tmp operator =(const shstr_tmp &sh) |
|
|
111 | { |
|
|
112 | s = sh.s; |
|
|
113 | |
|
|
114 | return *this; |
|
|
115 | } |
|
|
116 | |
|
|
117 | // this is used for informational messages and the like |
|
|
118 | const char *operator &() const { return s; } |
|
|
119 | |
|
|
120 | operator const char *() const { return s == null () ? 0 : s; } |
|
|
121 | |
|
|
122 | protected: |
|
|
123 | // dummy is there so it isn't used as type converter accidentally |
|
|
124 | shstr_tmp (int dummy, const char *s) |
|
|
125 | : s(s) |
|
|
126 | { |
|
|
127 | } |
|
|
128 | }; |
|
|
129 | |
|
|
130 | inline bool operator ==(const shstr_tmp &a, const shstr_tmp &b) |
|
|
131 | { |
|
|
132 | return a.s == b.s; |
|
|
133 | } |
|
|
134 | |
|
|
135 | inline bool operator !=(const shstr_tmp &a, const shstr_tmp &b) |
|
|
136 | { |
|
|
137 | return a.s != b.s; |
|
|
138 | } |
|
|
139 | |
|
|
140 | inline int strlen (shstr_tmp sh) |
|
|
141 | { |
|
|
142 | return sh.length (); |
|
|
143 | } |
|
|
144 | |
|
|
145 | static inline std::ostream &operator <<(std::ostream &o, shstr_tmp sh) |
|
|
146 | { |
|
|
147 | o.write (sh.s, sh.length ()); |
|
|
148 | |
|
|
149 | return o; |
|
|
150 | } |
|
|
151 | |
|
|
152 | struct shstr : shstr_tmp |
|
|
153 | { |
|
|
154 | static unsigned int &refcnt (const char *s) |
|
|
155 | { |
|
|
156 | return *((unsigned int *)s - 1); |
|
|
157 | } |
|
|
158 | |
|
|
159 | unsigned int &refcnt () const |
|
|
160 | { |
|
|
161 | return refcnt (s); |
|
|
162 | } |
|
|
163 | |
|
|
164 | shstr () |
|
|
165 | { |
|
|
166 | } |
|
|
167 | |
|
|
168 | static const char *find (const char *s); |
|
|
169 | static const char *intern (const char *s); |
|
|
170 | |
|
|
171 | static void gc (); // garbage collect a few strings |
|
|
172 | |
|
|
173 | shstr (const shstr &sh) |
|
|
174 | : shstr_tmp (sh) |
|
|
175 | { |
|
|
176 | ++refcnt (); |
|
|
177 | } |
|
|
178 | |
|
|
179 | shstr (const shstr_tmp &sh) |
|
|
180 | : shstr_tmp (sh) |
|
|
181 | { |
|
|
182 | ++refcnt (); |
|
|
183 | } |
|
|
184 | |
|
|
185 | explicit shstr (const char *str) |
|
|
186 | : shstr_tmp (0, is_constant (str) && !str ? null () : intern (str)) |
|
|
187 | { |
|
|
188 | } |
|
|
189 | |
|
|
190 | ~shstr () |
|
|
191 | { |
|
|
192 | --refcnt (); |
|
|
193 | } |
|
|
194 | |
|
|
195 | using shstr_tmp::operator &; |
|
|
196 | using shstr_tmp::operator const char *; |
|
|
197 | |
|
|
198 | // (note: not the copy constructor) |
|
|
199 | shstr &operator =(const shstr_tmp &sh) |
|
|
200 | { |
|
|
201 | --refcnt (); |
|
|
202 | s = sh.s; |
|
|
203 | ++refcnt (); |
|
|
204 | |
|
|
205 | return *this; |
|
|
206 | } |
|
|
207 | |
|
|
208 | // here it comes |
|
|
209 | shstr &operator =(const shstr &sh) |
|
|
210 | { |
|
|
211 | return (*this) = (shstr_tmp)sh; |
|
|
212 | } |
|
|
213 | |
|
|
214 | // shstr_tmp doesn't have this one |
|
|
215 | shstr &operator =(const char *str) |
|
|
216 | { |
|
|
217 | --refcnt (); |
|
|
218 | s = is_constant (str) && !str ? null () : intern (str); |
|
|
219 | |
|
|
220 | return *this; |
|
|
221 | } |
|
|
222 | }; |
|
|
223 | |
|
|
224 | // only good for mass comparisons to shstr objects, or very |
|
|
225 | // temporary passing, basically a non-refcounted shstr |
|
|
226 | struct shstr_cmp |
|
|
227 | { |
|
|
228 | const char *s; |
|
|
229 | |
|
|
230 | // initialies to the non-matching string (as opposed to the null string) |
|
|
231 | shstr_cmp () |
|
|
232 | { |
|
|
233 | s = 0; |
|
|
234 | } |
|
|
235 | |
|
|
236 | shstr_cmp (const char *str) |
|
|
237 | : s (shstr::find (str)) |
|
|
238 | { |
|
|
239 | } |
|
|
240 | |
|
|
241 | shstr_cmp (shstr_tmp sh) |
|
|
242 | : s (sh.s) |
|
|
243 | { |
|
|
244 | } |
|
|
245 | |
|
|
246 | // this is used for informational messages and the like |
|
|
247 | const char *operator &() const { return s; } |
|
|
248 | |
|
|
249 | operator const char *() const { return s; } |
|
|
250 | }; |
|
|
251 | |
|
|
252 | inline bool operator ==(const shstr_cmp &a, const shstr_tmp &b) |
|
|
253 | { |
|
|
254 | return a.s == b.s; |
|
|
255 | } |
|
|
256 | |
|
|
257 | inline bool operator ==(const shstr_tmp &a, const shstr_cmp &b) |
|
|
258 | { |
|
|
259 | return a.s == b.s; |
|
|
260 | } |
|
|
261 | |
|
|
262 | #define def2(id,str) extern const shstr id; |
|
|
263 | #define def(id) def2(shstr_ ## id, # id) |
|
|
264 | # include "shstrinc.h" |
|
|
265 | #undef def |
|
|
266 | #undef def2 |
|
|
267 | |
|
|
268 | // undefined external reference to catch people using str* functions when they shouldn't |
|
|
269 | //template<class any> void strcmp (const shstr_tmp &a, any b); |
|
|
270 | template<class any> void strstr (const shstr_tmp &a, any b); |
|
|
271 | |
17 | #endif |
272 | #endif |
18 | |
273 | |
19 | /* In the unlikely occurence that 16383 references to a string are too |
|
|
20 | * few, you can modify the below type to something bigger. |
|
|
21 | * (The top bit of "refcount" is used to signify that "u.array" points |
|
|
22 | * at the array entry.) |
|
|
23 | */ |
|
|
24 | #define REFCOUNT_TYPE int |
|
|
25 | |
|
|
26 | /* The offsetof macro is part of ANSI C, but many compilers lack it, for |
|
|
27 | * example "gcc -ansi" |
|
|
28 | */ |
|
|
29 | #if !defined (offsetof) |
|
|
30 | #define offsetof(type, member) (int)&(((type *)0)->member) |
|
|
31 | #endif |
|
|
32 | |
|
|
33 | /* SS(string) will return the address of the shared_string struct which |
|
|
34 | * contains "string". |
|
|
35 | */ |
|
|
36 | #define SS(x) ((shared_string *) ((x) - offsetof(shared_string, string))) |
|
|
37 | |
|
|
38 | #define SS_STATISTICS |
|
|
39 | |
|
|
40 | #define SS_DUMP_TABLE 1 |
|
|
41 | #define SS_DUMP_TOTALS 2 |
|
|
42 | |
|
|
43 | #ifdef SS_STATISTICS |
|
|
44 | static struct statistics { |
|
|
45 | int calls; |
|
|
46 | int hashed; |
|
|
47 | int strcmps; |
|
|
48 | int search; |
|
|
49 | int linked; |
|
|
50 | } add_stats, add_ref_stats, free_stats, find_stats, hash_stats; |
|
|
51 | #define GATHER(n) (++n) |
|
|
52 | #else /* !SS_STATISTICS */ |
|
|
53 | #define GATHER(n) |
|
|
54 | #endif /* SS_STATISTICS */ |
|
|
55 | |
|
|
56 | #define TOPBIT ((unsigned REFCOUNT_TYPE) 1 << (sizeof(REFCOUNT_TYPE) * CHAR_BIT - 1)) |
|
|
57 | |
|
|
58 | #define PADDING ((2 * sizeof(long) - sizeof(REFCOUNT_TYPE)) % sizeof(long)) + 1 |
|
|
59 | |
|
|
60 | typedef struct _shared_string { |
|
|
61 | union { |
|
|
62 | struct _shared_string **array; |
|
|
63 | struct _shared_string *previous; |
|
|
64 | } u; |
|
|
65 | struct _shared_string *next; |
|
|
66 | /* The top bit of "refcount" is used to signify that "u.array" points |
|
|
67 | * at the array entry. |
|
|
68 | */ |
|
|
69 | unsigned REFCOUNT_TYPE refcount; |
|
|
70 | /* Padding will be unused memory, since we can't know how large |
|
|
71 | * the padding when allocating memory. We assume here that |
|
|
72 | * sizeof(long) is a good boundary. |
|
|
73 | */ |
|
|
74 | char string[PADDING]; |
|
|
75 | } shared_string; |
|
|
76 | |
|
|
77 | extern void init_hash_table(void); |
|
|
78 | extern const char *add_string(const char *str); |
|
|
79 | extern const char *add_refcount(const char *str); |
|
|
80 | extern int query_refcount(const char *str); |
|
|
81 | extern const char *find_string(const char *str); |
|
|
82 | extern void free_string(const char *str); |
|
|
83 | extern void ss_dump_statistics(void); |
|
|
84 | extern const char *ss_dump_table(int what); |
|
|
85 | extern int buf_overflow(const char *buf1, const char *buf2, int bufsize); |
|
|
86 | |
|
|
87 | //// NEW STUFF |
|
|
88 | |
|
|
89 | #if 0 |
|
|
90 | struct shstr |
|
|
91 | { |
|
|
92 | const char *s; |
|
|
93 | operator const char *() const { return s; }; |
|
|
94 | shstr &operator =(const char *str) { s = str; }; |
|
|
95 | }; |
|
|
96 | #else |
|
|
97 | typedef const char *shstr; |
|
|
98 | #endif |
|
|
99 | |
|
|
100 | #endif |
|
|
101 | |
|
|