| 1 | #ifndef SHSTR_H__ |
1 | #ifndef SHSTR_H__ |
| 2 | #define SHSTR_H__ |
2 | #define SHSTR_H__ |
| 3 | |
3 | |
| 4 | /* The size of the shared strings hashtable. This must be smaller than |
4 | extern int buf_overflow(const char *buf1, const char *buf2, int bufsize); |
| 5 | * 32767, but 947 ought to be plenty enough. |
|
|
| 6 | */ |
|
|
| 7 | #define TABLESIZE 4133 |
|
|
| 8 | |
5 | |
| 9 | /* This specifies how many characters the hashing routine should look at. |
6 | struct shstr |
| 10 | * You may actually save CPU by increasing this number if the typical string |
7 | { |
| 11 | * is large. |
8 | const char *s; |
| 12 | */ |
|
|
| 13 | #ifndef MAXSTRING |
|
|
| 14 | #define MAXSTRING 20 |
|
|
| 15 | #endif |
|
|
| 16 | |
9 | |
| 17 | /* In the unlikely occurence that 16383 references to a string are too |
10 | static int &refcnt (const char *s) |
| 18 | * few, you can modify the below type to something bigger. |
11 | { |
| 19 | * (The top bit of "refcount" is used to signify that "u.array" points |
12 | return *((int *)s - 1); |
| 20 | * at the array entry.) |
13 | } |
| 21 | */ |
|
|
| 22 | #define REFCOUNT_TYPE int |
|
|
| 23 | |
14 | |
| 24 | /* The offsetof macro is part of ANSI C, but many compilers lack it, for |
15 | static int &length (const char *s) |
| 25 | * example "gcc -ansi" |
16 | { |
| 26 | */ |
17 | return *((int *)s - 2); |
| 27 | #if !defined (offsetof) |
18 | } |
| 28 | #define offsetof(type, member) (int)&(((type *)0)->member) |
|
|
| 29 | #endif |
|
|
| 30 | |
19 | |
| 31 | /* SS(string) will return the address of the shared_string struct which |
20 | int &refcnt () const |
| 32 | * contains "string". |
21 | { |
| 33 | */ |
22 | return refcnt (s); |
| 34 | #define SS(x) ((shared_string *) ((x) - offsetof(shared_string, string))) |
23 | } |
| 35 | |
24 | |
| 36 | #define SS_STATISTICS |
25 | int length () const |
|
|
26 | { |
|
|
27 | return s ? length (s) : 0; |
|
|
28 | } |
| 37 | |
29 | |
| 38 | #define SS_DUMP_TABLE 1 |
30 | static const char *find (const char *s); |
| 39 | #define SS_DUMP_TOTALS 2 |
31 | static const char *intern (const char *s); |
| 40 | |
32 | |
| 41 | #ifdef SS_STATISTICS |
33 | static void gc (); // garbage collect a few strings |
| 42 | static struct statistics { |
|
|
| 43 | int calls; |
|
|
| 44 | int hashed; |
|
|
| 45 | int strcmps; |
|
|
| 46 | int search; |
|
|
| 47 | int linked; |
|
|
| 48 | } add_stats, add_ref_stats, free_stats, find_stats, hash_stats; |
|
|
| 49 | #define GATHER(n) (++n) |
|
|
| 50 | #else /* !SS_STATISTICS */ |
|
|
| 51 | #define GATHER(n) |
|
|
| 52 | #endif /* SS_STATISTICS */ |
|
|
| 53 | |
34 | |
| 54 | #define TOPBIT ((unsigned REFCOUNT_TYPE) 1 << (sizeof(REFCOUNT_TYPE) * CHAR_BIT - 1)) |
35 | // this is used for informational messages and the like |
|
|
36 | const char *operator &() const { return s ? s : "<nil>"; } |
| 55 | |
37 | |
| 56 | #define PADDING ((2 * sizeof(long) - sizeof(REFCOUNT_TYPE)) % sizeof(long)) + 1 |
38 | const char &operator [](int i) const { return s[i]; } |
|
|
39 | operator const char *() const { return s; } |
| 57 | |
40 | |
| 58 | typedef struct _shared_string { |
41 | shstr () |
| 59 | union { |
42 | : s (0) |
| 60 | struct _shared_string **array; |
43 | { |
| 61 | struct _shared_string *previous; |
44 | } |
| 62 | } u; |
|
|
| 63 | struct _shared_string *next; |
|
|
| 64 | /* The top bit of "refcount" is used to signify that "u.array" points |
|
|
| 65 | * at the array entry. |
|
|
| 66 | */ |
|
|
| 67 | unsigned REFCOUNT_TYPE refcount; |
|
|
| 68 | /* Padding will be unused memory, since we can't know how large |
|
|
| 69 | * the padding when allocating memory. We assume here that |
|
|
| 70 | * sizeof(long) is a good boundary. |
|
|
| 71 | */ |
|
|
| 72 | char string[PADDING]; |
|
|
| 73 | } shared_string; |
|
|
| 74 | |
45 | |
| 75 | extern void init_hash_table(void); |
46 | shstr (const shstr &sh) |
| 76 | extern const char *add_string(const char *str); |
47 | : s (sh.s) |
| 77 | extern const char *add_refcount(const char *str); |
48 | { |
| 78 | extern int query_refcount(const char *str); |
49 | if (s) ++refcnt (); |
| 79 | extern const char *find_string(const char *str); |
50 | } |
| 80 | extern void free_string(const char *str); |
51 | |
| 81 | extern void ss_dump_statistics(void); |
52 | explicit shstr (const char *s) |
| 82 | extern const char *ss_dump_table(int what); |
53 | : s (intern (s)) |
| 83 | extern int buf_overflow(const char *buf1, const char *buf2, int bufsize); |
54 | { |
|
|
55 | } |
|
|
56 | |
|
|
57 | ~shstr () |
|
|
58 | { |
|
|
59 | if (s) --refcnt (); |
|
|
60 | } |
|
|
61 | |
|
|
62 | const shstr &operator =(const shstr &sh) |
|
|
63 | { |
|
|
64 | if (s) --refcnt (); |
|
|
65 | s = sh.s; |
|
|
66 | if (s) ++refcnt (); |
|
|
67 | |
|
|
68 | return *this; |
|
|
69 | } |
|
|
70 | |
|
|
71 | const shstr &operator =(const char *str) |
|
|
72 | { |
|
|
73 | if (s) --refcnt (); |
|
|
74 | s = intern (str); |
|
|
75 | |
|
|
76 | return *this; |
|
|
77 | } |
|
|
78 | }; |
|
|
79 | |
|
|
80 | inline int strlen (const shstr &sh) |
|
|
81 | { |
|
|
82 | return sh.length (); |
|
|
83 | } |
|
|
84 | |
|
|
85 | inline bool operator ==(const shstr &a, const shstr &b) |
|
|
86 | { |
|
|
87 | return a.s == b.s; |
|
|
88 | } |
|
|
89 | |
|
|
90 | inline bool operator !=(const shstr &a, const shstr &b) |
|
|
91 | { |
|
|
92 | return !(a == b); |
|
|
93 | } |
| 84 | |
94 | |
| 85 | #endif |
95 | #endif |
| 86 | |
96 | |
