| 1 | .\" -*-nroff-*- |
| 2 | .de VS |
| 3 | .sp 1 |
| 4 | .RS |
| 5 | .nf |
| 6 | .ft B |
| 7 | .. |
| 8 | .de VE |
| 9 | .ft R |
| 10 | .fi |
| 11 | .RE |
| 12 | .sp 1 |
| 13 | .. |
| 14 | .TH sym 3 "8 May 1999" "Straylight/Edgeware" "mLib utilities library" |
| 15 | .SH NAME |
| 16 | sym \- symbol table manager |
| 17 | .\" @sym_create |
| 18 | .\" @sym_destroy |
| 19 | .\" @sym_find |
| 20 | .\" @sym_remove |
| 21 | .\" @sym_mkiter |
| 22 | .\" @sym_next |
| 23 | .\" |
| 24 | .\" @SYM_NAME |
| 25 | .\" @SYM_LEN |
| 26 | .\" @SYM_HASH |
| 27 | .\" |
| 28 | .SH SYNOPSIS |
| 29 | .nf |
| 30 | .B "#include <mLib/sym.h>" |
| 31 | |
| 32 | .BI "void sym_create(sym_table *" t ); |
| 33 | .BI "void sym_destroy(sym_table *" t ); |
| 34 | |
| 35 | .BI "void *sym_find(sym_table *" t , |
| 36 | .BI " const char *" n ", long " l , |
| 37 | .BI " size_t " sz ", unsigned *" f ); |
| 38 | .BI "void sym_remove(sym_table *" t ", void *" b ); |
| 39 | |
| 40 | .BI "const char *SYM_NAME(const void *" p ); |
| 41 | .BI "size_t SYM_LEN(const void *" p ); |
| 42 | .BI "uint32 SYM_HASH(const void *" p ); |
| 43 | |
| 44 | .BI "void sym_mkiter(sym_iter *" i ", sym_table *" t ); |
| 45 | .BI "void *sym_next(sym_iter *" i ); |
| 46 | .fi |
| 47 | .SH "DESCRIPTION" |
| 48 | The |
| 49 | .B sym |
| 50 | functions implement a data structure often described as a dictionary, a |
| 51 | finite map, an associative array, or a symbol table. It associates |
| 52 | .I values |
| 53 | with |
| 54 | .I keys |
| 55 | such that the value corresponding to a given key can be found quickly. |
| 56 | Additionally, all stored associations can be enumerated. |
| 57 | .PP |
| 58 | The interface provides an |
| 59 | .I intrusive |
| 60 | symbol table. The data objects stored in the table must include a small |
| 61 | header used by the symbol table manager. This reduces the amount of |
| 62 | pointer fiddling that needs to be done, and in practice doesn't seem to |
| 63 | be much of a problem. It's also fairly easy to construct a |
| 64 | non-intrusive interface if you really want one. |
| 65 | .PP |
| 66 | There are three main data structures involved in the interface: |
| 67 | .TP |
| 68 | .B sym_table |
| 69 | Keeps track of the information associated with a particular table. |
| 70 | .TP |
| 71 | .B sym_base |
| 72 | The header which must be attached to the front of all the value |
| 73 | objects. |
| 74 | .TP |
| 75 | .B sym_iter |
| 76 | An iterator object, used for enumerating all of the associations stored |
| 77 | in a symbol table. |
| 78 | .PP |
| 79 | All of the above data structures should be considered |
| 80 | .IR opaque : |
| 81 | don't try looking inside. Representations have changed in the past, and |
| 82 | they may change again in the future. |
| 83 | .SS "Creation and destruction" |
| 84 | The |
| 85 | .B sym_table |
| 86 | object itself needs to be allocated by the caller. It is initialized by |
| 87 | passing it to the function |
| 88 | .BR sym_create . |
| 89 | After initialization, the table contains no entries. |
| 90 | .PP |
| 91 | Initializing a symbol table involves allocating some memory. If this |
| 92 | allocation fails, an |
| 93 | .B EXC_NOMEM |
| 94 | exception is raised. |
| 95 | .PP |
| 96 | When a symbol table is no longer needed, the memory occupied by the |
| 97 | values and other maintenance structures can be reclaimed by calling |
| 98 | .BR sym_destroy . |
| 99 | Any bits of user data attached to values should previously have been |
| 100 | destroyed. |
| 101 | .SS "Adding, searching and removing" |
| 102 | Most of the actual work is done by the function |
| 103 | .BR sym_find . |
| 104 | It does both lookup and creation, depending on its arguments. To do its |
| 105 | job, it needs to know the following bits of information: |
| 106 | .TP |
| 107 | .BI "sym_table *" t |
| 108 | A pointer to a symbol table to manipulate. |
| 109 | .TP |
| 110 | .BI "const char *" n |
| 111 | The address of the |
| 112 | .I key |
| 113 | to look up or create. Usually this will be a simple text string, |
| 114 | although it can actually be any arbitrary binary data. |
| 115 | .TP |
| 116 | .BI "long " l |
| 117 | The length of the key. If this is \-1, |
| 118 | .B sym_find |
| 119 | assumes that the key is a null-terminated string, and calculates its |
| 120 | length itself. This is entirely equivalent to passing |
| 121 | .BI strlen( n )\fR. |
| 122 | .TP |
| 123 | .BI "size_t " sz |
| 124 | The size of the value block to allocate if the key could not be found. |
| 125 | If this is zero, no value is allocated, and a null pointer is returned |
| 126 | to indicate an unsuccessful lookup. |
| 127 | .TP |
| 128 | .BI "unsigned *" f |
| 129 | The address of a `found' flag to set. This is an output parameter. On |
| 130 | exit, |
| 131 | .B sym_find |
| 132 | will set the value of |
| 133 | .BI * f |
| 134 | to zero if the key could not be found, or nonzero if it was found. This |
| 135 | can be used to tell whether the value returned has been newly allocated, |
| 136 | or whether it was already in the table. |
| 137 | .PP |
| 138 | A terminating null byte is appended to the copy of the symbol's name in |
| 139 | memory. This is not considered to be a part of the symbol's name, and |
| 140 | does not contribute to the name's length as reported by the |
| 141 | .B SYM_LEN |
| 142 | macro. |
| 143 | .PP |
| 144 | A symbol can be removed from the table by calling |
| 145 | .BR sym_remove , |
| 146 | passing the symbol table itself, and the value block that needs |
| 147 | removing. |
| 148 | .SS "Enquiries about symbols" |
| 149 | Three macros are provided to enable simple enquiries about a symbol. |
| 150 | Given a pointer |
| 151 | .I s |
| 152 | to a symbol table entry, |
| 153 | .BI SYM_LEN( s ) |
| 154 | returns the length of the symbol's name (excluding any terminating null |
| 155 | byte); |
| 156 | .BI SYM_NAME( s ) |
| 157 | returns a pointer to the symbol's name; and |
| 158 | .BI SYM_HASH( s ) |
| 159 | returns the symbol's hash value. |
| 160 | .SS "Enumerating symbols" |
| 161 | Enumerating the values in a symbol table is fairly simple. Allocate a |
| 162 | .B sym_iter |
| 163 | object from somewhere. Attach it to a symbol table by calling |
| 164 | .BR sym_mkiter , |
| 165 | and passing in the addresses of the iterator and the symbol table. |
| 166 | Then, each call to |
| 167 | .B sym_next |
| 168 | will return a different value from the symbol table, until all of them |
| 169 | have been enumerated, at which point, |
| 170 | .B sym_next |
| 171 | returns a null pointer. |
| 172 | .PP |
| 173 | It's safe to remove the symbol you've just been returned by |
| 174 | .BR sym_next . |
| 175 | However, it's not safe to remove any other symbol. So don't do that. |
| 176 | .PP |
| 177 | When you've finished with an iterator, it's safe to just throw it away. |
| 178 | You don't need to call any functions beforehand. |
| 179 | .SS "Use in practice" |
| 180 | In normal use, the keys are simple strings (usually identifiers from |
| 181 | some language), and the values are nontrivial structures providing |
| 182 | information about types and values. |
| 183 | .PP |
| 184 | In this case, you'd define something like the following structure for |
| 185 | your values: |
| 186 | .VS |
| 187 | typedef struct val { |
| 188 | sym_base _base; /* Symbol header */ |
| 189 | unsigned type; /* Type of this symbol */ |
| 190 | int dispoff; /* Which display variable is in */ |
| 191 | size_t frameoff; /* Offset of variable in frame */ |
| 192 | } val; |
| 193 | .VE |
| 194 | Given a pointer |
| 195 | .I v |
| 196 | to a |
| 197 | .BR val , |
| 198 | you can find the variable's name by calling |
| 199 | .BI SYM_NAME( v )\fR. |
| 200 | .PP |
| 201 | You can look up a name in the table by saying something like: |
| 202 | .VS |
| 203 | val *v = sym_find(t, name, -1, 0, 0); |
| 204 | if (!v) |
| 205 | error("unknown variable `%s'", name); |
| 206 | .VE |
| 207 | You can add in a new variable by saying something like |
| 208 | .VS |
| 209 | unsigned f; |
| 210 | val *v = sym_find(t, name, -1, sizeof(val), &f); |
| 211 | if (f) |
| 212 | error("variable `%s' already exists", name); |
| 213 | /* fill in v */ |
| 214 | .VE |
| 215 | You can examine all the variables in your symbol table by saying |
| 216 | something like: |
| 217 | .VS |
| 218 | sym_iter i; |
| 219 | val *v; |
| 220 | |
| 221 | for (sym_mkiter(&i, t); (v = sym_next(&i)) != 0; ) { |
| 222 | /* ... */ |
| 223 | } |
| 224 | .VE |
| 225 | That ought to be enough examples to be getting on with. |
| 226 | .SS Implementation |
| 227 | The symbol table is an extensible hashtable, using the universal hash |
| 228 | function described in |
| 229 | .BR unihash (3) |
| 230 | and the global hashing key. The hash chains are kept very short |
| 231 | (probably too short, actually). Every time a symbol is found, its block |
| 232 | is promoted to the front of its bin chain so it gets found faster next |
| 233 | time. |
| 234 | .SH SEE ALSO |
| 235 | .BR hash (3), |
| 236 | .BR mLib (3). |
| 237 | .SH AUTHOR |
| 238 | Mark Wooding, <mdw@distorted.org.uk> |