token.h

   1 #ifndef TOKEN_H
   2 #define TOKEN_H
   3 /*
   4  * Basic tokenization structures. NOTE! Those tokens had better
   5  * be pretty small, since we're going to keep them all in memory
   6  * indefinitely.
   7  *
   8  * Copyright (C) 2003 Transmeta Corp, all rights reserved.
   9  */
  10
  11 #include <sys/types.h>
  12 #include "lib.h"
  13
  14 /*
  15  * This describes the pure lexical elements (tokens), with
  16  * no semantic meaning. In other words, an identifier doesn't
  17  * have a type or meaning, it is only a specific string in
  18  * the input stream.
  19  *
  20  * Semantic meaning is handled elsewhere.
  21  */
  22
  23 struct stream {
  24         int fd;
  25         const char *name;
  26
  27         /* Use these to check for "already parsed" */
  28         int constant, nesting;
  29         struct ident *protect;
  30         dev_t dev;
  31         ino_t ino;
  32 };
  33
  34 extern int input_stream_nr;
  35 extern struct stream *input_streams;
  36
  37 extern int ident_hit, ident_miss;
  38
  39 struct ident {
  40         struct ident *next;     /* Hash chain of identifiers */
  41         struct symbol *symbols; /* Pointer to semantic meaning list */
  42         unsigned char len;      /* Length of identifier name */
  43         char name[];            /* Actual identifier */
  44 };
  45
  46 enum token_type {
  47         TOKEN_EOF,
  48         TOKEN_ERROR,
  49         TOKEN_IDENT,
  50         TOKEN_INTEGER,
  51         TOKEN_FP,
  52         TOKEN_CHAR,
  53         TOKEN_STRING,
  54         TOKEN_SPECIAL,
  55         TOKEN_STREAMBEGIN,
  56         TOKEN_STREAMEND,
  57 };
  58
  59 /* Combination tokens */
  60 #define COMBINATION_STRINGS {   \
  61         "+=", "++",             \
  62         "-=", "--", "->",       \
  63         "*=",                   \
  64         "/=", "/*", "//",       \
  65         "%=",                   \
  66         "..", "...",            \
  67         "<=", "<<", "<<=",      \
  68         ">=", ">>", ">>=",      \
  69         "==", "!=",             \
  70         "&&", "&=",             \
  71         "||", "|=",             \
  72         "^=", "##",             \
  73         " @ ",                  \
  74 }
  75
  76 enum special_token {
  77         SPECIAL_BASE = 256,
  78         SPECIAL_ADD_ASSIGN = 256,
  79         SPECIAL_INCREMENT,
  80         SPECIAL_SUB_ASSIGN,
  81         SPECIAL_DECREMENT,
  82         SPECIAL_DEREFERENCE,
  83         SPECIAL_MUL_ASSIGN,
  84         SPECIAL_DIV_ASSIGN,
  85         SPECIAL_COMMENT,
  86         SPECIAL_CPPCOMMENT,
  87         SPECIAL_MOD_ASSIGN,
  88         SPECIAL_DOTDOT,
  89         SPECIAL_ELLIPSIS,
  90         SPECIAL_LTE,
  91         SPECIAL_LEFTSHIFT,
  92         SPECIAL_SHL_ASSIGN,
  93         SPECIAL_GTE,
  94         SPECIAL_RIGHTSHIFT,
  95         SPECIAL_SHR_ASSIGN,
  96         SPECIAL_EQUAL,
  97         SPECIAL_NOTEQUAL,
  98         SPECIAL_LOGICAL_AND,
  99         SPECIAL_AND_ASSIGN,
 100         SPECIAL_LOGICAL_OR,
 101         SPECIAL_OR_ASSIGN,
 102         SPECIAL_XOR_ASSIGN,
 103         SPECIAL_HASHHASH,
 104         SPECIAL_ARG_SEPARATOR
 105 };
 106
 107 struct string {
 108         unsigned int length;
 109         char data[];
 110 };
 111
 112 /*
 113  * This is a very common data structure, it should be kept
 114  * as small as humanly possible. Big (rare) types go as
 115  * pointers.
 116  */
 117 struct token {
 118         struct position pos;
 119         struct token *next;
 120         union {
 121                 char *integer;
 122                 char *fp;
 123                 struct ident *ident;
 124                 unsigned int special;
 125                 struct string *string;
 126                 int character;
 127         };
 128 };
 129
 130 #define token_type(x) ((x)->pos.type)
 131
 132 /*
 133  * Last token in the stream - points to itself.
 134  * This allows us to not test for NULL pointers
 135  * when following the token->next chain..
 136  */
 137 extern int preprocessing;
 138 extern struct token eof_token_entry;
 139 #define eof_token(x) ((x) == &eof_token_entry)
 140
 141 extern int init_stream(const char *, int fd);
 142 extern struct ident *hash_ident(struct ident *);
 143 extern struct ident *built_in_ident(const char *);
 144 extern struct token *built_in_token(int, const char *);
 145 extern const char *show_special(int);
 146 extern const char *show_ident(const struct ident *);
 147 extern const char *show_string(const struct string *string);
 148 extern const char *show_token(const struct token *);
 149 extern struct token * tokenize(const char *, int, struct token *);
 150 extern struct token * tokenize_buffer(unsigned char *, unsigned long, struct token *);
 151
 152 extern void die(const char *, ...);
 153 extern void show_identifier_stats(void);
 154 extern struct token *preprocess(struct token *);
 155
 156 static inline int match_op(struct token *token, int op)
 157 {
 158         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 159 }
 160
 161 static inline int match_ident(struct token *token, struct ident *id)
 162 {
 163         return token->pos.type == TOKEN_IDENT && token->ident == id;
 164 }
 165
 166 #endif