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.
   9  *               2003 Linus Torvalds
  10  *
  11  *  Licensed under the Open Software License version 1.1
  12  */
  13
  14 #include <sys/types.h>
  15 #include "lib.h"
  16
  17 /*
  18  * This describes the pure lexical elements (tokens), with
  19  * no semantic meaning. In other words, an identifier doesn't
  20  * have a type or meaning, it is only a specific string in
  21  * the input stream.
  22  *
  23  * Semantic meaning is handled elsewhere.
  24  */
  25
  26 struct stream {
  27         int fd;
  28         const char *name;
  29
  30         /* Use these to check for "already parsed" */
  31         int constant, nesting;
  32         struct ident *protect;
  33         dev_t dev;
  34         ino_t ino;
  35 };
  36
  37 extern int input_stream_nr;
  38 extern struct stream *input_streams;
  39
  40 extern int ident_hit, ident_miss;
  41
  42 struct ident {
  43         struct ident *next;     /* Hash chain of identifiers */
  44         struct symbol *symbols; /* Pointer to semantic meaning list */
  45         unsigned char len;      /* Length of identifier name */
  46         char name[];            /* Actual identifier */
  47 };
  48
  49 enum token_type {
  50         TOKEN_EOF,
  51         TOKEN_ERROR,
  52         TOKEN_IDENT,
  53         TOKEN_INTEGER,
  54         TOKEN_FP,
  55         TOKEN_CHAR,
  56         TOKEN_STRING,
  57         TOKEN_SPECIAL,
  58         TOKEN_STREAMBEGIN,
  59         TOKEN_STREAMEND,
  60 };
  61
  62 /* Combination tokens */
  63 #define COMBINATION_STRINGS {   \
  64         "+=", "++",             \
  65         "-=", "--", "->",       \
  66         "*=",                   \
  67         "/=", "/*", "//",       \
  68         "%=",                   \
  69         "..", "...",            \
  70         "<=", "<<", "<<=",      \
  71         ">=", ">>", ">>=",      \
  72         "==", "!=",             \
  73         "&&", "&=",             \
  74         "||", "|=",             \
  75         "^=", "##",             \
  76         " @ ",                  \
  77 }
  78
  79 enum special_token {
  80         SPECIAL_BASE = 256,
  81         SPECIAL_ADD_ASSIGN = 256,
  82         SPECIAL_INCREMENT,
  83         SPECIAL_SUB_ASSIGN,
  84         SPECIAL_DECREMENT,
  85         SPECIAL_DEREFERENCE,
  86         SPECIAL_MUL_ASSIGN,
  87         SPECIAL_DIV_ASSIGN,
  88         SPECIAL_COMMENT,
  89         SPECIAL_CPPCOMMENT,
  90         SPECIAL_MOD_ASSIGN,
  91         SPECIAL_DOTDOT,
  92         SPECIAL_ELLIPSIS,
  93         SPECIAL_LTE,
  94         SPECIAL_LEFTSHIFT,
  95         SPECIAL_SHL_ASSIGN,
  96         SPECIAL_GTE,
  97         SPECIAL_RIGHTSHIFT,
  98         SPECIAL_SHR_ASSIGN,
  99         SPECIAL_EQUAL,
 100         SPECIAL_NOTEQUAL,
 101         SPECIAL_LOGICAL_AND,
 102         SPECIAL_AND_ASSIGN,
 103         SPECIAL_LOGICAL_OR,
 104         SPECIAL_OR_ASSIGN,
 105         SPECIAL_XOR_ASSIGN,
 106         SPECIAL_HASHHASH,
 107         SPECIAL_ARG_SEPARATOR
 108 };
 109
 110 struct string {
 111         unsigned int length;
 112         char data[];
 113 };
 114
 115 /*
 116  * This is a very common data structure, it should be kept
 117  * as small as humanly possible. Big (rare) types go as
 118  * pointers.
 119  */
 120 struct token {
 121         struct position pos;
 122         struct token *next;
 123         struct token *parent;
 124         union {
 125                 char *integer;
 126                 char *fp;
 127                 struct ident *ident;
 128                 unsigned int special;
 129                 struct string *string;
 130                 int character;
 131         };
 132 };
 133
 134 #define token_type(x) ((x)->pos.type)
 135
 136 /*
 137  * Last token in the stream - points to itself.
 138  * This allows us to not test for NULL pointers
 139  * when following the token->next chain..
 140  */
 141 extern int preprocessing, verbose;
 142 extern struct token eof_token_entry;
 143 #define eof_token(x) ((x) == &eof_token_entry)
 144
 145 extern int init_stream(const char *, int fd);
 146 extern struct ident *hash_ident(struct ident *);
 147 extern struct ident *built_in_ident(const char *);
 148 extern struct token *built_in_token(int, const char *);
 149 extern const char *show_special(int);
 150 extern const char *show_ident(const struct ident *);
 151 extern const char *show_string(const struct string *string);
 152 extern const char *show_token(const struct token *);
 153 extern struct token * tokenize(const char *, int, struct token *);
 154 extern struct token * tokenize_buffer(unsigned char *, unsigned long, struct token *);
 155
 156 extern void die(const char *, ...);
 157 extern void show_identifier_stats(void);
 158 extern struct token *preprocess(struct token *);
 159
 160 static inline int match_op(struct token *token, int op)
 161 {
 162         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 163 }
 164
 165 static inline int match_ident(struct token *token, struct ident *id)
 166 {
 167         return token->pos.type == TOKEN_IDENT && token->ident == id;
 168 }
 169
 170 #endif