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 enum constantfile {
  27   CONSTANT_FILE_MAYBE,    // To be determined, not inside any #ifs in this file
  28   CONSTANT_FILE_IFNDEF,   // To be determined, currently inside #ifndef
  29   CONSTANT_FILE_NOPE,     // No
  30   CONSTANT_FILE_YES       // Yes
  31 };
  32
  33 extern const char *includepath[];
  34
  35 struct stream {
  36         int fd;
  37         const char *name;
  38         const char *path;    // input-file path - see set_stream_include_path()
  39         const char **next_path;
  40
  41         /* Use these to check for "already parsed" */
  42         enum constantfile constant;
  43         int dirty;
  44         struct ident *protect;
  45         struct token *ifndef;
  46         struct token *top_if;
  47 };
  48
  49 extern int input_stream_nr;
  50 extern struct stream *input_streams;
  51
  52 struct ident {
  53         struct ident *next;     /* Hash chain of identifiers */
  54         struct symbol *symbols; /* Pointer to semantic meaning list */
  55         unsigned char len;      /* Length of identifier name */
  56         unsigned char tainted:1,
  57                       reserved:1,
  58                       keyword:1;
  59         char name[];            /* Actual identifier */
  60 };
  61
  62 enum token_type {
  63         TOKEN_EOF,
  64         TOKEN_ERROR,
  65         TOKEN_IDENT,
  66         TOKEN_ZERO_IDENT,
  67         TOKEN_NUMBER,
  68         TOKEN_CHAR,
  69         TOKEN_STRING,
  70         TOKEN_SPECIAL,
  71         TOKEN_STREAMBEGIN,
  72         TOKEN_STREAMEND,
  73         TOKEN_MACRO_ARGUMENT,
  74         TOKEN_STR_ARGUMENT,
  75         TOKEN_QUOTED_ARGUMENT,
  76         TOKEN_CONCAT,
  77         TOKEN_GNU_KLUDGE,
  78         TOKEN_UNTAINT,
  79         TOKEN_ARG_COUNT,
  80         TOKEN_IF,
  81         TOKEN_SKIP_GROUPS,
  82         TOKEN_ELSE,
  83 };
  84
  85 /* Combination tokens */
  86 #define COMBINATION_STRINGS {   \
  87         "+=", "++",             \
  88         "-=", "--", "->",       \
  89         "*=",                   \
  90         "/=",                   \
  91         "%=",                   \
  92         "<=", ">=",             \
  93         "==", "!=",             \
  94         "&&", "&=",             \
  95         "||", "|=",             \
  96         "^=", "##",             \
  97         "<<", ">>", "..",       \
  98         "<<=", ">>=", "...",    \
  99         "",                     \
 100         "<", ">", "<=", ">="    \
 101 }
 102
 103 extern unsigned char combinations[][4];
 104
 105 enum special_token {
 106         SPECIAL_BASE = 256,
 107         SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
 108         SPECIAL_INCREMENT,
 109         SPECIAL_SUB_ASSIGN,
 110         SPECIAL_DECREMENT,
 111         SPECIAL_DEREFERENCE,
 112         SPECIAL_MUL_ASSIGN,
 113         SPECIAL_DIV_ASSIGN,
 114         SPECIAL_MOD_ASSIGN,
 115         SPECIAL_LTE,
 116         SPECIAL_GTE,
 117         SPECIAL_EQUAL,
 118         SPECIAL_NOTEQUAL,
 119         SPECIAL_LOGICAL_AND,
 120         SPECIAL_AND_ASSIGN,
 121         SPECIAL_LOGICAL_OR,
 122         SPECIAL_OR_ASSIGN,
 123         SPECIAL_XOR_ASSIGN,
 124         SPECIAL_HASHHASH,
 125         SPECIAL_LEFTSHIFT,
 126         SPECIAL_RIGHTSHIFT,
 127         SPECIAL_DOTDOT,
 128         SPECIAL_SHL_ASSIGN,
 129         SPECIAL_SHR_ASSIGN,
 130         SPECIAL_ELLIPSIS,
 131         SPECIAL_ARG_SEPARATOR,
 132         SPECIAL_UNSIGNED_LT,
 133         SPECIAL_UNSIGNED_GT,
 134         SPECIAL_UNSIGNED_LTE,
 135         SPECIAL_UNSIGNED_GTE,
 136 };
 137
 138 struct string {
 139         unsigned int length;
 140         char data[];
 141 };
 142
 143 /* will fit into 32 bits */
 144 struct argcount {
 145         unsigned normal:10;
 146         unsigned quoted:10;
 147         unsigned str:10;
 148         unsigned vararg:1;
 149 };
 150
 151 /*
 152  * This is a very common data structure, it should be kept
 153  * as small as humanly possible. Big (rare) types go as
 154  * pointers.
 155  */
 156 struct token {
 157         struct position pos;
 158         struct token *next;
 159         union {
 160                 const char *number;
 161                 struct ident *ident;
 162                 unsigned int special;
 163                 struct string *string;
 164                 int character;
 165                 int argnum;
 166                 struct argcount count;
 167         };
 168 };
 169
 170 #define MAX_STRING 4095
 171
 172 static inline struct token *containing_token(struct token **p)
 173 {
 174         void *addr = (char *)p - ((char *)&((struct token *)0)->next - (char *)0);
 175         return addr;
 176 }
 177
 178 #define token_type(x) ((x)->pos.type)
 179
 180 /*
 181  * Last token in the stream - points to itself.
 182  * This allows us to not test for NULL pointers
 183  * when following the token->next chain..
 184  */
 185 extern struct token eof_token_entry;
 186 #define eof_token(x) ((x) == &eof_token_entry)
 187
 188 extern int init_stream(const char *, int fd, const char **next_path);
 189 extern const char *stream_name(int stream);
 190 extern struct ident *hash_ident(struct ident *);
 191 extern struct ident *built_in_ident(const char *);
 192 extern struct token *built_in_token(int, const char *);
 193 extern const char *show_special(int);
 194 extern const char *show_ident(const struct ident *);
 195 extern const char *show_string(const struct string *string);
 196 extern const char *show_token(const struct token *);
 197 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
 198 extern struct token * tokenize_buffer(void *, unsigned long, struct token *);
 199
 200 extern void show_identifier_stats(void);
 201 extern struct token *preprocess(struct token *);
 202
 203 static inline int match_op(struct token *token, int op)
 204 {
 205         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 206 }
 207
 208 static inline int match_ident(struct token *token, struct ident *id)
 209 {
 210         return token->pos.type == TOKEN_IDENT && token->ident == id;
 211 }
 212
 213 #endif