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 extern unsigned int tabstop;
  52
  53 struct ident {
  54         struct ident *next;     /* Hash chain of identifiers */
  55         struct symbol *symbols; /* Pointer to semantic meaning list */
  56         unsigned char len;      /* Length of identifier name */
  57         unsigned char tainted:1,
  58                       reserved:1,
  59                       keyword:1;
  60         char name[];            /* Actual identifier */
  61 };
  62
  63 enum token_type {
  64         TOKEN_EOF,
  65         TOKEN_ERROR,
  66         TOKEN_IDENT,
  67         TOKEN_ZERO_IDENT,
  68         TOKEN_NUMBER,
  69         TOKEN_CHAR,
  70         TOKEN_STRING,
  71         TOKEN_SPECIAL,
  72         TOKEN_STREAMBEGIN,
  73         TOKEN_STREAMEND,
  74         TOKEN_MACRO_ARGUMENT,
  75         TOKEN_STR_ARGUMENT,
  76         TOKEN_QUOTED_ARGUMENT,
  77         TOKEN_CONCAT,
  78         TOKEN_GNU_KLUDGE,
  79         TOKEN_UNTAINT,
  80         TOKEN_ARG_COUNT,
  81         TOKEN_IF,
  82         TOKEN_SKIP_GROUPS,
  83         TOKEN_ELSE,
  84 };
  85
  86 /* Combination tokens */
  87 #define COMBINATION_STRINGS {   \
  88         "+=", "++",             \
  89         "-=", "--", "->",       \
  90         "*=",                   \
  91         "/=",                   \
  92         "%=",                   \
  93         "<=", ">=",             \
  94         "==", "!=",             \
  95         "&&", "&=",             \
  96         "||", "|=",             \
  97         "^=", "##",             \
  98         "<<", ">>", "..",       \
  99         "<<=", ">>=", "...",    \
 100         "",                     \
 101         "<", ">", "<=", ">="    \
 102 }
 103
 104 extern unsigned char combinations[][4];
 105
 106 enum special_token {
 107         SPECIAL_BASE = 256,
 108         SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
 109         SPECIAL_INCREMENT,
 110         SPECIAL_SUB_ASSIGN,
 111         SPECIAL_DECREMENT,
 112         SPECIAL_DEREFERENCE,
 113         SPECIAL_MUL_ASSIGN,
 114         SPECIAL_DIV_ASSIGN,
 115         SPECIAL_MOD_ASSIGN,
 116         SPECIAL_LTE,
 117         SPECIAL_GTE,
 118         SPECIAL_EQUAL,
 119         SPECIAL_NOTEQUAL,
 120         SPECIAL_LOGICAL_AND,
 121         SPECIAL_AND_ASSIGN,
 122         SPECIAL_LOGICAL_OR,
 123         SPECIAL_OR_ASSIGN,
 124         SPECIAL_XOR_ASSIGN,
 125         SPECIAL_HASHHASH,
 126         SPECIAL_LEFTSHIFT,
 127         SPECIAL_RIGHTSHIFT,
 128         SPECIAL_DOTDOT,
 129         SPECIAL_SHL_ASSIGN,
 130         SPECIAL_SHR_ASSIGN,
 131         SPECIAL_ELLIPSIS,
 132         SPECIAL_ARG_SEPARATOR,
 133         SPECIAL_UNSIGNED_LT,
 134         SPECIAL_UNSIGNED_GT,
 135         SPECIAL_UNSIGNED_LTE,
 136         SPECIAL_UNSIGNED_GTE,
 137 };
 138
 139 struct string {
 140         unsigned int length;
 141         char data[];
 142 };
 143
 144 /* will fit into 32 bits */
 145 struct argcount {
 146         unsigned normal:10;
 147         unsigned quoted:10;
 148         unsigned str:10;
 149         unsigned vararg:1;
 150 };
 151
 152 /*
 153  * This is a very common data structure, it should be kept
 154  * as small as humanly possible. Big (rare) types go as
 155  * pointers.
 156  */
 157 struct token {
 158         struct position pos;
 159         struct token *next;
 160         union {
 161                 const char *number;
 162                 struct ident *ident;
 163                 unsigned int special;
 164                 struct string *string;
 165                 int character;
 166                 int argnum;
 167                 struct argcount count;
 168         };
 169 };
 170
 171 #define MAX_STRING 4095
 172
 173 static inline struct token *containing_token(struct token **p)
 174 {
 175         void *addr = (char *)p - ((char *)&((struct token *)0)->next - (char *)0);
 176         return addr;
 177 }
 178
 179 #define token_type(x) ((x)->pos.type)
 180
 181 /*
 182  * Last token in the stream - points to itself.
 183  * This allows us to not test for NULL pointers
 184  * when following the token->next chain..
 185  */
 186 extern struct token eof_token_entry;
 187 #define eof_token(x) ((x) == &eof_token_entry)
 188
 189 extern int init_stream(const char *, int fd, const char **next_path);
 190 extern const char *stream_name(int stream);
 191 extern struct ident *hash_ident(struct ident *);
 192 extern struct ident *built_in_ident(const char *);
 193 extern struct token *built_in_token(int, const char *);
 194 extern const char *show_special(int);
 195 extern const char *show_ident(const struct ident *);
 196 extern const char *show_string(const struct string *string);
 197 extern const char *show_token(const struct token *);
 198 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
 199 extern struct token * tokenize_buffer(void *, unsigned long, struct token **);
 200
 201 extern void show_identifier_stats(void);
 202 extern struct token *preprocess(struct token *);
 203
 204 static inline int match_op(struct token *token, int op)
 205 {
 206         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 207 }
 208
 209 static inline int match_ident(struct token *token, struct ident *id)
 210 {
 211         return token->pos.type == TOKEN_IDENT && token->ident == id;
 212 }
 213
 214 #endif