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;    // inputfile 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         char name[];            /* Actual identifier */
  59 };
  60
  61 enum token_type {
  62         TOKEN_EOF,
  63         TOKEN_ERROR,
  64         TOKEN_IDENT,
  65         TOKEN_ZERO_IDENT,
  66         TOKEN_NUMBER,
  67         TOKEN_CHAR,
  68         TOKEN_STRING,
  69         TOKEN_SPECIAL,
  70         TOKEN_STREAMBEGIN,
  71         TOKEN_STREAMEND,
  72         TOKEN_MACRO_ARGUMENT,
  73         TOKEN_STR_ARGUMENT,
  74         TOKEN_QUOTED_ARGUMENT,
  75         TOKEN_CONCAT,
  76         TOKEN_GNU_KLUDGE,
  77         TOKEN_UNTAINT,
  78         TOKEN_ARG_COUNT,
  79         TOKEN_IF,
  80         TOKEN_SKIP_GROUPS,
  81         TOKEN_ELSE,
  82 };
  83
  84 /* Combination tokens */
  85 #define COMBINATION_STRINGS {   \
  86         "+=", "++",             \
  87         "-=", "--", "->",       \
  88         "*=",                   \
  89         "/=",                   \
  90         "%=",                   \
  91         "..", "...",            \
  92         "<=", "<<", "<<=",      \
  93         ">=", ">>", ">>=",      \
  94         "==", "!=",             \
  95         "&&", "&=",             \
  96         "||", "|=",             \
  97         "^=", "##",             \
  98         "",                     \
  99         "<", ">", "<=", ">="    \
 100 }
 101
 102 extern unsigned char combinations[][3];
 103
 104 enum special_token {
 105         SPECIAL_BASE = 256,
 106         SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
 107         SPECIAL_INCREMENT,
 108         SPECIAL_SUB_ASSIGN,
 109         SPECIAL_DECREMENT,
 110         SPECIAL_DEREFERENCE,
 111         SPECIAL_MUL_ASSIGN,
 112         SPECIAL_DIV_ASSIGN,
 113         SPECIAL_MOD_ASSIGN,
 114         SPECIAL_DOTDOT,
 115         SPECIAL_ELLIPSIS,
 116         SPECIAL_LTE,
 117         SPECIAL_LEFTSHIFT,
 118         SPECIAL_SHL_ASSIGN,
 119         SPECIAL_GTE,
 120         SPECIAL_RIGHTSHIFT,
 121         SPECIAL_SHR_ASSIGN,
 122         SPECIAL_EQUAL,
 123         SPECIAL_NOTEQUAL,
 124         SPECIAL_LOGICAL_AND,
 125         SPECIAL_AND_ASSIGN,
 126         SPECIAL_LOGICAL_OR,
 127         SPECIAL_OR_ASSIGN,
 128         SPECIAL_XOR_ASSIGN,
 129         SPECIAL_HASHHASH,
 130         SPECIAL_ARG_SEPARATOR,
 131         SPECIAL_UNSIGNED_LT,
 132         SPECIAL_UNSIGNED_GT,
 133         SPECIAL_UNSIGNED_LTE,
 134         SPECIAL_UNSIGNED_GTE,
 135 };
 136
 137 struct string {
 138         unsigned int length;
 139         char data[];
 140 };
 141
 142 /* will fit into 32 bits */
 143 struct argcount {
 144         unsigned normal:10;
 145         unsigned quoted:10;
 146         unsigned str:10;
 147         unsigned vararg:1;
 148 };
 149
 150 /*
 151  * This is a very common data structure, it should be kept
 152  * as small as humanly possible. Big (rare) types go as
 153  * pointers.
 154  */
 155 struct token {
 156         struct position pos;
 157         struct token *next;
 158         union {
 159                 const char *number;
 160                 struct ident *ident;
 161                 unsigned int special;
 162                 struct string *string;
 163                 int character;
 164                 int argnum;
 165                 struct argcount count;
 166         };
 167 };
 168
 169 #define MAX_STRING 4095
 170
 171 static inline struct token *containing_token(struct token **p)
 172 {
 173         void *addr = (char*)p - ((char*)&((struct token *)0)->next - (char*)0);
 174         return addr;
 175 }
 176
 177 #define token_type(x) ((x)->pos.type)
 178
 179 /*
 180  * Last token in the stream - points to itself.
 181  * This allows us to not test for NULL pointers
 182  * when following the token->next chain..
 183  */
 184 extern struct token eof_token_entry;
 185 #define eof_token(x) ((x) == &eof_token_entry)
 186
 187 extern int init_stream(const char *, int fd, const char **next_path);
 188 extern const char *stream_name(int stream);
 189 extern struct ident *hash_ident(struct ident *);
 190 extern struct ident *built_in_ident(const char *);
 191 extern struct token *built_in_token(int, const char *);
 192 extern const char *show_special(int);
 193 extern const char *show_ident(const struct ident *);
 194 extern const char *show_string(const struct string *string);
 195 extern const char *show_token(const struct token *);
 196 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
 197 extern struct token * tokenize_buffer(void *, unsigned long, struct token *);
 198
 199 extern void show_identifier_stats(void);
 200 extern struct token *preprocess(struct token *);
 201
 202 static inline int match_op(struct token *token, int op)
 203 {
 204         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 205 }
 206
 207 static inline int match_ident(struct token *token, struct ident *id)
 208 {
 209         return token->pos.type == TOKEN_IDENT && token->ident == id;
 210 }
 211
 212 #endif