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 **next_path;
  39
  40         /* Use these to check for "already parsed" */
  41         enum constantfile constant;
  42         int nesting;
  43         struct ident *protect;
  44 };
  45
  46 extern int input_stream_nr;
  47 extern struct stream *input_streams;
  48
  49 extern int ident_hit, ident_miss;
  50
  51 struct ident {
  52         struct ident *next;     /* Hash chain of identifiers */
  53         struct symbol *symbols; /* Pointer to semantic meaning list */
  54         unsigned char len;      /* Length of identifier name */
  55         unsigned char tainted:1,
  56                       reserved:1;
  57         char name[];            /* Actual identifier */
  58 };
  59
  60 enum token_type {
  61         TOKEN_EOF,
  62         TOKEN_ERROR,
  63         TOKEN_IDENT,
  64         TOKEN_NUMBER,
  65         TOKEN_CHAR,
  66         TOKEN_STRING,
  67         TOKEN_SPECIAL,
  68         TOKEN_STREAMBEGIN,
  69         TOKEN_STREAMEND,
  70         TOKEN_MACRO_ARGUMENT,
  71         TOKEN_STR_ARGUMENT,
  72         TOKEN_QUOTED_ARGUMENT,
  73         TOKEN_CONCAT,
  74         TOKEN_GNU_KLUDGE,
  75         TOKEN_UNTAINT,
  76         TOKEN_ARG_COUNT,
  77 };
  78
  79 /* Combination tokens */
  80 #define COMBINATION_STRINGS {   \
  81         "+=", "++",             \
  82         "-=", "--", "->",       \
  83         "*=",                   \
  84         "/=",                   \
  85         "%=",                   \
  86         "..", "...",            \
  87         "<=", "<<", "<<=",      \
  88         ">=", ">>", ">>=",      \
  89         "==", "!=",             \
  90         "&&", "&=",             \
  91         "||", "|=",             \
  92         "^=", "##",             \
  93         "",                     \
  94         "<", ">", "<=", ">="    \
  95 }
  96
  97 enum special_token {
  98         SPECIAL_BASE = 256,
  99         SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
 100         SPECIAL_INCREMENT,
 101         SPECIAL_SUB_ASSIGN,
 102         SPECIAL_DECREMENT,
 103         SPECIAL_DEREFERENCE,
 104         SPECIAL_MUL_ASSIGN,
 105         SPECIAL_DIV_ASSIGN,
 106         SPECIAL_MOD_ASSIGN,
 107         SPECIAL_DOTDOT,
 108         SPECIAL_ELLIPSIS,
 109         SPECIAL_LTE,
 110         SPECIAL_LEFTSHIFT,
 111         SPECIAL_SHL_ASSIGN,
 112         SPECIAL_GTE,
 113         SPECIAL_RIGHTSHIFT,
 114         SPECIAL_SHR_ASSIGN,
 115         SPECIAL_EQUAL,
 116         SPECIAL_NOTEQUAL,
 117         SPECIAL_LOGICAL_AND,
 118         SPECIAL_AND_ASSIGN,
 119         SPECIAL_LOGICAL_OR,
 120         SPECIAL_OR_ASSIGN,
 121         SPECIAL_XOR_ASSIGN,
 122         SPECIAL_HASHHASH,
 123         SPECIAL_ARG_SEPARATOR,
 124         SPECIAL_UNSIGNED_LT,
 125         SPECIAL_UNSIGNED_GT,
 126         SPECIAL_UNSIGNED_LTE,
 127         SPECIAL_UNSIGNED_GTE,
 128 };
 129
 130 struct string {
 131         unsigned int length;
 132         char data[];
 133 };
 134
 135 /* will fit into 32 bits */
 136 struct argcount {
 137         unsigned normal:10;
 138         unsigned quoted:10;
 139         unsigned str:10;
 140         unsigned vararg:1;
 141 };
 142
 143 /*
 144  * This is a very common data structure, it should be kept
 145  * as small as humanly possible. Big (rare) types go as
 146  * pointers.
 147  */
 148 struct token {
 149         struct position pos;
 150         struct token *next;
 151         union {
 152                 const char *number;
 153                 struct ident *ident;
 154                 unsigned int special;
 155                 struct string *string;
 156                 int character;
 157                 int argnum;
 158                 struct argcount count;
 159         };
 160 };
 161
 162 #define MAX_STRING 4095
 163
 164 static inline struct token *containing_token(struct token **p)
 165 {
 166         void *addr = (char*)p - ((char*)&((struct token *)0)->next - (char*)0);
 167         return addr;
 168 }
 169
 170 #define token_type(x) ((x)->pos.type)
 171
 172 /*
 173  * Last token in the stream - points to itself.
 174  * This allows us to not test for NULL pointers
 175  * when following the token->next chain..
 176  */
 177 extern struct token eof_token_entry;
 178 #define eof_token(x) ((x) == &eof_token_entry)
 179
 180 extern int init_stream(const char *, int fd, const char **next_path);
 181 extern const char *stream_name(int stream);
 182 extern struct ident *hash_ident(struct ident *);
 183 extern struct ident *built_in_ident(const char *);
 184 extern struct token *built_in_token(int, const char *);
 185 extern const char *show_special(int);
 186 extern const char *show_ident(const struct ident *);
 187 extern const char *show_string(const struct string *string);
 188 extern const char *show_token(const struct token *);
 189 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
 190 extern struct token * tokenize_buffer(unsigned char *, unsigned long, struct token *);
 191
 192 extern void show_identifier_stats(void);
 193 extern struct token *preprocess(struct token *);
 194
 195 static inline int match_op(struct token *token, int op)
 196 {
 197         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 198 }
 199
 200 static inline int match_ident(struct token *token, struct ident *id)
 201 {
 202         return token->pos.type == TOKEN_IDENT && token->ident == id;
 203 }
 204
 205 #endif