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 nesting;
  44         struct ident *protect;
  45 };
  46
  47 extern int input_stream_nr;
  48 extern struct stream *input_streams;
  49
  50 struct ident {
  51         struct ident *next;     /* Hash chain of identifiers */
  52         struct symbol *symbols; /* Pointer to semantic meaning list */
  53         unsigned char len;      /* Length of identifier name */
  54         unsigned char tainted:1,
  55                       reserved:1;
  56         char name[];            /* Actual identifier */
  57 };
  58
  59 enum token_type {
  60         TOKEN_EOF,
  61         TOKEN_ERROR,
  62         TOKEN_IDENT,
  63         TOKEN_ZERO_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 extern unsigned char combinations[][3];
  98
  99 enum special_token {
 100         SPECIAL_BASE = 256,
 101         SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
 102         SPECIAL_INCREMENT,
 103         SPECIAL_SUB_ASSIGN,
 104         SPECIAL_DECREMENT,
 105         SPECIAL_DEREFERENCE,
 106         SPECIAL_MUL_ASSIGN,
 107         SPECIAL_DIV_ASSIGN,
 108         SPECIAL_MOD_ASSIGN,
 109         SPECIAL_DOTDOT,
 110         SPECIAL_ELLIPSIS,
 111         SPECIAL_LTE,
 112         SPECIAL_LEFTSHIFT,
 113         SPECIAL_SHL_ASSIGN,
 114         SPECIAL_GTE,
 115         SPECIAL_RIGHTSHIFT,
 116         SPECIAL_SHR_ASSIGN,
 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_ARG_SEPARATOR,
 126         SPECIAL_UNSIGNED_LT,
 127         SPECIAL_UNSIGNED_GT,
 128         SPECIAL_UNSIGNED_LTE,
 129         SPECIAL_UNSIGNED_GTE,
 130 };
 131
 132 struct string {
 133         unsigned int length;
 134         char data[];
 135 };
 136
 137 /* will fit into 32 bits */
 138 struct argcount {
 139         unsigned normal:10;
 140         unsigned quoted:10;
 141         unsigned str:10;
 142         unsigned vararg:1;
 143 };
 144
 145 /*
 146  * This is a very common data structure, it should be kept
 147  * as small as humanly possible. Big (rare) types go as
 148  * pointers.
 149  */
 150 struct token {
 151         struct position pos;
 152         struct token *next;
 153         union {
 154                 const char *number;
 155                 struct ident *ident;
 156                 unsigned int special;
 157                 struct string *string;
 158                 int character;
 159                 int argnum;
 160                 struct argcount count;
 161         };
 162 };
 163
 164 #define MAX_STRING 4095
 165
 166 static inline struct token *containing_token(struct token **p)
 167 {
 168         void *addr = (char*)p - ((char*)&((struct token *)0)->next - (char*)0);
 169         return addr;
 170 }
 171
 172 #define token_type(x) ((x)->pos.type)
 173
 174 /*
 175  * Last token in the stream - points to itself.
 176  * This allows us to not test for NULL pointers
 177  * when following the token->next chain..
 178  */
 179 extern struct token eof_token_entry;
 180 #define eof_token(x) ((x) == &eof_token_entry)
 181
 182 extern int init_stream(const char *, int fd, const char **next_path);
 183 extern const char *stream_name(int stream);
 184 extern struct ident *hash_ident(struct ident *);
 185 extern struct ident *built_in_ident(const char *);
 186 extern struct token *built_in_token(int, const char *);
 187 extern const char *show_special(int);
 188 extern const char *show_ident(const struct ident *);
 189 extern const char *show_string(const struct string *string);
 190 extern const char *show_token(const struct token *);
 191 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
 192 extern struct token * tokenize_buffer(void *, unsigned long, struct token *);
 193
 194 extern void show_identifier_stats(void);
 195 extern struct token *preprocess(struct token *);
 196
 197 static inline int match_op(struct token *token, int op)
 198 {
 199         return token->pos.type == TOKEN_SPECIAL && token->special == op;
 200 }
 201
 202 static inline int match_ident(struct token *token, struct ident *id)
 203 {
 204         return token->pos.type == TOKEN_IDENT && token->ident == id;
 205 }
 206
 207 #endif