build: allow easy override of GCC_BASE
[smatch.git] / token.h
bloba7ec77e05b15ba64fc3ba0e31b45d38dc21a2898
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.
8 * Copyright (C) 2003 Transmeta Corp.
9 * 2003 Linus Torvalds
11 * Licensed under the Open Software License version 1.1
14 #include <sys/types.h>
15 #include "lib.h"
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.
23 * Semantic meaning is handled elsewhere.
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
33 extern const char *includepath[];
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;
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;
49 extern int input_stream_nr;
50 extern struct stream *input_streams;
51 extern unsigned int tabstop;
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 */
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_WIDE_CHAR,
71 TOKEN_STRING,
72 TOKEN_WIDE_STRING,
73 TOKEN_SPECIAL,
74 TOKEN_STREAMBEGIN,
75 TOKEN_STREAMEND,
76 TOKEN_MACRO_ARGUMENT,
77 TOKEN_STR_ARGUMENT,
78 TOKEN_QUOTED_ARGUMENT,
79 TOKEN_CONCAT,
80 TOKEN_GNU_KLUDGE,
81 TOKEN_UNTAINT,
82 TOKEN_ARG_COUNT,
83 TOKEN_IF,
84 TOKEN_SKIP_GROUPS,
85 TOKEN_ELSE,
88 /* Combination tokens */
89 #define COMBINATION_STRINGS { \
90 "+=", "++", \
91 "-=", "--", "->", \
92 "*=", \
93 "/=", \
94 "%=", \
95 "<=", ">=", \
96 "==", "!=", \
97 "&&", "&=", \
98 "||", "|=", \
99 "^=", "##", \
100 "<<", ">>", "..", \
101 "<<=", ">>=", "...", \
102 "", \
103 "<", ">", "<=", ">=" \
106 extern unsigned char combinations[][4];
108 enum special_token {
109 SPECIAL_BASE = 256,
110 SPECIAL_ADD_ASSIGN = SPECIAL_BASE,
111 SPECIAL_INCREMENT,
112 SPECIAL_SUB_ASSIGN,
113 SPECIAL_DECREMENT,
114 SPECIAL_DEREFERENCE,
115 SPECIAL_MUL_ASSIGN,
116 SPECIAL_DIV_ASSIGN,
117 SPECIAL_MOD_ASSIGN,
118 SPECIAL_LTE,
119 SPECIAL_GTE,
120 SPECIAL_EQUAL,
121 SPECIAL_NOTEQUAL,
122 SPECIAL_LOGICAL_AND,
123 SPECIAL_AND_ASSIGN,
124 SPECIAL_LOGICAL_OR,
125 SPECIAL_OR_ASSIGN,
126 SPECIAL_XOR_ASSIGN,
127 SPECIAL_HASHHASH,
128 SPECIAL_LEFTSHIFT,
129 SPECIAL_RIGHTSHIFT,
130 SPECIAL_DOTDOT,
131 SPECIAL_SHL_ASSIGN,
132 SPECIAL_SHR_ASSIGN,
133 SPECIAL_ELLIPSIS,
134 SPECIAL_ARG_SEPARATOR,
135 SPECIAL_UNSIGNED_LT,
136 SPECIAL_UNSIGNED_GT,
137 SPECIAL_UNSIGNED_LTE,
138 SPECIAL_UNSIGNED_GTE,
141 struct string {
142 unsigned int length;
143 char data[];
146 /* will fit into 32 bits */
147 struct argcount {
148 unsigned normal:10;
149 unsigned quoted:10;
150 unsigned str:10;
151 unsigned vararg:1;
155 * This is a very common data structure, it should be kept
156 * as small as humanly possible. Big (rare) types go as
157 * pointers.
159 struct token {
160 struct position pos;
161 struct token *next;
162 union {
163 const char *number;
164 struct ident *ident;
165 unsigned int special;
166 struct string *string;
167 int character;
168 int argnum;
169 struct argcount count;
173 #define MAX_STRING 4095
175 static inline struct token *containing_token(struct token **p)
177 void *addr = (char *)p - ((char *)&((struct token *)0)->next - (char *)0);
178 return addr;
181 #define token_type(x) ((x)->pos.type)
184 * Last token in the stream - points to itself.
185 * This allows us to not test for NULL pointers
186 * when following the token->next chain..
188 extern struct token eof_token_entry;
189 #define eof_token(x) ((x) == &eof_token_entry)
191 extern int init_stream(const char *, int fd, const char **next_path);
192 extern const char *stream_name(int stream);
193 extern struct ident *hash_ident(struct ident *);
194 extern struct ident *built_in_ident(const char *);
195 extern struct token *built_in_token(int, const char *);
196 extern const char *show_special(int);
197 extern const char *show_ident(const struct ident *);
198 extern const char *show_string(const struct string *string);
199 extern const char *show_token(const struct token *);
200 extern struct token * tokenize(const char *, int, struct token *, const char **next_path);
201 extern struct token * tokenize_buffer(void *, unsigned long, struct token **);
203 extern void show_identifier_stats(void);
204 extern struct token *preprocess(struct token *);
206 static inline int match_op(struct token *token, int op)
208 return token->pos.type == TOKEN_SPECIAL && token->special == op;
211 static inline int match_ident(struct token *token, struct ident *id)
213 return token->pos.type == TOKEN_IDENT && token->ident == id;
216 #endif