terom@13: #ifndef LIB_LEXER_H
terom@13: #define LIB_LEXER_H
terom@13: 
terom@13: /*
terom@13:  * Simple FSM lexing
terom@13:  *
terom@13:  * The lexer is implemented as a Finite State Machine, consisting for a number of states, which then contain a set of
terom@13:  * transitions, which move the lexer from state to state based on each char of input at a time.
terom@13:  *
terom@13:  * Whenever the state changes, the token callback is triggered with the collected token data.
terom@13:  */
terom@13: 
terom@13: /*
terom@14:  * Transition flags
terom@14:  */
terom@14: enum lex_transition_flags {
terom@14:     LEX_TRANS_DEFAULT   = 0x01,
terom@14:     LEX_TRANS_FINAL     = 0x02,
terom@14: };
terom@14: 
terom@14: /*
terom@13:  * A transition from one state to another.
terom@13:  */
terom@13: struct lex_transition {
terom@13:     // applies to chars [left, right]
terom@13:     char left, right;
terom@13:     
terom@14:     // flags from lex_transition_flags
terom@14:     char flags;
terom@14:     
terom@13:     // next state to enter
terom@13:     int next_state;
terom@13: };
terom@13: 
terom@13: /*
terom@14:  * State flags
terom@14:  */ 
terom@14: enum lex_state_flags {
terom@14:     LEX_STATE_END       = 0x01;
terom@14: };
terom@14: 
terom@14: /*
terom@13:  * A state
terom@13:  */
terom@13: struct lex_state {
terom@13:     // the state name (for debugging)
terom@13:     const char *name;
terom@13: 
terom@14:     // flags from lex_state_flags
terom@14:     char flags;
terom@14: 
terom@13:     // list of transitions for this state, terminated by a transition with next_state=0
terom@13:     struct lex_transition *trans_list;
terom@13: };
terom@13: 
terom@13: /*
terom@13:  * Lex machine
terom@13:  */
terom@13: struct lex {
terom@13:     // number of states
terom@13:     size_t state_count;
terom@13: 
terom@13:     // array of lex_states, indexable by the state id.
terom@13:     struct lex_state *state_list;
terom@13: 
terom@13:     /*
terom@13:      * Core token handler. Everytime a full token is lexed (i.e. the state changes), this will be called.
terom@13:      * `this_token` represents the full token that was parsed, and `token_data` is the token's value. `next_token`
terom@13:      * is the state that terminated this token, and `prev_token` was the token before this one.
terom@13:      *
terom@13:      * `token_data` is a buffer allocated by the lexer that the actual input data is copied into. Thence, it can be
terom@13:      * modified, as its contents will be replaced by the next token. Hence, if you need to keep hold of it, copy it.
terom@13:      *
terom@13:      * Return zero to have lexing continue, nonzero to stop lexing.
terom@13:      */
terom@14:     int (*token_fn) (int this_token, char *token_data, int next_token, int prev_token, void *arg);
terom@13: 
terom@13:     /*
terom@13:      * Called on every char handled by the lexer. `this_token` is the state of the token that the char belongs to.
terom@13:      *
terom@13:      * Return zero to have lexing continue, nonzero to stop lexing.
terom@13:      */
terom@14:     int (*char_fn) (int this_token, char token_char, void *arg);
terom@13: 
terom@13:     /*
terom@13:      * Called when the end of input has been reached, `last_token` is the state that we terminated in.
terom@13:      *
terom@13:      * Return zero to indiciate that the input was valid, nonzero to indicate an error.
terom@13:      */
terom@14:     int (*end_fn) (int last_token, void *arg);
terom@13: };
terom@13: 
terom@13: /*
terom@13:  * Helper macros for building the state_list
terom@13:  */
terom@14: #define LEX_STATE(enum_val)     { #enum_val, 0,
terom@14: #define LEX_STATE_END(enum_val) { #enum_val, LEX_STATE_END,
terom@13: 
terom@14:     #define LEX_CHAR(c, to)         { c, c, 0, to },
terom@14:     #define LEX_RANGE(l, r, to)     { l, r, 0, to },
terom@13:     #define LEX_ALPHA(to)           LEX_RANGE('a', 'z', to), LEX_RANGE('A', 'Z', to)
terom@13:     #define LEX_NUMBER(to)          LEX_RANGE('0', '9', to)
terom@13:     #define LEX_ALNUM(to)           LEX_ALPHA(to), LEX_NUMBER(to), LEX_CHAR('-', to), LEX_CHAR('_', to)
terom@13:     #define LEX_WHITESPACE(to)      LEX_CHAR(' ', to), LEX_CHAR('\n', to), LEX_CHAR('\t', to)
terom@13: 
terom@14:     #define LEX_DEFAULT(to)         { 0, 0, LEX_TRANS_DEFAULT, to } \
terom@14:                                   }
terom@14:     #define LEX_END                 { 0, 0, 0, 0 } \
terom@14:                                   }
terom@13: 
terom@13: /*
terom@13:  * Lex it!
terom@13:  *
terom@13:  * Return zero to indiciate that the input was valid, nonzero otherwise.
terom@13:  */
terom@14: int lexer (const struct lex *lex, const char *input, void *arg);
terom@13: 
terom@13: #endif /* LIB_LEXER_H */