/*
 * pgbench Expression Tokenizer -- Public Interface
 *
 * Hand-written tokenizer that implements the lexical analysis from
 * PostgreSQL's src/bin/pgbench/exprscan.l, adapted for use with
 * a Lime-generated parser.
 *
 * The tokenizer handles:
 *   - Case-insensitive keywords (AND, OR, NOT, CASE, WHEN, etc.)
 *   - Variables (:name)
 *   - Integer constants (decimal)
 *   - Double constants (decimal with '.', 'e'/'E' exponent)
 *   - Boolean constants (TRUE, FALSE)
 *   - NULL constant
 *   - The special INT64_MIN overflow literal "9223372036854775808"
 *   - All arithmetic, comparison, bitwise, and logical operators
 *   - Function names (identifiers not matching a keyword)
 *
 * Token codes come from pgbench_expr.h (generated by Lime).
 *
 * Usage:
 *   PgbenchScanner *s = pgbench_scanner_create(input, len);
 *   TokenValue val;
 *   int tok;
 *   while ((tok = pgbench_scan(s, &val)) > 0) {
 *       pgbenchExprParse(parser, tok, val, pstate);
 *   }
 *   pgbenchExprParse(parser, 0, empty_val, pstate);
 *   pgbench_scanner_destroy(s);
 */

#ifndef PGBENCH_TOKENIZE_H
#define PGBENCH_TOKENIZE_H

#include <stdint.h>
#include <stddef.h>
#include <stdbool.h>

/* Token code defines from the Lime-generated parser header. */
#include "pgbench_expr.h"

#ifdef __cplusplus
extern "C" {
#endif


/* ======================================================================
 * Semantic value type
 *
 * Matches the %token_type in the Lime grammar.
 * ====================================================================== */

typedef union TokenValue {
    int64_t     ival;       /* Integer constants, function numbers */
    double      dval;       /* Double constants */
    bool        bval;       /* Boolean constants */
    char       *str;        /* Variable names, function names */
} TokenValue;


/* ======================================================================
 * Opaque scanner handle
 * ====================================================================== */

typedef struct PgbenchScanner PgbenchScanner;


/* ======================================================================
 * Scanner lifecycle
 * ====================================================================== */

/*
 * Create a scanner for the given input string.
 * The input must remain valid for the lifetime of the scanner.
 */
PgbenchScanner *pgbench_scanner_create(const char *input, size_t length);

/*
 * Destroy a scanner and free resources. Passing NULL is a no-op.
 */
void pgbench_scanner_destroy(PgbenchScanner *s);


/* ======================================================================
 * Token extraction
 * ====================================================================== */

/*
 * Extract the next token from the input.
 *
 * Returns:
 *   > 0  Token code from pgbench_expr.h
 *   0    End of input
 *   -1   Lexical error (call pgbench_scanner_error() for details)
 *
 * On success, *val is filled with the semantic value.
 * Caller owns any val->str pointer returned (must free when done).
 */
int pgbench_scan(PgbenchScanner *s, TokenValue *val);


/* ======================================================================
 * Error reporting
 * ====================================================================== */

/*
 * Return the last error message, or NULL if no error.
 */
const char *pgbench_scanner_error(const PgbenchScanner *s);

/*
 * Return the byte offset of the last error, or -1.
 */
int pgbench_scanner_error_location(const PgbenchScanner *s);


/* ======================================================================
 * Debugging
 * ====================================================================== */

/*
 * Return a human-readable name for a token code.
 */
const char *pgbench_token_name(int token);


#ifdef __cplusplus
}
#endif

#endif /* PGBENCH_TOKENIZE_H */