2013-12-21 18:17:45 +00:00
|
|
|
struct _mp_lexer_t;
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
// a mp_parse_node_t is:
|
2013-10-04 19:53:11 +01:00
|
|
|
// - 0000...0000: no node
|
|
|
|
// - xxxx...0001: an identifier; bits 4 and above are the qstr
|
|
|
|
// - xxxx...0011: a small integer; bits 4 and above are the signed value, 2's complement
|
|
|
|
// - xxxx...0101: an integer; bits 4 and above are the qstr holding the value
|
|
|
|
// - xxxx...0111: a decimal; bits 4 and above are the qstr holding the value
|
|
|
|
// - xxxx...1001: a string; bits 4 and above are the qstr holding the value
|
|
|
|
// - xxxx...1011: a string with triple quotes; bits 4 and above are the qstr holding the value
|
2013-12-21 18:17:45 +00:00
|
|
|
// - xxxx...1101: a token; bits 4 and above are mp_token_kind_t
|
|
|
|
// - xxxx...xxx0: pointer to mp_parse_node_struct_t
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-10-12 16:16:56 +01:00
|
|
|
// makes sure the top 5 bits of x are all cleared (positive number) or all set (negavite number)
|
|
|
|
// these macros can probably go somewhere else because they are used more than just in the parser
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_UINT_HIGH_5_BITS (~((~((machine_uint_t)0)) >> 5))
|
|
|
|
#define MP_FIT_SMALL_INT(x) (((((machine_uint_t)(x)) & MP_UINT_HIGH_5_BITS) == 0) || ((((machine_uint_t)(x)) & MP_UINT_HIGH_5_BITS) == MP_UINT_HIGH_5_BITS))
|
2013-10-12 16:16:56 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_PARSE_NODE_NULL (0)
|
|
|
|
#define MP_PARSE_NODE_ID (0x1)
|
|
|
|
#define MP_PARSE_NODE_SMALL_INT (0x3)
|
|
|
|
#define MP_PARSE_NODE_INTEGER (0x5)
|
|
|
|
#define MP_PARSE_NODE_DECIMAL (0x7)
|
|
|
|
#define MP_PARSE_NODE_STRING (0x9)
|
|
|
|
#define MP_PARSE_NODE_BYTES (0xb)
|
|
|
|
#define MP_PARSE_NODE_TOKEN (0xd)
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
typedef machine_uint_t mp_parse_node_t; // must be pointer size
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
typedef struct _mp_parse_node_struct_t {
|
2014-01-18 23:24:36 +00:00
|
|
|
uint32_t source_line; // line number in source file
|
2013-10-04 19:53:11 +01:00
|
|
|
uint32_t kind_num_nodes; // parse node kind, and number of nodes
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t nodes[]; // nodes
|
|
|
|
} mp_parse_node_struct_t;
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
// macros for mp_parse_node_t usage
|
2013-10-04 19:53:11 +01:00
|
|
|
// some of these evaluate their argument more than once
|
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_PARSE_NODE_IS_NULL(pn) ((pn) == MP_PARSE_NODE_NULL)
|
|
|
|
#define MP_PARSE_NODE_IS_LEAF(pn) ((pn) & 1)
|
|
|
|
#define MP_PARSE_NODE_IS_STRUCT(pn) ((pn) != MP_PARSE_NODE_NULL && ((pn) & 1) == 0)
|
|
|
|
#define MP_PARSE_NODE_IS_STRUCT_KIND(pn, k) ((pn) != MP_PARSE_NODE_NULL && ((pn) & 1) == 0 && MP_PARSE_NODE_STRUCT_KIND((mp_parse_node_struct_t*)(pn)) == (k))
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_PARSE_NODE_IS_ID(pn) (((pn) & 0xf) == MP_PARSE_NODE_ID)
|
|
|
|
#define MP_PARSE_NODE_IS_SMALL_INT(pn) (((pn) & 0xf) == MP_PARSE_NODE_SMALL_INT)
|
|
|
|
#define MP_PARSE_NODE_IS_TOKEN(pn) (((pn) & 0xf) == MP_PARSE_NODE_TOKEN)
|
|
|
|
#define MP_PARSE_NODE_IS_TOKEN_KIND(pn, k) ((pn) == (MP_PARSE_NODE_TOKEN | (k << 4)))
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_PARSE_NODE_LEAF_KIND(pn) ((pn) & 0xf)
|
2013-10-04 19:53:11 +01:00
|
|
|
// TODO should probably have int and uint versions of this macro
|
2013-12-21 18:17:45 +00:00
|
|
|
#define MP_PARSE_NODE_LEAF_ARG(pn) (((machine_int_t)(pn)) >> 4)
|
|
|
|
#define MP_PARSE_NODE_STRUCT_KIND(pns) ((pns)->kind_num_nodes & 0xff)
|
|
|
|
#define MP_PARSE_NODE_STRUCT_NUM_NODES(pns) ((pns)->kind_num_nodes >> 8)
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-12-21 18:17:45 +00:00
|
|
|
mp_parse_node_t mp_parse_node_new_leaf(machine_int_t kind, machine_int_t arg);
|
2014-01-25 13:51:19 +00:00
|
|
|
uint mp_parse_node_free(mp_parse_node_t pn);
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2014-01-19 11:48:48 +00:00
|
|
|
void mp_parse_node_print(mp_parse_node_t pn, int indent);
|
2013-10-04 19:53:11 +01:00
|
|
|
|
2013-10-18 19:58:12 +01:00
|
|
|
typedef enum {
|
2013-12-21 18:17:45 +00:00
|
|
|
MP_PARSE_SINGLE_INPUT,
|
|
|
|
MP_PARSE_FILE_INPUT,
|
|
|
|
MP_PARSE_EVAL_INPUT,
|
|
|
|
} mp_parse_input_kind_t;
|
2013-10-18 19:58:12 +01:00
|
|
|
|
2014-01-15 21:23:31 +00:00
|
|
|
// returns MP_PARSE_NODE_NULL on error, and then exc_id_out and exc_msg_out are valid
|
|
|
|
mp_parse_node_t mp_parse(struct _mp_lexer_t *lex, mp_parse_input_kind_t input_kind, qstr *exc_id_out, const char **exc_msg_out);
|