parse: Refactor parse node encoding to support full range of small ints.

Based on suggestion by @dpgeorge at
https://github.com/micropython/micropython/pull/313

1 files changed, 29 insertions, 27 deletions

diff --git a/py/parse.h b/py/parse.h
index 6e299ef69..6492f4d9e 100644
--- a/py/parse.h
+++ b/py/parse.h
@@ -2,29 +2,30 @@ struct _mp_lexer_t;
 
 // a mp_parse_node_t is:
 //  - 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
-//  - xxxx...1101: a token; bits 4 and above are mp_token_kind_t
-//  - xxxx...xxx0: pointer to mp_parse_node_struct_t
+//  - xxxx...xxx1: a small integer; bits 1 and above are the signed value, 2's complement
+//  - xxxx...xx00: pointer to mp_parse_node_struct_t
+//  - xx...x00010: an identifier; bits 5 and above are the qstr
+//  - xx...x00110: an integer; bits 5 and above are the qstr holding the value
+//  - xx...x01010: a decimal; bits 5 and above are the qstr holding the value
+//  - xx...x01110: a string; bits 5 and above are the qstr holding the value
+//  - xx...x10010: a string with triple quotes; bits 5 and above are the qstr holding the value
+//  - xx...x10110: a token; bits 5 and above are mp_token_kind_t
 
-// makes sure the top 5 bits of x are all cleared (positive number) or all set (negavite number)
+// TODO: these can now be unified with MP_OBJ_FITS_SMALL_INT(x)
+// makes sure the top 2 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
-#define MP_UINT_HIGH_5_BITS (~((~((machine_uint_t)0)) >> 5))
+#define MP_UINT_HIGH_2_BITS (~((~((machine_uint_t)0)) >> 2))
 // parser's small ints are different from VM small int
-#define MP_PARSE_FITS_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))
+#define MP_PARSE_FITS_SMALL_INT(x) (((((machine_uint_t)(x)) & MP_UINT_HIGH_2_BITS) == 0) || ((((machine_uint_t)(x)) & MP_UINT_HIGH_2_BITS) == MP_UINT_HIGH_2_BITS))
 
 #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)
+#define MP_PARSE_NODE_SMALL_INT (0x1)
+#define MP_PARSE_NODE_ID        (0x02)
+#define MP_PARSE_NODE_INTEGER   (0x06)
+#define MP_PARSE_NODE_DECIMAL   (0x0a)
+#define MP_PARSE_NODE_STRING    (0x0e)
+#define MP_PARSE_NODE_BYTES     (0x12)
+#define MP_PARSE_NODE_TOKEN     (0x16)
 
 typedef machine_uint_t mp_parse_node_t; // must be pointer size
 
@@ -38,18 +39,19 @@ typedef struct _mp_parse_node_struct_t {
 // some of these evaluate their argument more than once
 
 #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))
+#define MP_PARSE_NODE_IS_LEAF(pn) ((pn) & 3)
+#define MP_PARSE_NODE_IS_STRUCT(pn) ((pn) != MP_PARSE_NODE_NULL && ((pn) & 3) == 0)
+#define MP_PARSE_NODE_IS_STRUCT_KIND(pn, k) ((pn) != MP_PARSE_NODE_NULL && ((pn) & 3) == 0 && MP_PARSE_NODE_STRUCT_KIND((mp_parse_node_struct_t*)(pn)) == (k))
 
-#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)))
+#define MP_PARSE_NODE_IS_SMALL_INT(pn) (((pn) & 0x1) == MP_PARSE_NODE_SMALL_INT)
+#define MP_PARSE_NODE_IS_ID(pn) (((pn) & 0x1f) == MP_PARSE_NODE_ID)
+#define MP_PARSE_NODE_IS_TOKEN(pn) (((pn) & 0x1f) == MP_PARSE_NODE_TOKEN)
+#define MP_PARSE_NODE_IS_TOKEN_KIND(pn, k) ((pn) == (MP_PARSE_NODE_TOKEN | ((k) << 5)))
 
-#define MP_PARSE_NODE_LEAF_KIND(pn) ((pn) & 0xf)
+#define MP_PARSE_NODE_LEAF_KIND(pn) ((pn) & 0x1f)
 // TODO should probably have int and uint versions of this macro
-#define MP_PARSE_NODE_LEAF_ARG(pn) (((machine_int_t)(pn)) >> 4)
+#define MP_PARSE_NODE_LEAF_ARG(pn) (((machine_uint_t)(pn)) >> 5)
+#define MP_PARSE_NODE_LEAF_SMALL_INT(pn) (((machine_int_t)(pn)) >> 1)
 #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)