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

py/compile.c

@@ -86,8 +86,8 @@ mp_parse_node_t fold_constants(mp_parse_node_t pn) {
         switch (MP_PARSE_NODE_STRUCT_KIND(pns)) {
             case PN_shift_expr:
                 if (n == 3 && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
-                    int arg0 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
+                    int arg0 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[0]);
-                    int arg1 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
+                    int arg1 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[2]);
                     if (MP_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], MP_TOKEN_OP_DBL_LESS)) {
 #if MICROPY_EMIT_CPYTHON
                         // can overflow; enabled only to compare with CPython
@@ -105,8 +105,8 @@ mp_parse_node_t fold_constants(mp_parse_node_t pn) {
             case PN_arith_expr:
                 // overflow checking here relies on SMALL_INT being strictly smaller than machine_int_t
                 if (n == 3 && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
-                    machine_int_t arg0 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
+                    machine_int_t arg0 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[0]);
-                    machine_int_t arg1 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
+                    machine_int_t arg1 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[2]);
                     machine_int_t res;
                     if (MP_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], MP_TOKEN_OP_PLUS)) {
                         res = arg0 + arg1;
@@ -125,8 +125,8 @@ mp_parse_node_t fold_constants(mp_parse_node_t pn) {
 
             case PN_term:
                 if (n == 3 && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[2])) {
-                    int arg0 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
+                    int arg0 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[0]);
-                    int arg1 = MP_PARSE_NODE_LEAF_ARG(pns->nodes[2]);
+                    int arg1 = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[2]);
                     if (MP_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[1], MP_TOKEN_OP_STAR)) {
 #if MICROPY_EMIT_CPYTHON
                         // can overflow; enabled only to compare with CPython
@@ -149,7 +149,7 @@ mp_parse_node_t fold_constants(mp_parse_node_t pn) {
 
             case PN_factor_2:
                 if (MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[1])) {
-                    machine_int_t arg = MP_PARSE_NODE_LEAF_ARG(pns->nodes[1]);
+                    machine_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[1]);
                     if (MP_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[0], MP_TOKEN_OP_PLUS)) {
                         pn = mp_parse_node_new_leaf(MP_PARSE_NODE_SMALL_INT, arg);
                     } else if (MP_PARSE_NODE_IS_TOKEN_KIND(pns->nodes[0], MP_TOKEN_OP_MINUS)) {
@@ -169,10 +169,10 @@ mp_parse_node_t fold_constants(mp_parse_node_t pn) {
                 if (MP_PARSE_NODE_IS_SMALL_INT(pns->nodes[0]) && MP_PARSE_NODE_IS_NULL(pns->nodes[1]) && !MP_PARSE_NODE_IS_NULL(pns->nodes[2])) {
                     mp_parse_node_struct_t* pns2 = (mp_parse_node_struct_t*)pns->nodes[2];
                     if (MP_PARSE_NODE_IS_SMALL_INT(pns2->nodes[0])) {
-                        int power = MP_PARSE_NODE_LEAF_ARG(pns2->nodes[0]);
+                        int power = MP_PARSE_NODE_LEAF_SMALL_INT(pns2->nodes[0]);
                         if (power >= 0) {
                             int ans = 1;
-                            int base = MP_PARSE_NODE_LEAF_ARG(pns->nodes[0]);
+                            int base = MP_PARSE_NODE_LEAF_SMALL_INT(pns->nodes[0]);
                             for (; power > 0; power--) {
                                 ans *= base;
                             }
@@ -320,10 +320,14 @@ STATIC void cpython_c_print_quoted_str(vstr_t *vstr, qstr qstr, bool bytes) {
 
 STATIC void cpython_c_tuple_emit_const(compiler_t *comp, mp_parse_node_t pn, vstr_t *vstr) {
     assert(MP_PARSE_NODE_IS_LEAF(pn));
+    if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
+        vstr_printf(vstr, INT_FMT, MP_PARSE_NODE_LEAF_SMALL_INT(pn));
+        return;
+    }
+
     int arg = MP_PARSE_NODE_LEAF_ARG(pn);
     switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
         case MP_PARSE_NODE_ID: assert(0);
-        case MP_PARSE_NODE_SMALL_INT: vstr_printf(vstr, "%d", arg); break;
         case MP_PARSE_NODE_INTEGER: vstr_printf(vstr, "%s", qstr_str(arg)); break;
         case MP_PARSE_NODE_DECIMAL: vstr_printf(vstr, "%s", qstr_str(arg)); break;
         case MP_PARSE_NODE_STRING: cpython_c_print_quoted_str(vstr, arg, false); break;
@@ -421,11 +425,11 @@ void compile_generic_tuple(compiler_t *comp, mp_parse_node_struct_t *pns) {
 
 STATIC bool node_is_const_false(mp_parse_node_t pn) {
     return MP_PARSE_NODE_IS_TOKEN_KIND(pn, MP_TOKEN_KW_FALSE);
-    // untested: || (MP_PARSE_NODE_IS_SMALL_INT(pn) && MP_PARSE_NODE_LEAF_ARG(pn) == 1);
+    // untested: || (MP_PARSE_NODE_IS_SMALL_INT(pn) && MP_PARSE_NODE_LEAF_SMALL_INT(pn) == 0);
 }
 
 STATIC bool node_is_const_true(mp_parse_node_t pn) {
-    return MP_PARSE_NODE_IS_TOKEN_KIND(pn, MP_TOKEN_KW_TRUE) || (MP_PARSE_NODE_IS_SMALL_INT(pn) && MP_PARSE_NODE_LEAF_ARG(pn) == 1);
+    return MP_PARSE_NODE_IS_TOKEN_KIND(pn, MP_TOKEN_KW_TRUE) || (MP_PARSE_NODE_IS_SMALL_INT(pn) && MP_PARSE_NODE_LEAF_SMALL_INT(pn) == 1);
 }
 
 #if MICROPY_EMIT_CPYTHON
@@ -1464,7 +1468,8 @@ void compile_for_stmt_optimised_range(compiler_t *comp, mp_parse_node_t pn_var,
     // compile: if var <cond> end: goto top
     compile_node(comp, pn_var);
     compile_node(comp, pn_end);
-    if (MP_PARSE_NODE_LEAF_ARG(pn_step) >= 0) {
+    assert(MP_PARSE_NODE_IS_SMALL_INT(pn_step));
+    if (MP_PARSE_NODE_LEAF_SMALL_INT(pn_step) >= 0) {
         EMIT_ARG(binary_op, RT_BINARY_OP_LESS);
     } else {
         EMIT_ARG(binary_op, RT_BINARY_OP_MORE);
@@ -2514,11 +2519,13 @@ STATIC compile_function_t compile_function[] = {
 void compile_node(compiler_t *comp, mp_parse_node_t pn) {
     if (MP_PARSE_NODE_IS_NULL(pn)) {
         // pass
+    } else if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
+        machine_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pn);
+        EMIT_ARG(load_const_small_int, arg);
     } else if (MP_PARSE_NODE_IS_LEAF(pn)) {
-        machine_int_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
+        machine_uint_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
         switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
             case MP_PARSE_NODE_ID: EMIT_ARG(load_id, arg); break;
-            case MP_PARSE_NODE_SMALL_INT: EMIT_ARG(load_const_small_int, arg); break;
             case MP_PARSE_NODE_INTEGER: EMIT_ARG(load_const_int, arg); break;
             case MP_PARSE_NODE_DECIMAL: EMIT_ARG(load_const_dec, arg); break;
             case MP_PARSE_NODE_STRING: EMIT_ARG(load_const_str, arg, false); break;

py/emitinlinethumb.c

@@ -110,7 +110,7 @@ STATIC int get_arg_i(qstr op, mp_parse_node_t *pn_args, int wanted_arg_num, int
         printf("SyntaxError: '%s' expects an integer in position %d\n", qstr_str(op), wanted_arg_num);
         return 0;
     }
-    int i = MP_PARSE_NODE_LEAF_ARG(pn_args[wanted_arg_num]);
+    int i = MP_PARSE_NODE_LEAF_SMALL_INT(pn_args[wanted_arg_num]);
     if ((i & (~fit_mask)) != 0) {
         printf("SyntaxError: '%s' integer 0x%x does not fit in mask 0x%x\n", qstr_str(op), i, fit_mask);
         return 0;

py/parse.c

@@ -125,7 +125,10 @@ STATIC void pop_rule(parser_t *parser, const rule_t **rule, uint *arg_i, uint *s
 }
 
 mp_parse_node_t mp_parse_node_new_leaf(machine_int_t kind, machine_int_t arg) {
-    return (mp_parse_node_t)(kind | (arg << 4));
+    if (kind == MP_PARSE_NODE_SMALL_INT) {
+        return (mp_parse_node_t)(kind | (arg << 1));
+    }
+    return (mp_parse_node_t)(kind | (arg << 5));
 }
 
 //int num_parse_nodes_allocated = 0;
@@ -171,11 +174,13 @@ void mp_parse_node_print(mp_parse_node_t pn, int indent) {
     }
     if (MP_PARSE_NODE_IS_NULL(pn)) {
         printf("NULL\n");
+    } else if (MP_PARSE_NODE_IS_SMALL_INT(pn)) {
+        machine_int_t arg = MP_PARSE_NODE_LEAF_SMALL_INT(pn);
+        printf("int(" INT_FMT ")\n", arg);
     } else if (MP_PARSE_NODE_IS_LEAF(pn)) {
-        machine_int_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
+        machine_uint_t arg = MP_PARSE_NODE_LEAF_ARG(pn);
         switch (MP_PARSE_NODE_LEAF_KIND(pn)) {
             case MP_PARSE_NODE_ID: printf("id(%s)\n", qstr_str(arg)); break;
-            case MP_PARSE_NODE_SMALL_INT: printf("int(" INT_FMT ")\n", arg); break;
             case MP_PARSE_NODE_INTEGER: printf("int(%s)\n", qstr_str(arg)); break;
             case MP_PARSE_NODE_DECIMAL: printf("dec(%s)\n", qstr_str(arg)); break;
             case MP_PARSE_NODE_STRING: printf("str(%s)\n", qstr_str(arg)); break;

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...xxx1: a small integer; bits 1 and above are the signed value, 2's complement
-//  - xxxx...0011: a small integer; bits 4 and above are the signed value, 2's complement
+//  - xxxx...xx00: pointer to mp_parse_node_struct_t
-//  - xxxx...0101: an integer; bits 4 and above are the qstr holding the value
+//  - xx...x00010: an identifier; bits 5 and above are the qstr
-//  - xxxx...0111: a decimal; bits 4 and above are the qstr holding the value
+//  - xx...x00110: an integer; bits 5 and above are the qstr holding the value
-//  - xxxx...1001: a string; bits 4 and above are the qstr holding the value
+//  - xx...x01010: a decimal; bits 5 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
+//  - xx...x01110: a string; bits 5 and above are the qstr holding the value
-//  - xxxx...1101: a token; bits 4 and above are mp_token_kind_t
+//  - xx...x10010: a string with triple quotes; bits 5 and above are the qstr holding the value
-//  - xxxx...xxx0: pointer to mp_parse_node_struct_t
+//  - 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 (0x1)
-#define MP_PARSE_NODE_SMALL_INT (0x3)
+#define MP_PARSE_NODE_ID        (0x02)
-#define MP_PARSE_NODE_INTEGER   (0x5)
+#define MP_PARSE_NODE_INTEGER   (0x06)
-#define MP_PARSE_NODE_DECIMAL   (0x7)
+#define MP_PARSE_NODE_DECIMAL   (0x0a)
-#define MP_PARSE_NODE_STRING    (0x9)
+#define MP_PARSE_NODE_STRING    (0x0e)
-#define MP_PARSE_NODE_BYTES     (0xb)
+#define MP_PARSE_NODE_BYTES     (0x12)
-#define MP_PARSE_NODE_TOKEN     (0xd)
+#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_LEAF(pn) ((pn) & 3)
-#define MP_PARSE_NODE_IS_STRUCT(pn) ((pn) != MP_PARSE_NODE_NULL && ((pn) & 1) == 0)
+#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) & 1) == 0 && MP_PARSE_NODE_STRUCT_KIND((mp_parse_node_struct_t*)(pn)) == (k))
+#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) & 0x1) == MP_PARSE_NODE_SMALL_INT)
-#define MP_PARSE_NODE_IS_SMALL_INT(pn) (((pn) & 0xf) == 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) & 0xf) == MP_PARSE_NODE_TOKEN)
+#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 << 4)))
+#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)
 

tests/basics/int-small.py

@@ -1,5 +1,29 @@
 # This tests small int range for 32-bit machine
 
+# Small ints are variable-length encoded in MicroPython, so first
+# test that encoding works as expected.
+
+print(0)
+print(1)
+print(-1)
+# Value is split in 7-bit "subwords", and taking into account that all
+# ints in Python are signed, there're 6 bits of magnitude. So, around 2^6
+# there's "turning point"
+print(63)
+print(64)
+print(65)
+print(-63)
+print(-64)
+print(-65)
+# Maximum values of small ints on 32-bit platform
+print(1073741823)
+# Per python semantics, lexical integer is without a sign (i.e. positive)
+# and '-' is unary minus operation applied to it. That's why -1073741824
+# (min two-complement's negative value) is not allowed.
+print(-1073741823)
+
+# Operations tests
+
 a = 0x3fffff
 print(a)
 a *= 0x10