optimize for sets

Author: kumaraditya303

Include/internal/pycore_opcode_utils.h

@@ -74,7 +74,8 @@ extern "C" {
 #define CONSTANT_BUILTIN_ALL 3
 #define CONSTANT_BUILTIN_ANY 4
 #define CONSTANT_BUILTIN_LIST 5
-#define NUM_COMMON_CONSTANTS 6
+#define CONSTANT_BUILTIN_SET 6
+#define NUM_COMMON_CONSTANTS 7
 
 /* Values used in the oparg for RESUME */
 #define RESUME_AT_FUNC_START 0

Lib/opcode.py

@@ -40,7 +40,8 @@
 _intrinsic_2_descs = _opcode.get_intrinsic2_descs()
 _special_method_names = _opcode.get_special_method_names()
 _common_constants = [builtins.AssertionError, builtins.NotImplementedError,
-                     builtins.tuple, builtins.all, builtins.any, builtins.list]
+                     builtins.tuple, builtins.all, builtins.any, builtins.list,
+                     builtins.set]
 _nb_ops = _opcode.get_nb_ops()
 
 hascompare = [opmap["COMPARE_OP"]]

Lib/test/test_builtin.py

@@ -246,7 +246,7 @@ def test_any(self):
         S = [10, 20, 30]
         self.assertEqual(any(x > 42 for x in S), False)
 
-    def test_all_any_tuple_list_optimization(self):
+    def test_all_any_tuple_list_set_optimization(self):
         def f_all():
             return all(x-2 for x in [1,2,3])
 
@@ -259,7 +259,10 @@ def f_tuple():
         def f_list():
             return list(2*x for x in [1,2,3])
 
-        funcs = [f_all, f_any, f_tuple, f_list]
+        def f_set():
+            return set(2*x for x in [1,2,3])
+
+        funcs = [f_all, f_any, f_tuple, f_list, f_set]
 
         for f in funcs:
             # check that generator code object is not duplicated
@@ -269,33 +272,35 @@ def f_list():
 
         # check the overriding the builtins works
 
-        global all, any, tuple, list
-        saved = all, any, tuple, list
+        global all, any, tuple, list, set
+        saved = all, any, tuple, list, set
         try:
             all = lambda x : "all"
             any = lambda x : "any"
             tuple = lambda x : "tuple"
             list = lambda x : "list"
+            set = lambda x : "set"
 
             overridden_outputs = [f() for f in funcs]
         finally:
-            all, any, tuple, list = saved
+            all, any, tuple, list, set = saved
 
-        self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list'])
+        self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list', 'set'])
         # Now repeat, overriding the builtins module as well
-        saved = all, any, tuple, list
+        saved = all, any, tuple, list, set
         try:
             builtins.all = all = lambda x : "all"
             builtins.any = any = lambda x : "any"
             builtins.tuple = tuple = lambda x : "tuple"
             builtins.list = list = lambda x : "list"
+            builtins.set = set = lambda x : "set"
 
             overridden_outputs = [f() for f in funcs]
         finally:
-            all, any, tuple, list = saved
-            builtins.all, builtins.any, builtins.tuple, builtins.list = saved
+            all, any, tuple, list, set = saved
+            builtins.all, builtins.any, builtins.tuple, builtins.list, builtins.set = saved
 
-        self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list'])
+        self.assertEqual(overridden_outputs, ['all', 'any', 'tuple', 'list', 'set'])
 
     def test_ascii(self):
         self.assertEqual(ascii(''), '\'\'')

Python/codegen.c

@@ -3895,6 +3895,9 @@ maybe_optimize_function_call(compiler *c, expr_ty e, jump_target_label end)
     else if (_PyUnicode_EqualToASCIIString(func->v.Name.id, "list")) {
         const_oparg = CONSTANT_BUILTIN_LIST;
     }
+    else if (_PyUnicode_EqualToASCIIString(func->v.Name.id, "set")) {
+        const_oparg = CONSTANT_BUILTIN_SET;
+    }
     if (const_oparg != -1) {
         ADDOP_I(c, loc, COPY, 1); // the function
         ADDOP_I(c, loc, LOAD_COMMON_CONSTANT, const_oparg);
@@ -3904,6 +3907,8 @@ maybe_optimize_function_call(compiler *c, expr_ty e, jump_target_label end)
 
         if (const_oparg == CONSTANT_BUILTIN_TUPLE || const_oparg == CONSTANT_BUILTIN_LIST) {
             ADDOP_I(c, loc, BUILD_LIST, 0);
+        } else if (const_oparg == CONSTANT_BUILTIN_SET) {
+            ADDOP_I(c, loc, BUILD_SET, 0);
         }
         expr_ty generator_exp = asdl_seq_GET(args, 0);
         VISIT(c, expr, generator_exp);
@@ -3917,14 +3922,19 @@ maybe_optimize_function_call(compiler *c, expr_ty e, jump_target_label end)
         if (const_oparg == CONSTANT_BUILTIN_TUPLE || const_oparg == CONSTANT_BUILTIN_LIST) {
             ADDOP_I(c, loc, LIST_APPEND, 3);
             ADDOP_JUMP(c, loc, JUMP, loop);
+        } else if (const_oparg == CONSTANT_BUILTIN_SET) {
+            ADDOP_I(c, loc, SET_ADD, 3);
+            ADDOP_JUMP(c, loc, JUMP, loop);
         }
         else {
             ADDOP(c, loc, TO_BOOL);
             ADDOP_JUMP(c, loc, continue_jump_opcode, loop);
         }
 
         ADDOP(c, NO_LOCATION, POP_ITER);
-        if (const_oparg != CONSTANT_BUILTIN_TUPLE && const_oparg != CONSTANT_BUILTIN_LIST) {
+        if (const_oparg != CONSTANT_BUILTIN_TUPLE &&
+            const_oparg != CONSTANT_BUILTIN_LIST &&
+            const_oparg != CONSTANT_BUILTIN_SET) {
             ADDOP_LOAD_CONST(c, loc, initial_res == Py_True ? Py_False : Py_True);
         }
         ADDOP_JUMP(c, loc, JUMP, end);
@@ -3936,6 +3946,8 @@ maybe_optimize_function_call(compiler *c, expr_ty e, jump_target_label end)
             ADDOP_I(c, loc, CALL_INTRINSIC_1, INTRINSIC_LIST_TO_TUPLE);
         } else if (const_oparg == CONSTANT_BUILTIN_LIST) {
             // result is already a list
+        } else if (const_oparg == CONSTANT_BUILTIN_SET) {
+            // result is already a set
         }
         else {
             ADDOP_LOAD_CONST(c, loc, initial_res);

Python/pylifecycle.c

@@ -830,6 +830,7 @@ pycore_init_builtins(PyThreadState *tstate)
     interp->common_consts[CONSTANT_BUILTIN_ALL] = all;
     interp->common_consts[CONSTANT_BUILTIN_ANY] = any;
     interp->common_consts[CONSTANT_BUILTIN_LIST] = (PyObject*)&PyList_Type;
+    interp->common_consts[CONSTANT_BUILTIN_SET] = (PyObject*)&PySet_Type;
 
     for (int i=0; i < NUM_COMMON_CONSTANTS; i++) {
         assert(interp->common_consts[i] != NULL);