gh-105481: expose opcode metadata via the _opcode module

python · iritkatriel · Jul 14, 2023 · Jul 12, 2023 · Jul 12, 2023 · Jul 13, 2023
commit d65881b968f866beb625ee4802a283e3a910cb54
diff --git a/Include/internal/pycore_opcode_metadata.h b/Include/internal/pycore_opcode_metadata.h
diff --git a/Lib/test/test__opcode.py b/Lib/test/test__opcode.py
@@ -9,6 +9,70 @@
 
 class OpcodeTests(unittest.TestCase):
 
+    def check_bool_function_result(self, func, ops, expected):
+        for op in ops:
+            if isinstance(op, str):
+                op = dis.opmap[op]
+            with self.subTest(opcode=op, func=func):
+                self.assertIsInstance(func(op), bool)
+                self.assertEqual(func(op), expected)
+
+    def test_invalid_opcodes(self):
+        invalid = [-100, -1, 255, 512, 513, 1000]
+        self.check_bool_function_result(_opcode.is_valid, invalid, False)
+        self.check_bool_function_result(_opcode.has_arg, invalid, False)
+        self.check_bool_function_result(_opcode.has_const, invalid, False)
+        self.check_bool_function_result(_opcode.has_name, invalid, False)
+        self.check_bool_function_result(_opcode.has_jump, invalid, False)
+
+    def test_is_valid(self):
+        names = [
+            'CACHE',
+            'POP_TOP',
+            'IMPORT_NAME',
+            'JUMP',
+            'INSTRUMENTED_RETURN_VALUE',
+        ]
+        opcodes = [dis.opmap[opname] for opname in names]
+        self.check_bool_function_result(_opcode.is_valid, opcodes, True)
+
+    def test_has_arg(self):
+        has_arg = ['SWAP', 'LOAD_FAST', 'INSTRUMENTED_POP_JUMP_IF_TRUE', 'JUMP']
+        no_arg = ['SETUP_WITH', 'POP_TOP', 'NOP', 'CACHE']
+        self.check_bool_function_result(_opcode.has_arg, has_arg, True)
+        self.check_bool_function_result(_opcode.has_arg, no_arg, False)
+
+    def test_has_const(self):
+        has_const = ['LOAD_CONST', 'RETURN_CONST', 'KW_NAMES']
+        no_const = ['SETUP_WITH', 'POP_TOP', 'NOP', 'CACHE']
+        self.check_bool_function_result(_opcode.has_const, has_const, True)
+        self.check_bool_function_result(_opcode.has_const, no_const, False)
+
+    def test_has_name(self):
+        has_name = ['STORE_NAME', 'DELETE_ATTR', 'STORE_GLOBAL', 'IMPORT_FROM',
+                    'LOAD_FROM_DICT_OR_GLOBALS']
+        no_name = ['SETUP_WITH', 'POP_TOP', 'NOP', 'CACHE']
+        self.check_bool_function_result(_opcode.has_name, has_name, True)
+        self.check_bool_function_result(_opcode.has_name, no_name, False)
+
+    def test_has_jump(self):
+        has_jump = ['FOR_ITER', 'JUMP_FORWARD', 'JUMP', 'POP_JUMP_IF_TRUE', 'SEND']
+        no_jump = ['SETUP_WITH', 'POP_TOP', 'NOP', 'CACHE']
+        self.check_bool_function_result(_opcode.has_jump, has_jump, True)
+        self.check_bool_function_result(_opcode.has_jump, no_jump, False)
+
+    # the following test is part of the refactor, it will be removed soon
+    def test_against_legacy_bool_values(self):
+        # limiting to ops up to ENTER_EXECUTOR, because everything after that
+        # is not currently categorized correctly in opcode.py.
+        for op in range(0, opcode.opmap['ENTER_EXECUTOR']):
+            with self.subTest(op=op):
+                if opcode.opname[op] != f'<{op}>':
+                    self.assertEqual(op in dis.hasarg, _opcode.has_arg(op))
+                    self.assertEqual(op in dis.hasconst, _opcode.has_const(op))
+                    self.assertEqual(op in dis.hasname, _opcode.has_name(op))
+                    self.assertEqual(op in dis.hasjrel, _opcode.has_jump(op))
+
     def test_stack_effect(self):
         self.assertEqual(stack_effect(dis.opmap['POP_TOP']), -1)
         self.assertEqual(stack_effect(dis.opmap['BUILD_SLICE'], 0), -1)

diff --git a/Modules/_opcode.c b/Modules/_opcode.c
@@ -1,6 +1,9 @@
 #include "Python.h"
 #include "opcode.h"
 #include "internal/pycore_code.h"
+#define NEED_OPCODE_METADATA
+#include "internal/pycore_opcode_metadata.h"
+#undef NEED_OPCODE_METADATA
 
 /*[clinic input]
 module _opcode
@@ -59,6 +62,87 @@ _opcode_stack_effect_impl(PyObject *module, int opcode, PyObject *oparg,
     return effect;
 }
 
+/*[clinic input]
+
+_opcode.is_valid -> bool
+
+  opcode: int
+
+Return True if opcode is valid, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_is_valid_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=b0d918ea1d073f65 input=fe23e0aa194ddae0]*/
+{
+    return IS_VALID_OPCODE(opcode);
+}
+
+/*[clinic input]
+
+_opcode.has_arg -> bool
+
+  opcode: int
+
+Return True if the opcode uses its oparg, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_has_arg_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=7a062d3b2dcc0815 input=93d878ba6361db5f]*/
+{
+    return IS_VALID_OPCODE(opcode) && OPCODE_HAS_ARG(opcode);
+}
+
+/*[clinic input]
+
+_opcode.has_const -> bool
+
+  opcode: int
+
+Return True if the opcode accesses a constant, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_has_const_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=c646d5027c634120 input=a6999e4cf13f9410]*/
+{
+    return IS_VALID_OPCODE(opcode) && OPCODE_HAS_CONST(opcode);
+}
+
+/*[clinic input]
+
+_opcode.has_name -> bool
+
+  opcode: int
+
+Return True if the opcode accesses an attribute by name, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_has_name_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=b49a83555c2fa517 input=448aa5e4bcc947ba]*/
+{
+    return IS_VALID_OPCODE(opcode) && OPCODE_HAS_NAME(opcode);
+}
+
+/*[clinic input]
+
+_opcode.has_jump -> bool
+
+  opcode: int
+
+Return True if the opcode has a jump target, False otherwise.
+[clinic start generated code]*/
+
+static int
+_opcode_has_jump_impl(PyObject *module, int opcode)
+/*[clinic end generated code: output=e9c583c669f1c46a input=35f711274357a0c3]*/
+{
+    return IS_VALID_OPCODE(opcode) && OPCODE_HAS_JUMP(opcode);
+}
+
+
 /*[clinic input]
 
 _opcode.get_specialization_stats
@@ -80,6 +164,11 @@ _opcode_get_specialization_stats_impl(PyObject *module)
 static PyMethodDef
 opcode_functions[] =  {
     _OPCODE_STACK_EFFECT_METHODDEF
+    _OPCODE_IS_VALID_METHODDEF
+    _OPCODE_HAS_ARG_METHODDEF
+    _OPCODE_HAS_CONST_METHODDEF
+    _OPCODE_HAS_NAME_METHODDEF
+    _OPCODE_HAS_JUMP_METHODDEF
     _OPCODE_GET_SPECIALIZATION_STATS_METHODDEF
     {NULL, NULL, 0, NULL}
 };