pythongh-137308: Replace a single docstring with pass in -OO mode (python#137318)

This is required so we would never have empty node bodies.
Refs python#130087

Author: sobolevn

Lib/test/test_ast/test_ast.py

@@ -220,6 +220,131 @@ def test_negative_locations_for_compile(self):
                 # This also must not crash:
                 ast.parse(tree, optimize=2)
 
+    def test_docstring_optimization_single_node(self):
+        # https://github.com/python/cpython/issues/137308
+        class_example1 = textwrap.dedent('''
+            class A:
+                """Docstring"""
+        ''')
+        class_example2 = textwrap.dedent('''
+            class A:
+                """
+                Docstring"""
+        ''')
+        def_example1 = textwrap.dedent('''
+            def some():
+                """Docstring"""
+        ''')
+        def_example2 = textwrap.dedent('''
+            def some():
+                """Docstring
+                                       """
+        ''')
+        async_def_example1 = textwrap.dedent('''
+            async def some():
+                """Docstring"""
+        ''')
+        async_def_example2 = textwrap.dedent('''
+            async def some():
+                """
+                Docstring
+            """
+        ''')
+        for code in [
+            class_example1,
+            class_example2,
+            def_example1,
+            def_example2,
+            async_def_example1,
+            async_def_example2,
+        ]:
+            for opt_level in [0, 1, 2]:
+                with self.subTest(code=code, opt_level=opt_level):
+                    mod = ast.parse(code, optimize=opt_level)
+                    self.assertEqual(len(mod.body[0].body), 1)
+                    if opt_level == 2:
+                        pass_stmt = mod.body[0].body[0]
+                        self.assertIsInstance(pass_stmt, ast.Pass)
+                        self.assertEqual(
+                            vars(pass_stmt),
+                            {
+                                'lineno': 3,
+                                'col_offset': 4,
+                                'end_lineno': 3,
+                                'end_col_offset': 8,
+                            },
+                        )
+                    else:
+                        self.assertIsInstance(mod.body[0].body[0], ast.Expr)
+                        self.assertIsInstance(
+                            mod.body[0].body[0].value,
+                            ast.Constant,
+                        )
+
+                    compile(code, "a", "exec")
+                    compile(code, "a", "exec", optimize=opt_level)
+                    compile(mod, "a", "exec")
+                    compile(mod, "a", "exec", optimize=opt_level)
+
+    def test_docstring_optimization_multiple_nodes(self):
+        # https://github.com/python/cpython/issues/137308
+        class_example = textwrap.dedent(
+            """
+            class A:
+                '''
+                Docstring
+                '''
+                x = 1
+            """
+        )
+
+        def_example = textwrap.dedent(
+            """
+            def some():
+                '''
+                Docstring
+
+            '''
+                x = 1
+            """
+        )
+
+        async_def_example = textwrap.dedent(
+            """
+            async def some():
+
+                '''Docstring
+
+            '''
+                x = 1
+            """
+        )
+
+        for code in [
+            class_example,
+            def_example,
+            async_def_example,
+        ]:
+            for opt_level in [0, 1, 2]:
+                with self.subTest(code=code, opt_level=opt_level):
+                    mod = ast.parse(code, optimize=opt_level)
+                    if opt_level == 2:
+                        self.assertNotIsInstance(
+                            mod.body[0].body[0],
+                            (ast.Pass, ast.Expr),
+                        )
+                    else:
+                        self.assertIsInstance(mod.body[0].body[0], ast.Expr)
+                        self.assertIsInstance(
+                            mod.body[0].body[0].value,
+                            ast.Constant,
+                        )
+
+                    compile(code, "a", "exec")
+                    compile(code, "a", "exec", optimize=opt_level)
+                    compile(mod, "a", "exec")
+                    compile(mod, "a", "exec", optimize=opt_level)
+
     def test_slice(self):
         slc = ast.parse("x[::]").body[0].value.slice
         self.assertIsNone(slc.upper)

Misc/NEWS.d/next/Core_and_Builtins/2025-08-02-10-27-53.gh-issue-137308.at05p_.rst

@@ -0,0 +1,3 @@
+A standalone docstring in a node body is optimized as a :keyword:`pass`
+statement to ensure that the node's body is never empty. There was a
+:exc:`ValueError` in :func:`compile` otherwise.

Python/ast_preprocess.c

@@ -435,13 +435,38 @@ stmt_seq_remove_item(asdl_stmt_seq *stmts, Py_ssize_t idx)
     return 1;
 }
 
+static int
+remove_docstring(asdl_stmt_seq *stmts, Py_ssize_t idx, PyArena *ctx_)
+{
+    assert(_PyAST_GetDocString(stmts) != NULL);
+    // In case there's just the docstring in the body, replace it with `pass`
+    // keyword, so body won't be empty.
+    if (asdl_seq_LEN(stmts) == 1) {
+        stmt_ty docstring = (stmt_ty)asdl_seq_GET(stmts, 0);
+        stmt_ty pass = _PyAST_Pass(
+            docstring->lineno, docstring->col_offset,
+            // we know that `pass` always takes 4 chars and a single line,
+            // while docstring can span on multiple lines
+            docstring->lineno, docstring->col_offset + 4,
+            ctx_
+        );
+        if (pass == NULL) {
+            return 0;
+        }
+        asdl_seq_SET(stmts, 0, pass);
+        return 1;
+    }
+    // In case there are more than 1 body items, just remove the docstring.
+    return stmt_seq_remove_item(stmts, idx);
+}
+
 static int
 astfold_body(asdl_stmt_seq *stmts, PyArena *ctx_, _PyASTPreprocessState *state)
 {
     int docstring = _PyAST_GetDocString(stmts) != NULL;
     if (docstring && (state->optimize >= 2)) {
         /* remove the docstring */
-        if (!stmt_seq_remove_item(stmts, 0)) {
+        if (!remove_docstring(stmts, 0, ctx_)) {
             return 0;
         }
         docstring = 0;