bpo-41690: Use a loop to collect args in the parser instead of recursion (GH-22053)

This program can segfault the parser by stack overflow:

```
import ast

code = "f(" + ",".join(['a' for _ in range(100000)]) + ")"
print("Ready!")
ast.parse(code)
```

the reason is that the rule for arguments has a simple recursion when collecting args:

args[expr_ty]:
    [...]
    | a=named_expression b=[',' c=args { c }] {
        [...] }
This commit is contained in:
Pablo Galindo 2020-09-02 17:44:19 +01:00 committed by GitHub
parent 3940333637
commit 4a97b1517a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 916 additions and 803 deletions

View file

@ -535,22 +535,11 @@ arguments[expr_ty] (memo):
| a=args [','] &')' { a }
| incorrect_arguments
args[expr_ty]:
| a=starred_expression b=[',' c=args { c }] {
_Py_Call(_PyPegen_dummy_name(p),
(b) ? CHECK(_PyPegen_seq_insert_in_front(p, a, ((expr_ty) b)->v.Call.args))
: CHECK(_PyPegen_singleton_seq(p, a)),
(b) ? ((expr_ty) b)->v.Call.keywords : NULL,
EXTRA) }
| a=','.(starred_expression | named_expression !'=')+ b=[',' k=kwargs {k}] { _PyPegen_collect_call_seqs(p, a, b) }
| a=kwargs { _Py_Call(_PyPegen_dummy_name(p),
CHECK_NULL_ALLOWED(_PyPegen_seq_extract_starred_exprs(p, a)),
CHECK_NULL_ALLOWED(_PyPegen_seq_delete_starred_exprs(p, a)),
EXTRA) }
| a=named_expression b=[',' c=args { c }] {
_Py_Call(_PyPegen_dummy_name(p),
(b) ? CHECK(_PyPegen_seq_insert_in_front(p, a, ((expr_ty) b)->v.Call.args))
: CHECK(_PyPegen_singleton_seq(p, a)),
(b) ? ((expr_ty) b)->v.Call.keywords : NULL,
EXTRA) }
kwargs[asdl_seq*]:
| a=','.kwarg_or_starred+ ',' b=','.kwarg_or_double_starred+ { _PyPegen_join_sequences(p, a, b) }
| ','.kwarg_or_starred+