bpo-36440: include node names in ParserError messages, instead of numeric IDs (GH-12565)
The error messages in the parser module are referring to numeric IDs for the nodes. To improve readability, use the node names when reporting errors.
diff --git a/Lib/test/test_parser.py b/Lib/test/test_parser.py
index bfa0a5a..ff587c3 100644
--- a/Lib/test/test_parser.py
+++ b/Lib/test/test_parser.py
@@ -749,6 +749,22 @@
with self.assertRaises(UnicodeEncodeError):
parser.sequence2st(tree)
+ def test_invalid_node_id(self):
+ tree = (257, (269, (-7, '')))
+ self.check_bad_tree(tree, "negative node id")
+ tree = (257, (269, (99, '')))
+ self.check_bad_tree(tree, "invalid token id")
+ tree = (257, (269, (9999, (0, ''))))
+ self.check_bad_tree(tree, "invalid symbol id")
+
+ def test_ParserError_message(self):
+ try:
+ parser.sequence2st((257,(269,(257,(0,'')))))
+ except parser.ParserError as why:
+ self.assertIn("compound_stmt", str(why)) # Expected
+ self.assertIn("file_input", str(why)) # Got
+
+
class CompileTestCase(unittest.TestCase):
diff --git a/Misc/NEWS.d/next/Core and Builtins/2019-03-25-13-45-19.bpo-36440.gkvzhi.rst b/Misc/NEWS.d/next/Core and Builtins/2019-03-25-13-45-19.bpo-36440.gkvzhi.rst
new file mode 100644
index 0000000..372b1f7
--- /dev/null
+++ b/Misc/NEWS.d/next/Core and Builtins/2019-03-25-13-45-19.bpo-36440.gkvzhi.rst
@@ -0,0 +1,2 @@
+Include node names in ``ParserError`` messages, instead of numeric IDs.
+Patch by A. Skrobov.
diff --git a/Modules/parsermodule.c b/Modules/parsermodule.c
index fd330b5..a215c7e 100644
--- a/Modules/parsermodule.c
+++ b/Modules/parsermodule.c
@@ -24,10 +24,6 @@
* Py_[X]DECREF() and Py_[X]INCREF() macros. The lint annotations
* look like "NOTE(...)".
*
- * To debug parser errors like
- * "parser.ParserError: Expected node type 12, got 333."
- * decode symbol numbers using the automatically-generated files
- * Lib/symbol.h and Include/token.h.
*/
#include "Python.h" /* general Python API */
@@ -666,6 +662,13 @@
for (pos = 0; pos < nch; ++pos) {
node *ch = CHILD(tree, pos);
int ch_type = TYPE(ch);
+ if ((ch_type >= NT_OFFSET + _PyParser_Grammar.g_ndfas)
+ || (ISTERMINAL(ch_type) && (ch_type >= N_TOKENS))
+ || (ch_type < 0)
+ ) {
+ PyErr_Format(parser_error, "Unrecognized node type %d.", ch_type);
+ return 0;
+ }
if (ch_type == suite && TYPE(tree) == funcdef) {
/* This is the opposite hack of what we do in parser.c
(search for func_body_suite), except we don't ever
@@ -700,8 +703,10 @@
const char *expected_str = _PyParser_Grammar.g_ll.ll_label[a_label].lb_str;
if (ISNONTERMINAL(next_type)) {
- PyErr_Format(parser_error, "Expected node type %d, got %d.",
- next_type, ch_type);
+ PyErr_Format(parser_error, "Expected %s, got %s.",
+ _PyParser_Grammar.g_dfa[next_type - NT_OFFSET].d_name,
+ ISTERMINAL(ch_type) ? _PyParser_TokenNames[ch_type] :
+ _PyParser_Grammar.g_dfa[ch_type - NT_OFFSET].d_name);
}
else if (expected_str != NULL) {
PyErr_Format(parser_error, "Illegal terminal: expected '%s'.",