bpo-40585: Normalize errors messages in codeop when comparing them (GH-20030)

With the new parser, the error message contains always the trailing
newlines, causing the comparison of the repr of the error messages
in codeop to fail. This commit makes the new parser mirror the old parser's
behaviour regarding trailing newlines.

Lib/test/test_codeop.py

@@ -288,6 +288,15 @@ def test_invalid(self):
 
         ai("[i for i in range(10)] = (1, 2, 3)")
 
+    def test_invalid_exec(self):
+        ai = self.assertInvalid
+        ai("raise = 4", symbol="exec")
+        ai('def a-b', symbol='exec')
+        ai('await?', symbol='exec')
+        ai('=!=', symbol='exec')
+        ai('a await raise b', symbol='exec')
+        ai('a await raise b?+1', symbol='exec')
+
     def test_filename(self):
         self.assertEqual(compile_command("a = 1\n", "abc").co_filename,
                          compile("a = 1\n", "abc", 'single').co_filename)

Misc/NEWS.d/next/Core and Builtins/2020-05-11-00-19-42.bpo-40585.yusknY.rst

@@ -0,0 +1,2 @@
+Fixed a bug when using :func:`codeop.compile_command` that was causing
+exceptions to be swallowed with the new parser. Patch by Pablo Galindo

Parser/pegen/pegen.c

@@ -310,6 +310,12 @@ get_error_line(char *buffer, int is_file)
         newline = strchr(buffer, '\n');
     }
 
+    if (is_file) {
+        while (newline > buffer && newline[-1] == '\n') {
+            --newline;
+        }
+    }
+
     if (newline) {
         return PyUnicode_DecodeUTF8(buffer, newline - buffer, "replace");
     }