fix(debug): ast_dump failed on a few things

Author: nedbat

coverage/parser.py

@@ -1306,7 +1306,7 @@ def _is_simple_value(value):
     """Is `value` simple enough to be displayed on a single line?"""
     return (
         value in [None, [], (), {}, set()] or
-        isinstance(value, (str, int, float))
+        isinstance(value, (bytes, int, float, str))
     )
 
 def ast_dump(node, depth=0, print=print):   # pylint: disable=redefined-builtin
@@ -1352,7 +1352,10 @@ def ast_dump(node, depth=0, print=print):   # pylint: disable=redefined-builtin
             elif isinstance(value, list):
                 print(f"{prefix} [")
                 for n in value:
-                    ast_dump(n, depth + 8, print=print)
+                    if _is_simple_value(n):
+                        print(f"{next_indent}    {n!r}")
+                    else:
+                        ast_dump(n, depth + 8, print=print)
                 print(f"{next_indent}]")
             else:
                 print(prefix)

tests/stress_phystoken.tok

@@ -4,6 +4,7 @@
 # Here's some random Python so that test_tokenize_myself will have some
 # stressful stuff to try.  This file is .tok instead of .py so pylint won't
 # complain about it, check_eol won't look at it, etc.
+# Some lines are here to reproduce fixed bugs in ast_dump also.
 
 first_back = """\
 hey there!
@@ -50,6 +51,15 @@ a_long_string =              \
     "3 is longer"
 
 def hello():
+        global x    # ast_dump bug
         print("Hello world!")
 
 hello()
+
+# ast dump bugs:
+weird = {
+    **d,
+    **{'c': 7},
+    'd': 8,
+}
+self.hash.update(b'.')

tests/test_parser.py

@@ -4,7 +4,6 @@
 """Tests for coverage.py's code parsing."""
 
 import os.path
-import re
 import textwrap
 
 import pytest
@@ -14,7 +13,7 @@
 from coverage.parser import ast_dump, ast_parse, PythonParser
 
 from tests.coveragetest import CoverageTest, TESTS_DIR
-from tests.helpers import arcz_to_arcs
+from tests.helpers import arcz_to_arcs, re_lines
 
 
 class PythonParserTest(CoverageTest):
@@ -485,17 +484,23 @@ def test_ast_dump():
     # Run the AST_DUMP code to make sure it doesn't fail, with some light
     # assertions. Use parser.py as the test code since it is the longest file,
     # and fitting, since it's the AST_DUMP code.
-    parser_py = os.path.join(TESTS_DIR, "../coverage/parser.py")
-    with open(parser_py) as f:
-        ast_root = ast_parse(f.read())
-    result = []
-    ast_dump(ast_root, print=result.append)
-    assert len(result) > 10000
-    assert result[0] == "<Module"
-    assert result[-1] == ">"
-
-    def count(pat):
-        return sum(1 for line in result if re.search(pat, line))
-
-    assert count(r"^\s+>") > 2000
-    assert count(r"<Name @ \d+,\d+(:\d+)? id: '\w+'>") > 1000
+    files = [
+        os.path.join(TESTS_DIR, "../coverage/parser.py"),
+        os.path.join(TESTS_DIR, "stress_phystoken.tok"),
+    ]
+    for fname in files:
+        with open(fname) as f:
+            source = f.read()
+            num_lines = len(source.splitlines())
+            print(f"file {fname} has {num_lines} lines")
+            ast_root = ast_parse(source)
+        result = []
+        ast_dump(ast_root, print=result.append)
+        if num_lines < 100:
+            continue
+        assert len(result) > 5 * num_lines
+        assert result[0] == "<Module"
+        assert result[-1] == ">"
+        result_text = "\n".join(result)
+        assert len(re_lines(result_text, r"^\s+>")) > num_lines
+        assert len(re_lines(result_text, r"<Name @ \d+,\d+(:\d+)? id: '\w+'>")) > num_lines // 2

tests/test_phystokens.py

@@ -15,6 +15,7 @@
 from coverage.python import get_python_source
 
 from tests.coveragetest import CoverageTest, TESTS_DIR
+from tests.helpers import re_lines
 
 
 # A simple program and its token stream.
@@ -97,10 +98,15 @@ def test_tokenize_real_file(self):
 
     def test_stress(self):
         # Check the tokenization of a stress-test file.
+        # And check that those files haven't been incorrectly "fixed".
         stress = os.path.join(TESTS_DIR, "stress_phystoken.tok")
         self.check_file_tokenization(stress)
+        with open(stress) as fstress:
+            assert re_lines(fstress.read(), r"\s$"), f"{stress} needs a trailing space."
         stress = os.path.join(TESTS_DIR, "stress_phystoken_dos.tok")
         self.check_file_tokenization(stress)
+        with open(stress) as fstress:
+            assert re_lines(fstress.read(), r"\s$"), f"{stress} needs a trailing space."
 
 
 @pytest.mark.skipif(not env.PYBEHAVIOR.soft_keywords, reason="Soft keywords are new in Python 3.10")