Merge pull request #33 from xybaby/master

Add parameter 'filter' to some functions and a better representation for lambdas

Author: mgedmin

CHANGES.rst

@@ -6,7 +6,10 @@ Changes
 3.1.3 (unreleased)
 ------------------
 
-- Nothing changed yet.
+- New ``filter`` argument for :func:`typestats`, :func:`most_common_types`,
+  :func:`show_most_common_types`, :func:`show_growth`.
+
+- Show lambda function more human-readble with change to :func:`_short_repr`
 
 
 3.1.2 (2017-11-27)

objgraph.py

@@ -135,14 +135,18 @@ def count(typename, objects=None):
         del objects  # clear cyclic references to frame
 
 
-def typestats(objects=None, shortnames=True):
+def typestats(objects=None, shortnames=True, filter=None):
     """Count the number of instances for each type tracked by the GC.
 
     Note that the GC does not track simple objects like int or str.
 
     Note that classes with the same name but defined in different modules
     will be lumped together if ``shortnames`` is True.
 
+    If ``filter` is specified, it should be a function taking one argument and
+    returning a boolean. Objects for which ``filter(obj)`` returns ``False``
+    will be ignored.
+
     Example:
 
         >>> typestats()
@@ -158,6 +162,9 @@ def typestats(objects=None, shortnames=True):
     .. versionchanged:: 1.8
        New parameter: ``shortnames``.
 
+    .. versionchanged:: 3.1.3
+       New parameter: ``filter``.
+
     """
     if objects is None:
         objects = gc.get_objects()
@@ -168,21 +175,27 @@ def typestats(objects=None, shortnames=True):
             typename = _long_typename
         stats = {}
         for o in objects:
+            if filter and not filter(o):
+                continue
             n = typename(o)
             stats[n] = stats.get(n, 0) + 1
         return stats
     finally:
         del objects  # clear cyclic references to frame
 
 
-def most_common_types(limit=10, objects=None, shortnames=True):
+def most_common_types(limit=10, objects=None, shortnames=True, filter=None):
     """Count the names of types with the most instances.
 
     Returns a list of (type_name, count), sorted most-frequent-first.
 
     Limits the return value to at most ``limit`` items.  You may set ``limit``
     to None to avoid that.
 
+    If ``filter` is specified, it should be a function taking one argument and
+    returning a boolean. Objects for which ``filter(obj)`` returns ``False``
+    will be ignored.
+
     The caveats documented in :func:`typestats` apply.
 
     Example:
@@ -198,9 +211,13 @@ def most_common_types(limit=10, objects=None, shortnames=True):
     .. versionchanged:: 1.8
        New parameter: ``shortnames``.
 
+    .. versionchanged:: 3.1.3
+       New parameter: ``filter``.
+
     """
-    stats = sorted(typestats(objects, shortnames=shortnames).items(),
-                   key=operator.itemgetter(1), reverse=True)
+    stats = sorted(
+        typestats(objects, shortnames=shortnames, filter=filter).items(),
+        key=operator.itemgetter(1), reverse=True)
     if limit:
         stats = stats[:limit]
     return stats
@@ -210,9 +227,14 @@ def show_most_common_types(
         limit=10,
         objects=None,
         shortnames=True,
-        file=None):
+        file=None,
+        filter=None):
     """Print the table of types of most common instances.
 
+    If ``filter` is specified, it should be a function taking one argument and
+    returning a boolean. Objects for which ``filter(obj)`` returns ``False``
+    will be ignored.
+
     The caveats documented in :func:`typestats` apply.
 
     Example:
@@ -235,16 +257,21 @@ def show_most_common_types(
     .. versionchanged:: 3.0
        New parameter: ``file``.
 
+    .. versionchanged:: 3.1.3
+       New parameter: ``filter``.
+
     """
     if file is None:
         file = sys.stdout
-    stats = most_common_types(limit, objects, shortnames=shortnames)
+    stats = most_common_types(limit, objects, shortnames=shortnames,
+                              filter=filter)
     width = max(len(name) for name, count in stats)
     for name, count in stats:
         file.write('%-*s %i\n' % (width, name, count))
 
 
-def show_growth(limit=10, peak_stats={}, shortnames=True, file=None):
+def show_growth(limit=10, peak_stats={}, shortnames=True, file=None,
+                filter=None):
     """Show the increase in peak object counts since last call.
 
     Limits the output to ``limit`` largest deltas.  You may set ``limit`` to
@@ -254,6 +281,10 @@ def show_growth(limit=10, peak_stats={}, shortnames=True, file=None):
     seen peak object counts.  Usually you don't need to pay attention to this
     argument.
 
+    If ``filter` is specified, it should be a function taking one argument and
+    returning a boolean. Objects for which ``filter(obj)`` returns ``False``
+    will be ignored.
+
     The caveats documented in :func:`typestats` apply.
 
     Example:
@@ -272,9 +303,12 @@ def show_growth(limit=10, peak_stats={}, shortnames=True, file=None):
     .. versionchanged:: 2.1
        New parameter: ``file``.
 
+    .. versionchanged:: 3.1.3
+       New parameter: ``filter``.
+
     """
     gc.collect()
-    stats = typestats(shortnames=shortnames)
+    stats = typestats(shortnames=shortnames, filter=filter)
     deltas = {}
     for name, count in iteritems(stats):
         old_count = peak_stats.get(name, 0)
@@ -920,6 +954,9 @@ def _short_repr(obj):
             return name + ' (bound)'
         else:
             return name
+    if _isinstance(obj, types.LambdaType):
+        return 'lambda: %s:%s' % (os.path.basename(obj.__code__.co_filename),
+                                  obj.__code__.co_firstlineno)
     if _isinstance(obj, types.FrameType):
         return '%s:%s' % (obj.f_code.co_filename, obj.f_lineno)
     if _isinstance(obj, (tuple, list, dict, set)):

tests.py

@@ -281,6 +281,29 @@ def test_no_new_reference_cycles(self):
         self.assertEqual(len(gc.get_referrers(x)), 1)
 
 
+class TypestatsFilterArguTest(GarbageCollectedMixin, unittest.TestCase):
+    """Tests for the typestats function, especially for augument
+    ``filter`` which is added at version 3.1.3"""
+
+    def test_without_filter(self):
+        MyClass = type('MyClass', (), {'__module__': 'mymodule'})  # noqa
+        x, y = MyClass(), MyClass()
+        x.magic_attr = True
+        y.magic_attr = False
+        stats = objgraph.typestats(shortnames=False)
+        self.assertEqual(2, stats['mymodule.MyClass'])
+
+    def test_with_filter(self):
+        MyClass = type('MyClass', (), {'__module__': 'mymodule'})  # noqa
+        x, y = MyClass(), MyClass()
+        x.magic_attr = True
+        y.magic_attr = False
+        stats = objgraph.typestats(
+            shortnames=False,
+            filter=lambda e: isinstance(e, MyClass) and e.magic_attr)
+        self.assertEqual(1, stats['mymodule.MyClass'])
+
+
 class ByTypeTest(GarbageCollectedMixin, unittest.TestCase):
     """Tests for the by_test function."""
 
@@ -399,6 +422,12 @@ def test_edge_label_long_type_names(self):
             objgraph._edge_label(d, 1, shortnames=False),
             ' [label="mymodule\.MyClass\\n<mymodule\.MyClass object at .*"]')
 
+    def test_short_repr_lambda(self):
+        f = lambda x: x  # noqa
+        lambda_lineno = sys._getframe().f_lineno - 1
+        self.assertEqual('lambda: tests.py:%s' % lambda_lineno,
+                         objgraph._short_repr(f))
+
 
 class StubSubprocess(object):