allow to define custom serialiser

Author: mr-mixas

nested_diff/__init__.py

@@ -16,6 +16,8 @@
 
 """Recursive diff and patch for nested structures."""
 
+import pickle
+
 import nested_diff.handlers
 
 __all__ = ['Differ', 'Iterator', 'Patcher', 'diff', 'patch']
@@ -101,7 +103,7 @@ class Differ():
     default_differ = DEFAULT_HANDLER.diff
 
     def __init__(self, A=True, N=True, O=True, R=True, U=True,  # noqa: E501 E741 N803
-                 trimR=False, handlers=None):
+                 trimR=False, dumper=None, handlers=None):
         """Initialize Differ.
 
         Args:
@@ -111,6 +113,7 @@ def __init__(self, A=True, N=True, O=True, R=True, U=True,  # noqa: E501 E741 N8
             R: Enable/disable removed items.
             U: Enable/disable unchanged items.
             trimR: When enabled will replace removed data by None.
+            dumper: Optional objects serialiser.
             handlers: A list of type handlers.
 
         """
@@ -121,6 +124,8 @@ def __init__(self, A=True, N=True, O=True, R=True, U=True,  # noqa: E501 E741 N8
         self.op_u = U
         self.op_trim_r = trimR
 
+        self.dump = dumper or pickle.dumps
+
         self._differs = {}
 
         for handler in TYPE_HANDLERS if handlers is None else handlers:

nested_diff/handlers.py

@@ -18,7 +18,6 @@
 
 from difflib import SequenceMatcher
 from math import isnan
-from pickle import dumps
 
 
 class TypeHandler():
@@ -50,7 +49,7 @@ def diff(self, differ, a, b):
         diff = {}
         equal = True
 
-        if dumps(a, -1) == dumps(b, -1):
+        if differ.dump(a) == differ.dump(b):
             if differ.op_u:
                 diff['U'] = a
         else:
@@ -342,8 +341,8 @@ def diff(self, differ, a, b):
         (False, {'D': [{'R': 0}, {'N': 4, 'I': 3}, {'A': 5}]})
         >>>
         """
-        self.lcs.set_seq1(tuple(dumps(i, -1) for i in a))
-        self.lcs.set_seq2(tuple(dumps(i, -1) for i in b))
+        self.lcs.set_seq1(tuple(differ.dump(i) for i in a))
+        self.lcs.set_seq2(tuple(differ.dump(i) for i in b))
 
         diff = []
         equal = True

tests/test_diff.py

@@ -1,4 +1,5 @@
 import pytest
+import sys
 
 from nested_diff import Differ, diff, handlers
 
@@ -74,6 +75,43 @@ def test_different_object_attributes():
     assert got == expected
 
 
+@pytest.mark.skipif(sys.version_info < (3, 8),
+                    reason='reducer_override appeared in 3.8')
+def test_custom_dumper():
+    import io
+    import pickle
+
+    class ClassToTestDiff:
+        pass
+
+    class _Pickler(pickle.Pickler):
+        def reducer_override(self, obj):
+            if type(obj) is ClassToTestDiff:
+                return str, tuple('ClassToTestDiff obj, id: ' + str(id(obj)))
+
+            return NotImplemented
+
+    def _dumper(obj):
+        buf = io.BytesIO()
+        pickler = _Pickler(buf)
+        pickler.dump(obj)
+        return buf.getvalue()
+
+    a = ClassToTestDiff()
+    b = ClassToTestDiff()
+
+    expected = {
+        'D': [
+            {'U': a},
+            {'O': a, 'N': b},
+            {'U': 42},
+        ],
+    }
+    _, got = Differ(dumper=_dumper).diff([a, a, 42], [a, b, 42])
+
+    assert got == expected
+
+
 def test_dicts_with_same_data_but_different_sequence_u_disabled():
     # for example pickle.dumps({1: 1, 2: 2}) != pickle.dumps({2: 2, 1: 1})
     a = {1: 1, 2: 2}