hashing: support multi-type sorting

esrftaskgraph/esrftaskgraph/hashing.py

@@ -12,6 +12,26 @@ def classhashdata(cls):
     return qualname(cls).encode()
 
 
+def multitype_sorted(sequence, key=None):
+    try:
+        return sorted(sequence, key=key)
+    except TypeError:
+        pass
+    if key is None:
+        key = lambda item: item
+
+    adict = dict()
+    for item in sequence:
+        typename = type(key(item)).__name__
+        adict.setdefault(typename, list()).append(item)
+
+    return [
+        item
+        for _, items in sorted(adict.items(), key=lambda tpl: tpl[0])
+        for item in sorted(items, key=key)
+    ]
+
+
 def uhash(value, _hash=None):
     """Universial hash (as opposed to python's hash).
     This is an example. Must find something better.
@@ -42,13 +62,12 @@ def uhash(value, _hash=None):
     elif isinstance(value, (numpy.ndarray, numpy.number)):
         _hash.update(value.tobytes())
     elif isinstance(value, Mapping):
-        keys, values = zip(*sorted(value.items(), key=lambda item: item[0]))
+        keys, values = zip(*multitype_sorted(value.items(), key=lambda item: item[0]))
         uhash(keys, _hash=_hash)
         uhash(values, _hash=_hash)
     elif isinstance(value, Set):
-        # Unordered
-        # TODO: the values in the set are no necessarily sortable
-        raise TypeError(value, type(value))
+        values = multitype_sorted(value)
+        uhash(values, _hash=_hash)
     elif isinstance(value, Iterable):
         # Ordered
         for v in value:

esrftaskgraph/tests/test_hashing.py

@@ -25,6 +25,12 @@ def test_hashing_unique():
     assert hashing.uhash(alist) == hashing.uhash(list(alist))
     assert hashing.uhash(alist) != hashing.uhash(alist[::-1])
     assert hashing.uhash(alist) != hashing.uhash(tuple(alist))
+    assert hashing.uhash(alist) != hashing.uhash(set(alist))
+
+    aset = set(unique_values)
+    assert hashing.uhash(aset) == hashing.uhash(set(aset))
+    assert hashing.uhash(aset) != hashing.uhash(tuple(aset))
+    assert hashing.uhash(aset) != hashing.uhash(list(alist))
 
     andarray = numpy.arange(10)
     assert hashing.uhash(andarray) == hashing.uhash(andarray.copy())