adding tests for hash_value, also for nested objects

Author: djarecka

pydra/engine/helpers.py

@@ -10,7 +10,7 @@
 import subprocess as sp
 
 from .specs import Runtime, File, attr_fields
-from .helpers_file import is_existing_file, hash_file, copyfile, is_existing_file
+from .helpers_file import hash_file, copyfile, is_existing_file
 
 
 def ensure_list(obj, tuple2list=False):
@@ -444,21 +444,21 @@ def hash_function(obj):
 
 def hash_value(value, tp=None, metadata=None):
     """calculating hash or returning values recursively"""
+    if metadata is None:
+        metadata = {}
     if isinstance(value, (tuple, list)):
         return [hash_value(el, tp, metadata) for el in value]
     elif isinstance(value, dict):
         dict_hash = {k: hash_value(v, tp, metadata) for (k, v) in value.items()}
         # returning a sorted object
-        return sorted(dict_hash.items(), key=lambda x: x[0])
+        return [list(el) for el in sorted(dict_hash.items(), key=lambda x: x[0])]
     else:  # not a container
         if (
             (tp is File or "pydra.engine.specs.File" in str(tp))
             and is_existing_file(value)
             and "container_path" not in metadata
         ):
             return hash_file(value)
-        elif isinstance(value, tuple):
-            return list(value)
         else:
             return value
 

pydra/engine/tests/test_helpers.py

@@ -4,17 +4,18 @@
 import cloudpickle as cp
 
 from .utils import multiply
-from .. import helpers
+from ..helpers import hash_value, hash_function, save, create_pyscript
 from .. import helpers_file
+from ..specs import File
 
 
 def test_save(tmpdir):
     outdir = Path(tmpdir)
     with pytest.raises(ValueError):
-        helpers.save(tmpdir)
+        save(tmpdir)
     foo = multiply(name="mult", x=1, y=2)
     # save task
-    helpers.save(outdir, task=foo)
+    save(outdir, task=foo)
     del foo
     # load saved task
     task_pkl = outdir / "_task.pklz"
@@ -24,7 +25,7 @@ def test_save(tmpdir):
     # execute task and save result
     res = foo()
     assert res.output.out == 2
-    helpers.save(outdir, result=res)
+    save(outdir, result=res)
     del res
     # load saved result
     res_pkl = outdir / "_result.pklz"
@@ -35,10 +36,10 @@ def test_save(tmpdir):
 def test_create_pyscript(tmpdir):
     outdir = Path(tmpdir)
     with pytest.raises(Exception):
-        helpers.create_pyscript(outdir, "checksum")
+        create_pyscript(outdir, "checksum")
     foo = multiply(name="mult", x=1, y=2)
-    helpers.save(outdir, task=foo)
-    pyscript = helpers.create_pyscript(outdir, foo.checksum)
+    save(outdir, task=foo)
+    pyscript = create_pyscript(outdir, foo.checksum)
     assert pyscript.exists()
 
 
@@ -64,11 +65,62 @@ def test_hashfun_float():
         == pi_50.as_integer_ratio()
         == pi_15.as_integer_ratio()
     )
-    assert (
-        helpers.hash_function(math.pi)
-        == helpers.hash_function(pi_15)
-        == helpers.hash_function(pi_50)
-    )
+    assert hash_function(math.pi) == hash_function(pi_15) == hash_function(pi_50)
     # comparing for x that have different x.as_integer_ratio()
     assert math.pi.as_integer_ratio() != pi_10.as_integer_ratio()
-    assert helpers.hash_function(math.pi) != helpers.hash_function(pi_10)
+    assert hash_function(math.pi) != hash_function(pi_10)
+
+
+def test_hash_value_dict():
+    dict1 = {"a": 10, "b": 5}
+    dict2 = {"b": 5, "a": 10}
+    assert (
+        hash_value(dict1)
+        == hash_value(dict2)
+        == [["a", hash_value(10)], ["b", hash_value(5)]]
+        == [["a", 10], ["b", 5]]
+    )
+
+
+def test_hash_value_list_tpl():
+    lst = [2, 5.6, "ala"]
+    tpl = (2, 5.6, "ala")
+    assert hash_value(lst) == [hash_value(2), hash_value(5.6), hash_value("ala")] == lst
+    assert hash_value(lst) == hash_value(tpl)
+
+
+def test_hash_value_list_dict():
+    lst = [2, {"a": "ala", "b": 1}]
+    hash_value(lst)
+    assert (
+        hash_value(lst)
+        == [hash_value(2), hash_value([["a", "ala"], ["b", 1]])]
+        == [2, [["a", "ala"], ["b", 1]]]
+    )
+
+
+def test_hash_value_files(tmpdir):
+    file_1 = tmpdir.join("file_1.txt")
+    file_2 = tmpdir.join("file_2.txt")
+    with open(file_1, "w") as f:
+        f.write("hello")
+    with open(file_2, "w") as f:
+        f.write("hello")
+
+    assert hash_value(file_1, tp=File) == hash_value(file_2, tp=File)
+    assert hash_value(file_1, tp=str) != hash_value(file_2, tp=str)
+    assert hash_value(file_1) != hash_value(file_2)
+
+
+def test_hash_value_files_list(tmpdir):
+    file_1 = tmpdir.join("file_1.txt")
+    file_2 = tmpdir.join("file_2.txt")
+    with open(file_1, "w") as f:
+        f.write("hello")
+    with open(file_2, "w") as f:
+        f.write("hi")
+
+    assert hash_value([file_1, file_2], tp=File) == [
+        hash_value(file_1, tp=File),
+        hash_value(file_2, tp=File),
+    ]

pydra/engine/tests/test_specs.py

@@ -289,7 +289,7 @@ def test_input_file_hash_4(tmpdir):
 
     # checking specific hash value
     hash1 = inputs(in_file=[{"file": file, "int": 3}]).hash
-    assert hash1 == "0148961a66106e15e55de3ab7e8c49e6227b6db7cdc614dcbae20406feb2548a"
+    assert hash1 == "e0555e78a40a02611674b0f48da97cdd28eee7e9885ecc17392b560c14826f06"
 
     # the same file, but int field changes
     hash1a = inputs(in_file=[{"file": file, "int": 5}]).hash