ENH Add check_call_in_cache method to check cache without calling fun…

…ction (#820)

Co-authored-by: cottrell <david@g18e.com>
Co-authored-by: Olivier Grisel <olivier.grisel@gmail.com>
Co-authored-by: Thomas Moreau <thomas.moreau.2010@gmail.com>

CHANGES.rst

@@ -17,6 +17,9 @@ Development version
 1.0.1
 -----
 
+- Add check_call_in_cache method to check cache without calling function.
+  https://github.com/joblib/joblib/pull/820
+
 - dask: avoid redundant scattering of large arguments to make a more
   efficient use of the network resources and avoid crashing dask with
   "OSError: [Errno 55] No buffer space available"

doc/memory.rst

@@ -427,7 +427,7 @@ objects that, in addition of behaving like normal functions, expose
 methods useful for cache exploration and management.
 
 .. autoclass:: MemorizedFunc
-    :members: __init__, call, clear
+    :members: __init__, call, clear, check_call_in_cache
 
 
 ..

joblib/memory.py

@@ -19,7 +19,6 @@
 import traceback
 import warnings
 import inspect
-import sys
 import weakref
 
 from tokenize import open as open_py_source
@@ -33,7 +32,6 @@
 from ._store_backends import StoreBackendBase, FileSystemStoreBackend
 
 
-
 FIRST_LINE_TEXT = "# first line:"
 
 # TODO: The following object should have a data store object as a sub
@@ -136,7 +134,6 @@ def _store_backend_factory(backend, location, verbose=0, backend_options=None):
             "supported by joblib. Returning None instead.".format(
                 location.__class__.__name__), UserWarning)
 
-
     return None
 
 
@@ -361,6 +358,12 @@ def clear(self, warn=True):
         # Argument "warn" is for compatibility with MemorizedFunc.clear
         pass
 
+    def call(self, *args, **kwargs):
+        return self.func(*args, **kwargs)
+
+    def check_call_in_cache(self, *args, **kwargs):
+        return False
+
 
 ###############################################################################
 # class `MemorizedFunc`
@@ -606,6 +609,21 @@ def __getstate__(self):
 
         return state
 
+    def check_call_in_cache(self, *args, **kwargs):
+        """Check if function call is in the memory cache.
+
+        Does not call the function or do any work besides func inspection
+        and arg hashing.
+
+        Returns
+        -------
+        is_call_in_cache: bool
+            Whether or not the result of the function has been cached
+            for the input arguments that have been passed.
+        """
+        func_id, args_id = self._get_output_identifiers(*args, **kwargs)
+        return self.store_backend.contains_item((func_id, args_id))
+
     # ------------------------------------------------------------------------
     # Private interface
     # ------------------------------------------------------------------------
@@ -683,8 +701,8 @@ def _check_previous_func_code(self, stacklevel=2):
                 extract_first_line(
                     self.store_backend.get_cached_func_code([func_id]))
         except (IOError, OSError):  # some backend can also raise OSError
-                self._write_func_code(func_code, first_line)
-                return False
+            self._write_func_code(func_code, first_line)
+            return False
         if old_func_code == func_code:
             return True
 
@@ -821,8 +839,6 @@ def _persist_input(self, duration, args, kwargs, this_duration_limit=0.5):
                           % this_duration, stacklevel=5)
         return metadata
 
-    # XXX: Need a method to check if results are available.
-
     # ------------------------------------------------------------------------
     # Private `object` interface
     # ------------------------------------------------------------------------

joblib/test/test_memory.py

@@ -609,6 +609,19 @@ def test_persistence(tmpdir):
     gp(1)
 
 
+def test_check_call_in_cache(tmpdir):
+    for func in (MemorizedFunc(f, tmpdir.strpath),
+                 Memory(location=tmpdir.strpath, verbose=0).cache(f)):
+        result = func.check_call_in_cache(2)
+        assert not result
+        assert isinstance(result, bool)
+        assert func(2) == 5
+        result = func.check_call_in_cache(2)
+        assert result
+        assert isinstance(result, bool)
+        func.clear()
+
+
 def test_call_and_shelve(tmpdir):
     # Test MemorizedFunc outputting a reference to cache.