add can_read method to HDMFIO and HDF5IO (#875)

Co-authored-by: Ryan Ly <[email protected]>

Author: bendichter

CHANGELOG.md

@@ -1,17 +1,17 @@
 # HDMF Changelog
 
-## HMDF 3.7.0 (Upcoming)
+## HDMF 3.7.0 (Upcoming)
 
 ### New features and minor improvements
 - Updated `ExternalResources` to have EntityKeyTable with updated tests/documentation and minor bug fix to ObjectKeyTable. @mavaylon1 [#872](https://github.com/hdmf-dev/hdmf/pull/872)
+- Added abstract static method `HDMFIO.can_read()` and concrete static method `HDF5IO.can_read()`. @bendichter [#875](https://github.com/hdmf-dev/hdmf/pull/875)
 - Added warning for `DynamicTableRegion` links that are not added to the same parent as the original container object. @mavaylon1 [#891](https://github.com/hdmf-dev/hdmf/pull/891)
 - Added the `TermSet` class along with integrated validation methods for any child of `AbstractContainer`, e.g., `VectorData`, `Data`, `DynamicTable`. @mavaylon1 [#880](https://github.com/hdmf-dev/hdmf/pull/880)
 - Allow for `datetime.date` to be used instead of `datetime.datetime`. @bendichter [#874](https://github.com/hdmf-dev/hdmf/pull/874)
 - Updated `HDMFIO` and `HDF5IO` to support `ExternalResources`. @mavaylon1 [#895](https://github.com/hdmf-dev/hdmf/pull/895)
 - Dropped Python 3.7 support. @rly [#897](https://github.com/hdmf-dev/hdmf/pull/897)
 
 ### Documentation and tutorial enhancements:
-
 - Added tutorial for the new `TermSet` class @mavaylon1 [#880](https://github.com/hdmf-dev/hdmf/pull/880)
 
 ## Bug fixes

src/hdmf/backends/hdf5/h5tools.py

@@ -36,6 +36,17 @@ class HDF5IO(HDMFIO):
 
     __ns_spec_path = 'namespace'  # path to the namespace dataset within a namespace group
 
+    @staticmethod
+    def can_read(path):
+        """Determines whether a given path is readable by the HDF5IO class"""
+        if not os.path.isfile(path):
+            return False
+        try:
+            with h5py.File(path, "r"):
+                return True
+        except IOError:
+            return False
+
     @docval({'name': 'path', 'type': (str, Path), 'doc': 'the path to the HDF5 file', 'default': None},
             {'name': 'mode', 'type': str,
              'doc': ('the mode to open the HDF5 file with, one of ("w", "r", "r+", "a", "w-", "x"). '
@@ -82,8 +93,8 @@ def __init__(self, **kwargs):
         self.__file = file_obj
         super().__init__(manager, source=path, external_resources_path=external_resources_path)
         # NOTE: source is not set if path is None and file_obj is passed
-        self.__built = dict()       # keep track of each builder for each dataset/group/link for each file
-        self.__read = dict()        # keep track of which files have been read. Key is the filename value is the builder
+        self.__built = dict() # keep track of each builder for each dataset/group/link for each file
+        self.__read = dict() # keep track of which files have been read. Key is the filename value is the builder
         self.__ref_queue = deque()  # a queue of the references that need to be added
         self.__dci_queue = HDF5IODataChunkIteratorQueue()  # a queue of DataChunkIterators that need to be exhausted
         ObjectMapper.no_convert(Dataset)
@@ -603,7 +614,7 @@ def __read_group(self, h5obj, name=None, ignore=set()):
                             builder = self.__read_dataset(target_obj, builder_name)
                         else:
                             builder = self.__read_group(target_obj, builder_name, ignore=ignore)
-                        self.__set_built(sub_h5obj.file.filename,  target_obj.id, builder)
+                        self.__set_built(sub_h5obj.file.filename, target_obj.id, builder)
                     link_builder = LinkBuilder(builder=builder, name=k, source=os.path.abspath(h5obj.file.filename))
                     link_builder.location = h5obj.name
                     self.__set_written(link_builder)
@@ -648,7 +659,7 @@ def __read_dataset(self, h5obj, name=None):
             name = str(os.path.basename(h5obj.name))
         kwargs['source'] = os.path.abspath(h5obj.file.filename)
         ndims = len(h5obj.shape)
-        if ndims == 0:                                       # read scalar
+        if ndims == 0:  # read scalar
             scalar = h5obj[()]
             if isinstance(scalar, bytes):
                 scalar = scalar.decode('UTF-8')
@@ -678,7 +689,7 @@ def __read_dataset(self, h5obj, name=None):
                 elif isinstance(elem1, Reference):
                     d = BuilderH5ReferenceDataset(h5obj, self)
                     kwargs['dtype'] = d.dtype
-            elif h5obj.dtype.kind == 'V':    # table / compound data type
+            elif h5obj.dtype.kind == 'V':  # table / compound data type
                 cpd_dt = h5obj.dtype
                 ref_cols = [check_dtype(ref=cpd_dt[i]) or check_dtype(vlen=cpd_dt[i]) for i in range(len(cpd_dt))]
                 d = BuilderH5TableDataset(h5obj, self, ref_cols)
@@ -708,7 +719,7 @@ def __compound_dtype_to_list(cls, h5obj_dtype, dset_dtype):
     def __read_attrs(self, h5obj):
         ret = dict()
         for k, v in h5obj.attrs.items():
-            if k == SPEC_LOC_ATTR:     # ignore cached spec
+            if k == SPEC_LOC_ATTR:  # ignore cached spec
                 continue
             if isinstance(v, RegionReference):
                 raise ValueError("cannot read region reference attributes yet")
@@ -925,14 +936,14 @@ def set_attributes(self, **kwargs):
                     self.logger.debug("Setting %s '%s' attribute '%s' to %s"
                                       % (obj.__class__.__name__, obj.name, key, value.__class__.__name__))
                     obj.attrs[key] = value
-                elif isinstance(value, (Container, Builder, ReferenceBuilder)):           # a reference
+                elif isinstance(value, (Container, Builder, ReferenceBuilder)):  # a reference
                     self.__queue_ref(self._make_attr_ref_filler(obj, key, value))
                 else:
                     self.logger.debug("Setting %s '%s' attribute '%s' to %s"
                                       % (obj.__class__.__name__, obj.name, key, value.__class__.__name__))
                     if isinstance(value, np.ndarray) and value.dtype.kind == 'U':
                         value = np.array(value, dtype=H5_TEXT)
-                    obj.attrs[key] = value                   # a regular scalar
+                    obj.attrs[key] = value  # a regular scalar
             except Exception as e:
                 msg = "unable to write attribute '%s' on object '%s'" % (key, obj.name)
                 raise RuntimeError(msg) from e
@@ -1079,7 +1090,7 @@ def write_dataset(self, **kwargs):  # noqa: C901
         name = builder.name
         data = builder.data
         dataio = None
-        options = dict()   # dict with additional
+        options = dict()  # dict with additional
         if isinstance(data, H5DataIO):
             options['io_settings'] = data.io_settings
             dataio = data

src/hdmf/backends/io.py

@@ -10,6 +10,13 @@
 
 
 class HDMFIO(metaclass=ABCMeta):
+
+    @staticmethod
+    @abstractmethod
+    def can_read(path):
+        """Determines whether a given path is readable by this HDMFIO class"""
+        pass
+
     @docval({'name': 'manager', 'type': BuildManager,
              'doc': 'the BuildManager to use for I/O', 'default': None},
             {"name": "source", "type": (str, Path),

tests/unit/test_io_hdf5_h5tools.py

@@ -3228,6 +3228,10 @@ def test_non_manager_container(self):
 
         class OtherIO(HDMFIO):
 
+            @staticmethod
+            def can_read(path):
+                pass
+
             def read_builder(self):
                 pass
 
@@ -3257,6 +3261,10 @@ def test_non_HDF5_src_link_data_true(self):
 
         class OtherIO(HDMFIO):
 
+            @staticmethod
+            def can_read(path):
+                pass
+
             def __init__(self, manager):
                 super().__init__(manager=manager)
 
@@ -3570,3 +3578,9 @@ def test_dataio_shape_then_data(self):
         dataio = H5DataIO(shape=(10, 10), dtype=int)
         with self.assertRaisesRegex(ValueError, "Setting data when dtype and shape are not None is not supported"):
             dataio.data = list()
+
+
+def test_hdf5io_can_read():
+    assert not HDF5IO.can_read("not_a_file")
+    assert HDF5IO.can_read("tests/unit/back_compat_tests/1.0.5.h5")
+    assert not HDF5IO.can_read(__file__)  # this file is not an HDF5 file