Support hdf5 files external links

dvc/repo/add.py

@@ -18,6 +18,7 @@
 from ..repo.scm_context import scm_context
 from ..utils import LARGE_DIR_SIZE, glob_targets, resolve_output, resolve_paths
 from ..utils.collections import ensure_list, validate
+from ..utils.hdf5 import get_external_links
 from . import locked
 
 if TYPE_CHECKING:
@@ -237,6 +238,8 @@ def _find_all_targets(
             if not repo.scm.belongs_to_scm(path)
             if not repo.scm.is_tracked(path)
         )
+    elif (target.find(".h5") != -1):
+        yield from get_external_links(target)
     else:
         yield target
 

dvc/utils/hdf5.py

@@ -0,0 +1,21 @@
+
+import h5py
+
+class H5ls:
+    def __init__(self):
+        # Store an empty list for link names
+        self.names = []
+
+    def __call__(self, name, obj):
+        for key, val in obj.items():
+            p = obj.get(name=key,default=None, getclass=False, getlink=True)
+            if (isinstance(p, h5py.ExternalLink)):
+                self.names.append(p.filename)
+
+def get_external_links(target):
+    h5ls = H5ls()
+    h5ls.names.append(target)
+    with h5py.File(target,'r') as hdf:
+        hdf.visititems(h5ls)
+        return set(h5ls.names)
+