From 94a8376c3d5ac9b6e28cbe257e4af351ab7e7655 Mon Sep 17 00:00:00 2001
From: Annmary Justine K <annmary.roy@hpe.com>
Date: Thu, 23 Sep 2021 00:56:59 -0500
Subject: [PATCH] Support hdf5 files external links

---
 dvc/repo/add.py   |  3 +++
 dvc/utils/hdf5.py | 21 +++++++++++++++++++++
 2 files changed, 24 insertions(+)
 create mode 100644 dvc/utils/hdf5.py

diff --git a/dvc/repo/add.py b/dvc/repo/add.py
index 7318cb320e..ecd74e8938 100644
--- a/dvc/repo/add.py
+++ b/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
 
diff --git a/dvc/utils/hdf5.py b/dvc/utils/hdf5.py
new file mode 100644
index 0000000000..73bec5f20c
--- /dev/null
+++ b/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)
+