Support for hdf5 proxies in lazy operands

Author: FrancescAlted

caterva2/hdf5.py

@@ -405,6 +405,11 @@ def cparams(self):
     def dparams(self):
         return self.b2arr.dparams
 
+    @property
+    def schunk(self):
+        # This is basically needed to certificate that it is an NDArray in the LazyArray machinery
+        return self.b2arr.schunk
+
     @property
     def fields(self) -> Mapping[str, numpy.dtype]:
         return self.b2arr.fields

caterva2/services/srv_utils.py

@@ -181,10 +181,16 @@ def get_relpath(ndarr, cache, personal, shared, public):
     path = pathlib.Path(ndarr.schunk.urlpath)
     if shared is not None and path.is_relative_to(shared):
         # Shared: /.../<shared>/<subpath> to <path> (i.e. no change)
-        return path
+        path = path.relative_to(shared)
+        parts = list(path.parts)
+        parts[0] = "@shared"
+        return pathlib.Path(*parts)
     elif public is not None and path.is_relative_to(public):
         # Shared: /.../<public>/<subpath> to <path> (i.e. no change)
-        return path
+        path = path.relative_to(public)
+        parts = list(path.parts)
+        parts[0] = "@public"
+        return pathlib.Path(*parts)
     try:
         # Cache: /.../<root>/<subpath> to <root>/<subpath>
         path = path.relative_to(cache)

caterva2/services/sub.py

@@ -293,8 +293,12 @@ def open_b2(abspath, path):
             for key, value in operands.items():
                 if value is None:
                     raise ValueError(f'Missing operand "{key}"')
-                if "proxy-source" in value.schunk.meta:
-                    # Save operand as Proxy, see blosc2.open doc for more info
+                vlmetaval = value.schunk.vlmeta
+                if "proxy-source" in value.schunk.meta or (
+                    "_ftype" in vlmetaval and vlmetaval["_ftype"] == "hdf5"
+                ):
+                    # Save operand as Proxy, see blosc2.open doc for more info.
+                    # Or, it can be an HDF5 dataset too (which should be handled in the next call)
                     relpath = srv_utils.get_relpath(
                         value, settings.cache, settings.personal, settings.shared, settings.public
                     )