Support retrieving multiple attributes by name

Author: mattjala

h5pyd/_hl/attrs.py

@@ -166,7 +166,7 @@ def __getitem__(self, name):
 
         return arr
 
-    def get_attributes(self, pattern=None, limit=None, marker=None, use_cache=True):
+    def get_attributes(self, names=None, pattern=None, limit=None, marker=None, use_cache=True):
         """
         Get all attributes or a subset of attributes from the target object.
         If 'use_cache' is True, use the objdb cache if available.
@@ -179,6 +179,9 @@ def get_attributes(self, pattern=None, limit=None, marker=None, use_cache=True):
         if use_cache and (pattern or limit or marker):
             raise ValueError("use_cache cannot be used with pattern, limit, or marker parameters")
 
+        if names and (pattern or limit or marker or use_cache):
+            raise ValueError("names cannot be used with pattern, limit, marker, or cache")
+
         if self._objdb_attributes is not None:
             # use the objdb cache
             out = {}
@@ -189,8 +192,8 @@ def get_attributes(self, pattern=None, limit=None, marker=None, use_cache=True):
 
         # Omit trailing slash
         req = self._req_prefix[:-1]
-
         req += "?IncludeData=1"
+        body = {}
 
         if pattern:
             req += "&pattern=" + pattern
@@ -199,7 +202,21 @@ def get_attributes(self, pattern=None, limit=None, marker=None, use_cache=True):
         if marker:
             req += "&Marker=" + marker
 
-        rsp = self._parent.GET(req)
+        if names:
+            if isinstance(names, list):
+                names = [name.decode('utf-8') if isinstance(name, bytes) else name for name in names]
+            else:
+                if isinstance(names, bytes):
+                    names = names.decode("utf-8")
+                names = [names]
+
+            body['attr_names'] = names
+
+        if body:
+            rsp = self._parent.POST(req, body=body)
+        else:
+            rsp = self._parent.GET(req)
+
         attrs_json = rsp['attributes']
         names = [attr['name'] for attr in attrs_json]
         values = [attr['value'] for attr in attrs_json]

test/hl/test_attribute.py

@@ -198,7 +198,8 @@ def test_get_multiple(self):
         self.assertTrue("attr5" in values_out)
         self.assertTrue(np.array_equal(values_out["attr5"], values[5]))
 
-        # get attributes that match the pattern 'att*' (all attributes)
+        # get only attributes that match the pattern 'att*'
+        g1.attrs['new_attr'] = np.arange(100)
         pattern = "att*"
         values_out = g1.attrs.get_attributes(pattern=pattern, use_cache=False)
 
@@ -228,6 +229,18 @@ def test_get_multiple(self):
             self.assertTrue(names[i] in values_out)
             self.assertTrue(np.array_equal(values_out[names[i]], values[i]))
 
+        # get set of attributes by name
+        names = ['attr5', 'attr7', 'attr9']
+
+        values_out = g1.attrs.get_attributes(names=names, use_cache=False)
+
+        self.assertEqual(len(values_out), 3)
+
+        for name in names:
+            self.assertTrue(name in values_out)
+            i = int(name[4])
+            self.assertTrue(np.array_equal(values_out[name], values[i]))
+
     def test_delete_multiple(self):
         if config.get('use_h5py') or self.hsds_version() < "0.9.0":
             return
@@ -266,6 +279,16 @@ def test_delete_multiple(self):
             self.assertTrue(names[i] in g1.attrs)
             self.assertTrue(np.array_equal(g1.attrs[names[i]], values[i]))
 
+        # delete attributes with name that must be URL-encoded
+        names = ['attr with spaces', 'attr%', 'unicode八attr']
+        for name in names:
+            g1.attrs[name] = np.arange(100)
+
+        del g1.attrs[names]
+
+        for name in names:
+            self.assertTrue(name not in g1.attrs)
+
 
 if __name__ == '__main__':
     loglevel = logging.ERROR