update datasetvalues for in init

Author: jreadey

src/h5json/hsdsstore/hsds_writer.py

@@ -436,14 +436,31 @@ def updateValues(self, dset_ids):
             if getCollectionForId(dset_id) != "datasets":
                 continue  # ignore groups and datatypes
             dset_json = self.db.getObjectById(dset_id)
-            if "updates" not in dset_json:
+            dset_shape = dset_json["shape"]
+            dset_class = dset_shape['class']
+            if dset_class == "H5S_NULL":
+                # no data to update
                 continue
-            updates = dset_json["updates"]
-            if updates:
-                self.log.debug(f"hsds_writer> {dset_id} update count: {len(updates)}")
-                for (sel, arr) in updates:
-                    self.updateValue(dset_id, sel, arr)
-                updates.clear()
+            if self._init:
+                # get all data for the dataset
+                # TBD: do this by chunks
+                if dset_class == "H5S_SCALAR":
+                    dset_dims = []
+                else:
+                    dset_dims = dset_shape["dims"]
+                sel_all = selections.select(dset_dims, ...)
+                arr = self.db.getDatasetValues(dset_id, sel_all)
+                if arr is not None:
+                    self.updateValue(dset_id, sel_all, arr)
+            else:
+                if "updates" not in dset_json:
+                    continue
+                updates = dset_json["updates"]
+                if updates:
+                    self.log.debug(f"hsds_writer> {dset_id} update count: {len(updates)}")
+                    for (sel, arr) in updates:
+                        self.updateValue(dset_id, sel, arr)
+                    updates.clear()
 
     def flush(self):
         """ Write dirty items """
@@ -472,6 +489,9 @@ def flush(self):
             self.createObjects(obj_ids)
             dirty_ids.update(obj_ids)
             dirty_ids.add(root_id)  # add back root for attribute and link creation
+            if not self._no_data:
+                # initialize dataset values
+                self.updateValues(obj_ids)
             self._init = False
         elif self.db.new_objects:
             self.log.debug(f"hsds_writer> {len(self.db.new_objects)} objects to create")

test/unit/hsds_writer_test.py

@@ -199,6 +199,46 @@ def testH5PyToHS(self):
         self.assertEqual(g1_json["attributeCount"], 0)
         self.assertEqual(g1_json["linkCount"], 2)
 
+        # get the g1.1 link
+        http_rsp = http_conn.GET(f"/groups/{g1_id}/links/g1.1")
+        self.assertEqual(http_rsp.status_code, 200)
+        rsp_json = http_rsp.json()
+        g1_1_link = rsp_json["link"]
+        g1_1_id = g1_1_link["id"]
+
+        # Get the g1.1 json
+        http_rsp = http_conn.GET(f"/groups/{g1_1_id}")
+        self.assertEqual(http_rsp.status_code, 200)
+        g1_json = http_rsp.json()
+        self.assertEqual(g1_json["attributeCount"], 0)
+        self.assertEqual(g1_json["linkCount"], 2)
+
+        # get the dset1.1.1 link
+        http_rsp = http_conn.GET(f"/groups/{g1_1_id}/links/dset1.1.1")
+        self.assertEqual(http_rsp.status_code, 200)
+        rsp_json = http_rsp.json()
+        dset1_1_1_link = rsp_json["link"]
+        dset1_1_1_id = dset1_1_1_link["id"]
+
+        # get the dset1.1.1 json
+        http_rsp = http_conn.GET(f"/datasets/{dset1_1_1_id}")
+        self.assertEqual(http_rsp.status_code, 200)
+        dset1_1_1_json = http_rsp.json()
+        dset1_1_1_shape = dset1_1_1_json["shape"]
+        self.assertEqual(dset1_1_1_shape["class"], "H5S_SIMPLE")
+
+        # get the dset1_1_1 data
+        http_rsp = http_conn.GET(f"/datasets/{dset1_1_1_id}/value")
+        self.assertEqual(http_rsp.status_code, 200)
+        rsp_json = http_rsp.json()
+        dset1_1_1_value = rsp_json["value"]
+        self.assertEqual(len(dset1_1_1_value), 10)
+        for i in range(10):
+            row = dset1_1_1_value[i]
+            self.assertEqual(len(row), 10)
+            for j in range(10):
+                self.assertEqual(row[j], i * j)
+
         db.close()