switch to json metadata format and add a test

apps/hiopy/worker.py

 #!/usr/bin/env python3
 
+import json
 import logging
 from argparse import ArgumentParser
 from itertools import chain, groupby
 
 import numpy as np
-import yaml
 import zarr
 from coyote import (
     Coyote,
@@ -149,7 +149,7 @@ def main():
             time_methods2yac = {"point": 0, "sum": 1, "mean": 2, "min": 3, "max": 4}
 
             collection_size = v.shape[1] if len(v.shape) == 3 else 1
-            src_name = v.attrs.get("hiopy::src_name", default=v.name.split("/")[-1])
+            src_name = v.attrs.get("hiopy::src_name", default=v.basename)
             if "hiopy::yac_source" in v.attrs:
                 src_comp, src_grid = v.attrs["hiopy::yac_source"]
             else:
@@ -159,7 +159,7 @@ def main():
                     for gid, vg, _, gvars in chain(*distributed_data_vars)
                     for groupvar in gvars
                     if groupvar.name.split("/")[-2] == v.group.attrs["hiopy::parent"]
-                    and groupvar.name.split("/")[-1] == src_name
+                    and groupvar.name.split("/")[-1] == v.basename
                 ][0]
                 src_comp = src_grid = f"{args.process_group}_{gid}_{src_vgroup}"
             time_method = v.attrs.get("hiopy::time_method", "point")
@@ -196,23 +196,31 @@ def main():
                 frac_mask=frac_mask,
             )
 
-    def get_source_triple(v):
-        if "hiopy::parent" in v.attrs:
-            return get_source_triple(v.group.parent[v.attrs["hiopy::parent"]][v.basename])
-        else:
+    def get_source_triple(v, group):
+        if "hiopy::parent" in group.attrs:
+            parent_field = group.parent[group.attrs["hiopy::parent"]][v.basename]
+            pgroup = group.parent[group.attrs["hiopy::parent"]]
+            pgroup.parent = group.parent
+            return get_source_triple(parent_field, pgroup)
+        elif "hiopy::yac_source" in v.attrs:
             src_comp, src_grid = v.attrs["hiopy::yac_source"]
-            src_field = v.attrs["hiopy::src_name"] if "src_name" else v.basename
+            src_field = v.attrs.get("hiopy::src_name", v.basename)
             return src_comp, src_grid, src_field
+        else:
+            raise RuntimeError("Invalid attributes: " + str(dict(v.attrs)))
 
     ensure_enddef()
     if group_comm_rank() == 0:
         for v in all_data_vars:
             if "hiopy::copy_metadata" in v.attrs:
-                comp, grid, field = get_source_triple(v)
-                metadata = yaml.safe_load(get_field_metadata(comp, grid, field))
-                for key, value in metadata["cf"]:
+                comp, grid, field = get_source_triple(v, v.group)
+                md_str = get_field_metadata(comp, grid, field)
+                print(md_str)
+                metadata = json.loads(md_str)
+                print(metadata)
+                for key, value in metadata["cf"].items():
                     v.attrs[key] = value
-                    del v.attrs["hiopy::copy_metadata"]  # copy only once
+                del v.attrs["hiopy::copy_metadata"]  # copy only once
 
     run()
 

tests/simple_source.c

@@ -45,7 +45,8 @@ int main(int argc, char** argv) {
     yac_cdef_datetime("2000-01-01T00:00:00.00Z", enddate);
     printf("enddate: %s\n", enddate);
     int comp_id, grid_id, point_id;
-    yac_cdef_comp("simple_source", &comp_id);
+    const char* comp_name = "simple_source";
+    yac_cdef_comp(comp_name, &comp_id);
     int nbr_points[] = {2 * resolution, resolution};
     int cyclic[]     = {1, 0};
     double lon[nbr_points[0]];
@@ -56,7 +57,8 @@ int main(int argc, char** argv) {
     for (size_t i = 0; i < nbr_points[1]; ++i) {
         lat[i] = -M_PI_2 + (i + 0.5) * M_PI / nbr_points[1];
     }
-    yac_cdef_grid_reg2d("simple_grid", nbr_points, cyclic, lon, lat, &grid_id);
+    const char* grid_name = "simple_grid";
+    yac_cdef_grid_reg2d(grid_name, nbr_points, cyclic, lon, lat, &grid_id);
     int glb_idx[nbr_points[0] * nbr_points[1]];
     for (int i = 0; i < nbr_points[0] * nbr_points[1]; ++i)
         glb_idx[i] = i;
@@ -65,13 +67,57 @@ int main(int argc, char** argv) {
 
     int x_field_id = -1;
     yac_cdef_field("x", comp_id, &point_id, 1, 1, "PT1S", YAC_TIME_UNIT_ISO_FORMAT, &x_field_id);
+    yac_cdef_field_metadata(
+        comp_name,
+        grid_name,
+        "x",
+        "{\n"
+        "  \"cf\": {\n"
+        "    \"standard_name\": \"x\",\n"
+        "    \"long_name\": \"this is a very long x\",\n"
+        "    \"units\": \"1\"\n"
+        "  }\n"
+        "}");
     int y_field_id = -1;
     yac_cdef_field("y", comp_id, &point_id, 1, 1, "PT1S", YAC_TIME_UNIT_ISO_FORMAT, &y_field_id);
+    yac_cdef_field_metadata(
+        comp_name,
+        grid_name,
+        "y",
+        "{\n"
+        "  \"cf\": {\n"
+        "    \"standard_name\": \"y\",\n"
+        "    \"long_name\": \"this is a very long y\",\n"
+        "    \"units\": \"1\"\n"
+        "  }\n"
+        "}");
     int z_field_id = -1;
     yac_cdef_field("z", comp_id, &point_id, 1, 1, "PT1S", YAC_TIME_UNIT_ISO_FORMAT, &z_field_id);
+    yac_cdef_field_metadata(
+        comp_name,
+        grid_name,
+        "z",
+        "{\n"
+        "  \"cf\": {\n"
+        "    \"standard_name\": \"z\",\n"
+        "    \"long_name\": \"this is a very long z\",\n"
+        "    \"units\": \"1\"\n"
+        "  }\n"
+        "}");
     int clock_field_id = -1;
     yac_cdef_field(
         "clock", comp_id, &point_id, 1, 1, "PT1S", YAC_TIME_UNIT_ISO_FORMAT, &clock_field_id);
+    yac_cdef_field_metadata(
+        comp_name,
+        grid_name,
+        "clock",
+        "{\n"
+        "  \"cf\": {\n"
+        "    \"standard_name\": \"t\",\n"
+        "    \"long_name\": \"time\",\n"
+        "    \"units\": \"s\"\n"
+        "  }\n"
+        "}");
 
     yac_cenddef();