test keylookup

src/hub/dataload/sources/chembl/parser.py

@@ -0,0 +1,28 @@
+import json
+
+
+def parse_data(data):
+    for item in data["targets"]:
+        output = {
+            "chembl_target": item["target_chembl_id"],
+            "xrefs": {
+                "accession": (
+                    [
+                        target_component.get("accession")
+                        for target_component in item["target_components"]
+                        if "accession" in target_component
+                    ]
+                ),
+            },
+        }
+        if output["xrefs"]["accession"]:
+            yield output
+
+
+def load_data(target_filepaths):
+    for file in target_filepaths:
+        with open(file) as f:
+            content = f.read()
+            json_data = json.loads(content)
+            parsed_data = parse_data(json_data)
+            return parsed_data

src/hub/dataload/sources/chembl/upload.py

@@ -1,15 +1,47 @@
+import glob
+import os
+
+import biothings.hub.dataload.storage as storage
 import biothings.hub.dataload.uploader as uploader
 
+from hub.datatransform.keylookup import MyGeneKeyLookup
+
+from .parser import load_data
+
 
-class ChemblUploader(uploader.DummySourceUploader):
+class ChemblUploader(uploader.BaseSourceUploader):
     name = "chembl"
 
+    storage_class = storage.RootKeyMergerStorage
+    TARGET_FILENAME_PATTERN = "target.*.json"
+
+    keylookup = MyGeneKeyLookup(
+        [
+            ("uniprot", "chembl.xrefs.accession"),
+        ],
+        skip_on_failure=True,
+    )
+
+    def load_data(self, data_folder):
+        target_filepaths = glob.iglob(
+            os.path.join(data_folder, self.TARGET_FILENAME_PATTERN)
+        )
+        return self.keylookup(load_data(target_filepaths), skip_on_failure=True)
+
     @classmethod
     def get_mapping(klass):
         mapping = {
             "chembl_target": {
                 "type": "keyword",
                 "normalizer": "keyword_lowercase_normalizer",
-            }
+            },
+            "xrefs": {
+                "properties": {
+                    "accession": {
+                        "type": "keyword",
+                        "normalizer": "keyword_lowercase_normalizer",
+                    },
+                }
+            },
         }
         return mapping