DTR integration

doglib/doglib.py

@@ -1,9 +1,10 @@
 import json
+import logging
 import os
 from typing import List, Union, Optional
 
 from . import curl
-from .dtr import expand_datatype
+from .dtr import expand_datatype, DataTypeNotFoundException
 from .pid import pid_factory, PID
 from .repos import JSONParser, XMLParser
 from .repos import RegRepo, warn_europeana
@@ -78,9 +79,16 @@ def fetch(self, pid_string: Union[str, PID], format: str = 'dict',
                 return ""
         fetch_result: dict = self._fetch(pid)
         if dtr:
+            unique_data_types: set = set()
             for entry in fetch_result['ref_files']:
                 for resource in entry['ref_resources']:
-                    resource['data_type_taxonomy'] = expand_datatype(resource['data_type'])
+                    data_type = resource['data_type']
+                    unique_data_types.add(data_type)
+            for unique_type in unique_data_types:
+                try:
+                    fetch_result['dtr'][unique_type] = expand_datatype(unique_type)
+                except DataTypeNotFoundException:
+                    continue
         if format == 'dict':
             return fetch_result
         elif format == 'jsons' or format == 'str':
@@ -308,3 +316,6 @@ def is_pid(self, pid_string: Union[str, PID]) -> bool:
         """
         pid = pid_factory(pid_string)
         return True if pid is not None else False
+
+    def get_repository_report(self):
+

doglib/dtr.py

@@ -2,10 +2,12 @@
 import json
 from typing import List, Set, Union
 
+from requests import RequestException
+
 from .curl import get
 
 
-class DataTypeNotFound(Exception):
+class DataTypeNotFoundException(Exception):
     def __init__(self, message):
         self.message = message
 
@@ -25,13 +27,17 @@ def get_dtr_taxonomy_by_type(data_type: str) -> dict:
     :return: dict, a dictionary representation of the type's taxonomy
     """
     dtr_type_search_endpoint = f"http://typeapi.lab.pidconsortium.net/v1/taxonomy/search?query={data_type}&name={data_type}"
-    url, dtr_taxonomy_search_response, header = get(dtr_type_search_endpoint)
+    try:
+        url, dtr_taxonomy_search_response, header = get(dtr_type_search_endpoint)
+    except RequestException as error:
+        raise DataTypeNotFoundException(f"DataType <{data_type}> doesn't exist in the DTR taxonomy") from error
+
     dtr_taxonomy_json = json.loads(dtr_taxonomy_search_response)
 
     try:
         dtr_type_id = dtr_taxonomy_json[0]["id"]
     except (IndexError, KeyError) as error:
-        raise DataTypeNotFound(f"DataType <{data_type}> doesn't exist in the DTR taxonomy") from error
+        raise DataTypeNotFoundException(f"DataType <{data_type}> doesn't exist in the DTR taxonomy") from error
     parents = dtr_taxonomy_json[0]["parents"]
     if parents:
         for parent_id, parent_name in parents.items():
@@ -53,7 +59,7 @@ def get_taxonomy_root_node_by_id(data_type_id: str) -> str:
     try:
         dtr_type_id = dtr_taxonomy_json["id"]
     except (IndexError, KeyError) as error:
-        raise DataTypeNotFound(f"DataType with id <{data_type_id}> doesn't exist in the DTR taxonomy") from error
+        raise DataTypeNotFoundException(f"DataType with id <{data_type_id}> doesn't exist in the DTR taxonomy") from error
     parents = dtr_taxonomy_json["parents"]
     # Assumption of single parent
     taxonomy_root_id = dtr_type_id