add envo term api capabiliites and doct strings

Author: brynnz22

support_code/nmdc/nom/api_info_retriever.py

@@ -1,15 +1,16 @@
 import requests
 from typing import Optional
+import yaml
 
 # TODO: Update og_url to be regular nmdc api url when berkeley is implemented
 
-class ApiInfoRetriever:
+class NmdcApiInfoRetriever:
     def __init__(self, collection_name: str):
         self.collection_name = collection_name
 
     def get_id_by_slot_from_collection(self, slot_name: str, slot_field_value: str):
         """
-        Retrieve the identifier from a specified collection based on a slot name and field value.
+        Retrieve the NMDC identifier from a specified collection based on a slot name and field value.
 
         Parameters
         ----------
@@ -51,3 +52,73 @@ def get_id_by_slot_from_collection(self, slot_name: str, slot_field_value: str):
 
         return identifier
 
+class BioOntologyInfoRetriever:
+    """
+    Client for retrieving ENVO term information from BioPortal API.
+
+    A class to handle authentication and retrieval of Environmental Ontology (ENVO)
+    terms using the BioPortal REST API service.
+
+    Parameters
+    ----------
+    config_path : str
+        Path to YAML configuration file containing BioPortal API credentials
+
+    Notes
+    -----
+    The configuration file should contain an 'api_key' field with a valid
+    BioPortal API key.
+
+    Examples
+    --------
+    >>> retriever = BioOntologyInfoRetriever('config.yaml')
+    >>> envo_terms = retriever.get_envo_terms('ENVO:00002042')
+    >>> print(envo_terms)
+    {'ENVO:00002042': 'surface water'}
+    """
+    def __init__(self, config_path: str):
+        self.config = config_path
+
+    def get_envo_terms(self, envo_id: dict):
+        """
+        Look up an ENVO term label using BioPortal API.
+
+        Parameters
+        ----------
+        envo_id : str
+            The ENVO identifier to look up (e.g., 'ENVO:00002042')
+
+        Returns
+        -------
+        dict
+            Dictionary with envo_id as key and term label as value
+            Example: {'ENVO:00002042': 'surface water'}
+
+        Raises
+        ------
+        requests.exceptions.RequestException
+            If the API request fails
+        KeyError
+            If the response doesn't contain expected data format
+        yaml.YAMLError
+            If the config file cannot be parsed
+        FileNotFoundError
+            If the config file is not found
+
+        Notes
+        -----
+        Makes an authenticated request to BioPortal API to retrieve the
+        preferred label (prefLabel) for the given ENVO term.
+        """
+        
+        config = yaml.safe_load(open(self.config))
+        api_key = config['api_key']
+
+        url = f"http://data.bioontology.org/ontologies/ENVO/classes/{envo_id}"
+        headers = {"Authorization": f"apikey token={api_key}"}
+
+        response = requests.get(url, headers=headers)
+        response.raise_for_status()
+
+        data = response.json()
+        return {envo_id: data['prefLabel']}

support_code/nmdc/nom/main.py

@@ -39,8 +39,10 @@ def main():
         - 12
         - 15
         - 21
-    - minting_client_config_path: str
-        The path to the NMDC minting client configuration file. Defaults to 'enviroMS/nmdc_metadata_generation/config.yaml'.
+    - config_path: str
+        The path to the NMDC minting client and Bioportal API key configuration yaml file. Defaults to 'enviroMS/nmdc_metadata_generation/config.yaml'.
+        Should have client_id and client_secret for the NMDC minting client. And api_key for BioPortal API access. To get a BioPortal api_key register here:
+        https://bioportal.bioontology.org/accounts/new and go to settings
 
     Notes
     -----
@@ -66,7 +68,7 @@ def main():
         execution_resource=config_data['execution_resource'],
         field_strength=config_data['field_strength'],
         workflow_version=config_data['workflow_version'],
-        minting_client_config_path=config_data['minting_client_config_path']
+        config_path=config_data['minting_client_config_path']
     )
 
     generator.run()