Updates

Update some of the logic and refactor the code.

loadtestutils/invenio_records_cache.py

 # vim: ts=4
 
 import urllib.parse
-from random import randrange
+from random import choice
 
 from requests import get
 
 
 class InvenioRecordsCache:
-    def __init__(self, base_url, max_records):
+    """A cache for visible records in an Invenio instance."""
+
+    def __init__(self, base_url, max_records, verify_ssl):
+        """Constructor."""
         self.base_url = base_url
         self.records_endpoint = urllib.parse.urljoin(self.base_url, "/api/records")
         self.max_records = max_records
+        self.verify_ssl = verify_ssl
         self.records = []
+        self.initialized = False
+
+    def load_records(self, force=False):
+        """Initialize the set of records if it hasn't been initialized yet."""
+        if self.initialized and not force:
+            return
 
-    def load_records(self):
         result = get(
             self.records_endpoint,
             headers={"Accept": "application/json"},
-            params="size={}".format(self.max_records),
+            params=f"size={self.max_records}",
+            verify=self.verify_ssl,
         )
 
-        self.records = self.__extract_records__(result.json())
+        self.records = self._extract_records(result.json())
+        self.initialized = True
 
-    def __extract_records__(self, result):
+    def _extract_records(self, result):
+        """Extract the records from the search result."""
         _extracted_records = []
         for record in result["hits"]["hits"]:
             _new_record = {
                 "id": record["id"],
                 "title": record["metadata"]["title"],
                 "url": record["links"]["self"],
-                "file_urls": self.__load_resource_urls__(record),
+                "file_urls": self._load_file_urls(record),
             }
             _extracted_records.append(_new_record)
 
         return _extracted_records
 
-    @staticmethod
-    def __load_resource_urls__(record):
-        record_files = get(record["links"]["files"]).json()
+    def _load_file_urls(self, record):
+        """Look up the record's file URLs from its JSON."""
+        if record["access"]["files"].lower() != "public":
+            # TODO check if we have access
+            return []
+
+        if "files" not in record["links"]:
+            return []
+
+        record_files = get(record["links"]["files"], verify=self.verify_ssl).json()
 
         if "entries" not in record_files.keys():
             return []
@@ -48,5 +67,12 @@ class InvenioRecordsCache:
         return file_urls
 
     def get_random_record(self):
-        record_index = randrange(0, len(self.records))
-        return self.records[record_index]
+        """Select a random record from the cached records.
+
+        Load the records if they're not loaded yet.
+        """
+        self.load_records()
+        if not self.records:
+            return None
+
+        return choice(self.records)