small fix for uuid

compass_sdk/compass.py

@@ -35,7 +35,7 @@
     DEFAULT_MAX_CHUNKS_PER_REQUEST,
     DEFAULT_MAX_ERROR_RATE,
     DEFAULT_MAX_RETRIES,
-    DEFAULT_SLEEP_RETRY_SECONDS,
+    DEFAULT_SLEEP_RETRY_SECONDS, UUID_NAMESPACE,
 )
 from compass_sdk.utils import open_document
 
@@ -281,22 +281,28 @@ def parse_and_insert_doc(
         *,
         index_name: str,
         filename: str,
-        file_id: Optional[str] = None,
+        file_uuid: Optional[str] = None,
         max_retries: int = DEFAULT_MAX_RETRIES,
         sleep_retry_seconds: int = DEFAULT_SLEEP_RETRY_SECONDS,
     ) -> Optional[str]:
         """
         Parse and insert a document into an index in Compass
         :param index_name: the name of the index
         :param filename: the filename of the document
-        :param file_id: the id of the document (optional)
+        :param file_uuid: the id of the document (optional)
         :param max_retries: the maximum number of times to retry a request if it fails
         :param sleep_retry_seconds: the number of seconds to wait before retrying an API request
         :return: an error message if the request failed, otherwise None
         """
+        def is_valid_uuid(fileid: Optional[str]) -> bool:
+            try:
+                uuid.UUID(fileid)
+                return True
+            except ValueError:
+                return False
 
         def generate_uuid(b64_string: str) -> uuid.UUID:
-            namespace = uuid.UUID("00000000-0000-0000-0000-000000000000")
+            namespace = uuid.UUID(UUID_NAMESPACE)
             return uuid.uuid5(namespace, b64_string)
 
         doc = open_document(filename)
@@ -313,10 +319,15 @@ def generate_uuid(b64_string: str) -> uuid.UUID:
 
         # Open the file and read the bytes. Get some metadata and send it to Compass
         b64 = base64.b64encode(doc.filebytes).decode("utf-8")
-        file_id = file_id or generate_uuid(b64)
+        if file_uuid and not is_valid_uuid(file_uuid):
+            err = f"Invalid UUID: {file_uuid}. Namespace: {UUID_NAMESPACE}"
+            logger.error(err)
+            return err
+        file_uuid = file_uuid or generate_uuid(b64)
         content_type = get_content_type(file=BytesIO(doc.filebytes), filepath=filename)
+
         put_doc_input = ParseableDocument(
-            id=file_id,
+            id=file_uuid,
             filename=filename,
             bytes=b64,
             content_type=content_type,

compass_sdk/constants.py

@@ -33,3 +33,5 @@
     "date",
     "authors",
 ]
+
+UUID_NAMESPACE = "00000000-0000-0000-0000-000000000000"