[FIX] Delete Vector DB nodes after extraction (#100)

Delete nodes after extraction

Signed-off-by: Deepak <[email protected]>

Author: Deepak-Kesavan

src/unstract/sdk/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.48.1"
+__version__ = "0.48.2"
 
 
 def get_sdk_version():

src/unstract/sdk/index.py

@@ -203,20 +203,7 @@ def index(
                     level=LogLevel.ERROR,
                 )
 
-            if doc_id_found and reindex:
-                # Delete the nodes for the doc_id
-                try:
-                    vector_db.delete(ref_doc_id=doc_id)
-                    self.tool.stream_log(f"Deleted nodes for {doc_id}")
-                except Exception as e:
-                    self.tool.stream_log(
-                        f"Error deleting nodes for {doc_id}: {e}",
-                        level=LogLevel.ERROR,
-                    )
-                    raise SdkError(f"Error deleting nodes for {doc_id}: {e}") from e
-                doc_id_found = False
-
-            if doc_id_found:
+            if doc_id_found and not reindex:
                 self.tool.stream_log(f"File was indexed already under {doc_id}")
                 return doc_id
 
@@ -287,6 +274,18 @@ def index(
                 documents.append(document)
             self.tool.stream_log(f"Number of documents: {len(documents)}")
 
+            if doc_id_found:
+                # Delete the nodes for the doc_id
+                try:
+                    vector_db.delete(ref_doc_id=doc_id)
+                    self.tool.stream_log(f"Deleted nodes for {doc_id}")
+                except Exception as e:
+                    self.tool.stream_log(
+                        f"Error deleting nodes for {doc_id}: {e}",
+                        level=LogLevel.ERROR,
+                    )
+                    raise SdkError(f"Error deleting nodes for {doc_id}: {e}") from e
+
             try:
                 if chunk_size == 0:
                     parser = SentenceSplitter.from_defaults(