Merge pull request #65 from oracle-samples/cdb-fix

Cdb fix

Author: corradodebari

.gitignore

@@ -43,3 +43,4 @@ temp/rag_agent.ipynb
 temp/tools.ipynb
 temp/tools.py
 temp/json-dual.sql
+env.sh

app/src/content/split_embed.py

@@ -10,6 +10,7 @@
 import os
 import re
 import requests
+import time
 
 # Streamlit
 import streamlit as st
@@ -316,6 +317,15 @@ def main():
             progress_text = "Operation in progress. Please wait."
             progress_bar = st.progress(0, text=progress_text)
             progress_ind = 100 / len(process_list)
+
+            try:
+                utilities.oci_create_bucket(state.oci_config, state.oci_namespace, oci_compartments[bucket_compartment], dst_bucket)
+                time.sleep(5)
+            except oci.exceptions.ServiceError as ex:
+                logger.info("WARNING: bucket exists")
+
+     
+            
             for index, f in process_list.iterrows():
                 progress_stat = int((((index + 1) * progress_ind) / 2))
                 progress_bar.progress(progress_stat, text=progress_text)

app/src/modules/split.py

@@ -185,7 +185,7 @@ def load_and_split_documents(
         all_split_docos += split_docos
     logger.info("Total Number of Chunks: %i", len(all_split_docos))
 
-    return split_docos, split_files
+    return all_split_docos, split_files
 
 
 ##########################################

app/src/modules/utilities.py

@@ -621,19 +621,27 @@ def oci_get_object(config, namespace, bucket_name, directory, object_name, retri
     return file_path
 
 
+
 def oci_put_object(config, namespace, compartment, bucket_name, file_path, retries=True):
     """Upload file to Object Storage"""
+    MAX_RETRY = 2
     file_name = os.path.basename(file_path)
     logger.info("Uploading %s to %s as %s ", file_path, bucket_name, file_name)
     client = oci_init_client(oci.object_storage.ObjectStorageClient, config, retries)
     upload_manager = oci.object_storage.UploadManager(client, allow_parallel_uploads=True, parallel_process_count=10)
     try:
         upload_manager.upload_file(namespace, bucket_name, file_name, file_path)
+        logger.info("Uploaded successfully %s to %s", file_name, bucket_name)
     except oci.exceptions.ServiceError as ex:
         logger.exception(ex, exc_info=False)
-        oci_create_bucket(config, namespace, compartment, bucket_name)
-        upload_manager.upload_file(namespace, bucket_name, file_name, file_path)
-    logger.info("Uploaded %s to %s", file_name, bucket_name)
+        for i in range(1, MAX_RETRY):
+            try:
+                logger.info("RETRY: upload_manager.upload_file()")
+                upload_manager.upload_file(namespace, bucket_name, file_name, file_path)
+                logger.info("Uploaded successfully %s to %s", file_name, bucket_name)
+                break
+            except oci.exceptions.ServiceError as ex:
+                logger.exception(ex, exc_info=False)
 
     os.remove(file_path)