Unable to get the status of document node resolved due to leading spaces (#1064)

* Unable to get the status of document node resolved due to leading and trailing spaces

* removed trailing leading spaces while creating source node

* graph driver connection resolved

Author: kaustubh-darekar

backend/src/document_sources/web_pages.py

@@ -6,7 +6,7 @@ def get_documents_from_web_page(source_url:str):
   try:
     pages = WebBaseLoader(source_url, verify_ssl=False).load()
     try:
-      file_name = pages[0].metadata['title']
+      file_name = pages[0].metadata['title'].strip()
       if not file_name:
         file_name = last_url_segment(source_url)      
     except:

backend/src/main.py

@@ -123,7 +123,7 @@ def create_source_node_graph_web_url(graph, model, source_url, source_type):
       message = f"Unable to read data for given url : {source_url}"
       raise LLMGraphBuilderException(message)
     try:
-      title = pages[0].metadata['title']
+      title = pages[0].metadata['title'].strip()
       if not title:
         title = last_url_segment(source_url)
       language = pages[0].metadata['language']

backend/src/post_processing.py

@@ -3,6 +3,7 @@
 import time
 from langchain_neo4j import Neo4jGraph
 import os
+from src.graph_query import get_graphDB_driver
 from src.shared.common_fn import load_embedding_model
 from langchain_core.output_parsers import JsonOutputParser
 from langchain_core.prompts import ChatPromptTemplate
@@ -137,7 +138,7 @@ def create_vector_fulltext_indexes(uri, username, password, database):
     logging.info("Starting the process of creating full-text indexes.")
 
     try:
-        driver = GraphDatabase.driver(uri, auth=(username, password), database=database)
+        driver = get_graphDB_driver(uri, username, password,database)
         driver.verify_connectivity()
         logging.info("Database connectivity verified.")
     except Exception as e: