And random suffixes to decreases any clash probability when writing tmp files to disc

src/khoj/processor/content/docx/docx_to_entries.py

@@ -1,6 +1,7 @@
 import logging
 import os
 from datetime import datetime
+from random import random
 from typing import Dict, List, Tuple
 
 from langchain_community.document_loaders import Docx2txtLoader
@@ -94,7 +95,8 @@ class DocxToEntries(TextToEntries):
         """Extract text from specified DOCX file"""
         try:
             timestamp_now = datetime.utcnow().timestamp()
-            tmp_file = f"tmp_docx_file_{timestamp_now}.docx"
+            random_suffix = random.randint(0, 1000)
+            tmp_file = f"tmp_docx_file_{timestamp_now}_{random_suffix}.docx"
             docx_entry_by_pages = []
             with open(tmp_file, "wb") as f:
                 bytes_content = docx_file

src/khoj/processor/content/pdf/pdf_to_entries.py

@@ -2,6 +2,7 @@ import base64
 import logging
 import os
 from datetime import datetime
+from random import random
 from typing import Dict, List, Tuple
 
 from langchain_community.document_loaders import PyMuPDFLoader
@@ -98,7 +99,8 @@ class PdfToEntries(TextToEntries):
         try:
             # Write the PDF file to a temporary file, as it is stored in byte format in the pdf_file object and the PDF Loader expects a file path
             timestamp_now = datetime.utcnow().timestamp()
-            tmp_file = f"tmp_pdf_file_{timestamp_now}.pdf"
+            random_suffix = random.randint(0, 1000)
+            tmp_file = f"tmp_pdf_file_{timestamp_now}_{random_suffix}.pdf"
             pdf_entry_by_pages = []
             with open(f"{tmp_file}", "wb") as f:
                 f.write(pdf_file)