replace unicode errors

Author: mmz-001

knowledge_gpt/core/parsing.py

@@ -88,7 +88,7 @@ def from_bytes(cls, file: BytesIO) -> "PdfFile":
 class TxtFile(File):
     @classmethod
     def from_bytes(cls, file: BytesIO) -> "TxtFile":
-        text = file.read().decode("utf-8")
+        text = file.read().decode("utf-8", errors="replace")
         text = strip_consecutive_newlines(text)
         file.seek(0)
         doc = Document(page_content=text.strip())

knowledge_gpt/main.py

@@ -63,7 +63,7 @@
 try:
     file = read_file(uploaded_file)
 except Exception as e:
-    display_file_read_error(e)
+    display_file_read_error(e, file_name=uploaded_file.name)
 
 chunked_file = chunk_file(file, chunk_size=300, chunk_overlap=0)
 

knowledge_gpt/ui.py

@@ -36,9 +36,9 @@ def is_file_valid(file: File) -> bool:
     return True
 
 
-def display_file_read_error(e: Exception) -> NoReturn:
+def display_file_read_error(e: Exception, file_name: str) -> NoReturn:
     st.error("Error reading file. Make sure the file is not corrupted or encrypted")
-    logger.error(f"{e.__class__.__name__}: {e}")
+    logger.error(f"{e.__class__.__name__}: {e}. Extension: {file_name.split('.')[-1]}")
     st.stop()