Merge pull request #4 from punitarani/more-loaders

Add support for docx, txt and images

Author: punitarani

app.py

@@ -4,8 +4,8 @@
 from langchain.schema import Document
 
 from rabbithole import summarize_document
-from rabbithole.loader import load_file
-from rabbithole.mp3 import SUPPORTED_FILE_TYPES
+from rabbithole.loader import load_file, SUPPORTED_IMG_FILE_TYPES
+from rabbithole.mp3 import SUPPORTED_AV_FILE_TYPES
 
 # Global variables
 results = {}
@@ -36,7 +36,9 @@ def run_summarization(document: list[Document], doc_name: str):
 
 st.title("RabbitHole")
 
-uploaded_files = st.file_uploader("Upload content", type=["pdf", *SUPPORTED_FILE_TYPES], accept_multiple_files=True)
+uploaded_files = st.file_uploader("Upload content",
+                                  type=["docx", "pdf", "txt", *SUPPORTED_IMG_FILE_TYPES, *SUPPORTED_AV_FILE_TYPES],
+                                  accept_multiple_files=True)
 
 if st.button("Summarize"):
     if not uploaded_files:

packages.txt

@@ -0,0 +1,2 @@
+ffmpeg
+tesseract