Spaces:

zamal
/

Multimodal-Chat-PDF

Running

anand004 commited on Jun 23, 2024

Commit

68093d8

verified ·

1 Parent(s): f55139c

Update app.py

Files changed (1) hide show

app.py CHANGED Viewed

@@ -122,7 +122,7 @@ def extract_data_from_pdfs(docs, session, include_images, progress=gr.Progress()
     images = []
     for doc in docs:
-        ocrmypdf.ocr(doc, "ocr.pdf", deskew=True, skip_text=True, output_type="pdf")
         text = extract_text("ocr.pdf")
         all_text += clean_text(text) + "\n\n"
         if include_images == "Include Images":

     images = []
     for doc in docs:
+        ocrmypdf.ocr(doc, "ocr.pdf", deskew=True, force_ocr=True)
         text = extract_text("ocr.pdf")
         all_text += clean_text(text) + "\n\n"
         if include_images == "Include Images":