Update app.py

app.py

@@ -8,6 +8,10 @@ import re
 import requests
 import trafilatura
 from bs4 import BeautifulSoup
+import fitz
+import pytesseract
+from PIL import Image
+import io
 
 import gradio as gr
 import numpy as np
@@ -60,26 +64,25 @@ def safe_download_pdf(url: str) -> str:
     return tmp_path
 
 
-def pdf_to_text(path: str, start_page: int = 1, end_page: int | None = None):
-    """
-    Extract per-page text from a PDF using PyMuPDF, with light cleaning.
-    Returns a list of strings (one per page).
-    """
+
+def pdf_to_text(path):
     doc = fitz.open(path)
-    total_pages = doc.page_count
+    pages_text = []
+
+    for page in doc:
+        # 1) Best text extractor for academic PDFs
+        text = page.get_textpage().extractText()
 
-    if end_page is None:
-        end_page = total_pages
+        # 2) If text is empty or garbage → OCR fallback
+        if len(text.strip()) < 50:
+            pix = page.get_pixmap(dpi=300)
+            img = Image.open(io.BytesIO(pix.tobytes("png")))
+            text = pytesseract.image_to_string(img)
 
-    texts = []
-    for i in range(start_page - 1, end_page):
-        page = doc.load_page(i)
-        txt = page.get_text("text")
-        txt = preprocess(txt)
-        texts.append(txt)
+        pages_text.append(text)
 
     doc.close()
-    return texts
+    return pages_text
 
 
 def text_to_chunks(texts, word_length=WORD_CHUNK_SIZE, start_page=1):