Update app.py

app.py

@@ -1,43 +1,40 @@
 import os
 import fitz  # PyMuPDF for PDF handling
-import easyocr  # OCR for text extraction
+import pytesseract  # OCR for text extraction
+from PIL import Image
 import tempfile
 import streamlit as st
 
-# Disable GStreamer to prevent OpenCV-related errors
-os.environ["OPENCV_VIDEOIO_PRIORITY_GSTREAMER"] = "0"
-
-def extract_text_with_ocr(pdf_path):
+def extract_text_with_tesseract(pdf_path):
     """
-    Extract text with bounding box positions using OCR for both English and Arabic text.
+    Extract text with bounding box positions using Tesseract OCR.
     :param pdf_path: Path to the input PDF file.
     :return: List of dictionaries containing text and positions for each page.
     """
     extracted_data = []
     doc = fitz.open(pdf_path)
 
-    # Convert each PDF page to an image for OCR processing
     for page_num in range(len(doc)):
         page = doc.load_page(page_num)
-        pix = page.get_pixmap(dpi=300)  # Convert PDF page to image
+        pix = page.get_pixmap(dpi=300)  # Convert PDF page to high-resolution image
         image_path = f"temp_page_{page_num}.png"
         pix.save(image_path)
 
-        # Perform OCR on the image
-        reader = easyocr.Reader(['en'])  # Supports English (add 'ar' for Arabic if needed)
-        results = reader.readtext(image_path, detail=1)  # detail=1 returns bounding box info
+        # Perform OCR using Tesseract
+        img = Image.open(image_path)
+        ocr_result = pytesseract.image_to_data(img, output_type=pytesseract.Output.DICT)
 
-        # Extract text and positions
         page_data = []
-        for (bbox, text, confidence) in results:
-            (x0, y0), (x1, y1) = bbox[0], bbox[2]
-            page_data.append({
-                "text": text,
-                "x0": x0,
-                "y0": y1,  # Adjust to bottom-left corner (PDF coordinates)
-                "font_size": y1 - y0,  # Approximate font size
-                "confidence": confidence
-            })
+        for i in range(len(ocr_result["text"])):
+            if ocr_result["text"][i].strip():  # Ignore empty text
+                page_data.append({
+                    "text": ocr_result["text"][i],
+                    "x0": ocr_result["left"][i],
+                    "y0": ocr_result["top"][i],
+                    "x1": ocr_result["left"][i] + ocr_result["width"][i],
+                    "y1": ocr_result["top"][i] + ocr_result["height"][i],
+                    "font_size": ocr_result["height"][i]
+                })
 
         extracted_data.append(page_data)
 
@@ -49,29 +46,26 @@ def extract_text_with_ocr(pdf_path):
 
 def overlay_text_with_fonts(pdf_path, extracted_data, output_pdf_path):
     """
-    Overlay extracted text onto the original PDF using fonts from different font families.
+    Overlay extracted text onto the original PDF using PyMuPDF.
     :param pdf_path: Path to the input PDF file.
-    :param extracted_data: List of extracted text with positions.
+    :param extracted_data: Extracted text and positions.
     :param output_pdf_path: Path to save the output PDF file.
     """
     doc = fitz.open(pdf_path)
 
-    # Define default font settings
-    default_font = "Helvetica"  # You can replace it with specific fonts like "Arial" or others.
+    default_font = "Helvetica"
 
     for page_num, page_data in enumerate(extracted_data):
         page = doc[page_num]
 
         for item in page_data:
-            if item["confidence"] > 0.8:  # Only overlay high-confidence text
-                page.insert_text(
-                    (item["x0"], item["y0"]),
-                    item["text"],
-                    fontsize=item["font_size"],
-                    fontname=default_font,
-                    color=(0, 0, 0),  # Black text
-                    render_mode=0  # Ensure text is not outlined
-                )
+            page.insert_text(
+                (item["x0"], item["y0"]),
+                item["text"],
+                fontsize=item["font_size"] / 2,  # Adjust font size for better scaling
+                fontname=default_font,
+                color=(0, 0, 0)  # Black text
+            )
 
     doc.save(output_pdf_path)
     print(f"PDF saved to: {output_pdf_path}")
@@ -79,29 +73,24 @@ def overlay_text_with_fonts(pdf_path, extracted_data, output_pdf_path):
 
 def process_pdf(uploaded_pdf, output_pdf_path):
     """
-    Process the uploaded PDF to extract text using OCR and overlay it as editable text.
-    :param uploaded_pdf: The uploaded PDF file.
+    Process the uploaded PDF to extract text using Tesseract and overlay it.
+    :param uploaded_pdf: Uploaded PDF file.
     :param output_pdf_path: Path to save the output PDF file.
     """
     with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as temp_pdf:
         temp_pdf.write(uploaded_pdf.read())
         temp_pdf_path = temp_pdf.name
 
-    # Step 1: Extract text using OCR
-    extracted_data = extract_text_with_ocr(temp_pdf_path)
-
-    # Step 2: Overlay extracted text onto the original PDF
+    extracted_data = extract_text_with_tesseract(temp_pdf_path)
     overlay_text_with_fonts(temp_pdf_path, extracted_data, output_pdf_path)
 
-    # Cleanup temporary file
-    if os.path.exists(temp_pdf_path):
-        os.remove(temp_pdf_path)
+    os.remove(temp_pdf_path)
 
 
 # Streamlit App
 def main():
-    st.title("PDF Text Conversion Tool")
-    st.write("Upload a PDF to convert vector text into regular, editable text.")
+    st.title("PDF OCR and Text Conversion Tool")
+    st.write("Upload a PDF to extract and overlay text as editable layers.")
 
     uploaded_file = st.file_uploader("Upload PDF", type=["pdf"])
     if uploaded_file:
@@ -121,10 +110,8 @@ def main():
                 mime="application/pdf"
             )
 
-        # Cleanup the processed output PDF
-        if os.path.exists(output_pdf_path):
-            os.remove(output_pdf_path)
+        os.remove(output_pdf_path)
 
 
 if __name__ == "__main__":
-    main()
+    main()