Update app.py

app.py

@@ -225,14 +225,6 @@ def get_page_md_and_regions(page_result):
     return md, regions
 
 
-def is_duplicate(text, existing_text):
-    """Check if text is already contained in existing_text using first 100 chars as fingerprint."""
-    if not text or not existing_text:
-        return False
-    fingerprint = text[:100].strip()
-    return fingerprint in existing_text
-
-
 def run_ocr(uploaded_file):
     if uploaded_file is None:
         return "Please upload a file."
@@ -301,8 +293,7 @@ def run_ocr(uploaded_file):
                         ftr = extract_pdf_text_in_band(path, page_num, PDF_FOOTER_BAND_FRAC, 1.0)
                 if not (ftr and ftr.strip()):
                     ftr = ocr_zone(img_path, fs, 1.0)
-                # Only append footer if NOT already present in page_md (avoids duplication)
-                if ftr and ftr.strip() and not is_duplicate(ftr, page_md):
+                if ftr and ftr.strip():
                     parts.append(ftr.strip())
 
             if parts:
@@ -323,4 +314,4 @@ with gr.Blocks(title="GLM-OCR") as demo:
     run_btn.click(fn=run_ocr, inputs=file_in, outputs=out)
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()