Update extract_pdf_data.py

extract_pdf_data.py

@@ -30,4 +30,8 @@ def extract_pdf_full_text(pdf_path, txt_path):
     print(f"✅ Saved deduped full text to {txt_path}")
 
 if __name__ == "__main__":
-    extract_pdf_full_text("test1.pdf", "pdf_all_text_full.txt")
+    import sys
+    # Usage: python extract_pdf_data.py input.pdf output.txt
+    input_pdf = sys.argv[1]
+    output_txt = sys.argv[2]
+    extract_pdf_full_text(input_pdf, output_txt)