Upload debug_conversion_v2.py with huggingface_hub

debug_conversion_v2.py

@@ -0,0 +1,41 @@
+from pyhwp.html_converter import HwpToHtmlConverter
+from hwp5.hwp5html import HTMLTransform
+from hwp5.xmlmodel import Hwp5File
+from contextlib import closing
+import os
+import sys
+
+def main():
+    hwp_file = 'test.hwp'
+    output_path = 'test.html'
+    
+    print("Step 1: Init HTMLTransform", flush=True)
+    transformer = HTMLTransform()
+    output_dir = os.path.dirname(os.path.abspath(output_path))
+    
+    print(f"Step 2: Opening {hwp_file}", flush=True)
+    with closing(Hwp5File(hwp_file)) as hwp5file:
+        print("Step 3: Creating temp XHWP5", flush=True)
+        with transformer.transformed_xhwp5_at_temp(hwp5file) as xhwp5path:
+            print(f"Step 4: Temp XHWP5 at {xhwp5path}", flush=True)
+            
+            print("Step 5: Generating HTML...", flush=True)
+            with open(output_path, 'wb') as f:
+                transformer.transform_xhwp5_to_xhtml(xhwp5path, f)
+            print("Step 5: HTML Done", flush=True)
+            
+            print("Step 6: Generating CSS...", flush=True)
+            css_path = os.path.join(output_dir, 'styles.css')
+            with open(css_path, 'wb') as f:
+                transformer.transform_xhwp5_to_css(xhwp5path, f)
+            print("Step 6: CSS Done", flush=True)
+
+        print("Step 7: Extracting BinData...", flush=True)
+        bindata_dir = os.path.join(output_dir, 'bindata')
+        transformer.extract_bindata_dir(hwp5file, bindata_dir)
+        print("Step 7: BinData Done", flush=True)
+
+    print("All Done", flush=True)
+
+if __name__ == "__main__":
+    main()