🐛 Write page_info Parquet file again

src/mods4pandas/mods4pandas.py

@@ -461,18 +461,15 @@ def process(mets_files: List[str], output_file: str, output_page_info: str):
 
     # Convert the mods_info SQL to a pandas DataFrame
     mods_info_df = pd.read_sql_query("SELECT * FROM mods_info", con, index_col="recordInfo_recordIdentifier")
-
-    # Save the DataFrame
     logger.info('Writing DataFrame to {}'.format(output_file))
     mods_info_df.to_parquet(output_file)
 
-    # Convert page_info
+    if output_page_info:
-    # TODO
+          # Convert page_info SQL to a pandas DataFrama
-    # if output_page_info:
+          page_info_df = pd.read_sql_query("SELECT * FROM page_info", con_page_info, index_col=["ppn", "ID"])
-    #     page_info_df = dicts_to_df(page_info, index_column=("ppn", "ID"))
+          # Save the DataFrame
-    #     # Save the DataFrame
+          logger.info('Writing DataFrame to {}'.format(output_page_info))
-    #     logger.info('Writing DataFrame to {}'.format(output_page_info))
+          page_info_df.to_parquet(output_page_info)
-    #     page_info_df.to_parquet(output_page_info)
 
 
 def main():