A question about the function generate_tgt_parse() in data-processing.py

[XZhang00]

The weighted ROUGE scores should been calculated by two parses, but in the function generate_tgt_parse() the code is "possible_drawn = step2_rouge(level_, freq, src_lines[i], level)[3]". Is the src_lines[i] parameter an error and should be replaced with src_pure_parses? Are there any errors in the results obtained in the paper?
Looking forward to your reply!