JetBrains-Research/commit-message-editing-visualization
0
1import diff_match_patch as dmp_module2from tqdm import tqdm3 4from api_wrappers import hf_data_loader5 6 7def get_annotated_diff(start_text, end_text):8 dmp = dmp_module.diff_match_patch()9 dmp_mapping = {-1: "-", 0: None, 1: "+"}10 11 diff = dmp.diff_main(start_text, end_text)12 dmp.diff_cleanupSemantic(diff)13 14 result = [[w, dmp_mapping[t]] for t, w in diff]15 16 return result17 18 19def annotated_diff_for_row(row):20 if "commit_msg_start" in row:21 start = row["commit_msg_start"]22 else:23 start = row["G_text"]24 if "commit_msg_end" in row:25 end = row["commit_msg_end"]26 else:27 end = row["E_text"]28 return get_annotated_diff(start, end)29 30 31def data_with_annotated_diffs():32 tqdm.pandas()33 34 df = hf_data_loader.load_synthetic_as_pandas()35 df = df.loc[df.is_related].copy()36 annotated = df.progress_apply(annotated_diff_for_row, axis=1)37 df["annotated_diff"] = annotated38 return df39 