generate_annotated_diffs.py

import json
from datetime import datetime

import gradio as gr

import hf_data_loader


def group_changes(changes):
    groups = {}
    for change in changes:
        group = datetime.fromisoformat(change['ts'])
        if group not in groups:
            groups[group] = []
        groups[group].append(change)

    grouped_changes = []
    for group in sorted(groups.keys()):
        groups[group].sort(key=lambda x: x['p'])
        grouped_changes.append(groups[group])

    return grouped_changes


def get_annotated_diff(initial_text, changes):
    grouped_changes = group_changes(changes)
    text = [(c, None) for c in initial_text]
    for change_group in grouped_changes:
        next_text = []
        text_pointer = 0
        real_text_ind = 0
        change_pointer = 0
        while text_pointer < len(text):
            if change_pointer >= len(change_group) or real_text_ind < change_group[change_pointer]['p']:
                next_text.append(text[text_pointer])
                real_text_ind += 1
                text_pointer += 1
            elif change_group[change_pointer]['t'] == '+':
                if not (text[text_pointer][1] == '-' and text[text_pointer][0] == change_group[change_pointer]['c']):
                    next_text.append((change_group[change_pointer]['c'], '+'))
                else:
                    text_pointer += 1

                real_text_ind += 1
                change_pointer += 1
            elif change_group[change_pointer]['t'] == '-':
                if not (text[text_pointer][1] == '+' and text[text_pointer][0] == change_group[change_pointer]['c']):
                    next_text.append((text[text_pointer][0], '-'))
                    text_pointer += 1

                real_text_ind += 1
                change_pointer += 1
            else:
                raise RuntimeError("Unexpected branch")
        text = next_text
    return text


def annotated_diff_for_row(row):
    start = row['commit_msg_start']
    changes = json.loads(row['commit_msg_history'])
    return get_annotated_diff(start, changes)


if __name__ == '__main__':
    df = hf_data_loader.load_raw_dataset_as_pandas()
    annotated = df.apply(annotated_diff_for_row, axis=1)
    with gr.Blocks(theme=gr.themes.Soft()) as application:
        gr.Highlightedtext(value=annotated[0], combine_adjacent=True, color_map={'+': "green", '-': "red"})
        gr.Markdown(value=df.iloc[0]['commit_msg_start'])
        gr.Markdown(value=df.iloc[0]['commit_msg_end'])
    application.launch()