app.py

from datetime import datetime
import os
import random
import uuid

import gradio as gr
from datasets import load_dataset

HF_TOKEN = os.environ.get('HF_TOKEN')
HF_DATASET = os.environ.get('HF_DATASET')

configuration = "commitchronicle-py-long"  # select a configuration
dataset = load_dataset("JetBrains-Research/lca-cmg",
                       configuration,
                       split="test",
                       cache_dir="data")
n_samples = len(dataset)

saver = gr.HuggingFaceDatasetSaver(HF_TOKEN, HF_DATASET, private=True)


def convert_diff_to_unified(diff):
    result = "\n".join(
        [
            f'--- {modified_file["old_path"]}\n'
            f'+++ {modified_file["new_path"]}\n'
            f'{modified_file["diff"]}'
            for modified_file in diff
        ]
    )

    return result


def get_diff2html_view(raw_diff):
    html = f"""
    <div style='width:100%; height:720px; overflow:auto; position: relative'>
        <div id='diff-raw' hidden>{raw_diff}</div> 
        <div class="d2h-view-wrapper">
            <div id='diff-view'></div>
        </div>
    </div>
    """

    return html


def update_commit_view(sample_ind):
    if sample_ind >= n_samples:
        return None

    record = dataset[sample_ind]

    diff_view = get_diff2html_view(convert_diff_to_unified(record['mods']))
    commit_msg = record['message']
    repo_val = record['repo']
    hash_val = record['hash']
    diff_loaded_timestamp = datetime.now().isoformat()
    return diff_view, commit_msg, repo_val, hash_val, diff_loaded_timestamp


def next_sample(current_sample_ind, shuffled_idx):
    if current_sample_ind == n_samples:
        return None

    current_sample_ind += 1
    updated_view = update_commit_view(shuffled_idx[current_sample_ind])
    return (current_sample_ind,) + updated_view


with open("head.html") as head_file:
    head_html = head_file.read()

with gr.Blocks(theme=gr.themes.Soft(), head=head_html, css="style_overrides.css") as application:
    repo_val = gr.Textbox(interactive=False, label='repo', visible=False)
    hash_val = gr.Textbox(interactive=False, label='hash', visible=False)
    shuffled_idx_val = gr.JSON(visible=False)

    with gr.Row():
        current_sample_sld = gr.Slider(minimum=0, maximum=n_samples, step=1,
                                       value=0,
                                       interactive=False,
                                       label='sample_ind',
                                       info=f"Samples labeled/skipped (out of {n_samples})",
                                       show_label=False,
                                       container=False,
                                       scale=5)
        with gr.Column(scale=1):
            skip_btn = gr.Button("Skip the current sample")
    with gr.Row():
        with gr.Column(scale=2):
            diff_view = gr.HTML()
        with gr.Column(scale=1):
            commit_msg = gr.Textbox(label="Commit message",
                                    interactive=False,
                                    )
            gr.Markdown("## Please, rate your level of agreement with each statement\n"
                        "\n"
                        "*1 - strongly disagree, 2 - disagree, 3 - agree, 4 - strongly agree*")

            is_correct = gr.Slider(
                info='The information provided in the commit message is consistent with the code changes.',
                label='is_correct',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            has_what = gr.Slider(
                info='The commit message answers the question of WHAT changes have been made.',
                label='has_what',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            has_why = gr.Slider(
                info='The commit message answers the question of WHY these changes have been made.',
                label='has_why',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            is_not_verbose = gr.Slider(
                info='The commit message can be substantially shortened without loss of important information.',
                label='is_not_verbose',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            has_headline = gr.Slider(
                info='The commit message includes a short headline that provides a good overview of the changes.',
                label='has_headline',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            easy_to_read = gr.Slider(
                info='The commit message is easy to read and to understand.',
                label='easy_to_read',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=4)

            overall_rating = gr.Slider(
                info='Please, describe your overall impression of the commit message (1 - very bad, 5 - very good)',
                label='overall_rating',
                show_label=False,
                minimum=1,
                step=1,
                interactive=True,
                maximum=5)

            comments = gr.Textbox(
                info='Additional comments on the commit message',
                label='comments',
                show_label=False,
                interactive=True)

            submit_btn = gr.Button("Submit and continue")
            session_val = gr.Textbox(info='Session', interactive=False, container=True, show_label=False,
                                     label='session')

            with gr.Row(visible=False):
                sample_loaded_timestamp = gr.Textbox(info="Sample loaded", label='loaded_ts', interactive=False,
                                                     container=True, show_label=False)
                sample_submitted_timestamp = gr.Textbox(info="Current time",
                                                        interactive=False, container=True, show_label=False,
                                                        value=lambda: datetime.now().isoformat(), every=1.0,
                                                        label='submitted_ts')

    commit_view = [
        diff_view,
        commit_msg,
        repo_val,
        hash_val,
        sample_loaded_timestamp
    ]

    feedback_form = [
        session_val,
        repo_val,
        hash_val,
        sample_loaded_timestamp,
        sample_submitted_timestamp,
        is_correct,
        has_what,
        has_why,
        is_not_verbose,
        has_headline,
        easy_to_read,
        overall_rating,
        comments
    ]

    saver.setup([current_sample_sld] + feedback_form, "feedback")

    skip_btn.click(next_sample, inputs=[current_sample_sld, shuffled_idx_val],
                   outputs=[current_sample_sld] + commit_view)


    def submit(current_sample, shuffled_idx, *args):
        saver.flag((current_sample,) + args)
        return next_sample(current_sample, shuffled_idx)


    submit_btn.click(submit, inputs=[current_sample_sld, shuffled_idx_val] + feedback_form,
                     outputs=[current_sample_sld] + commit_view)


    def init_session(current_sample):
        session = str(uuid.uuid4())
        shuffled_idx = list(range(n_samples))
        random.shuffle(shuffled_idx)
        return (session, shuffled_idx) + update_commit_view(shuffled_idx[current_sample])


    application.load(init_session,
                     inputs=[current_sample_sld],
                     outputs=[session_val, shuffled_idx_val] + commit_view, )

application.launch()