app.py

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 get_github_link(repo, hash):
    repo_url = f"https://github.com/{repo}/commit/{hash}"
    return repo_url


def get_diff2html_demo_iframe(github_link):
    diff2html_link = (f"https://diff2html.xyz/demo.html?matching=none"
                      f"&matchWordsThreshold=0.25"
                      f"&maxLineLengthHighlight=10000"
                      f"&diffStyle=word"
                      f"&colorScheme=light"
                      f"&renderNothingWhenEmpty=0"
                      f"&matchingMaxComparisons=2500"
                      f"&maxLineSizeInBlockForComparison=200"
                      f"&outputFormat=line-by-line"
                      f"&drawFileList=1&synchronisedScroll=1"
                      f"&highlight=1"
                      f"&fileListToggle=1"
                      f"&fileListStartVisible=0"
                      f"&highlightLanguages=[object%20Map]"
                      f"&smartSelection=1"
                      f"&fileContentToggle=1"
                      f"&stickyFileHeaders=1"
                      f"&diff={github_link}")

    iframe_html = (f"<iframe "
                   f"src=\"{diff2html_link}\" "
                   f"title=\"diff2html Demo Diff Viewer\" "
                   f"style='width:100%; height:720px; overflow:auto'>"
                   f"</iframe>")

    return iframe_html


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

    record = dataset[sample_ind]
    github_link = get_github_link(record['repo'], record['hash'])

    github_link_md = f"[See the commit on GitHub]({github_link})"
    diff_view = get_diff2html_demo_iframe(github_link)
    commit_msg = record['message']
    repo_val = record['repo']
    hash_val = record['hash']
    return github_link_md, diff_view, commit_msg, repo_val, hash_val


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 gr.Blocks(theme=gr.themes.Soft()) as demo:
    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):
            github_link = gr.Markdown()
            diff_view = gr.HTML()
        with gr.Column(scale=1):
            commit_msg = gr.Textbox(label="Commit message",
                                    interactive=False,
                                    )
            gr.Markdown("## Please, answer the questions below")
            verbosity_feedback = gr.Radio(info='How can you describe the length of the commit message above?',
                                          label='verbosity',
                                          show_label=False,
                                          choices=[
                                              ('Too short', 0),
                                              ('Just right', 1),
                                              ('Too verbose', 2)])
            correctness_feedback = gr.Radio(info='Is the commit message factually correct?',
                                            label='is_correct',
                                            show_label=False,
                                            choices=[
                                                ('Yes', True),
                                                ('No', False)])
            format_feedback = gr.Slider(info='Rate the commit message\'s format (1 - very bad, 5 - very good)',
                                        label='format_score',
                                        show_label=False,
                                        minimum=1,
                                        step=1,
                                        interactive=True,
                                        maximum=5)
            submit_btn = gr.Button("Submit and continue")
            session_val = gr.Textbox(info='Session', interactive=False, container=True, show_label=False,
                                     label='session')

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

    feedback_form = [
        session_val,
        repo_val,
        hash_val,
        verbosity_feedback,
        correctness_feedback,
        format_feedback
    ]

    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])


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

demo.launch()