app.py

import os
import threading
from pathlib import Path

try:
    from huggingface_hub import CommitScheduler
except Exception:  # pragma: no cover
    CommitScheduler = None  # type: ignore


class _LocalScheduler:
    """Fallback logger lock when HF_TOKEN is unavailable."""

    def __init__(self):
        self.lock = threading.Lock()


def make_scheduler(repo_id: str, log_folder: Path):
    log_folder.mkdir(parents=True, exist_ok=True)
    token = os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACE_HUB_TOKEN")
    if token and CommitScheduler is not None:
        try:
            return CommitScheduler(
                repo_id=repo_id,
                repo_type="dataset",
                folder_path=log_folder,
                path_in_repo="data",
                every=2,
                token=token,
            )
        except Exception as exc:
            print(f"CommitScheduler disabled: {exc}")
    return _LocalScheduler()


# Use a pipeline as a high-level helper
from transformers import pipeline
import torch
import gradio as gr

from pathlib import Path

import os
import uuid
import joblib
import json

# Prepare the logging functionality

log_file = Path("logs/") / f"data_{uuid.uuid4()}.json"
log_folder = log_file.parent

scheduler = make_scheduler("text-summarization-logs", log_folder)


text_summary = None

def get_summarizer():
    global text_summary
    if text_summary is None:
        text_summary = pipeline("summarization", model="sshleifer/distilbart-cnn-12-6")
    return text_summary

# model_path = "Model/models--sshleifer--distilbart-cnn-12-6/snapshots/a4f8f3ea906ed274767e9906dbaede7531d660ff"

# text_summary = pipeline("summarization", model=model_path, torch_dtype=torch.bfloat16)

# text="Why You Can Trust Forbes Advisor Small Business \
# The Forbes Advisor Small Business team is committed to bringing you unbiased \
# rankings and information with full editorial independence. We use product data, \
# strategic methodologies and expert insights to inform all of our content and guide \
# you in making the best decisions for your business journey.\
# We reviewed 11 systems to help you find the best blogging platform for your blog or \
# small business. Our ratings looked at factors that included the platform’s starting \
# price (including whether it offered a free trial or free version); useful general features,\
# such as drag-and-drop functionality and search engine optimization (SEO) tools; unique features, \
# how well the blogging platform fared on third-party review sites and a final review by our experts.\
# All ratings are determined solely by our editorial team."

# print(text_summary(text)[0])

def summary(input):
    output = get_summarizer()(input)

    with scheduler.lock:
        with log_file.open("a") as f:
            f.write(json.dumps(
                {
                    'Input Text': input,
                    'Summary':output[0]['summary_text']
                }
            ))
            f.write("\n")
    
    return output[0]['summary_text']

gr.close_all()

# demo = gr.Interface(fn=summary,inputs="text",outputs='text',title='Text Summarization Gradio Huggingface')
demo = gr.Interface(fn=summary,
                    inputs=[gr.Textbox(label="Input text to summarization", lines=6)],
                    outputs=[gr.Textbox(label="Summarized text", lines=4)],
                    title='Text Summarization',
                    description='This application will be used to summarize the text',
                    theme=gr.themes.Soft(),
                    concurrency_limit=16)

demo.launch(share=True)