app.py

import gradio as gr
from Summary import Summary
from NamedEntity import NER

entity_sample_text = \
    ("Mr Roberts had taken his dog for a walk in Hyde Park at around 9pm. "
     "He saw a group of people shouting at Stephen - a guy who would shortly "
     "have his Rolex watch and iPhone stolen by the same group of people "
     "that had surrounded him. A lady named Fiona Walker was crossing the High "
     "Street that runs alongside the park. She heard Mr Roberts shout for help "
     "and called the police to assist.\n\n Constable Robbins arrived after about "
     "20 minutes by which time the group had dispersed. Mr Roberts was able to "
     "give a description of the people who had stolen Stephen's Rolex watch and iPhone. "
     "He said that one of the people was wearing a blue Adidas t-shirt and another "
     "was wearing a red Arsenal football cap. "
     "It turned out the gang members hailed from Paddington and Mayfair and used Uber to "
     "move around the area.\n\n"
     "The gang leader had to appear at "
     "the Old Bailey on 1st January 2021. He was sentenced to 3 years in prison "
     "for robbery and assault by Judge Jennifer Sanderson."
     )

summary_sample_text = \
    ("The City of London, often simply referred to as The City, is a historic and iconic part of "
     "the British capital, London. With a rich history dating back over 2,000 years, it stands as a "
     "testament to the enduring legacy of British culture and finance. Covering an area of "
     "approximately 1.12 square miles (2.9 square kilometers), it may be small in size, but it packs "
     "a punch in terms of its global significance. One of the most notable features of The City is "
     "its status as the financial heart of London and, indeed, the world. The area is home to the "
     "Bank of England, the London Stock Exchange, and numerous multinational banks and financial "
     "institutions. The towering skyscrapers and modern architecture that dot the skyline serve as "
     "a symbol of the city's economic power and influence. The City's historic role in finance "
     "dates back to the Middle Ages when it became the hub of international trade and commerce. "
     "Today, it remains a hub for global finance, attracting professionals from all corners of the "
     "globe. The City's historic and architectural heritage is another captivating aspect. Wandering "
     "through its labyrinthine streets, one can marvel at the blend of old and new. Ancient structures "
     "like the Tower of London and St. Paul's Cathedral coexist with sleek modern office buildings. "
     "The contrast in architectural styles is a testament to London's ability to embrace its rich "
     "history while continually evolving to meet the demands of the future. Culturally, The City "
     "offers a unique blend of tradition and innovation. It hosts various cultural events and "
     "festivals throughout the year, attracting both locals and tourists. The City's vibrant food "
     "scene is another highlight, with a multitude of restaurants catering to diverse tastes, from "
     "classic British fare to international cuisine. Despite its bustling urban environment, The City "
     "also boasts several green spaces. One can escape the hustle and bustle of the financial district "
     "by strolling along the banks of the River Thames, enjoying the lush gardens of Postman's Park, "
     "or exploring the serene Barbican Conservatory. Transportation in The City is well-developed, "
     "making it easily accessible. The London Underground, buses, and extensive pedestrian walkways "
     "ensure that both residents and visitors can navigate the area efficiently. In conclusion, The "
     "City of London is a city within a city, a captivating blend of history, finance, culture, and "
     "architecture. Its enduring importance on the global stage, its rich heritage, and its vibrant "
     "cultural scene make it a must-visit destination for anyone exploring the dynamic and diverse city "
     "of London. Whether you are drawn by its financial prowess, architectural beauty, or cultural "
     "riches, The City has something to offer every visitor, and its enduring appeal is sure to stand "
     "the test of time."
     )


entity_desc = ("This demo uses the [DSLIM BERT model](https://huggingface.co/dslim/bert-base-NER) "
               "to identify named entities in a piece of text. It has been trained to recognise "
               "four types of entities: location (LOC), organisations (ORG), person (PER) and "
               "Miscellaneous (MISC). The model size is approximately 430Mb. \n\n"
               "This model is free for commercial use. \n\n"
               "A [larger model](https://huggingface.co/dslim/bert-large-NER) is also available (~1.3Gb)."
               )

summary_desc = ("This demo uses the "
                "[legal-bert-base-uncased model](https://huggingface.co/nlpaueb/legal-bert-base-uncased) "
                "intended to assist legal NLP research, computational law, and legal technology "
                "applications. The model size is approximately 500Mb. \n\n "
                "The model was trained using 12Gb of diverse English legal text across a number of fields. "
                "This model is free for commercial use. \n\n"
                )


class GlobalVariables:
    def __init__(self):
        self.entities = None
        self.summary = None


app_globals = GlobalVariables()


def process_entities(txt_data):
    if txt_data is None or len(txt_data.strip()) == 0:
        raise gr.Error("Text to analyse cannot be empty")

    app_globals.entities = NER(txt_data)
    app_globals.entities.entity_markdown()

    entity_list = '\n'.join(app_globals.entities.unique_entities)

    heading = 'Entities highlighted in the original text'
    output = f'## {heading} \n\n {app_globals.entities.markdown}'

    return entity_list, output


def session_data(txt_data):
    pass


def process_summary(txt_data):
    if txt_data is None or len(txt_data.strip()) == 0:
        raise gr.Error("Text to summarise cannot be empty")

    app_globals.summary = Summary(txt_data)
    result = app_globals.summary.result

    source_text_length = len(txt_data.split(' '))
    summary_text_length = len(result.split(' '))

    info = 'Words in source text: ' + str(source_text_length)
    info += '\nWords in summary: ' + str(summary_text_length)
    info += ('\nSource text shortened by a factor of: ' +
             str(round(source_text_length/summary_text_length, 1)) + ' times')

    return info, result


with gr.Blocks() as demo:
    # The legal summary appliation tab.
    with gr.Tab('Summaries'):
        gr.Markdown("# Summarising text")
        with gr.Accordion("See Details", open=False):
            gr.Markdown(summary_desc)
        text_summary_source = gr.Textbox(label="Text to summarise", lines=10)
        text_summary = gr.Textbox(label="Summary", lines=3)
        text_info = gr.Textbox(label="Related information", lines=5)

        with gr.Row():
            btn_sample_summary = gr.Button("Load Sample Text")
            btn_clear_summary = gr.Button("Clear Summary Data")
            btn_summary = gr.Button("Get Summary", variant='primary')

        # Event Handler
        btn_sample_summary.click(fn=lambda: summary_sample_text, outputs=[text_summary_source])
        btn_clear_summary.click(fn=lambda: ('', '', ''), outputs=[text_summary_source, text_summary, text_info])
        btn_summary.click(fn=process_summary, inputs=[text_summary_source], outputs=[text_info, text_summary])

    with gr.Tab('Entities'):
        gr.Markdown("# Extracting named entities")
        with gr.Accordion("See Details", open=False):
            gr.Markdown(entity_desc)
        text_source = gr.Textbox(label="Text to analyse", lines=10)
        text_entities = gr.Textbox(label="Unique entities", lines=3)
        mk_output = gr.Markdown(label="Entities Highlighted", value='Highlighted entities appear here')
        with gr.Row():
            btn_sample_entity = gr.Button("Load Sample Text")
            btn_clear_entity = gr.Button("Clear Data")
            btn_entities = gr.Button("Get Entities", variant='primary')

        # Event Handlers
        btn_sample_entity.click(fn=lambda: entity_sample_text, outputs=[text_source])
        btn_entities.click(fn=process_entities, inputs=[text_source], outputs=[text_entities, mk_output])
        btn_clear_entity.click(fn=lambda: ('', '', ''), outputs=[text_source, text_entities, mk_output])

demo.launch()