app.py

import gradio as gr
from NamedEntity import NER

text2 = ("UK law presumes the author of a copyright work to be the first owner. The law "
         "also recognises that copyright works may be the product of joint authors and "
         "co-authors. This means that as soon as the work is created, copyright in the work "
         "belongs to the person or persons who created it. This is true even if the author "
         "was hired to make the copyright work under a contract for services, such as a "
         "wedding photographer. In the absence of an assignment of copyright via contract, "
         "the wedding photographer retains copyright in the photographs, and the happy couple "
         "merely gains physical prints of the pictures, and a right to privacy preventing the "
         "issue, communication or exhibition of copies of the pictures to the public. There "
         "are two exceptions to the presumption of first ownership: 1. The owner of copyright "
         "in a work created by an employee in the course of his employment will be the "
         "employer, unless there is an agreement to the contrary. This applies to literary, "
         "dramatic, musical or artistic works, and films. It is not sufficient for the work "
         "to have been created during working hours by an employee for the employer to own the "
         "work, it must have been created as part of the job that employee was hired to do. "
         "However, the employer may be able to make some claim to the work if the employee should "
         "have been working for the employer at the time when he created the work, or if the "
         "nature or subject matter of the work is so closely related to the type of employment "
         "that the line between employment and private time becomes blurred. For these reasons "
         "it is important to address copyright in employment contracts where employees are likely "
         "to be creating copyright works. 2. Her Majesty the Queen is the first owner of any "
         "copyright in works created by officers or servants of the Crown. This includes any "
         "copyright works created by civil servants, such as this copyright notice."
         )

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."
        )

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"
                )


def process_entities(txt_data):
    ner = NER(txt_data)
    ner.entity_markdown()

    entity_list = '\n'.join(ner.unique_entities)

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

    return entity_list, output


def process_summary(txt_data):
    return 'The Summary'


with gr.Blocks() as demo:
    with gr.Tab('Entities'):
        gr.Markdown("# Named Entity Recognition")
        with gr.Accordion("See Details", open=False):
            gr.Markdown(entity_desc)
        text_source = gr.Textbox(label="Text to analyse", value=text, 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: 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()