app.py

from transformers import pipeline
import matplotlib.pyplot as plt
import twitter_scraper as ts
import pandas as pd

import gradio as gr

pretrained_sentiment = "w11wo/indonesian-roberta-base-sentiment-classifier"
pretrained_ner = "cahya/bert-base-indonesian-NER"

sentiment_pipeline = pipeline(
    "sentiment-analysis",
    model=pretrained_sentiment,
    tokenizer=pretrained_sentiment,
    return_all_scores=True
)

ner_pipeline = pipeline(
    "ner",
    model=pretrained_ner,
    tokenizer=pretrained_ner,
    grouped_entities=True
)

examples = [
    "Jokowi sangat kecewa dengan POLRI atas kerusuhan yang terjadi di Malang",
    "Lesti marah terhadap perlakuan KDRT yang dilakukan oleh Bilar",
    "Ungkapan rasa bahagia diutarakan oleh Coki Pardede karena kebabasannya dari penjara"
]

def sentiment_analysis(text):
    output = sentiment_pipeline(text)
    return {elm["label"]: elm["score"] for elm in output[0]}

def ner(text):
    output = ner_pipeline(text)
    for elm in output:
        elm['entity'] = elm['entity_group']
    return {"text": text, "entities": output}

def sentiment_ner(text):
    return sentiment_analysis(text), ner(text)

def sentiment_df(df):
    text_list = list(df["Text"].astype(str).values)
    result = [sentiment_analysis(text) for text in text_list]
    labels = []
    scores = []
    for pred in result:
        idx = list(pred.values()).index(max(list(pred.values())))
        labels.append(list(pred.keys())[idx])
        scores.append(round(list(pred.values())[idx], 3))
    df['Label'] = labels
    df['Score'] = scores
    return df

def ner_df(df):
    text_list = list(df["Text"].astype(str).values)
    label_list = list(df["Label"].astype(str).values)
    result = [ner(text) for text in text_list]
    terms = []
    sentiments = []
    ent = ['PER', 'NOR']
    for i, preds in enumerate(result):
        for pred in preds['entities']:
            if pred['entity_group'] in ent:
                terms.append(pred['word'])
                sentiments.append(label_list[i])
    df_ner = pd.DataFrame(columns=['Entity', 'Sentiment'])
    df_ner['Entity'] = terms
    df_ner['Sentiment'] = sentiments
    return df_ner


def twitter_analyzer(keyword, max_tweets):
    df = ts.scrape_tweets(keyword, max_tweets=max_tweets)
    df["Text"] = df["Text"].apply(ts.preprocess_text)
    df = sentiment_df(df)

    # df_ner = ner_df(df)
    # df_ner = df_ner[df_ner.Entity != keyword]

    fig = plt.figure()
    df.groupby(["Label"])["Text"].count().plot.pie(autopct="%.1f%%", figsize=(6,6))
    return fig, df[["URL", "Text", "Label", "Score"]]

if __name__ == "__main__":

    with gr.Blocks() as demo:

        gr.Markdown("""<h1 style="text-align:center">Entity Based Sentiment Analysis Indonesia</h1>""")

        gr.Markdown(
            """
            Creator: Wira Indra Kusuma
            """
            )

        with gr.Tab("Single Input"):
            with gr.Blocks():
                with gr.Row():
                    with gr.Column():
                        input_text = gr.Textbox(label="Input Text")
                        analyze_button = gr.Button(label="Analyze")
                        examples_bar = gr.Examples(examples=examples, inputs=input_text)
                    with gr.Column():
                        sent_output = gr.Label(label="Sentiment Analysis")
                        ner_output = gr.HighlightedText(label="Named Entity Recognition")

        with gr.Tab("Twitter"):
            with gr.Blocks():
                with gr.Row():
                    with gr.Column():
                        keyword_textbox = gr.Textbox(lines=1, label="Keyword")
                        max_tweets_component = gr.Number(value=10, label="Total of Tweets to Scrape", precision=0)
                        submit_button = gr.Button("Submit")

                    plot_component = gr.Plot(label="Pie Chart of Sentiments")
                dataframe_component = gr.DataFrame(type="pandas",
                                                label="Dataframe",
                                                max_rows=(20,'fixed'),
                                                overflow_row_behaviour='paginate',
                                                wrap=True)

                # df_ner = gr.DataFrame(type="pandas",
                #                                 label="Dataframe",
                #                                 max_rows=(20,'fixed'),
                #                                 overflow_row_behaviour='paginate',
                #                                 wrap=True)


        analyze_button.click(sentiment_ner, input_text, [sent_output, ner_output])
        submit_button.click(twitter_analyzer,
                    inputs=[keyword_textbox, max_tweets_component],
                    outputs=[plot_component, dataframe_component])

        gr.Markdown(
                """

                """

            )

    demo.launch(inbrowser=True)