app.py

import gradio as gr
import requests
import emoji
import re

API_URL = "https://api-inference.huggingface.co/models/Dabid/test2"
headers = {"Authorization": "Bearer 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"}

profanities = ['bobo', 'bobong', 'bwiset', 'bwisit', 'buwisit', 'buwiset', 'bwesit', 'gago', 'gagong', 'kupal',
               'pakshet', 'pakyu', 'pucha', 'puchang',
               'punyeta', 'punyetang', 'puta', 'putang', 'putangina', 'putanginang', 'tanga', 'tangang', 'tangina',
               'tanginang', 'tarantado', 'tarantadong', 'ulol']

contractions = {
    'di': 'hindi',
    'to': 'ito',
    'no': 'ano',
    'kundi': 'kung hindi',
    'nya': 'niya',
    'nyo': 'ninyo',
    'niyo': 'ninyo',
    'pano': 'paano',
    'sainyo': 'sa inyo',
    'sayo': 'sa iyo',
    'pag': 'kapag',
    'kesa': 'kaysa',
    'dun': 'doon',
    'ganto': 'ganito',
    'nandun': 'nandoon',
    'saka': 'tsaka',
    'ung': 'yung',
    'wag': 'huwag',
    'sya': 'siya',
    'bat': 'bakit',
    'yon': 'iyon',
    'yun': 'iyon',
    'dyan': 'diyan',
    'jan': 'diyan',
    'andito': 'nandito',
    'tanginamo': 'tangina mo',
    'putanginamo': 'putangina mo',
    'san': 'saan',
    'ganun': 'ganoon',
    'gagong': 'gago na',
    'bobong': 'bobo na',
    'tangang': 'tanga na',
    'kelan': 'kailan',
    'raw': 'daw',
    'tanginang': 'tangina na',
    'tarantadong': 'tarantado na',
    'putang ina': 'putangina',
    'putang inang': 'putangina',
    'putanginang': 'putangina',
    'itong': 'ito ang',
    'lng': 'lang',
    'bwisit': 'bwiset',
    'bwesit': 'bwiset',
    'buwisit': 'bwiset',
    'buwesit': 'bwiset'
}


def preprocess(row):
    laugh_texts = ['hahaha', 'wahaha', 'hahaa', 'ahha', 'haaha', 'hahah', 'ahah', 'hha']
    symbols = ['@', '#']

    # Lowercase
    row = row.lower()

    # Remove emojis
    row = emoji.replace_emoji(row, replace='')

    # Replace elongated words 'grabeee' -> 'grabe' (not applicable on 2 corresponding letter)
    row = re.sub(r'(.)\1{2,}', r'\1', row)

    # Split sentence into list of words
    row_split = row.split()

    for index, word in enumerate(row_split):

        # Remove words with symbols (e.g. @username, #hashtags)
        if any(x in word for x in symbols):
            row_split[index] = ''

        # Remove links
        if 'http' in word:
            row_split[index] = ''

        # Unify laugh texts format to 'haha'
        if any(x in word for x in laugh_texts):
            row_split[index] = 'haha'

        # Remove words with digits (4ever)
        if any(x.isdigit() for x in word):
            row_split[index] = ''

    # Combine list of words back to sentence
    combined_text = ' '.join(filter(None, row_split))

    # Check if output contains single word then return null
    if len(combined_text.split()) == 1:
        return combined_text

    # Filter needed characters
    combined_text = re.sub(r"[^A-Za-z ]+", '', combined_text)

    # Expand Contractions
    for i in contractions.items():
        combined_text = re.sub(rf"\b{i[0]}\b", i[1], combined_text)

    return combined_text


def query(payload):
    response = requests.post(API_URL, headers=headers, json=payload)
    return response.json()


def predict(text):
    output = query(preprocess(text))
    print(preprocess(text))
    
    if 'error' in output:
        return output['error'], 'Error occured. Try again later.'
    else:
        output = [tuple(i.values()) for i in output[0]]
        output = dict((x, y) for x, y in output)

        predicted_label = list(output.keys())[0]

        if predicted_label == 'Abusive':
            output_text = text
            for profanity in profanities:
                compiled = re.compile(re.escape(profanity), re.IGNORECASE)
                mask = ""
                for i in profanity:
                    mask += "*" if i != " " else " "
                output_text = compiled.sub(mask, output_text)
            return output, output_text
        else:
            return output, text
    
    


hf_writer = gr.HuggingFaceDatasetSaver('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', 'tagalog-profanity-feedbacks')


demo = gr.Interface(
    fn=predict,

    inputs=[gr.components.Textbox(lines=5, placeholder='Enter your input here', label='INPUT')],

    outputs=[gr.components.Label(num_top_classes=2, label="PREDICTION"),
             gr.components.Text(label='OUTPUT')],

    examples=['Tangina mo naman sobrang yabang mo gago!!😠😤 @davidrafael',
              'Napakainit ngayong araw pakshet namaaan!!',
              'Napakabagal naman ng wifi tangina #PLDC #HelloDITO',
              'Bobo ka ba? napakadali lang nyan eh... 🤡',
              'Uy gago laptrip yung nangyare samen kanina HAHAHA😂😂'],

    allow_flagging="manual",
    flagging_callback=hf_writer,
    flagging_options=['Good bot', 'Bad bot']
)

demo.launch()