app.py

from src.pipelines.completePipeline import Pipeline
import gradio as gr
import spaces
import os

# os.system("apt-get update -y")
# os.system("apt-get upgrade -y")
# os.system("apt install poppler-utils -y")

chain = None
pipeline = Pipeline()

@spaces.GPU
def getTextResponse(text: str, inputQuery: str):
    global chain
    if chain is None:
        chain = pipeline.plainText(text = text)
    else:
        pass
    response = chain.invoke(
        {
            "question": inputQuery
        }
    )
    return response


@spaces.GPU
def getSearchablePdfResponse(path: str, inputQuery: str):
    global chain
    if chain is None:
        chain = pipeline.searchablePdf(path = path)
    else:
        pass
    response = chain.invoke(
        {
            "question": inputQuery
        }
    )
    return response

@spaces.GPU
def getScannablePdfResponse(path: str, inputQuery: str):
    global chain
    if chain is None:
        chain = pipeline.scannablePdf(path = path)
    else:
        pass
    response = chain.invoke(
        {
            "question": inputQuery
        }
    )
    return response

def clearFunction():
    global chain
    chain = None

with gr.Blocks() as textInterface:
    with gr.Row():
        inputText = gr.Textbox(
            label = "Input Text",
            placeholder = "Enter you text here"
        )
    with gr.Row():
        question = gr.Textbox(
            label = "Question",
            placeholder = "Enter your question here"
        )
        answer = gr.Textbox(
            label = "Response",
            interactive = False
        )
    with gr.Row():
        submitButton = gr.Button(
            value = "Submit",
            variant = "primary"
        )
        clearButton = gr.ClearButton(
            components = [inputText, question, answer],
            value = "Clear",
            variant = "secondary"
        )
    submitButton.click(
        fn = getTextResponse,
        inputs = [inputText, question],
        outputs = [answer]
    )
    clearButton.click(
        fn = clearFunction
    )


with gr.Blocks() as searchablePdf:
    with gr.Row():
        inputFile = gr.File(
            file_types = [".pdf"],
            file_count = "single",
            label = "Select PDF"
        )
    with gr.Row():
        question = gr.Textbox(
            label = "Question",
            placeholder = "Enter your question here"
        )
        answer = gr.Textbox(
            label = "Response",
            interactive = False
        )
    with gr.Row():
        submitButton = gr.Button(
            value = "Submit",
            variant = "primary"
        )
        clearButton = gr.ClearButton(
            components = [inputFile, question, answer],
            value = "Clear",
            variant = "secondary"
        )
    submitButton.click(
        fn = getSearchablePdfResponse,
        inputs = [inputFile, question],
        outputs = [answer]
    )
    clearButton.click(
        fn = clearFunction
    )


with gr.Blocks() as scannablePdf:
    with gr.Row():
        inputFile = gr.File(
            file_types = [".pdf"],
            file_count = "single",
            label = "Select PDF"
        )
    with gr.Row():
        question = gr.Textbox(
            label = "Question",
            placeholder = "Enter your question here"
        )
        answer = gr.Textbox(
            label = "Response",
            interactive = False
        )
    with gr.Row():
        submitButton = gr.Button(
            value = "Submit",
            variant = "primary"
        )
        clearButton = gr.ClearButton(
            components = [inputFile, question, answer],
            value = "Clear",
            variant = "secondary"
        )
    submitButton.click(
        fn = getScannablePdfResponse,
        inputs = [inputFile, question],
        outputs = [answer]
    )
    clearButton.click(
        fn = clearFunction
    )


def getLinksButtonFn(baseUrl: str):
    links = pipeline.webCrawler.getLinks(url = baseUrl)
    checkboxes = gr.CheckboxGroup(
        choices = links,
        label = "Fetched Links",
        visible = True
    )
    row2 = gr.Row(visible = True)
    row3 = gr.Row(visible = True)
    return (
        checkboxes,
        row2,
        row3
    )

def getWebsiteResponse(links: list[str], inputQuery: str):
    global chain
    if chain is None:
        print(links)
        chain = pipeline.webCrawl(urls = links)
    else:
        pass
    response = chain.invoke(
        {
            "question": inputQuery
        }
    )
    return response

def clearWebsiteResponse():
    global chain
    chain = None
    checkboxes = gr.CheckboxGroup(
        choices = [],
        label = "Fetched Links",
        visible = False
    )
    return checkboxes

with gr.Blocks() as websiteCrawler:
    with gr.Row():
        inputUrl = gr.Textbox(
            label = "Base URL",
            placeholder = "Enter the Base URL to fetch other links",
            scale = 3
        )
        getLinksButton = gr.Button(
            value = "Get Links",
            variant = "primary",
            scale = 1
        )
    checkboxes = gr.CheckboxGroup(
        choices = [],
        label = "Fetched Links",
    )
    with gr.Row(visible = False) as row2:
        question = gr.Textbox(
            label = "Question",
            placeholder = "Enter your question here"
        )
        answer = gr.Textbox(
            label = "Response",
            interactive = False
        )
    with gr.Row(visible = False) as row3:
        submitButton = gr.Button(
            value = "Submit",
            variant = "primary"
        )
        clearButton = gr.ClearButton(
            components = [question, answer],
            value = "Clear",
            variant = "secondary"
        )
    getLinksButton.click(
        fn = getLinksButtonFn,
        inputs = [inputUrl],
        outputs = [checkboxes, row2, row3]
    )
    submitButton.click(
        fn = getWebsiteResponse,
        inputs = [checkboxes, question],
        outputs = [answer]
    )
    clearButton.click(
        fn = clearWebsiteResponse,
        inputs = None,
        outputs = [checkboxes]
    )


def getYoutubeResponse(links: str, inputQuery: str):
    global chain
    links = [link.strip() for link in links.split(",")]
    if chain is None:
        chain = pipeline.youtubeLinks(urls = links)
    else:
        pass
    response = chain.invoke(
        {
            "question": inputQuery
        }
    )
    return response


with gr.Blocks() as youtubeInterface:
    with gr.Row():
        inputLinks = gr.Textbox(
            label = "Youtube Links",
            placeholder = 'Enter comma(,)-separated youtube video links'
        )
    with gr.Row():
        question = gr.Textbox(
            label = "Question",
            placeholder = "Enter your question here"
        )
        answer = gr.Textbox(
            label = "Response",
            interactive = False
        )
    with gr.Row():
        submitButton = gr.Button(
            value = "Submit",
            variant = "primary"
        )
        clearButton = gr.ClearButton(
            components = [inputLinks, question, answer],
            value = "Clear",
            variant = "secondary"
        )
    submitButton.click(
        fn = getYoutubeResponse,
        inputs = [inputLinks, question],
        outputs = [answer]
    )
    clearButton.click(
        fn = clearFunction
    )


application = gr.TabbedInterface(
    [textInterface, searchablePdf, scannablePdf, websiteCrawler, youtubeInterface],
    ["Text", "Searchable PDF", "Scannable PDF", "Website Text", "Youtube Transcripts"]
)

application.launch()