app.py

# GNU
import gradio as gr
import PyPDF2
import yake

demo = gr.Blocks()

def generate(input):
    # Step 1: Import

    pdfFileObj = open(input.name, 'rb')

    pdfReader = PyPDF2.PdfReader(pdfFileObj)
    
    length = len(pdfReader.pages)

    f = open("t.txt", "w")

    pages = []

    for x in range(length):
        pages.append(pdfReader.pages[x].extract_text())

    f.writelines(pages)

    # Step 2: Process for Keywords
    r = open("t.txt", "r")
    read = r.read()

    kw_extractor = yake.KeywordExtractor(top = 50)
    keywords = kw_extractor.extract_keywords(read)

    kw_list = []
    for kw in keywords:
        kw_list.append(kw[0])

    # Step 3: Process for Assignment
    output = {}
    for kw in kw_list:
        output[kw] = []

    for x in range(length):
        pageText = pdfReader.pages[x].extract_text()

        for kw in kw_list:
            if pageText.find(kw) != -1:
                output[kw].append(x+1)


    # Step 4: Output List

    clean_output = {}
    for kw in kw_list:
        clean_output[kw] = []

    for kw in output:
        clean_pages = []
        if (len(output[kw]) == 0):
            print('detected but no pages!:', kw, output[kw])
            continue
        start = output[kw][0]
        end = output[kw][0]

        for num in output[kw]:
            if num > end + 1:
                if start == end:
                    clean_pages.append(str(start))
                else:
                    clean_pages.append(f'{start}-{end}')
                start = num
            end = num

        if start == end:
            clean_pages.append(str(start))
        else:
            clean_pages.append(f'{start}-{end}')
        
        clean_output[kw] = clean_pages

    return str(clean_output)


with demo:

    gr.Markdown("# PDF to Index")

    with gr.Column():

        gr.Markdown("## Load Inputs")

        uploaded_file = gr.File(
            label="Upload a PDF file",
            file_count="single",
            type="file"
        )

        gr.Markdown("---")

    with gr.Column():
            gr.Markdown("## Index From PDF")
            convert_button = gr.Button("Index From PDF!", variant="primary")
            out_placeholder = gr.HTML("<p><em>Output will appear below, with PyPDF2 for preprocessing and yake for processing:</em></p>")
            gr.Markdown("### Raw Index")
            index = gr.Textbox(
                label="Raw Index", placeholder="The index will appear here"
            )

    convert_button.click(
        fn=generate,
        inputs=[uploaded_file],
        outputs=[index],
    )

demo.launch()