Create app.py

app.py

@@ -0,0 +1,88 @@
+import gradio as gr
+import base64
+import os 
+api_key = os.getenv('API_KEY')
+
+def predict(input, file_input):
+    print("input:", input)
+    print("file_input:", file_input.name)
+    from gradio_client import Client
+    
+    client = Client(api_key)
+    extract_result = client.predict(
+        input,
+        file_input.name,
+        fn_index=1
+    )
+    if extract_result:
+        print(extract_result)
+        return extract_result
+    else:
+        return "Too many user, please wait a monument!"
+
+
+def view_pdf(pdf_file):
+    with open(pdf_file.name, 'rb') as f:
+        pdf_data = f.read()
+    # print("pdf_file", pdf_file)
+    # pdf_data = pdf_file
+    b64_data = base64.b64encode(pdf_data).decode()
+    # print("b64_data", b64_data)
+    return f"<embed src='data:application/pdf;base64,{b64_data}' type='application/pdf' width='100%' height='700px' />"
+
+
+en_1 = ["""could you please help me extract the information of 'title'/'journal'/'year'/'author'/'institution'/'email' from the previous content in a markdown table format?
+If any of this information was not available in the paper, please replaced it with the string `""`, If the property contains multiple entities, please use a list to contain.
+"""]
+
+en_2 = ["""could you please help me extract the information of 'title'/'journal'/'year'/'author'/'institution'/'email' from the previous content in a json format?
+If any of this information was not available in the paper, please replaced it with the string `""`, If the property contains multiple entities, please use a list to contain.
+"""]
+
+examples = [en_1, en_2]
+
+with gr.Blocks(title="ChatPaperGPT") as demo:
+    gr.Markdown(
+        '''<p align="center" width="100%">
+        <img src="https://big-cheng.com/img/pdf.png" alt="pdf-logo" width="50"/>
+        <p>
+
+        <h1 align="center"> Paper Extract GPT </h1>
+        <p> How to use:
+        <br> <strong>#1</strong>: Upload your pdf.
+        <br> <strong>#2</strong>: Click the View PDF button to view it.
+        <br> <strong>#3</strong>: Enter your extraction prompt in the input box (of course, you can click example to test).
+        <br> <strong>#4</strong>: Click Generate to extract, and the extracted information will be displayed in markdown form.
+        </p>
+        '''
+    )
+    with gr.Row():
+        with gr.Column():
+            gr.Markdown('## Upload PDF')
+            file_input = gr.File(type="filepath")
+            viewer_button = gr.Button("View PDF")
+            file_out = gr.HTML()
+        with gr.Column():
+            with gr.Row():
+                model_input = gr.Textbox(lines=7, placeholder='Input prompt about extract information from paper',
+                                         label='Input')
+            with gr.Row():
+                gen = gr.Button("Generate")
+                clr = gr.Button("Clear")
+            example = gr.Examples(examples=examples, inputs=model_input)
+
+            with gr.Row():
+                outputs = gr.Markdown(label='Output', show_label=True, value="""| Title                                       | Journal            | Year | Author                                        | Institution                                           | Email                 |
+|---------------------------------------------|--------------------|------|-----------------------------------------------|-------------------------------------------------------|-----------------------|
+| Paleomagnetic Study of Deccan Traps from Jabalpur to Amarkantak, Central India | J. Geomag. Geoelectr. | 1973 | R. K. VERMA, G. PULLAIAH, G.R. ANJANEYULU, P. K. MALLIK | National Geophysical Research Institute, Hyderabad, and Indian School o f Mines, Dhanbad | "" |
+""")
+
+    inputs = [model_input, file_input]
+    gen.click(fn=predict, inputs=inputs, outputs=outputs)
+    clr.click(fn=lambda value: [gr.update(value=""), gr.update(value="")], inputs=clr,
+              outputs=[model_input, outputs])
+
+    viewer_button.click(view_pdf, inputs=file_input, outputs=file_out)
+    # parser_button.click(extract_text, inputs=file_input, outputs=[xml_out, md_out, rich_md_out])
+
+demo.launch()