Spaces:
Running
Running
| import fitz # PyMuPDF | |
| import openai | |
| import gradio as gr | |
| class IncompleteSentenceFinder: | |
| """ | |
| This class finds and displays incomplete sentences in a PDF document using OpenAI's GPT-3. | |
| Args: | |
| api_key (str): Your OpenAI API key. | |
| """ | |
| def __init__(self): | |
| """ | |
| Initialize the IncompleteSentenceFinder with the PDF file and OpenAI API key. | |
| Args: | |
| api_key (str): Your OpenAI API key. | |
| """ | |
| # openai.api_key = openai_api_key | |
| pass | |
| def _check_incomplete_sentence(self, text: str) -> str: | |
| """ | |
| Use OpenAI's GPT-3 to identify incomplete sentences in the given text. | |
| Args: | |
| text (str): Text to check for incomplete sentences. | |
| Returns: | |
| str: Incomplete sentences identified by GPT-3. | |
| """ | |
| # Create a request to OpenAI's GPT-3 engine to identify incomplete sentences. | |
| response = openai.Completion.create( | |
| engine="text-davinci-003", | |
| prompt=f"list out the incomplete sentences in the following text:\n{text}", | |
| max_tokens=1000, | |
| ) | |
| # Extract and strip the text of identified incomplete sentences from the GPT-3 response. | |
| incomplete_sentences = response.choices[0].text.strip() | |
| print("incomplete_sentences Extracted Successfully!") | |
| return incomplete_sentences | |
| def get_incomplete_sentence(self,pdf_file) -> str: | |
| """ | |
| Extract text from the PDF document and find incomplete sentences. | |
| Returns: | |
| str: Incomplete sentences identified by GPT-3. | |
| """ | |
| try: | |
| # Open the PDF file using PyMuPDF's fitz library | |
| doc = fitz.open(pdf_file.name) | |
| incomplete_text = "" | |
| # Iterate through each page in the PDF document and extract the text | |
| for page in doc: | |
| text = page.get_text() | |
| incomplete_text += self._check_incomplete_sentence(text) | |
| return incomplete_text | |
| except Exception as e: | |
| print(f"An error occurred: {str(e)}") | |
| def file_output_fnn(self,file_path): | |
| file_path = file_path.name | |
| return file_path | |
| def gradio_interface(self): | |
| with gr.Blocks(css="style.css",theme='xiaobaiyuan/theme_brief') as demo: | |
| with gr.Row(elem_id = "col-container",scale=0.80): | |
| with gr.Column(elem_id = "col-container",scale=0.80): | |
| file1 = gr.File(label="File",elem_classes="filenameshow") | |
| with gr.Column(elem_id = "col-container",scale=0.20): | |
| upload_button1 = gr.UploadButton( | |
| "Browse File",file_types=[".txt", ".pdf", ".doc", ".docx",".json",".csv"], | |
| elem_classes="uploadbutton") | |
| incomplete_sentence_btn = gr.Button("Get Headings",elem_classes="uploadbutton") | |
| with gr.Row(elem_id = "col-container",scale=0.60): | |
| headings = gr.Textbox(label = "Headings") | |
| upload_button1.upload(self.file_output_fnn,upload_button1,file1) | |
| incomplete_sentence_btn.click(self.get_incomplete_sentence,upload_button1,headings) | |