from langchain_community.document_loaders import PyPDFLoader from langchain.text_splitter import RecursiveCharacterTextSplitter from langchain_community.vectorstores import FAISS from langchain_openai import OpenAIEmbeddings import PyPDF2 from PyPDF2 import PdfReader import pdfplumber from PIL import Image import pytesseract from pdf2image import convert_from_path from pdfminer.high_level import extract_pages, extract_text from pdfminer.layout import LTTextContainer, LTChar, LTRect, LTFigure import os from dotenv import load_dotenv load_dotenv() OPENAI_API_KEY = os.getenv('OPENAI_API_KEY') # Extracting tables from the page def extract_table(pdf_path, page_num, table_num): # Open the pdf file pdf = pdfplumber.open(pdf_path) # Find the examined page table_page = pdf.pages[page_num] # Extract the appropriate table table = table_page.extract_tables()[table_num] return table # Convert table into appropriate fromat def table_converter(table): table_string = '' # Iterate through each row of the table for row_num in range(len(table)): row = table[row_num] # Remove the line breaker from the wrapted texts cleaned_row = [item.replace('\n', ' ') if item is not None and '\n' in item else 'None' if item is None else item for item in row] # Convert the table into a string table_string+=('|'+'|'.join(cleaned_row)+'|'+'\n') # Removing the last line break table_string = table_string[:-1] return table_string # Create a function to check if the element is in any tables present in the page def is_element_inside_any_table(element, page ,tables): x0, y0up, x1, y1up = element.bbox # Change the cordinates because the pdfminer counts from the botton to top of the page y0 = page.bbox[3] - y1up y1 = page.bbox[3] - y0up for table in tables: tx0, ty0, tx1, ty1 = table.bbox if tx0 <= x0 <= x1 <= tx1 and ty0 <= y0 <= y1 <= ty1: return True return False # Function to find the table for a given element def find_table_for_element(element, page ,tables): x0, y0up, x1, y1up = element.bbox # Change the cordinates because the pdfminer counts from the botton to top of the page y0 = page.bbox[3] - y1up y1 = page.bbox[3] - y0up for i, table in enumerate(tables): tx0, ty0, tx1, ty1 = table.bbox if tx0 <= x0 <= x1 <= tx1 and ty0 <= y0 <= y1 <= ty1: return i # Return the index of the table return None def text_extraction(element): # Extracting the text from the in line text element line_text = element.get_text() # Find the formats of the text # Initialize the list with all the formats appeared in the line of text line_formats = [] for text_line in element: if isinstance(text_line, LTTextContainer): # Iterating through each character in the line of text for character in text_line: if isinstance(character, LTChar): # Append the font name of the character #line_formats.append(character.fontname) # Append the font size of the character #line_formats.append(character.size) line_formats.append("") # Find the unique font sizes and names in the line format_per_line = list(set(line_formats)) # Return a tuple with the text in each line along with its format return (line_text, format_per_line) # Create a function to crop the image elements from PDFs def crop_image(element, pageObj): # Get the coordinates to crop the image from PDF [image_left, image_top, image_right, image_bottom] = [element.x0,element.y0,element.x1,element.y1] # Crop the page using coordinates (left, bottom, right, top) pageObj.mediabox.lower_left = (image_left, image_bottom) pageObj.mediabox.upper_right = (image_right, image_top) # Save the cropped page to a new PDF cropped_pdf_writer = PyPDF2.PdfWriter() cropped_pdf_writer.add_page(pageObj) # Save the cropped PDF to a new file with open('cropped_image.pdf', 'wb') as cropped_pdf_file: cropped_pdf_writer.write(cropped_pdf_file) # Create a function to convert the PDF to images def convert_to_images(input_file,): images = convert_from_path(input_file) image = images[0] output_file = 'PDF_image.png' image.save(output_file, 'PNG') # Create a function to read text from images def image_to_text(image_path): # Read the image img = Image.open(image_path) # Extract the text from the image text = pytesseract.image_to_string(img) return text def read_file_get_prompts(file_name): if file_name is not None: # Find the PDF path pdf_path = file_name # '/content/data/'+file_name+".pdf" pdfReaded = PyPDF2.PdfReader(file_name) # Create the dictionary to extract text from each image text_per_page = {} # Create a boolean variable for image detection image_flag = False number_of_pages = len(list(extract_pages(file_name))) result = '' # We extract the pages from the PDF for pagenum, page in enumerate(extract_pages(file_name)): # Initialize the variables needed for the text extraction from the page pageObj = pdfReaded.pages[pagenum] page_text = [] line_format = [] text_from_images = [] text_from_tables = [] page_content = [] # Initialize the number of the examined tables table_in_page= -1 # Open the pdf file pdf = pdfplumber.open(pdf_path) # Find the examined page page_tables = pdf.pages[pagenum] # Find the number of tables in the page tables = page_tables.find_tables() if len(tables)!=0: table_in_page = 0 # Extracting the tables of the page for table_num in range(len(tables)): # Extract the information of the table table = extract_table(pdf_path, pagenum, table_num) # Convert the table information in structured string format table_string = table_converter(table) # Append the table string into a list text_from_tables.append(table_string) # Find all the elements page_elements = [(element.y1, element) for element in page._objs] # Sort all the element as they appear in the page page_elements.sort(key=lambda a: a[0], reverse=True) # Find the elements that composed a page for i,component in enumerate(page_elements): # Extract the element of the page layout element = component[1] # Check the elements for tables if table_in_page == -1: pass else: if is_element_inside_any_table(element, page ,tables): table_found = find_table_for_element(element,page ,tables) if table_found == table_in_page and table_found != None: page_content.append(text_from_tables[table_in_page]) #page_text.append('table') #line_format.append('table') table_in_page+=1 # Pass this iteration because the content of this element was extracted from the tables continue if not is_element_inside_any_table(element,page,tables): # Check if the element is text element if isinstance(element, LTTextContainer): # Use the function to extract the text and format for each text element (line_text, format_per_line) = text_extraction(element) # Append the text of each line to the page text page_text.append(line_text) # Append the format for each line containing text line_format.append(format_per_line) page_content.append(line_text) # Check the elements for images if isinstance(element, LTFigure): # Crop the image from PDF crop_image(element, pageObj) # Convert the croped pdf to image convert_to_images('cropped_image.pdf') # Extract the text from image image_text = image_to_text('PDF_image.png') image_text = "" # removed to remove the errors with image text_from_images.append(image_text) page_content.append(image_text) # Add a placeholder in the text and format lists #page_text.append('image') #line_format.append('image') # Update the flag for image detection image_flag = True # Create the key of the dictionary dctkey = 'Page_'+str(pagenum) print(dctkey) # Add the list of list as value of the page key #text_per_page[dctkey]= [page_text, line_format, text_from_images,text_from_tables, page_content] text_per_page[dctkey]= [page_text, text_from_images,text_from_tables, page_content] #result = result.join(page_text).join(line_format).join(text_from_images).join(text_from_tables).join(page_content) result = " " for t in range(number_of_pages): page = 'Page_'+str(t) #result = result.join(map(str, text_per_page[page])) for q in range(len(text_per_page[page])): #print(f"{''.join(map(str, text_per_page[page][q]))}") result = result + f"{''.join(map(str, text_per_page[page][q]))}" return result return True def save_to_vector_store(text): text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200) docs = text_splitter.create_documents(text) vectorstore = FAISS.from_documents(documents=docs, embedding=OpenAIEmbeddings(api_key=OPENAI_API_KEY)) vectorstore.save_local(DB_FAISS_PATH, index_name="njmvc_Index") #create a new file named vectorstore in your current directory. if __name__=="__main__": DB_FAISS_PATH = 'vectorstore/db_faiss' file_name = "./data/drivermanual-2-small.pdf" #loader=read_file_get_prompts(file_name) text=read_file_get_prompts(file_name) #pdfReaded = PyPDF2.PdfReader(file_name) #docs=loader.load() save_to_vector_store(text) #save_to_vector_store(text)