|
| 1 | +import PyPDF2 |
| 2 | +import fitz |
| 3 | +import os |
| 4 | + |
| 5 | +def pdf_to_text(pdf_file, output_folder): |
| 6 | + # Open the PDF file |
| 7 | + with open(pdf_file, 'rb') as file: |
| 8 | + pdf_reader = PyPDF2.PdfFileReader(file) |
| 9 | + |
| 10 | + # Create output folder if it doesn't exist |
| 11 | + if not os.path.exists(output_folder): |
| 12 | + os.makedirs(output_folder) |
| 13 | + |
| 14 | + # Extract text from each page |
| 15 | + for page_num in range(pdf_reader.numPages): |
| 16 | + page = pdf_reader.getPage(page_num) |
| 17 | + text = page.extractText() |
| 18 | + |
| 19 | + # Write extracted text to a text file |
| 20 | + text_file_path = os.path.join(output_folder, f"page_{page_num + 1}.txt") |
| 21 | + with open(text_file_path, 'w') as text_file: |
| 22 | + text_file.write(text) |
| 23 | + |
| 24 | + print("Text extraction completed. Text files saved in:", output_folder) |
| 25 | + |
| 26 | +def pdf_to_images(pdf_file, output_folder): |
| 27 | + # Open the PDF file |
| 28 | + pdf_document = fitz.open(pdf_file) |
| 29 | + |
| 30 | + # Create output folder if it doesn't exist |
| 31 | + if not os.path.exists(output_folder): |
| 32 | + os.makedirs(output_folder) |
| 33 | + |
| 34 | + # Iterate through each page and save as image |
| 35 | + for page_num in range(len(pdf_document)): |
| 36 | + page = pdf_document[page_num] |
| 37 | + image_path = os.path.join(output_folder, f"page_{page_num + 1}.png") |
| 38 | + pix = page.get_pixmap() |
| 39 | + pix.writePNG(image_path) |
| 40 | + |
| 41 | + print("Image conversion completed. Images saved in:", output_folder) |
| 42 | + |
| 43 | +def main(): |
| 44 | + pdf_file = 'sample.pdf' # Change this to the path of your PDF file |
| 45 | + output_folder = 'output' # Output folder where converted files will be saved |
| 46 | + |
| 47 | + # Convert PDF to text |
| 48 | + pdf_to_text(pdf_file, output_folder) |
| 49 | + |
| 50 | + # Convert PDF to images |
| 51 | + pdf_to_images(pdf_file, output_folder) |
| 52 | + |
| 53 | +if __name__ == "__main__": |
| 54 | + main() |
0 commit comments