from PIL import Image
import PyPDF2
tess.pytesseract.tesseract_cmd = r'C:\Users\szaid\AppData\Local\Programs\Tesseract-OCR\tesseract'
translator=Translator()

project_dir = os.path.dirname(os.path.abspath(__file__))

app = Flask(__name__,
            static_url_path='',
            static_folder='static',
            template_folder='template')
photos = UploadSet('photos', IMAGES)
pdf = UploadSet('pdf', DOCUMENTS)
app.config['DEBUG'] = True
app.config['UPLOAD_FOLDER']= 'images'

class Get(object):
        def __init__(self, file): enter code here
                if render_template('image.html'):
                        self.file = tess.image_to_string(Image.open(project_dir   '/images/'  file))
                elif render_template('pdf.html'):
                        self.file = PyPDF2.PdfFileReader(project_dir '/pdff/'  file)

And below, this is backend code to save the file in 'images' folder but its giving me an error. I mean there is two different html pages and in both pages, there is an option to upload file and image. but dont know how to manage in flask...

@app.route('/pdf', methods=["GET","POST"])
def pdf():
        if request.method == 'POST':
                
                if 'pdf' not in request.files:
                        return 'there is no photo'
                name1 = request.form['pdf-name']   '.pdf'
                pdf = request.files['pdf']
                path = os.path.join(app.config['UPLOAD_FOLDER'], name1)
                pdf.save(path)
                TextObject = Get(name1)
                return TextObject.file

        return render_template('pdf.html')

error

CodePudding user response：

You're getting this error because your path to the PDF file is wrong. You've added the file extension but the path is missing the file name. You also have a problem with your slashes in the path. A correct path would be like:

"D:\images\yourFileName.pdf"