read_image_to_text.py

import io
import os
import re
import string

import FreeSimpleGUI as sg

try:
    from PIL import Image, ImageEnhance
except ImportError:
    import Image

import pytesseract as ocr

"""
For Windows Only
1 - You need to have Tesseract OCR installed on your computer.
      get it from here. https://github.com/UB-Mannheim/tesseract/wiki
      Download the suitable version.
2 - Add Tesseract path to your System Environment. i.e. Edit system variables.
3 - Run pip install pytesseract and pip install tesseract
4 - Add this line to your python script every time
pytesseract.pytesseract.tesseract_cmd = 'C:/OCR/Tesseract-OCR/tesseract.exe'  # your path may be different
5 - Run the code.
"""


class Gui:
    def __init__(self):
        self.file_types = [
            ("JPG (*.jpg)", "*.jpg"),
            ("JPEG (*.jpeg)", "*.jpeg"),
            ("PNG (*.png)", "*.png"),
            ("All files (*.*)", "*.*"),
        ]

    def list_folder(self):
        layout = [
            [sg.T("Escolha a IMAGEM")],
            [sg.Image(key="-IMAGE-")],
            [
                sg.Input(size=(25, 1), key="-FILE-"),
                sg.FileBrowse(file_types=self.file_types),
                sg.Button("Preview", key="preview"),
            ],
            [sg.Button("Submit"), sg.Exit()],
            [
                sg.Multiline(
                    font="Calibri",
                    key="resultado_final",
                    size=(100, 30),
                    disabled=True,
                    border_width=0,
                )
            ],
        ]
        window = sg.Window("Transformar IMG em Texto", grab_anywhere=False).Layout(layout)
        while True:
            event, values = window.read()
            if "Submit" in event:
                if values["-FILE-"]:
                    img_selected = values["-FILE-"]
                    if os.name != "posix":  # Windows
                        text_in_the_image = self.get_text_to_imgr_by_windows(img_selected)
                    else:
                        text_in_the_image = self.get_text_to_imgr_by_linux(img_selected)
                    # ngrams = self.cleanInput(text_in_the_image)
                    # window["resultado_final"].update(f"{' '.join(ngrams)}")
                    window["resultado_final"].update(f"{' '.join(text_in_the_image)}")
            elif "preview" in event:
                filename = values["-FILE-"]
                if os.path.exists(filename):
                    image = Image.open(values["-FILE-"])
                    image.thumbnail((400, 600))
                    bio = io.BytesIO()
                    image.save(bio, format="PNG")
                    window["-IMAGE-"].update(data=bio.getvalue())
            elif event == sg.WIN_CLOSED or "Exit" in event:
                sg.popup_auto_close("Exit...", auto_close_duration=0.5)
                break
        window.close()

    def cleanInput(self, input):
        input = re.sub(r"\n", " ", input)
        input = re.sub(r"\[[0-9]*\]", "", input)
        input = re.sub(" +", " ", input)
        input = bytes(input, "UTF-8")
        input = input.decode("ascii", "ignore")
        cleanInput = []
        input = input.split(" ")
        for item in input:
            item1 = item.strip(string.punctuation)
            if len(item1) > 1 or item1.lower() in ["a", "i"]:
                cleanInput.append(item1.lower())
        return cleanInput

    def get_text_to_imgr_by_windows(self, filename):
        homepath = os.path.expanduser(os.getenv("USERPROFILE"))
        desktoppath = "Desktop"
        # 'C:/Users/zicad/OneDrive/Área de Trabalho/deere.png'
        path_to_desktop = os.path.join(homepath, desktoppath)
        os.chdir(path_to_desktop)
        # path_to_img = os.path.abspath("fpython/ImageToText/img")
        # path = os.path.join(path_to_img, filename)
        path = os.path.join(path_to_desktop, filename)
        # ocr.pytesseract.tesseract_cmd = "C:/Users/luxu/AppData/Local/Programs/Tesseract-OCR/tesseract.exe"
        ocr.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR/tesseract.exe"

        with Image.open(path) as img:
            img.verify()  # Retorna None se estiver tudo bem
            print("A imagem é válida.")

        # Carrega a imagem novamente para manipulação
        img = Image.open(path)
        img = img.convert("RGB")  # Caso precise converter o formato
        print("A imagem foi carregada e está pronta para uso.")

        # Melhorar o contraste
        enhancer = ImageEnhance.Contrast(img)
        img = enhancer.enhance(2)  # Aumenta o contraste, ajuste conforme necessário
        print("A imagem teve aumento de constraste.")

        img = img.convert("L")  # Converte para tons de cinza

        try:
            # img = Image.open(path)
            # is_img_valid = img.verify()
            image = ocr.image_to_string(img, lang="por")
            # image = ocr.image_to_string(img)
            tamanho = len(image)
            if tamanho > 0:
                print("A imagem foi carregada com sucesso.")
            else:
                print("Texto está sem nada.")
            return image
        except Exception as e:
            raise ValueError(f"Error: {e}") from e


if __name__ == "__main__":
    tela = Gui()
    img = tela.list_folder()