import cv2

import pytesseract

import numpy as np

import os

#tesseract dependencies

pytesseract.pytesseract.tesseract_cmd = 'C:\\Program Files\\Tesseract-OCR\\tesseract.exe'

tessdata_dir_config = '--tessdata-dir "C:\\Program Files\\Tesseract-OCR\\tessdata"'

cv2.resizeWindow('output', 400, 400)

# OCR in text block

def ACABS(imgg):

    img = cv2.imread(imgg)

    output = img.copy()

    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

    #crop 10 colums one the right and left of the image

    #avoid bugs with the segmentation algorithm on the border of the image

    h, w = gray.shape

    gray = np.array(gray[0:0 + h, 10: w - 10])

    # clean the image using otsu method with the inversed binarized image

    ret1, th1 = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)

    boxs = create_bounding_box(th1, output)

    crop_top, crop_bottom = crop(boxs, gray, 0.10)

    text = ocr(gray, crop_top, crop_bottom, boxs)

    return text

# apply dilatation and erosion

# create bounding box

def create_bounding_box(thresh, output):

    boxs = []

    # assign a rectangle kernel size

    k1 = np.ones((15, 15), 'uint8')

    par_img = cv2.dilate(thresh, k1, iterations=3)

    par_img2 = cv2.erode(par_img, kernel = np.ones((5,5), 'uint8'), iterations=3)

    (contours, _) = cv2.findContours(par_img2.copy(), cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)

    #create the bounding box

    for cnt in reversed(contours):

        x, y, w, h = cv2.boundingRect(cnt)

        boxs.append([x, y, w, h])

        #draw bounding box on image

        #cv2.rectangle(output, (x, y), (x + w, y + h), (0, 255, 0), 1)

    #cv2.imwrite("../segmentation/adaptative_reding_exemple/repport_9_box.jpg", output)

    return boxs

#define line of cropping

def crop(boxs, gray, percent):

    h, w = gray.shape

    #define percent of Y crop on the image

    top_crop_percent = percent

    bottom_crop_percent = percent

    crop_top = h * top_crop_percent

    crop_bottom = h * bottom_crop_percent

    #test

    output = gray.copy()

    #Adapt crop to avoid cropping in the middle of a bounding box

    #if the crop is made between Y and Y + H (bounding box) the crop become Y + H

    # box[0] : x

    # box[1] : y

    # box[2] : w

    # box[3] : h

    change_bottom = False

    for box in boxs:

        if (inbetween(box[1], crop_top, box[1] + box[3]) == crop_top):

            crop_top = box[1] + box[3]

        if (inbetween(box[1], h-crop_bottom, box[1] + box[3]) == h-crop_bottom):

            crop_bottom = box[1] + box[3]

            change_bottom = True

        cv2.rectangle(output, (box[0], box[1]), (box[0] + box[2], box[1] + box[3]), (0, 255, 0), 10)

    #round to avoid error

    crop_top = round(crop_top)

    crop_bottom = round(crop_bottom)

    #if bottom crop haven't change

    if not change_bottom:

        crop_bottom = h-crop_bottom

    #print bottom and top threshold lines on the image

    # bottom

    #output = cv2.line(output, (0, crop_bottom), (w, crop_bottom), (0, 0, 255), 5)

    # top

    #output = cv2.line(output, (0, crop_top), (w, crop_top), (0, 0, 255), 5)

    #cv2.imwrite("../segmentation/adaptative_reding_exemple/repport_9_box.jpg", output)

    return crop_top, crop_bottom

#find if val is between min and max

def inbetween(min,val,max):

    return sorted([min,val,max])[1]

#extract text for each bounding box

def ocr(gray, crop_top, crop_bottom, cropped_boxs):

        text = ""

        for box in cropped_boxs:

            #box[0] : x

            #box[1] : y

            #box[2] : w

            #box[3] : h

            if (box[1] > crop_top) and (box[1] + box[3] <= crop_bottom):

                ocr_box = np.array(gray[box[1]:box[1] + box[3], box[0]:box[0] + box[2]])

                #check if array contain only 255

                #if(all(all(p == 255 for p in lines)for lines in ocr_box)):

                #check if 95% of the array is 255

                flattened = np.ravel(ocr_box)

                if not (np.sum((flattened == 255)) / len(flattened) > 0.95):

                    #cv2.imshow('ok', ocr_box)

                    #cv2.waitKey(0)

                    data = (pytesseract.image_to_string(ocr_box, lang='fra'))

                    #\jump used as a delimiter of each text block

                    #Used in the NER annotator software

                    data = data.replace(' \n\n', '')

                    data = data.replace(' \n\x0c', '')

                    data = data.replace('\x0c', '')

                    text += data

        return text