[not finished] a script for playing http://sxiao.4399.com/4399swf/upload_swf/ftp25/csya/20180602/1/index.html

rabbit_jump.py

#coding:utf-8

from selenium import webdriver
from selenium.webdriver.common.action_chains import ActionChains

import base64
import time

import numpy as np
import cv2

driver = webdriver.Firefox(executable_path="./geckodriver")

url = "http://sxiao.4399.com/4399swf/upload_swf/ftp25/csya/20180602/1/index.html"

driver.get(url)

time.sleep(10)

canvas = driver.find_element_by_id('#canvas')

ITEMS = {
    'empty': (0, 300),
    'stick': (2800, 3200),
    'rotten_stick': (3500, 3800),
    'rock': (4100, 4400),
    'turtle': (4700, 5300),
    'crocodile': (5800, 6300)
}

# x_range = [1380, 1580, 1780, 1980]
x_range = [1380, 1580]
y_range = [800, 1000]

def get_screenshot():
    global driver
    img_base64 = driver.get_screenshot_as_base64()
    img_decode = base64.b64decode(img_base64)
    img_array = np.fromstring(img_decode, np.uint8)
    img = cv2.imdecode(img_array, cv2.COLOR_BGR2GRAY)
    return img

def count_adaptive_pixels(img):
    img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    img_blur = cv2.medianBlur(img_gray, 5)
    thresh = cv2.adaptiveThreshold(img_gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
    rows, _ = np.where(thresh != 255)
    return rows.shape[0]

def predict_item(nb_pixels):
    if nb_pixels >= ITEMS['empty'][0] and nb_pixels < ITEMS['empty'][1]:
        return 'empty'
    elif nb_pixels >= ITEMS['stick'][0] and nb_pixels < ITEMS['stick'][1]:
        return 'stick'
    elif nb_pixels >= ITEMS['rotten_stick'][0] and nb_pixels < ITEMS['rotten_stick'][1]:
        return 'rotten_stick'
    elif nb_pixels >= ITEMS['rock'][0] and nb_pixels < ITEMS['rock'][1]:
        return 'rock'
    elif nb_pixels >= ITEMS['turtle'][0] and nb_pixels < ITEMS['turtle'][1]:
        return 'turtle'
    elif nb_pixels >= ITEMS['crocodile'][0] and nb_pixels < ITEMS['crocodile'][1]:
        return 'crocodile'
    else:
        return 'unknown'

def predict_item_in_image(img):
    pixels = count_adaptive_pixels(img[y_range[0]:y_range[1],x_range[0]:x_range[1]])
    return predict_item(pixels)

def jump(ty):
    global driver
    global canvas
    if ty == 1:
        ActionChains(driver).click(canvas).perform()
    else:
        ActionChains(driver).context_click(canvas).perform()

def take_action(item):
    if item in ['empty', 'rotten_stick']:
        print("jump 2 spaces")
        jump(2)
    else:
        print("jump 1 space")
        jump(1)

k = 0
while k < 100:
    try:
        img = get_screenshot()
        item = predict_item_in_image(img)
        print(item)
        take_action(item)
    except:
        pass
    k += 1
    time.sleep(0.5)