Giter Club home page Giter Club logo

baiduindex's Introduction

百度指数抓取,再用图像识别得到指数

前言:

土福曾说,百度指数很难抓,在淘宝上面是20块1个关键字:

哥那么叼的人怎么会被他吓到,于是乎花了零零碎碎加起来大约2天半搞定,在此鄙视一下土福

安装的库很多:

谷歌图像识别tesseract-ocr

pip3 install pillow

pip3 install pyocr

selenium2.45

Chrome47.0.2526.106 m or Firebox32.0.1

chromedriver.exe

图像识别验证码请参考我的博客:

python图像识别--验证码

selenium用法请参考我的博客:

python之selenium

进入百度指数需要登陆,登陆的账号密码写在文本account里面:

万能登陆代码如下:

# 打开浏览器
def openbrowser():
    global browser

    # https://passport.baidu.com/v2/?login
    url = "https://passport.baidu.com/v2/?login&tpl=mn&u=http%3A%2F%2Fwww.baidu.com%2F"
    # 打开谷歌浏览器
    # Firefox()
    # Chrome()
    browser = webdriver.Chrome()
    # 输入网址
    browser.get(url)
    # 打开浏览器时间
    # print("等待10秒打开浏览器...")
    # time.sleep(10)

    # 找到id="TANGRAM__PSP_3__userName"的对话框
    # 清空输入框
    browser.find_element_by_id("TANGRAM__PSP_3__userName").clear()
    browser.find_element_by_id("TANGRAM__PSP_3__password").clear()

    # 输入账号密码
    # 输入账号密码
    account = []
    try:
        fileaccount = open("../baidu/account.txt")
        accounts = fileaccount.readlines()
        for acc in accounts:
            account.append(acc.strip())
        fileaccount.close()
    except Exception as err:
        print(err)
        input("请正确在account.txt里面写入账号密码")
        exit()
    browser.find_element_by_id("TANGRAM__PSP_3__userName").send_keys(account[0])
    browser.find_element_by_id("TANGRAM__PSP_3__password").send_keys(account[1])

    # 点击登陆登陆
    # id="TANGRAM__PSP_3__submit"
    browser.find_element_by_id("TANGRAM__PSP_3__submit").click()

    # 等待登陆10秒
    # print('等待登陆10秒...')
    # time.sleep(10)
    print("等待网址加载完毕...")

    select = input("请观察浏览器网站是否已经登陆(y/n):")
    while 1:
        if select == "y" or select == "Y":
            print("登陆成功!")
            print("准备打开新的窗口...")
            # time.sleep(1)
            # browser.quit()
            break

        elif select == "n" or select == "N":
            selectno = input("账号密码错误请按0,验证码出现请按1...")
            # 账号密码错误则重新输入
            if selectno == "0":

                # 找到id="TANGRAM__PSP_3__userName"的对话框
                # 清空输入框
                browser.find_element_by_id("TANGRAM__PSP_3__userName").clear()
                browser.find_element_by_id("TANGRAM__PSP_3__password").clear()

                # 输入账号密码
                account = []
                try:
                    fileaccount = open("../baidu/account.txt")
                    accounts = fileaccount.readlines()
                    for acc in accounts:
                        account.append(acc.strip())
                    fileaccount.close()
                except Exception as err:
                    print(err)
                    input("请正确在account.txt里面写入账号密码")
                    exit()

                browser.find_element_by_id("TANGRAM__PSP_3__userName").send_keys(account[0])
                browser.find_element_by_id("TANGRAM__PSP_3__password").send_keys(account[1])
                # 点击登陆sign in
                # id="TANGRAM__PSP_3__submit"
                browser.find_element_by_id("TANGRAM__PSP_3__submit").click()

            elif selectno == "1":
                # 验证码的id为id="ap_captcha_guess"的对话框
                input("请在浏览器中输入验证码并登陆...")
                select = input("请观察浏览器网站是否已经登陆(y/n):")

        else:
            print("请输入“y”或者“n”!")
            select = input("请观察浏览器网站是否已经登陆(y/n):")

登陆的页面:

登陆过后需要打开新的窗口,也就是打开百度指数,并且切换窗口,在selenium用:

# 新开一个窗口,通过执行js来新开一个窗口
js = 'window.open("http://index.baidu.com");'
browser.execute_script(js)
# 新窗口句柄切换,进入百度指数
# 获得当前打开所有窗口的句柄handles
# handles为一个数组
handles = browser.window_handles
# print(handles)
# 切换到当前最新打开的窗口
browser.switch_to_window(handles[-1])
    

清空输入框,构造点击天数:

# 清空输入框
browser.find_element_by_id("schword").clear()
# 写入需要搜索的百度指数
browser.find_element_by_id("schword").send_keys(keyword)
# 点击搜索
# <input type="submit" value="" id="searchWords" onclick="searchDemoWords()">
browser.find_element_by_id("searchWords").click()
time.sleep(2)
# 最大化窗口
browser.maximize_window()
# 构造天数
sel = int(input("查询7天请按0,30天请按1,90天请按2,半年请按3:"))
day = 0
if sel == 0:
    day = 7
elif sel == 1:
    day = 30
elif sel == 2:
    day = 90
elif sel == 3:
    day = 180
sel = '//a[@rel="' + str(day) + '"]'
browser.find_element_by_xpath(sel).click()
# 太快了
time.sleep(2)

天数也就是这里:

找到图形框:

xoyelement = browser.find_elements_by_css_selector("#trend rect")[2]

图形框就是:

根据坐标点的不同构造偏移量:

选取7天的坐标来观察:

第一个点的横坐标为1031.66666

第二个点的横坐标为1234

所以7天两个坐标之间的差为:202.33,其他的天数类似

用selenium库来模拟鼠标滑动悬浮:

from selenium.webdriver.common.action_chains import ActionChains
ActionChains(browser).move_to_element_with_offset(xoyelement,x_0,y_0).perform()

但是这样子确定的点指出是在这个位置:

也就是矩形的左上角,这里是不会加载js显示弹出框的,所以要给横坐标+1:

x_0 = 1
y_0 = 0

写个按照天数的循环,让横坐标累加:

# 按照选择的天数循环
for i in range(day):
    # 构造规则
    if day == 7:
        x_0 = x_0 + 202.33
    elif day == 30:
        x_0 = x_0 + 41.68
    elif day == 90:
        x_0 = x_0 + 13.64
    elif day == 180:
        x_0 = x_0 + 6.78

鼠标横移时会弹出框,在网址里面找到这个框:

selenium自动识别之...:

# <div class="imgtxt" style="margin-left:-117px;"></div>
imgelement = browser.find_element_by_xpath('//div[@id="viewbox"]')

并且确定这个框的大小位置:

# 找到图片坐标
locations = imgelement.location
print(locations)
# 找到图片大小
sizes = imgelement.size
print(sizes)
# 构造指数的位置
rangle = (int(locations['x']), int(locations['y']), int(locations['x'] + sizes['width']),
          int(locations['y'] + sizes['height']))

截取的图形为:

下面的思路就是:

  1. 将整个屏幕截图下来
  2. 打开截图用上面得到的这个坐标rangle进行裁剪

但是最后裁剪出来的是上面的那个黑框,我想要的效果是:

本次更新加入了对于关键词长度的判断,能够自动识别关键词长度而进行截取:

add_length = (len(keyword) - 2) * sizes['width'] / 15

找到位置:

# 跨浏览器兼容
scroll = browser.execute_script("return window.scrollY;")
top = locations['y'] - scroll
# 构造指数的位置
rangle = (
    int(locations['x'] + sizes['width'] / 4 + add_length), int(top + sizes['height'] / 2),
    int(locations['x'] + sizes['width'] * 2 / 3), int(top + sizes['height']))

后面的完整代码是:

# <div class="imgtxt" style="margin-left:-117px;"></div>
imgelement = browser.find_element_by_xpath('//div[@id="viewbox"]')
# 找到图片坐标
locations = imgelement.location
print(locations)
# 找到图片大小
sizes = imgelement.size
print(sizes)
# 构造指数的位置
rangle = (int(locations['x'] + sizes['width']/3), int(locations['y'] + sizes['height']/2), int(locations['x'] + sizes['width']*2/3),
          int(locations['y'] + sizes['height']))
# 截取当前浏览器
path = "../baidu/" + str(num)
browser.save_screenshot(str(path) + ".png")
# 打开截图切割
img = Image.open(str(path) + ".png")
jpg = img.crop(rangle)
jpg.save(str(path) + ".jpg")

但是后面发现裁剪的图片太小,识别精度太低,所以需要对图片进行扩大:

# 将图片放大一倍
# 原图大小73.29
jpgzoom = Image.open(str(path) + ".jpg")
(x, y) = jpgzoom.size
x_s = 146
y_s = 58
out = jpgzoom.resize((x_s, y_s), Image.ANTIALIAS)
out.save(path + 'zoom.jpg', 'png', quality=95)

原图大小请 右键->属性->详细信息 查看,我的是长73像素,宽29像素

最后就是图像识别

# 图像识别
index = []
image = Image.open(str(path) + "zoom.jpg")
code = pytesseract.image_to_string(image)
if code:
    index.append(code)

最后效果图:

详细解说请观看我的博客:

TTyb

更新日志:

2017-10-23修复截图位置不对的bug,优化关键词自动识别长度的漏洞

baiduindex's People

Contributors

ttyb avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

baiduindex's Issues

百度指数的按照时间来选择是怎么做的

我把输入的js信息改成了如下的xi形式:

days = "-".join(str(day).split(","))
params = {'time':days,'word':keyword}
querystring = urlencode(params)
print (querystring)
url = 'http://index.baidu.com/?tpl=trend&type=0&area=0&'+querystring
url.replace("-","%7C")
print(url)
js = 'window.open("'+url+'");'
browser.execute_script(js)

在模拟的浏览器上就成了:(图片上传不了),就是在地址栏里面word后面是中文,而不是urlencode后的字符,而且在输入框里面的是乱码。

请问:有什么比较不麻烦的时间选择特定时间获取数据的方式么,操作选择框,填充时间这个我没有尝试成功。

抠取指数的位置好像不太对

获取指数位置,发现有些对的上,有些对不上。我发现它的位置和曲线上点的相对位置并不是固定滴。请问大神你是怎么解决的。

请问效率怎么样?

如题,比如取一个关键词的90天指数,也就是登陆+截90张图+分析需要多久? ^_^ 谢谢了老哥!

鼠标移动问题

    while True:
        try:
            xoyelement = self.driver.find_elements_by_css_selector("#trend rect")[2]
            break
        except NoSuchElementException as e:
            print (e.message)
        except IndexError:
            pass
    print ("提取图片信息")

    for i in range(Time_difference+1):
        x = i*(MAX_WIDTH/Time_difference)
        y = 10
        ActionChains(self.driver).move_to_element_with_offset(xoyelement, x, y).perform()

这是我的代码 按照你的代码 选取trend rect 的下标为2的 但是并没有弹出对应数据导致后面这句话报错
这里的x y 的值我可以确定是没问题的
imgelement = self.driver.find_element_by_xpath('//div[@id="viewbox"]')

不知道你有没有遇到过

鼠标移动之后抓到的图不对

鼠标自动移动的时候,没有截取到指数数值的图。 看起来图的定位不正确。
用的Chrome Driver 2.32.498537版本。
Chrome: Version 60.0.3112.113 (Official Build) (64-bit)。
Python 3.6

browser.find_element_by_id("TANGRAM__PSP_3__userName").clear()位置报错

是不是因为登录百度时候优先出现二维码,所以报错了。
selenium.common.exceptions.InvalidElementStateException: Message: invalid element state: Element is not currently interactable and may not be manipulated
(Session info: chrome=63.0.3239.108)
(Driver info: chromedriver=2.34.522940 (1a76f96f66e3ca7b8e57d503b4dd3bccfba87af1),platform=Windows NT 10.0.10586 x86_64)

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.