# kivy 沒支援目前版本(4.9.2)的 lxml 的 alternative 方案 ＝》beautifulSoup, xml2dict, pyquery ...
---

需求:
[kivy 使用 kivy ,Camera + autopredictor 拍照後傳回辨識檔案,使用 python kivy套件寫 iOS 和 Android APP](https://hackmd.io/@cssmiley/rk3T64_l3)
lxml 模組錯誤暫時無解,安裝到模擬器跑一直報錯只好改用bs4模組完成一樣的作用

lxml 原本的作用

# 解析 response HTML 取得 _token, cookies
from lxml import html
tree = html.fromstring(response.text)
_token =tree.xpath('//input[@name="_token"]/@value')

錯誤: dynamic module does not define module export function (PyInit_etree)

[lxml alternatives and similar packages](https://python.libhunt.com/lxml-alternatives)

解決:
參考下面的方式改成使用 BeautifulSoup 解析
- [can we use XPath with BeautifulSoup?](https://stackoverflow.com/questions/11465555/can-we-use-xpath-with-beautifulsoup)
BeautifulSoup has a function named findNext from current element directed childern,so:

	father.findNext('div',{'class':'class_value'}).findNext('div',{'id':'id_value'}).findAll('a') 

Above code can imitate the following xpath:

	div[class=class_value]/div[id=id_value]

- [lxml.html](https://lxml.de/lxmlhtml.html)
- [xpath即時轉換](http://xpather.com)
- 重要[【Day 09】- 大家都愛的 BeautifulSoup](https://ithelp.ithome.com.tw/articles/10271319?sc=iThelpR)
-重要 [什麼是XPath](https://matthung0807.blogspot.com/2017/12/xpath.html)
- [CSS2XPath](https://css2xpath.github.io)
- 重要[安装解析器 html.parser](https://www.crummy.com/software/BeautifulSoup/bs4/doc.zh/#id13)
- 將原本 lxml 的 code 修改成使用 BeautifulSoup的方式:

    # 暫時先使用mock data 和 BeautifulSoup代替上面的lxml
    soup = BeautifulSoup(html_doc, 'html.parser') # 使用內建的 html.parser 解析,不用另外裝 html5lib
    #soup = BeautifulSoup(html_doc, 'html5lib')
    print(soup.prettify())
    _token = soup.input.findNext("input",{"name":"_token"})["value"]