将html转换为txt,Python 将HTML转换为TXT

CODE:

#!/usr/bin/python

# -*- coding: utf-8 -*-

'''

Created on 2014-9-5

@author: guaguastd

@name: html_to_text.py

'''

from login import google_api_request

from html import cleanHtml

while True:

query = raw_input("Input query(None to quit): ")

if query.strip() == '':

break

#people_feed = google_login_http("people", "query", query)

#people_feed = google_api.people().search(query=query).execute()

#print json.dumps(people_feed['items'], indent=1)

people_feed = google_api_request(0, action='search', query=query)

for user in people_feed['items']:

userId = user['id']

activity_feed = google_api_request(1, action='list', collection='public', maxResults='100', userId=userId)

#print json.dumps(activity_feed, indent=1)

print '\nPrimary content including HTML tag:'

print activity_feed['items'][0]['object']['content']

print '\nConverted content without HTML tag:'

print cleanHtml(activity_feed['items'][0]['object']['content'])

RESULT:

Input query(None to quit): Tim O'Reilly

Primary content including HTML tag:

ABC World News covers @HomeDepot credit card data breach, recommends @BillGuard to protect yourself! BillGuard on ABC World News Tonight With David Muir … (Minute 1:25)

Converted content without HTML tag:

ABC World News covers @HomeDepot credit card data breach, recommends @BillGuard to protect yourself! BillGuard on ABC World News Tonight With David Muir … (Minute 1:25 )