Upload #photos to renren.com and fetch news #python #sh

renews.sh

#!/bin/sh

### get news from your friends at renren.com
# Copyright: (c) 2009,2010 Jakukyo Friel <[email protected]>
# License: GPL v2

# Config area

# We put everything in an hg repo.  You should set it up already.
renrepo=$HOME/repo/renren
# Login to m.renren.com with your acccount and get the bookmark url.
privateURL='http://m.renren.com/home.do?bm=yourPrivateURLHere'
# The cookies file.
rencookie=rencookie.txt

# Main script

# functions

# fetch pages: fetch url > outputfile
fetch() {
    wget -O - --load-cookies $rencookie \
    --save-cookies rencookie.txt --keep-session-cookies $1
}

# get the link to the next page:
# It's dirty to use global virable, but afterall this is a sh script.
getlink() {
    local rawurl=$(grep -Eo '<a[^>]+>下一页</a>'  |
    grep -Eo 'href="[^"]+"' |
    sed -e 's/^href=//' |
    sed -e 's/"//g')
    if (echo $rawurl | grep -e '^http://' > /dev/null); then
        echo $rawurl
    else
        echo http://3g.renren.com$rawurl
    fi
}


# procedures

cd $renrepo

# fetch front page
fetch $privateURL >0.html


# fetch more pages
newspage=0.html
i=1
while (grep '下一页' $newspage); do
    output=$i.html
    fetch $(cat $newspage | getlink) >$output
    newspage=$output
    i=$(expr $i + 1)
done

fetch $(cat $newspage | getlink) >$i.html

# We have fetched all the pages we want.


# convert to text file
for file in *.html; do
w3m -dump $file > $file.txt
done

# extract news
for file in [0-9]*.html.txt; do
    sed -n '/新鲜事/,/[上下]一页/p' $file >> news.all
done

sed -e '/新鲜事/d' -e '/[上下]一页/d' news.all | sort | uniq > news

# commit in changes
# you should have already run 'hg init' and 'hg add news' before.
hg ci -m 'automated commit'

rrfoto.py

#!/usr/bin/python2.6

# by weakish <[email protected]>, licensed under GPL v2.

'''
Upload photos to renren.com

This script uploads all jpeg files at the current dir to an existing album
at renren.com.

Todo:
- read albumid from command line arguments
- support files other than jpg
'''

import glob
from twill.commands import *

# config
email = '[email protected]'
password = 'password'
albumid = 'albumid'

# auth
go('http://www.renren.com/')
formvalue('1', 'email', email)
formvalue('1', 'password', password)
submit()
code('200')

# upload
for photo in glob.glob('*.jpg'): 
    go('http://upload.renren.com/addphotoPlain.do?id='+albumid)
    formfile('uploadForm', 'photo1', photo)
    submit()
    code('200')