def fetch_core(uid, access_token):
uid = str(uid)
# Initialize the needed modules
CHandler = urllib2.HTTPCookieProcessor(cookielib.CookieJar())
browser = urllib2.build_opener(CHandler)
browser.addheaders = [('User-agent', 'InFB - [email protected] - http://ruel.me')]
urllib2.install_opener(browser)
# print 'Using access token: %s' % access_token + '\n'
url = 'https://graph.facebook.com/%s?access_token=%s' % (uid, access_token)
print url
res = browser.open(url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
FileUtility.write('data/profile/%s.json' % uid,fres)
# fetch profile picture
if os.path.isfile('data/profile/%s_picture_large.png'%uid) == False:
f = open('data/profile/%s_picture_large.png'%uid,"wb")
f.write(urllib.urlopen('https://graph.facebook.com/%s/picture?type=large'%uid ).read())
f.close()
if os.path.isfile('data/profile/%s_picture.png'%uid) == False:
f = open('data/profile/%s_picture.png'%uid,"wb")
f.write(urllib.urlopen('https://graph.facebook.com/%s/picture'%uid ).read())
f.close()
def fetch_core(file_dir, url, access_token):
# Initialize the needed modules
CHandler = urllib2.HTTPCookieProcessor(cookielib.CookieJar())
browser = urllib2.build_opener(CHandler)
browser.addheaders = [('User-agent', 'InFB - [email protected] - http://ruel.me')]
urllib2.install_opener(browser)
# print 'Using access token: %s' % access_token + '\n'
url = 'https://graph.facebook.com/'+url+ ('?limit=100&access_token=%s' % access_token)
print url
res = browser.open(url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
i=1
FileUtility.write(file_dir+"/data"+str(i)+".json",fres)
while len(jdata['data'])>0:
data = jdata['data']
for sub_data in data:
pid = sub_data['id']
FetchPostsDetail.fetch_core(pid,access_token, browser)
url = jdata['paging']['next']
print url
res = browser.open(url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
i=i+1
FileUtility.write(file_dir+"/data"+str(i)+".json", fres)
def fetch_core(post_id, access_token, browser):
if os.path.isdir("data/posts/"+post_id) == False:
os.mkdir("data/posts/"+post_id)
else:
print 'File has aleary existed, skip!!'
return
print 'In FetchPostsDetail.py parse' + post_id
url = 'https://graph.facebook.com/'+post_id+ ('&access_token=%s' % access_token)
res = try_brower(browser, url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
FileUtility.write("data/posts/"+post_id+"/content.json",fres)
url = 'https://graph.facebook.com/'+post_id+'/likes'+ ('&access_token=%s' % access_token)
res = try_brower(browser, url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
FileUtility.write("data/posts/"+post_id+"/likes.json",fres)
url = 'https://graph.facebook.com/'+post_id+'/comments'+ ('&access_token=%s' % access_token)
res = try_brower(browser, url)
fres = res.read()
jdata = json.loads(fres)
fres = json.dumps(jdata, ensure_ascii=False)
FileUtility.write("data/posts/"+post_id+"/comments.json",fres)
