def get_info_history(session):
geo_num = 63
starttime = convert_time('2015', '1', '1', '0')
starttime = str(starttime)
starttime = starttime[:-2]
page_count = 50
endtime = starttime
sort = 0
offset = 0
inter_lat = INTER_LAT
inter_lon = INTER_LON
for p_id in range(0, geo_num):
geo_range = DISTANCE
p = 1
index = [0]*50
view_list = fourtree(session, QUERY_COORDINATE_LIST[p_id], starttime, geo_range, inter_lat, inter_lon)
for view in view_list:
p = 1
while p < 20:
coordinate = view['coordinate']
geo_range = view['geo_range']
temp_time = random.randint(10, 15)
wait_time(temp_time)
info_list = get_weibo_by_coordinate(session, coordinate, starttime, endtime, geo_range, 0, page_count, p, 0)
if info_list:
pd = save_data_by_db(info_list)
if not pd:
break
p += 1
else:
break
sleep_time = random.randint(10, 20)
wait_time(sleep_time)
def search_info(session,
keyword="",
start_time="",
end_time="",
num=1,
location=0):
content_text = str()
id_text = str()
haslink = str()
if location != 0:
haslink = "&haslink=1"
for i in range(START_PAGE, START_PAGE + TOTAL_PAGE):
url = 'http://s.weibo.com/weibo/' + keyword + '&scope=ori' + haslink + '×cope=custom:' + start_time + ':' + end_time + '&page=' + str(
i) + '&rd=newTips'
sleep_time = random.randint(10, 30)
wait_time(sleep_time)
get_text = session.get(url).text
get_text = u'' + get_text
get_text = get_text.encode('utf-8')
content_text = save_catch_page(get_text)
pd = out_page(content_text)
if not pd:
return num
num = get_page_info(content_text, session, location, num)
return num
def fourtree(session, coordinate, starttime, geo_range, inter_lat, inter_lon):
temp_time = random.randint(2, 5)
wait_time(temp_time)
info_list = get_weibo_by_coordinate(session, coordinate, starttime, 0,
geo_range, 0, 50, 20, 0)
if info_list:
print(geo_range)
save_data_by_db(info_list)
else:
print(0)
if info_list:
inter_lat = round(inter_lat / 2, 6)
inter_lon = round(inter_lon / 2, 6)
coordinate1 = dict()
coordinate2 = dict()
coordinate3 = dict()
coordinate4 = dict()
geo_range = float(geo_range)
geo_range = int(round(geo_range / 2 * 1.3))
if geo_range < 100:
geo_range = 100
coordinate1['latitude'] = str(
round(float(coordinate['latitude']) + inter_lat, 6))
coordinate1['longitude'] = str(
round(float(coordinate['longitude']) + inter_lat, 6))
coordinate2['latitude'] = str(
round(float(coordinate['latitude']) - inter_lon, 6))
coordinate2['longitude'] = str(
round(float(coordinate['longitude']) + inter_lon, 6))
coordinate3['latitude'] = str(
round(float(coordinate['latitude']) + inter_lat, 6))
coordinate3['longitude'] = str(
round(float(coordinate['longitude']) - inter_lat, 6))
coordinate4['latitude'] = str(
round(float(coordinate['latitude']) - inter_lon, 6))
coordinate4['longitude'] = str(
round(float(coordinate['longitude']) - inter_lon, 6))
l1 = fourtree(session, coordinate1, starttime, geo_range, inter_lat,
inter_lon)
l2 = fourtree(session, coordinate2, starttime, geo_range, inter_lat,
inter_lon)
l3 = fourtree(session, coordinate3, starttime, geo_range, inter_lat,
inter_lon)
l4 = fourtree(session, coordinate4, starttime, geo_range, inter_lat,
inter_lon)
l5 = [{'coordinate': coordinate, 'geo_range': geo_range}]
return l1 + l2 + l3 + l4 + l5
else:
return [{'coordinate': coordinate, 'geo_range': geo_range}]
def wblogin(username, password):
resp = session.get(
'http://login.sina.com.cn/sso/prelogin.php?'
'entry=sso&callback=sinaSSOController.preloginCallBack&'
'su=%s&rsakt=mod&client=%s' %
(base64.b64encode(username.encode('utf-8')), WBCLIENT)
)
wait_time(3)
pre_login_str = re.match(r'[^{]+({.+?})', resp.text).group(1)
pre_login = json.loads(pre_login_str)
pre_login = json.loads(pre_login_str)
data = {
'entry': 'weibo',
'gateway': 1,
'from': '',
'savestate': 7,
'userticket': 1,
'ssosimplelogin': 1,
'su': base64.b64encode(requests.utils.quote(username).encode('utf-8')),
'service': 'miniblog',
'servertime': pre_login['servertime'],
'nonce': pre_login['nonce'],
'vsnf': 1,
'vsnval': '',
'pwencode': 'rsa2',
'sp': encrypt_passwd(
password, pre_login['pubkey'], pre_login['servertime'], pre_login['nonce']
),
'rsakv': pre_login['rsakv'],
'encoding':'UTF-8',
'prelt': '115',
'url': 'http://weibo.com/ajaxlogin.php?framelogin=1&callback=parent.si'
'naSSOController.feedBackUrlCallBack',
'returntype': 'META',
}
resp = session.post(
'http://login.sina.com.cn/sso/login.php?client=%s' % WBCLIENT, data=data
)
login_url = re.search(r'replace\([\"\']([^\'\"]+)[\"\']', resp.text).group(1)
resp = session.get(login_url)
login_str = re.match(r'[^{]+({.+?}})', resp.text).group(1)
return json.loads(login_str)
def search_info(session, keyword="", start_time="", end_time="", num=1, location=0):
content_text = str()
id_text = str()
haslink = str()
if location != 0:
haslink = "&haslink=1"
for i in range(START_PAGE, START_PAGE+TOTAL_PAGE):
url = 'http://s.weibo.com/weibo/'+keyword+'&scope=ori'+haslink+'×cope=custom:'+start_time+':'+end_time+'&page='+str(i)+'&rd=newTips'
sleep_time = random.randint(10, 30)
wait_time(sleep_time)
get_text = session.get(url).text
get_text = u'' + get_text
get_text = get_text.encode('utf-8')
content_text = save_catch_page(get_text)
pd = out_page(content_text)
if not pd:
return num
num = get_page_info(content_text, session, location, num)
return num
def get_info_history(session):
geo_num = 63
starttime = convert_time('2015', '1', '1', '0')
starttime = str(starttime)
starttime = starttime[:-2]
page_count = 50
endtime = starttime
sort = 0
offset = 0
inter_lat = INTER_LAT
inter_lon = INTER_LON
for p_id in range(0, geo_num):
geo_range = DISTANCE
p = 1
index = [0] * 50
view_list = fourtree(session, QUERY_COORDINATE_LIST[p_id], starttime,
geo_range, inter_lat, inter_lon)
for view in view_list:
p = 1
while p < 20:
coordinate = view['coordinate']
geo_range = view['geo_range']
temp_time = random.randint(10, 15)
wait_time(temp_time)
info_list = get_weibo_by_coordinate(session, coordinate,
starttime, endtime,
geo_range, 0, page_count,
p, 0)
if info_list:
pd = save_data_by_db(info_list)
if not pd:
break
p += 1
else:
break
sleep_time = random.randint(10, 20)
wait_time(sleep_time)
def fourtree(session, coordinate, starttime, geo_range, inter_lat, inter_lon):
temp_time = random.randint(2, 5)
wait_time(temp_time)
info_list = get_weibo_by_coordinate(session, coordinate, starttime, 0, geo_range, 0, 50, 20, 0)
if info_list:
print (geo_range)
save_data_by_db(info_list)
else:
print (0)
if info_list:
inter_lat = round(inter_lat/2, 6)
inter_lon = round(inter_lon/2, 6)
coordinate1 = dict()
coordinate2 = dict()
coordinate3 = dict()
coordinate4 = dict()
geo_range = float(geo_range)
geo_range = int(round(geo_range/2*1.3))
if geo_range < 100:
geo_range = 100
coordinate1['latitude'] = str(round(float(coordinate['latitude']) + inter_lat, 6))
coordinate1['longitude'] = str(round(float(coordinate['longitude']) + inter_lat, 6))
coordinate2['latitude'] = str(round(float(coordinate['latitude']) - inter_lon, 6))
coordinate2['longitude'] = str(round(float(coordinate['longitude']) + inter_lon, 6))
coordinate3['latitude'] = str(round(float(coordinate['latitude']) + inter_lat, 6))
coordinate3['longitude'] = str(round(float(coordinate['longitude']) - inter_lat, 6))
coordinate4['latitude'] = str(round(float(coordinate['latitude']) - inter_lon, 6))
coordinate4['longitude'] = str(round(float(coordinate['longitude']) - inter_lon, 6))
l1 = fourtree(session, coordinate1, starttime, geo_range, inter_lat, inter_lon)
l2 = fourtree(session, coordinate2, starttime, geo_range, inter_lat, inter_lon)
l3 = fourtree(session, coordinate3, starttime, geo_range, inter_lat, inter_lon)
l4 = fourtree(session, coordinate4, starttime, geo_range, inter_lat, inter_lon)
l5 = [{'coordinate': coordinate, 'geo_range': geo_range}]
return l1+l2+l3+l4 +l5
else:
return [{'coordinate': coordinate, 'geo_range': geo_range}]
def get_weibo_by_coordinate(session, coordinate, starttime, endtime, range=2000, sort=0, count=20, page=1, offset=0):
if log_date.log_date.year != datetime.datetime.now():
log_date.change_log_date()
init_log()
num = 0
pd_403 = [0] * len(APP_SOURCE_LIST)
end_403 = [1] * len(APP_SOURCE_LIST)
while True:
try:
app_id = random.randint(0, len(APP_SOURCE_LIST)-1)
url = "http://api.weibo.com/2/place/nearby_timeline.json?"
url += "source="+APP_SOURCE_LIST[app_id]
url += "&lat="+coordinate['latitude']+"&long="+coordinate['longitude']
url += "&starttime="+str(starttime)+"&range="+str(range)+"&sort="+str(sort)
url += "&count="+str(count)+"&page="+str(page)+"&offset="+str(offset)
text = session.get(url)
if text.status_code == 403:
pd_403[app_id] = 1
if pd_403 == end_403:
sleep_time = 15600
else:
sleep_time = random.randint(12, 30)
wait_time(sleep_time)
continue
break
except ConnectionError:
num += 1
lg_warning(ConnectionError)
lg_debug('connect fail'+str(num))
sleep_time = random.randint(6, 10)
wait_time(str(sleep_time))
continue
except Exception:
num += 1
print('Connection reset by peer error')
lg_warning(Exception)
lg_debug('Connection reset by peer'+str(num))
sleep_time = random.randint(10, 20)
wait_time(str(sleep_time))
continue
text_dict = None
text_list_dict = None
try:
text_dict = text.json()
if text_dict.has_key('statuses'):
text_list_dict = text_dict['statuses']
lg_debug('success catch the info_list')
else:
lg_debug("get_weibo_by_coordinate: No Json")
except Exception:
lg_warning(Exception.message)
lg_debug("get_weibo_by_coordinate: No Json")
return text_list_dict
def wblogin(username, password):
resp = session.get('http://login.sina.com.cn/sso/prelogin.php?'
'entry=sso&callback=sinaSSOController.preloginCallBack&'
'su=%s&rsakt=mod&client=%s' %
(base64.b64encode(username.encode('utf-8')), WBCLIENT))
wait_time(3)
pre_login_str = re.match(r'[^{]+({.+?})', resp.text).group(1)
pre_login = json.loads(pre_login_str)
pre_login = json.loads(pre_login_str)
data = {
'entry':
'weibo',
'gateway':
1,
'from':
'',
'savestate':
7,
'userticket':
1,
'ssosimplelogin':
1,
'su':
base64.b64encode(requests.utils.quote(username).encode('utf-8')),
'service':
'miniblog',
'servertime':
pre_login['servertime'],
'nonce':
pre_login['nonce'],
'vsnf':
1,
'vsnval':
'',
'pwencode':
'rsa2',
'sp':
encrypt_passwd(password, pre_login['pubkey'], pre_login['servertime'],
pre_login['nonce']),
'rsakv':
pre_login['rsakv'],
'encoding':
'UTF-8',
'prelt':
'115',
'url':
'http://weibo.com/ajaxlogin.php?framelogin=1&callback=parent.si'
'naSSOController.feedBackUrlCallBack',
'returntype':
'META',
}
resp = session.post('http://login.sina.com.cn/sso/login.php?client=%s' %
WBCLIENT,
data=data)
login_url = re.search(r'replace\([\"\']([^\'\"]+)[\"\']',
resp.text).group(1)
resp = session.get(login_url)
login_str = re.match(r'[^{]+({.+?}})', resp.text).group(1)
return json.loads(login_str)
def get_weibo_by_coordinate(session,
coordinate,
starttime,
endtime,
range=2000,
sort=0,
count=20,
page=1,
offset=0):
if log_date.log_date.year != datetime.datetime.now():
log_date.change_log_date()
init_log()
num = 0
pd_403 = [0] * len(APP_SOURCE_LIST)
end_403 = [1] * len(APP_SOURCE_LIST)
while True:
try:
app_id = random.randint(0, len(APP_SOURCE_LIST) - 1)
url = "http://api.weibo.com/2/place/nearby_timeline.json?"
url += "source=" + APP_SOURCE_LIST[app_id]
url += "&lat=" + coordinate['latitude'] + "&long=" + coordinate[
'longitude']
url += "&starttime=" + str(starttime) + "&range=" + str(
range) + "&sort=" + str(sort)
url += "&count=" + str(count) + "&page=" + str(
page) + "&offset=" + str(offset)
text = session.get(url)
if text.status_code == 403:
pd_403[app_id] = 1
if pd_403 == end_403:
sleep_time = 15600
else:
sleep_time = random.randint(12, 30)
wait_time(sleep_time)
continue
break
except ConnectionError:
num += 1
lg_warning(ConnectionError)
lg_debug('connect fail' + str(num))
sleep_time = random.randint(6, 10)
wait_time(str(sleep_time))
continue
except Exception:
num += 1
print('Connection reset by peer error')
lg_warning(Exception)
lg_debug('Connection reset by peer' + str(num))
sleep_time = random.randint(10, 20)
wait_time(str(sleep_time))
continue
text_dict = None
text_list_dict = None
try:
text_dict = text.json()
if text_dict.has_key('statuses'):
text_list_dict = text_dict['statuses']
lg_debug('success catch the info_list')
else:
lg_debug("get_weibo_by_coordinate: No Json")
except Exception:
lg_warning(Exception.message)
lg_debug("get_weibo_by_coordinate: No Json")
return text_list_dict
starttime = convert_time('2015', '7', '1', '0')
endtime = starttime
starttime = str(starttime)
starttime = starttime[:-2]
info_list = get_weibo_by_coordinate(session, QUERY_COORDINATE_LIST[p_id],
starttime, endtime, geo_range, 0, page_count, p, 0)
save_data_by_db(info_list)
if not info_list:
continue
else:
pass
length = len(info_list)
cmpstr1 = info_list[length - 1]['mid']
cmpstr2 = index_num[p_id]
index_num[p_id] = info_list[0]['mid']
while arbitrary_precision_compare(cmpstr1, cmpstr2) == 1:
p += 1
info_list = get_weibo_by_coordinate(session, QUERY_COORDINATE_LIST[p_id],
starttime, endtime, geo_range, 0, page_count, p, 0)
save_data_by_db(info_list)
if not info_list:
break
else:
pass
length = len(info_list)
cmpstr1 = info_list[length - 1]['mid']
if p >= 20:
break
sleep_time = random.randint(10, 20)
wait_time(sleep_time)
