def get_data_projects(url):
try:
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
# r = session.get(url, headers=headers_xml, timeout=30)
r = session.get(url, headers=headers, timeout=20)
if r.status_code != 200:
# logger.debug('status_code', r.status_code)
return None
soup = BeautifulSoup(r.content, 'html.parser')
project_links = soup.find('div', {'id': 'search-results'}).find_all('a')
result = set()
for link in project_links:
if link['href']:
result.add(link['href'])
# logger.debug(result, len(result))
except requests.exceptions.RequestException as e:
# logger.debug(e)
return None
def get_data_graph(url, month_target_1, month_target_2):
"""
13.07 add:
Area's median sale price/sqm 1 year ago, Area's median sale price/sqm 8 months ago
"""
try:
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
# r = session.get(url, headers=headers_xml, timeout=30)
# print(url)
r = session.get(url, headers=headers_xml, timeout=35)
if r.status_code != 200:
logger.debug('status_code', r.status_code)
return None
data = r.json()['msg']
# print(r.text)
try:
median_sale_prices_sqm = data.split("'sqmSale': {")[1].split(
"data: [")[1].split("],")[0].split(',')
except:
median_sale_prices_sqm = []
try:
months = data.split("labels: [")[1].split("],")[0].split(",")
months = [x.replace('"', '') for x in months]
logger.debug("months {}".format(len(months)))
target_month_position = months.index(month_target_1)
logger.debug(target_month_position)
result1 = median_sale_prices_sqm[target_month_position]
except:
result1 = ''
try:
median_rent_prices_sqm = data.split("'sqmRent': {")[1].split(
"data:[")[1].split("],")[0].split(',')
target_month_position_2 = months.index(month_target_2)
result2 = median_rent_prices_sqm[target_month_position_2]
except:
result2 = ''
try:
area_median_sale_price_sqm_1_year_ago = median_sale_prices_sqm[
-13:-12][0]
#print(area_median_sale_price_sqm_1_year_ago)
except:
area_median_sale_price_sqm_1_year_ago = ''
try:
area_median_sale_price_sqm_8_month_ago = median_sale_prices_sqm[
-9:-8][0]
#print(area_median_sale_price_sqm_8_month_ago)
except:
area_median_sale_price_sqm_8_month_ago = ''
#try:
# proj_media_sale_price = \
# data.split('borderColor: "rgb(215, 219, 221)",')[1].split("label: htmlDecode('Average enquiry sale price/sqm.'")[0].replace("data: [", "").replace("],", "").replace("\n", "").replace("\t", "").split(',')
# proj_media_sale_price = [d.strip(" ") for d in proj_media_sale_price]
# # print(proj_media_sale_price)
#except:
# proj_media_sale_price = []
#
#try:
# proj_media_sale_price_sqm_1_year_ago = proj_media_sale_price[-12]
#except:
# proj_media_sale_price_sqm_1_year_ago = ''
#
#try:
# proj_media_sale_price_sqm_8_month_ago = proj_media_sale_price[-8]
#except:
# proj_media_sale_price_sqm_8_month_ago = ''
# logger.debug(f'{r.url}:\n\t Result: {result1}; {result2}; {area_median_sale_price_sqm_1_year_ago}; {area_median_sale_price_sqm_8_month_ago};') # {proj_media_sale_price_sqm_1_year_ago}; {proj_media_sale_price_sqm_8_month_ago}')
"""
The data in these columns should be interchanged, so the 3 digit numbers in here and the more than 5 -6 digit numbers in column CC
"""
return result2, result1, area_median_sale_price_sqm_1_year_ago, area_median_sale_price_sqm_8_month_ago, # proj_media_sale_price_sqm_1_year_ago, proj_media_sale_price_sqm_8_month_ago
except:
return '', '', '', ''
def get_data_projects(urls):
links_to_save = []
column_i = 2
# result = []
for url in urls:
result = []
# print(url)
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
#r = session.get(url, headers=headers_xml, timeout=30)
r = session.get(url, headers=headers, timeout=20)
if r.status_code != 200:
print('status_code', r.status_code)
return None
soup = BeautifulSoup(r.content, 'html.parser')
try:
number_of_projects = int(
soup.find('span', {
'id': 'properties_total'
}).text)
except:
number_of_projects = 0
# print('number_of_projects', number_of_projects)
result.append(number_of_projects)
if number_of_projects > 0:
try:
count_pages = math.ceil(number_of_projects / 20)
except:
count_pages = 0
# print('count_pages', count_pages)
project_links = soup.find('div', {
'id': 'search-results'
}).find_all('a')
for link in project_links:
if link['href'] and link['href'] not in result:
result.append(link['href'])
links_to_save.append(link['href'])
if count_pages > 1:
for i in range(2, count_pages + 1):
url_n_page = "{}?page={}".format(url, i)
# print(url_n_page)
# https://www.dotproperty.co.th/en/condos/all/Chiang-Mai?page=2
r = session.get(url_n_page, headers=headers, timeout=20)
if r.status_code != 200:
print('status_code', r.status_code)
return None
# TODO: add attempts here and everywhere in project
soup = BeautifulSoup(r.content, 'html.parser')
project_links = soup.find('div', {
'id': 'search-results'
}).find_all('a')
for link in project_links:
if link['href'] and link['href'] not in result:
result.append(link['href'])
links_to_save.append(link['href'])
# print(result, len(result))
# post here
wks.update_col(column_i, result, row_offset=2)
column_i += 1
# return result
# TODO: Save to file cont
file_object = open('data/links_th_condo_thai_prov.json',
'w') # Thai province URLs (condominums)
json.dump(links_to_save, file_object, indent=4)
def get_data_condo(url):
if url == '/cdn-cgi/l/email-protection':
return "", "", "", "", ""
try:
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
# r = session.get(url, headers=headers_xml, timeout=30)
r = session.get(url, headers=headers, timeout=20)
if r.status_code != 200:
logger.debug('status_code', r.status_code, url)
return ["", "", "", "", ""]
soup = BeautifulSoup(r.content, 'html.parser')
# print(soup)
if input_tab_name.split(" ")[0] == "Vietnam":
# print("vi")
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
vi_url = url.replace("/en", "")
vi_r = session.get(vi_url, headers=headers, timeout=20)
# print(vi_url)
vi_soup = BeautifulSoup(vi_r.content, 'html.parser')
try:
breadcrumbs = vi_soup.find('ol', {'class': 'breadcrumb'}).find_all('li')
except:
breadcrumbs = ''
try:
province = breadcrumbs[3].text.strip()
except:
province = ''
if len(breadcrumbs) > 5:
city = breadcrumbs[4].text.strip()
# print(city)
else:
city = ''
if len(breadcrumbs) > 6:
area = breadcrumbs[5].text.strip()
else:
area = ''
else:
try:
breadcrumbs = soup.find('ol', {'class': 'breadcrumb'}).find_all('li')
except:
breadcrumbs = ''
try:
province = breadcrumbs[3].text.strip()
except:
province = ''
if len(breadcrumbs) > 5:
city = breadcrumbs[4].text.strip()
else:
city = ''
if len(breadcrumbs) > 6:
area = breadcrumbs[5].text.strip()
else:
area = ''
try:
try:
condo_name = soup.find('div', {'class': 'row top-navigation-bar add-padding'}).find('a').text.strip()
except:
condo_name = ''
try:
developer_name = soup.find('div', {'class': 'col-sm-6 nav-top-btngroups text-right'}).find('li').find(
'p').text.strip()
except:
developer_name = ''
# total_units = soup.find('div', {'class': 'col-md-12 col-lg-8 project-content'}).find('section').text.strip().split('contains ')[1].split(' total')[0]
total_units_raw = \
soup.find('div', {'class': 'col-md-12 col-lg-8 project-content'}).find('section').text.strip().split(
' total units')[0].split(' ')[-1]
if any(char.isdigit() for char in total_units_raw):
total_units = total_units_raw
else:
total_units = ''
except:
total_units = ''
try:
units_for_rent = soup.find('a', {'id': 'open-tab-rent'}).text.strip().split(' ')[0]
except:
units_for_rent = ''
number_of_studios = scrape_room_types_prices(soup)
start_from, sqm = scrape_rent_units_listing(soup)
lowest_ask_price, sqm_ask = scrape_rent_sale_listing(soup)
other_projects_nearby = scrape_other_projects_nearby(soup)
popular_condos_in_area = scrape_popular_condos_in_area(soup)
room_types_prices = scrape_room_types_prices_ext(soup)
s_n_of_units_for_rent, s_av_rent, s_av_ask_price, bd1_n_of_units_for_rent, bd1_av_rent, bd1_av_ask_price, bd2_n_of_units_for_rent, bd2_av_rent, bd2_av_ask_price, bd3_n_of_units_for_rent, bd3_av_rent, bd3_av_ask_price, bd4_n_of_units_for_rent, bd4_av_rent, bd4_av_ask_price = '', '', '', '', '', '', '', '', '', '', '', '', '', '', ''
s_n_of_units_for_sale, bd1_n_of_units_for_sale, bd2_n_of_units_for_sale, bd3_n_of_units_for_sale, bd4_n_of_units_for_sale = '', '', '', '', ''
s_size, bd1_size, bd2_size, bd3_size, bd4_size = '', '', '', '', ''
for t in room_types_prices:
if t['type'] == 'Studio':
s_size = t['size']
s_n_of_units_for_rent = t['number of units for rent']
s_av_rent = t['average rent']
s_av_ask_price = t['average ask price']
s_n_of_units_for_sale = t['number of units for sale']
if t['type'] == '1 Bedroom' or t['type'] == '1 Phòng Ngủ':
bd1_size = t['size']
bd1_n_of_units_for_rent = t['number of units for rent']
bd1_av_rent = t['average rent']
bd1_av_ask_price = t['average ask price']
bd1_n_of_units_for_sale = t['number of units for sale']
if t['type'] == '2 Bedrooms' or t['type'] == '2 Phòng Ngủ':
bd2_size = t['size']
bd2_n_of_units_for_rent = t['number of units for rent']
bd2_av_rent = t['average rent']
bd2_av_ask_price = t['average ask price']
bd2_n_of_units_for_sale = t['number of units for sale']
if t['type'] == '3 Bedrooms' or t['type'] == '3 Phòng Ngủ':
bd3_size = t['size']
bd3_n_of_units_for_rent = t['number of units for rent']
bd3_av_rent = t['average rent']
bd3_av_ask_price = t['average ask price']
bd3_n_of_units_for_sale = t['number of units for sale']
if t['type'] == '4 Bedrooms' or t['type'] == '4 Phòng Ngủ':
bd4_size = t['size']
bd4_n_of_units_for_rent = t['number of units for rent']
bd4_av_rent = t['average rent']
bd4_av_ask_price = t['average ask price']
bd4_n_of_units_for_sale = t['number of units for sale']
graph_link = resolve_graph_link(url)
# logger.debug(graph_link)
try:
median_rent_price_sqm, median_sale_price_sqm, earliest_median_sale_price_sqm, earliest_month, \
earliest_median_rent_price_sqm, earliest_month_1, project_median_sale_price_sqm_1_year_ago, \
project_median_sale_price_sqm_8_month_ago, median_sale_price = get_data_graph(graph_link)
except TypeError as err:
median_rent_price_sqm, median_sale_price_sqm, earliest_median_sale_price_sqm, earliest_month, \
earliest_median_rent_price_sqm, earliest_month_1, project_median_sale_price_sqm_1_year_ago, \
project_median_sale_price_sqm_8_month_ago, median_sale_price = '', '', '', '', '', '', '', '', ''
logger.debug(err)
logger.debug("get data graph err")
# logger.debug('DEBUG s_n_of_units_for_rent:', s_n_of_units_for_rent)
try:
gps_str = soup.find('a', {'id': 'go-to-map-mobile'}).find('img')['src'].split('map_')[1].split('.jpg')[0]
gps_lat, gps_long = gps_str.split('_')
except:
gps_lat, gps_long = '', ''
result_bulk_1 = {
'url': url,
'condo_name': condo_name,
'developer_name': developer_name,
'province': province,
'city': city
}
# print(result_bulk_1)
result_bulk_2 = {
'area': area,
}
result_bulk_3 = {
'median_rent_price_sqm': median_rent_price_sqm, # L
's_size': s_size, #
'bd1_size': bd1_size,
'bd2_size': bd2_size,
'bd3_size': bd3_size,
'bd4_size': bd4_size, # Q
'median_sale_price_sqm': median_sale_price_sqm,
's_n_of_units_for_sale': s_n_of_units_for_sale,
'bd1_n_of_units_for_sale': bd1_n_of_units_for_sale,
'bd2_n_of_units_for_sale': bd2_n_of_units_for_sale,
'bd3_n_of_units_for_sale': bd3_n_of_units_for_sale,
'bd4_n_of_units_for_sale': bd4_n_of_units_for_sale, # W
'earliest_median_rent_price_sqm': earliest_median_rent_price_sqm, # X
'earliest_month_1': earliest_month_1,
'earliest_median_sale_price_sqm': earliest_median_sale_price_sqm, # Z
'earliest_month': earliest_month,
'total_units': total_units, # AB col
'part2_28_09-1': s_n_of_units_for_rent,
'part2_28_09-2': check_outlier(s_av_rent, filter_values_int[0], filter_values_int[1]),
'part2_28_09-3': check_outlier(s_av_ask_price, filter_values_int[2], filter_values_int[3]),
'part2_28_09-4': bd1_n_of_units_for_rent,
'part2_28_09-5': check_outlier(bd1_av_rent, filter_values_int[4], filter_values_int[5]),
'part2_28_09-6': check_outlier(bd1_av_ask_price, filter_values_int[6], filter_values_int[7]),
'part2_28_09-7': bd2_n_of_units_for_rent,
'part2_28_09-8': check_outlier(bd2_av_rent, filter_values_int[8], filter_values_int[9]),
'part2_28_09-9': check_outlier(bd2_av_ask_price, filter_values_int[10], filter_values_int[11]),
'part2_28_09-10': bd3_n_of_units_for_rent,
'part2_28_09-11': check_outlier(bd3_av_rent, filter_values_int[12], filter_values_int[13]),
'part2_28_09-12': check_outlier(bd3_av_ask_price, filter_values_int[14], filter_values_int[15]),
'part2_28_09-13': bd4_n_of_units_for_rent,
'part2_28_09-14': check_outlier(bd4_av_rent, filter_values_int[16], filter_values_int[17]),
'part2_28_09-15': check_outlier(bd4_av_ask_price, filter_values_int[18], filter_values_int[19]),
'start_from': start_from, # Lowest rent # AR col
'sqm': sqm,
'lowest_ask_price': lowest_ask_price, # AT col
'sqm_ask': sqm_ask,
'number_of_studios': number_of_studios,
'lat': gps_lat,
'long': gps_long,
'other_projects_nearby': other_projects_nearby,
}
result_bulk_4 = {
'project_median_sale_price_sqm_1_year_ago': project_median_sale_price_sqm_1_year_ago,
'project_median_sale_price_sqm_8_month_ago': project_median_sale_price_sqm_8_month_ago,
}
result_bulk_5 = {
'median_sale_price': median_sale_price
}
"""result = []
result.append(result_bulk_1)
result.append(result_bulk_2)
result.append(result_bulk_3)
result.append(result_bulk_4)
result.append(result_bulk_5)"""
return result_bulk_1, result_bulk_2, result_bulk_3, result_bulk_4, result_bulk_5
except requests.exceptions.RequestException as e:
# return {}, {}, {}, {}, {}
logger.debug(e)
def get_data_graph(url):
"""
13.07: add Project's median sale price/sqm 1 year ago,
Project's median sale price/sqm 8 months ago,
Median sale price
"""
try:
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
# r = session.get(url, headers=headers_xml, timeout=30)
r = session.get(url, headers=headers_xml, timeout=30)
if r.status_code != 200:
# logger.debug('status_code', r.status_code)
return None
data = r.json()['msg']
#print(r.text)
if r.url == "https://www.dotproperty.co.th/en":
return '', '', '', '', '', '', '', '', ''
# logger.debug(data)
try:
median_rent_price_sqm = data.split("'sqmRent': {")[1].split("data:[")[1].split("],")[0].split(',')[-1]
if median_rent_price_sqm is None:
median_rent_price_sqm = ''
except:
median_rent_price_sqm = ''
try:
earliest_median_rent_prices_sqm = data.split("'sqmRent': {")[1].split("data:[")[1].split("],")[0].split(',')
# print(earliest_median_rent_prices_sqm)
if earliest_median_rent_prices_sqm is None:
earliest_median_rent_prices_sqm = ''
except:
earliest_median_rent_price_sqm = ''
try:
median_sale_price_sqm = data.split("'sqmSale': {")[1].split("data:[")[1].split("],")[0].split(',')[-1]
if median_sale_price_sqm is None:
median_sale_price_sqm = ''
except:
median_sale_price_sqm = ''
try:
earliest_median_sale_prices_sqm = data.split("'sqmSale': {")[1].split("data:[")[1].split("],")[0].split(',')
# print(earliest_median_sale_prices_sqm)
if earliest_median_sale_prices_sqm is None:
earliest_median_sale_prices_sqm = ''
except:
earliest_median_sale_prices_sqm = ''
# earliest_median_sale_price_sqm = ''
try:
# TODO soup find earliest_median_sale_price_sqm
earliest_median_sale_price_sqm = ''
if earliest_median_sale_prices_sqm == '':
raise Exception
month_num = 0
# print(earliest_median_sale_prices_sqm)
for price in earliest_median_sale_prices_sqm:
month_num += 1
if price != '':
earliest_median_sale_price_sqm = price
break
months = data.split("labels: [")[1].split("],")[0].split(",")
earliest_month = months[month_num - 1].replace('"', '')
except:
earliest_median_sale_price_sqm = ''
earliest_month = ''
earliest_median_rent_price_sqm = ''
try:
# TODO soup find
earliest_median_rent_price_sqm = ''
if earliest_median_rent_prices_sqm == '':
raise Exception
month_num = 0
# print(earliest_median_rent_prices_sqm)
for price in earliest_median_rent_prices_sqm:
month_num += 1
if price != '' and price is not None:
earliest_median_rent_price_sqm = price
break
months = data.split("labels: [")[1].split("],")[0].split(",")
earliest_month_1 = months[month_num - 1].replace('"', '')
if earliest_month_1 is None:
earliest_month_1 = ''
except:
earliest_median_rent_price_sqm = ''
earliest_month_1 = ''
try:
project_median_sale_price_sqm_1_year_ago = \
data.split("'sqmSale': {")[1].split("data:[")[1].split("],")[0].split(',')[-12] #[0]
# print("1-{}".format(project_median_sale_price_sqm_1_year_ago))
if project_median_sale_price_sqm_1_year_ago is None:
project_median_sale_price_sqm_1_year_ago = ''
except:
project_median_sale_price_sqm_1_year_ago = ''
try:
project_median_sale_price_sqm_8_month_ago = \
data.split("'sqmSale': {")[1].split("data:[")[1].split("],")[0].split(',')[-8] #[0]
# print("8-{}".format(project_median_sale_price_sqm_8_month_ago))
if project_median_sale_price_sqm_8_month_ago is None:
project_median_sale_price_sqm_8_month_ago = ''
except:
project_median_sale_price_sqm_8_month_ago = ''
try:
median_sale_price = data.split("'sale': {")[1].split("data:[")[1].split("],")[0].split(',')[-1]
# print("median sale price-{}".format(median_sale_price))
if median_sale_price is None:
median_sale_price = ''
except:
median_sale_price = ''
# logger.debug("get_data_graph:" ,median_rent_price_sqm, median_sale_price_sqm, earliest_median_sale_price_sqm, earliest_month, earliest_median_rent_price_sqm, earliest_month_1)
return median_rent_price_sqm, median_sale_price_sqm, earliest_median_sale_price_sqm, earliest_month, earliest_median_rent_price_sqm, earliest_month_1, project_median_sale_price_sqm_1_year_ago, project_median_sale_price_sqm_8_month_ago, median_sale_price
except requests.exceptions.RequestException as e:
logger.debug(e)
return '', '', '', '', '', '', '', '', ''
def get_data_projects(url):
# links_to_save = []
# result = []
result = []
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
r = session.get(url, headers=headers, timeout=20, allow_redirects=True)
if r.status_code == 503:
logger.debug('request was block - {}'.format(r.status_code))
if r.status_code != 200:
logger.debug('status_code {}'.format(r.status_code))
return None
soup = BeautifulSoup(r.content, 'html.parser')
try:
number_of_projects = int(
soup.find('span', {
'id': 'properties_total'
}).text.replace(',', ''))
# print(number_of_projects)
except:
number_of_projects = 0
logger.debug('number_of_projects {}'.format(number_of_projects))
result.append(number_of_projects)
if number_of_projects > 0:
try:
count_pages = math.ceil(number_of_projects / 20)
except:
count_pages = 0
logger.debug('count_pages {}'.format(count_pages))
project_links = soup.find('div', {
'id': 'search-results'
}).find_all('a')
for link in project_links:
if link['href'] and link['href'] not in result:
result.append(link['href'])
links_to_save.append(link['href'])
if count_pages > 1:
for i in range(2, count_pages + 1):
logger.debug('page {}'.format(i))
url_n_page = "{}?page={}".format(url, i)
logger.debug(url_n_page)
r = session.get(url_n_page, headers=headers, timeout=20)
if r.status_code != 200:
logger.debug('status_code', r.status_code)
return None
# TODO: add attempts here and everywhere in project
soup = BeautifulSoup(r.content, 'html.parser')
project_links = soup.find('div', {
'id': 'search-results'
}).find_all('a')
for link in project_links:
if link['href'] and link['href'] not in result:
result.append(link['href'])
links_to_save.append(link['href'])
def get_data_graph(url):
logger.debug(url)
if url is None:
logger.debug("url is NONE")
return
session = requests.Session()
adapter = requests.adapters.HTTPAdapter(max_retries=20)
session.mount('https://', adapter)
session.mount('http://', adapter)
session.proxies.update(get_proxy())
r = session.get(url, headers=headers_xml, timeout=30)
if r.status_code != 200:
logger.debug('status_code {}'.format(r.status_code))
return None
data = r.json()['msg']
try:
median_rent_price_sqm = data.split("'sqmRent': {")[1].split(
"data:[")[1].split("],")[0].split(',')[-1]
except:
median_rent_price_sqm = ''
try:
median_sale_price_sqm = data.split("'sqmSale': {")[1].split(
"data: [")[1].split("],")[0].split(',')[-1]
except:
median_sale_price_sqm = ''
try:
earliest_median_sale_prices_sqm = data.split("'sqmSale': {")[1].split(
"data: [")[1].split("],")[0].split(',')
except:
earliest_median_sale_prices_sqm = ''
try:
earliest_median_rent_prices_sqm = data.split("'sqmRent': {")[1].split(
"data:[")[1].split("],")[0].split(',')
except:
earliest_median_rent_prices_sqm = ''
try:
earliest_median_sale_price_sqm = ''
if earliest_median_sale_prices_sqm == '':
raise Exception
month_num = 0
for price in earliest_median_sale_prices_sqm:
month_num += 1
if price != '':
earliest_median_sale_price_sqm = price
break
months = data.split("labels: [")[1].split("],")[0].split(",")
earliest_month = months[month_num - 1].replace('"', '')
except:
earliest_month = ''
try:
earliest_median_rent_price_sqm = ''
if earliest_median_rent_prices_sqm == '':
raise Exception
month_num = 0
for price in earliest_median_rent_prices_sqm:
month_num += 1
if price != '':
earliest_median_rent_price_sqm = price
break
months = data.split("labels: [")[1].split("],")[0].split(",")
if earliest_median_rent_price_sqm == '':
earliest_month_1 = ''
else:
earliest_month_1 = months[month_num - 1].replace('"', '')
except:
earliest_month_1 = ''
# print(data)
if input_tab_name == "Vietnam townhouses":
try:
part_2_1 = data.split("htmlDecode('Giá bán trung bình")[1].split(
"data: [")[1].split("],")[0].split(',')[-1]
except:
part_2_1 = ''
try:
part_2_2 = data.split("htmlDecode('Giá bán trung bình'\n")[
1].split("data:[")[1].split("],")[0].split(',')[-1]
except:
part_2_2 = ''
try:
part_2_3 = data.split("htmlDecode('Giá cho thuê trung bình")[
1].split("type: 'bar'")[1].split("data:[")[1].split(
"],")[0].split(',')[-1]
except:
part_2_3 = ''
try:
part_2_4 = data.split("htmlDecode('Giá cho thuê trung bình'\n")[
1].split("data:[")[1].split("],")[0].split(',')[-1]
except:
part_2_4 = ''
else:
try:
part_2_1 = data.split("htmlDecode('Median sale price")[1].split(
"data: [")[1].split("],")[0].split(',')[-1]
except:
part_2_1 = ''
try:
part_2_2 = data.split("htmlDecode('Median sale price'\n")[1].split(
"data:[")[1].split("],")[0].split(',')[-1]
except:
part_2_2 = ''
try:
part_2_3 = data.split("htmlDecode('Median rent price")[1].split(
"type: 'bar'")[1].split("data:[")[1].split("],")[0].split(
',')[-1]
except:
part_2_3 = ''
try:
part_2_4 = data.split("htmlDecode('Median rent price'\n")[1].split(
"data:[")[1].split("],")[0].split(',')[-1]
except:
part_2_4 = ''
try:
part_2_5 = data.split("'sale': {")[1].split("data:[")[1].split(
"],")[0].split(',')[-1]
except:
part_2_5 = ''
# return median_rent_price_sqm, earliest_median_sale_price_sqm, earliest_month, median_sale_price_sqm, earliest_median_rent_price_sqm, earliest_month_1, '', part_2_1, part_2_2, part_2_3, part_2_4,'',part_2_5
logger.debug(part_2_4, median_rent_price_sqm, part_2_3,
earliest_median_rent_price_sqm, earliest_month_1, part_2_5,
part_2_2, median_sale_price_sqm, part_2_1,
earliest_median_sale_price_sqm, earliest_month)
return part_2_4, median_rent_price_sqm, part_2_3, earliest_median_rent_price_sqm, earliest_month_1, part_2_5, part_2_2, median_sale_price_sqm, part_2_1, earliest_median_sale_price_sqm, earliest_month
