def do_scrape(
download_dir="download", download_processes_num=8, except_log_file_name="request_exceptions_discoverireland.log"
):
global img_dir
global processes_num
processes_num = download_processes_num
img_dir = site_subdir_creator(img_dir)(download_dir)
except_log_file = open(except_log_file_name, "w")
start_time = time.time()
discoverireland_data = process_site_layers(site_layers_description_list, except_log_file)
print "discoverireland.ie scrapping time: ", str(time.time() - start_time)
save_data(discoverireland_data, "discoverireland.dat")
except_log_file.close()
discoverireland_data = get_saved_data("discoverireland.dat")
to_csv(discoverireland_data)
global img_dir
global processes_num
processes_num = download_processes_num
img_dir = site_subdir_creator(img_dir)(download_dir)
except_log_file = open(except_log_file_name, "w")
start_time = time.time()
discoverireland_data = process_site_layers(site_layers_description_list, except_log_file)
print "discoverireland.ie scrapping time: ", str(time.time() - start_time)
save_data(discoverireland_data, "discoverireland.dat")
except_log_file.close()
discoverireland_data = get_saved_data("discoverireland.dat")
to_csv(discoverireland_data)
if __name__ == "__main__":
create_dirs(img_dir)
exceptions_log_file = open("discoverireland_exceptions.log", "w")
start_time = time.time()
discoverireland_data = process_site_layers(site_layers_description_list, exceptions_log_file)
print "discoverireland.ie scrapping time: ", str(time.time() - start_time)
save_data(discoverireland_data, "discoverireland.dat")
exceptions_log_file.close()
discoverireland_data = get_saved_data("discoverireland.dat")
to_csv(discoverireland_data)
# exceptions_log_file = open('discoverireland_exceptions.log', 'w')
# first_layer_processor(site_layers_description_list[0], exceptions_log_file)
# exceptions_log_file.close()
