from spiders.fdb_spider import *
|
|
|
|
|
|
import sys
|
|
|
|
config = "spiders/config.yaml"
|
|
list_of_fdbs = eval(sys.argv[1])
|
|
print(list_of_fdbs)
|
|
#list_of_fdbs = ["foerderinfo.bund.de-bekanntmachungen"]
|
|
|
|
|
|
# doing the crawling of government websites
|
|
|
|
spider = fdb_spider(config)
|
|
|
|
#spider.download_entry_list_pages_of_funding_databases(list_of_fdbs)
|
|
|
|
#spider.find_config_parameter(list_of_fdbs)
|
|
|
|
#spider.parse_entry_list_data2dictionary(list_of_fdbs)
|
|
|
|
spider.download_entry_data_htmls(list_of_fdbs)
|
|
|
|
spider.parse_entry_data2dictionary(list_of_fdbs)
|
|
|