def test_companies_parse_to_csv(self):
srconf_path = os.path.join(WEB_SOURCES_CONFIG_DIR,
'web_statgov_companies.json')
jobconf_path = os.path.join(JOBS_CONFIG_DIR, 'to_csv.json')
src_json = Utils.read_file(srconf_path)
job_json = Utils.read_file(jobconf_path)
download_handler = HandlersFactory.get_handler(
Downloader.handler_name(src_json))
service = Downloader(src_json, download_handler)
downloaded_file = service.download()
extract_handler = HandlersFactory.get_handler(
Extractor.handler_name(src_json))
service = Extractor(src_json, downloaded_file, self.temp_path,
extract_handler)
service.extract()
xlspaths = service.path(src_json, self.temp_path)
parse_handler = HandlersFactory.get_handler(
XLSParser.handler_name(src_json, job_json))
service = XLSParser(src_json, job_json, xlspaths, self.data_path,
parse_handler)
csvfile = service.path(src_json, job_json, self.data_path)
rows_cnt = service.parse()
self.assertTrue(os.path.exists(csvfile))
self.assertGreater(rows_cnt, 0)
def test_wrong_address_to_csv(self):
srconf_path = os.path.join(WEB_SOURCES_CONFIG_DIR,
'web_kgdgov_wrong_address.json')
jobconf_path = os.path.join(JOBS_CONFIG_DIR, 'to_csv.json')
src_json = Utils.read_file(srconf_path)
job_json = Utils.read_file(jobconf_path)
download_handler = HandlersFactory.get_handler(
Downloader.handler_name(src_json))
service = Downloader(src_json, download_handler)
downloaded_file = service.download()
parse_handler = HandlersFactory.get_handler(
XLSParser.handler_name(job_json))
service = XLSParser(src_json, job_json, downloaded_file,
self.data_path, parse_handler)
csvfile = service.path(src_json, job_json, self.data_path)
rows_cnt = service.parse()
self.assertTrue(os.path.exists(csvfile))
self.assertGreater(rows_cnt, 0)