def run(self):
src_file = os.path.join(WEB_SOURCES_CONFIG_DIR, str(self.sourcefile))
src_conf = Utils.read_file(src_file)
job_file = os.path.join(JOBS_CONFIG_DIR, str(self.jobfile))
job_conf = Utils.read_file(job_file)
handler = HandlersFactory.get_handler(
XLSParser.handler_name(src_conf, job_conf))
service = XLSParser(src_conf, job_conf, [f.path for f in self.input()],
WEB_DATA_PATH, handler)
service.parse()
def output(self):
src_file = os.path.join(WEB_SOURCES_CONFIG_DIR, str(self.sourcefile))
src_conf = Utils.read_file(src_file)
job_file = os.path.join(JOBS_CONFIG_DIR, str(self.jobfile))
job_conf = Utils.read_file(job_file)
return luigi.LocalTarget(
XLSParser.path(src_conf, job_conf, WEB_DATA_PATH))
def test_kurk_parse_to_csv(self):
srconf_path = os.path.join(WEB_SOURCES_CONFIG_DIR,
'web_statgov_kurk.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(src_json, 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)