Exemplo n.º 1
0
 def test_list_warcs(self):
     harvester = BaseHarvester(self.working_path, host="localhost")
     write_fake_warc(
         self.working_path,
         "test_1-20151109195229879-00000-97528-GLSS-F0G5RP-8000.warc.gz")
     write_fake_warc(
         self.working_path,
         "test_1-20151109195229879-00001-97528-GLSS-F0G5RP-8000.warc")
     write_fake_warc(
         self.working_path,
         "test_1-20151109195229879-00002-97528-GLSS-F0G5RP-8000")
     os.mkdir(
         os.path.join(
             self.working_path,
             "test_1-20151109195229879-00003-97528-GLSS-F0G5RP-8000.warc.gz"
         ))
     warc_dirs = harvester._list_warcs(self.working_path)
     self.assertSetEqual(
         {
             "test_1-20151109195229879-00000-97528-GLSS-F0G5RP-8000.warc.gz",
             "test_1-20151109195229879-00001-97528-GLSS-F0G5RP-8000.warc"
         }, set(warc_dirs))