def test_thesis(self):
workdir = get_working_dir()
for testname in ['T001', 'T002', 'T003', 'T004', 'T005', 'T006']:
for domain in sorted(self.journals):
journal = self.journals[domain]
maschine = MaschineLSH(config, journal, outdir=workdir, testname=testname)
maschine.process()
def test_journal_001(self):
test_domain = self.test_domain
test_path = config.get(test_domain, 'check_path1')
test_ts = config.getint(test_domain, 'check_ts11')
test_wlid = config.get(test_domain, 'check_uid11')
workdir = get_working_dir()
maschine = MaschineLSH(config, self.journal, outdir=workdir)
self.assertEqual(test_domain, self.journal.domain)
journal_entry = maschine.journal.get_entry(test_path, test_ts)
self.assertEqual(test_wlid, journal_entry['warc_entry'].wlid)
cross = maschine.get_cross(journal_entry)
self.assertEqual(
config.get(test_domain, 'check_cross11'),
cross.get_unique_id()
)
self.assertEqual((1, 1, 1, 1, 1), cross.to_bits())
self.assertEqual(test_wlid, cross.cur['warc_entry'].wlid)
diffmethod = DBCEMethodLSH(config, cross)
def test_process_001(self):
test_domain = self.test_domain
test_path = config.get(test_domain, 'check_path1')
test_ts = config.getint(test_domain, 'check_ts11')
test_wlid = config.get(test_domain, 'check_uid11')
workdir = get_working_dir()
for testname in ['T001', 'T002', 'T006']:
maschine = MaschineLSH(config, self.journal, outdir=workdir, testname=testname)
maschine.process()
parser.error('missing arguments')
elif args.report_dir:
try:
erc = load_eval_results(config, args.report_dir)
except:
import sys
import traceback
atype, value, tb = sys.exc_info()
traceback.print_exc()
pdb.post_mortem(tb)
elif args.mode == 'idx':
journals = idx_to_journals(config, args.idx_sources)
if args.lsh_only and not args.manual_annotate:
workdir = get_working_dir()
for domain in sorted(journals):
journal = journals[domain]
try:
maschine = MaschineLSH(config, journal, outdir=workdir)
maschine.process()
except:
import sys
import traceback
atype, value, tb = sys.exc_info()
traceback.print_exc()
pdb.post_mortem(tb)
elif args.manual_annotate:
if len(args.domain) != 1:
parser.error('manual annotation needs exactly one domain')