def fetch(fmgr):
"""
Fetch verdicts.
"""
crawler = FJUDCrawler()
while fmgr.has_fjud_record():
rec = fmgr.pop_fjud_record()
crawler.fetch(court=rec.court,
needle=rec.needle,
year=rec.get_start_year())
fmgr.remove_fjud_record(rec)
print("Sleep 60s for polite interaction with the server.")
time.sleep(1*60)
crawler.quit()
class FJUDCrawlerTest(unittest.TestCase):
def setUp(self):
self.crawler = FJUDCrawler(enable_unittest=True)
self.crawler.driver = WebdriverStub()
def tearDown(self):
self.crawler.quit()
def test_is_already_written(self):
test_dir = os.path.abspath(os.path.join(__file__, ".."))
self.assertFalse(fjud.is_already_written(ur"89,台上,2581", test_dir))
self.assertTrue(fjud.is_already_written(ur"query_result.html",
test_dir))
self.assertFalse(fjud.is_already_written(ur"query_result.html"))
def test_save_as(self):
output_path = fjud.get_default_output_path()
serial = ur"89,台上,2581____test_only"
fjud.save_as(serial, "test only")
output_file = os.path.join(output_path,
fjud.normalize_serial(serial))
self.assertTrue(os.path.isfile(output_file))
os.unlink(output_file)
self.assertFalse(os.path.isfile(output_file))
def test_get_courts(self):
all_courts = fjud.get_courts()
self.assertTrue(len(all_courts) > 0)
def test_normalize(self):
serial = ur"89,台上,2581"
self.assertEqual(
serial,
fjud.denormalize_serial(fjud.normalize_serial(serial)))
@unittest.skip("disable temporarily")
def disable_test_query(self):
crawler = FJUDCrawler()
ret = crawler.query()
with file("ret.txt", "w") as fp:
fp.write(ret)
print ret
def main():
"""
Search for bad verdicts and download them again.
"""
# If <pre> exists, the file has verdict text in it.
serials = files_without_word(u"<pre>")
for serial in serials:
serial = serial.decode("utf-8")
crawler = FJUDCrawler()
if u"台上" in serial:
crawler.fetch(court=u"最高法院", serial=serial, overwrite=True)
else:
crawler.fetch(court=u"臺灣高等法院", serial=serial, overwrite=True)
crawler.quit()
time.sleep(3)
def setUp(self):
self.crawler = FJUDCrawler(enable_unittest=True)
self.crawler.driver = WebdriverStub()
def disable_test_query(self):
crawler = FJUDCrawler()
ret = crawler.query()
with file("ret.txt", "w") as fp:
fp.write(ret)
print ret
