def test_builder_noindex(self):
'''test basic function'''
for f in glob.glob("mdx/Vocabulary*.db"):
os.remove(f)
print("***without sql index***\n")
start = time.time()
bd = IndexBuilder(self._mdx_file, sql_index=False, check=True)
print("takes {0} seconds to build without sql index\n".format(
time.time() - start))
start = time.time()
word = 'dedicate'
for i in range(self._repeat):
self.assertTrue(bd.mdx_lookup(word))
print("takes {0} second to lookup {1} {2} times\n".format(
time.time() - start, word, self._repeat))
for i in range(self._repeat):
bd.get_mdx_keys("dedi*")
print("takes {0} second to lookup {1} {2} times\n".format(
time.time() - start, "dedi*", self._repeat))
temp[i] = re.sub('\r\n(\s)+', '', contents[i])
eng = temp[0] + ' ' + temp[1] + ' ' + temp[2]
elif len(contents) == 2:
print(222)
print(span)
temp = ['', '']
for i in range(2):
if hasattr(contents[i], 'text'):
if contents[i].findChild():
continue
temp[i] = contents[i].text
else:
temp[i] = re.sub('\r\n(\s)+', '', contents[i])
eng = temp[0] + ' ' + temp[1]
if not eng.endswith("\n"):
eng += "\n"
u.append(eng + chn + "\r\n")
return u
keys = list(OrderedDict.fromkeys(builder.get_mdx_keys()))
"""
for key in keys:
write (key,"".join(getSentList(key)))
"""
for key in keys[1100:1120]:
print(key)
print("".join(getSentList(key)))
#if sys.argv.__len__() < 3:
# sys.exit(1)
#print sys.argv
dict = {}
query_type = sys.argv[2]
query_word = sys.argv[3].strip()
builder = IndexBuilder('/Users/david/Desktop/G/ciku/Longman Dictionary of Contemporary English.mdx')
if query_type == "key":
dict[query_word] = builder.mdx_lookup(query_word, True)
elif query_type == "wildcard":
keys = builder.get_mdx_keys(query_word)
count = 0
for key in keys:
count += 1
dict[key] = builder.mdx_lookup(key)
if count > 10:
break
elif query_type == "wildcardcount":
keys = builder.get_mdx_keys(query_word)
print keys.__len__()
sys.exit(0)
print json.dumps(dict)
#result_text = builder.mdx_lookup('dedication')
#print result_text
from mdict_query import IndexBuilder
bd = IndexBuilder("mdx\\oed.mdx")
keys = bd.get_mdx_keys("ded*")
result = bd.mdx_lookup('a')
pass