def test():
for testn in sorted(listdir('cases')):
if not exists(join('cases', testn, 'input')):
continue
print
print '-' * 40
print
print 'Running test', testn, '...'
with codecs.open(join('cases', testn, 'input'), 'r', 'utf-8') as f:
inp = f.read()
print 'input size:', len(inp)
parserules = True
for var in VALIDVARIANTS:
print
print 'testing', var, '...',
start = time()
try:
convhandler = ConverterHandler(var)
outp = convhandler.convert(inp, parserules)
except Exception:
print 'error occurred.'
print_exc()
else:
print '%.3f sec.' % (time() - start)
if exists(join('cases', testn, 'output.' + var)):
print 'comparing converter\'s output with expected output ...'
with codecs.open(join('cases', testn, 'output.' + var),
'r', 'utf-8') as f:
cp = f.read()
if cp == outp:
print 'OK'
else:
print 'FAILED'
def test():
for testn in sorted(listdir('cases')):
if not exists(join('cases', testn, 'input')):
continue
print
print '-' * 40
print
print 'Running test', testn, '...'
with codecs.open(join('cases', testn, 'input'), 'r', 'utf-8') as f:
inp = f.read()
print 'input size:', len(inp)
parserules = True
for var in VALIDVARIANTS:
print
print 'testing', var, '...',
start = time()
try:
convhandler = ConverterHandler(var)
outp = convhandler.convert(inp, parserules)
except Exception:
print 'error occurred.'
print_exc()
else:
print '%.3f sec.' % (time() - start)
if exists(join('cases', testn, 'output.' + var)):
print 'comparing converter\'s output with expected output ...'
with codecs.open(join('cases', testn, 'output.' + var), 'r', 'utf-8') as f:
cp = f.read()
if cp == outp:
print 'OK'
else:
print 'FAILED'
def genLoc(filePath):
c = open(filePath)
conv = ConverterHandler('zh-hant')
word = u'女装连衣裙'
for x in xrange(100000):
line = word
if len(line):
keyword = ''
keyword = urllib.quote(conv.convert(line.strip()).encode('utf-8'))
if not len(keyword):
continue
url = 'http://example.com/?q=%s' % keyword
yield url
DB = MySQLdb.connect(user="******",passwd="nishixian",
db="kancolle_wiki",unix_socket="/opt/lampp/var/mysql/mysql.sock",charset="utf8")
cursor = DB.cursor()
query = "SELECT ship_id,ship_name FROM kancolle_ship_info";
conv_hack = [(u'巻',u'卷'), (u'蔵',u'藏'), (u'黒',u'黑'), (u'暁',u'晓'), (u'満',u'满'), (u'皐',u'皋'), (u'歳',u'岁')]
if cursor.execute(query) > 0:
result = list(cursor.fetchall())
for item in result:
kanid = item[0]
kanname = item[1]
convhandler = ConverterHandler('zh-hans')
output = convhandler.convert(kanname,False)
for conv in conv_hack:
output = output.replace(conv[0],conv[1])
print kanname,'==>',output
query = "UPDATE kancolle_ship_info SET ship_name_sim = '%s' WHERE ship_id = %s" % (output, kanid)
try:
n = cursor.execute(query)
except Exception:
errcode,errordesc = e
print 'Error!',errordesc
DB.rollback()
else:
if n>0 :
print 'Convert Success!'
DB.commit()
else: