python unicode file read
def UnicodeReadFile(fileName):if os.path.exists(fileName) == False:
#print fileName + ": Not exist"
return ";"
CODEC = 'utf-8'
FILE = 'unicode.txt'
emails = ""
file_object = codecs.open(fileName, 'r', "utf-16")
for line in file_object:
try:
bytes_out = line.encode(CODEC)
except:
continue
#print bytes_out
bytes_out = bytes_out.strip('\r\n')
mat = emailPattern.match(bytes_out)
if mat:
#print "match unicode"
email = mat.group(1)
emails = emails + " " + email
file_object.close()
return emails + ";"
页:
[1]