|
|
@ -2,6 +2,7 @@ |
|
|
|
s = "asdf©qwer" |
|
|
|
for i in range(len(s)): |
|
|
|
print("s[%d]: %s %X"%(i, s[i], ord(s[i]))) |
|
|
|
|
|
|
|
# Test all three forms of Unicode escape, and |
|
|
|
# all blocks of UTF-8 byte patterns |
|
|
|
s = "a\xA9\xFF\u0123\u0800\uFFEE\U0001F44C" |
|
|
@ -11,3 +12,7 @@ for i in range(-len(s), len(s)): |
|
|
|
for j in range(i, len(s)): |
|
|
|
print("s[%d:%d]: %d chars, '%s'"%(i, j, len(s[i:j]), s[i:j])) |
|
|
|
print("s[%d:]: %d chars, '%s'"%(i, len(s[i:]), s[i:])) |
|
|
|
|
|
|
|
# Test UTF-8 encode and decode |
|
|
|
enc = s.encode() |
|
|
|
print(enc, enc.decode() == s) |
|
|
|