Previously: `Händler` would be tokenized as `ndler` or `ändler` depending on python version
Rather than the expected `händler`

Solution: use `regexp` rather than `re`.
This gives us the ability to use unicode character clasess such as `[[:upper:]]` and `[[:lower:]]`

Fixes myint#35

Previously: `Händler` would be tokenized as `ndler` or `ändler` depending on python version
Rather than the expected `händler`

Solution: use `regexp` rather than `re`.
This gives us the ability to use unicode character clasess such as `[[:upper:]]` and `[[:lower:]]`

Fixes myint#35

Previously: `Händler` would be tokenized as `ndler` or `ändler` depending on python version
Rather than the expected `händler`

Solution: use `regexp` rather than `re`.
This gives us the ability to use unicode character clasess such as `[[:upper:]]` and `[[:lower:]]`

Fixes myint#35

Previously: `Händler` would be tokenized as `ndler` or `ändler` depending on python version
Rather than the expected `händler`

Solution: use `regexp` rather than `re`.
This gives us the ability to use unicode character clasess such as `[[:upper:]]` and `[[:lower:]]`

unicodedata.normalize is because travis was working differently than my mac

Fixes myint#35

Previously: `Händler` would be tokenized as `ndler` or `ändler` depending on python version
Rather than the expected `händler`

Solution: use `regexp` rather than `re`.
This gives us the ability to use unicode character clasess such as `[[:upper:]]` and `[[:lower:]]`

Fixes myint#35