#!/usr/bin/env python3

import re
import saxo

pattern = "(?ims)<td>Version</td>.*?<td>(?:<[^>]+>)([^<]+)"

@saxo.command()
def uniwhen(arg):
    if not arg:
        return "o hai. uniwhen is the provenance of a unicode character. k thx"
    arg = arg.lower()
    if len(arg) == 1:
        arg = "%04x" % ord(arg)
    elif arg.startswith("u+"):
        arg = arg[2:]
    url = "http://www.fileformat.info/info/unicode/char/%s/index.htm" % arg
    page = saxo.request(url)
    text = page["text"]
    for v in re.findall(pattern, text):
        return v
    return "Sorry, couldn't find the provenance of that character!"