#!/usr/bin/python

import os, tempfile, re

def check(offset):
    f = tempfile.NamedTemporaryFile("wb", delete=False)
    f.write(b"a"*offset)
    f.write(b"\xe6\x96\x87")
    f.write(b"b"*100)
    f.close()

    # this tries to use ascii, and fails. The error message contains the wrong
    # offset. It appears to emit offset%4096 .
    try:
        lines = open(f.name, encoding="ascii").readlines()
        #data = open(f.name).read()
    except UnicodeDecodeError as e:
        s = str(e)
        mo = re.search(r"in position (\d+):", s)
        found = int(mo.group(1))
        if found != offset:
            print(s)
            print("Hey, found at %d, should have been at %d" % (found, offset))
            #print("diff is %d" % ((offset%4096)-found))
            return False
    finally:
        os.unlink(f.name)
    return True

if True:
    where = 0
    while True:
        if not check(where):
            break
        if where > 10000:
            print("all good")
            break
        where += 1

if False:
    import random
    for i in range(1000):
        l = random.randint(1, 50000)
        check(l)