bad routing where there are non-english characters in page url

[mderazon]

HI,
I have just updated Harp to the latest version (0.19.0) and I am noticing a problem with my hebrew pages.

In my _data.json I have

"משהו-משהו": {
    "title": "משהו משהו",
    "date": "26 Nov 2013",
    "tags": ["hebrew"],
    "published": true
  },

My public folder has a משהו-משהו with an index.md that contain the post and when I go to /משהו-משהו I get a 404 page. It probably has to do with a recent change to harp as my older version was working correctly.

Any idea ?

[mderazon]

looks like the url path is not being normalized well (I think): https://github.com/sintaxi/harp/blob/master/lib/helpers.js#L17

req.url comes url-encoded as %D7%9E%D7%A9%D7%94%D7%95-%D7%9E%D7%A9%D7%94%D7%95 and then after the normalize step I see it as �ש��-�ש�� instead of משהו-משהו. Therefor file is not found

[mderazon]

So initially I thought it might be related to this nodejs/node#2165

And was trying to follow the guideline on how to work with different filesystems: https://nodejs.org/en/docs/guides/working-with-different-filesystems/#avoid-a-lowest-common-denominator-approach

After trying different approaches, I tried replacing unescape function with decodeURIComponent
(see this SO post: http://stackoverflow.com/a/619428/599912) and everything worked perfectly.

@sintaxi Do you want me to send a PR ? Do you think of any problematic consequence for this ? I tested this on my blog and it worked okay.