converting \u0040 to @ in C#

Question

The Facebook graph API's return to me the user's email address as

foo\u0040bar.com.

in a JSON object. I need to convert it to

foo@bar.com.

There must be a built in method in .NET that changes the Unicode character expression (\u1234) to the actual unicode symbol.

Do you know what it is?

Note: I prefer not to use JSON.NET or JavaScriptSerializer for performance issues.

I think the problem is in my StreamReader:

        requestUrl = "https://graph.facebook.com/me?access_token=" + accessToken;
        request = WebRequest.Create(requestUrl) as HttpWebRequest;
        try
        {
            using (HttpWebResponse response2 = request.GetResponse() as HttpWebResponse)
            {
                // Get the response stream  
                reader = new StreamReader(response2.GetResponseStream(),System.Text.Encoding.UTF8);
                string json = reader.ReadToEnd();

I tried different encodings for the StreamReader, UTF8, UTF7, Unicode, ... none worked.

Many thanks!

Thanks to L.B for correcting me. The problem was not in the StreamReader.

Looks like something went wrong when reading/decoding this. Fixing it after the fact will not be so straightforward. — H H, Nov 30 '11 at 18:08
Most not trivial - how to extract `\u0040` symbol from the string — sll, Nov 30 '11 at 18:09
looking back at it it looks like I need to give the StreamReader that reads the JSON the right encoding — Barka, Nov 30 '11 at 18:15
`I prefer not to use JSON.NET or JavaScriptSerializer for performance issues` How many requests do you make to facebook per second? How many CPU cycles do you need more? — L.B, Nov 30 '11 at 18:46
L.B I think and @Henk Holterman also thinks that my StreamReader is at fault. So performance aside, going to other libraries will not solve the problem. — Barka, Nov 30 '11 at 18:51
No, I said 'something'. And that something probably is _not using_ a proper deserializer lib. — H H, Nov 30 '11 at 19:06

Guffa · Answer 1 · 2011-11-30T19:16:20.573

11

Yes, there is some built in method for that, but that would involve something like using a compiler to parse the string as code...

Use a simple replace:

s = s.Replace(@"\u0040", "@");

For a more flexible solution, you can use a regular expression that can handle any unicode character:

s = Regex.Replace(s, @"\\u([\dA-Fa-f]{4})", v => ((char)Convert.ToInt32(v.Groups[1].Value, 16)).ToString());

edited Nov 30 '11 at 19:16

answered Nov 30 '11 at 18:19

Guffa

687,336
108
737
1,005

The latter is exactly what I was typing. +1 – J. Holmes Nov 30 '11 at 18:26
This is an excellent answer, however looking back I think the fix really should be in my StreamReader. I edited the question to reflect my new thinking. Many thanks! – Barka Nov 30 '11 at 18:32
@user277498: I don't know who's version you used, it was edited several times after i posted it. Anyway, I have fixed it now. – Guffa Nov 30 '11 at 19:17

L.B · Accepted Answer · 2011-11-30T19:51:16.240

3

Json responses are not binary data to convert to a string using some encodings. Instead they are strings correctly decoded by your browser or by HttpWebResponse as in your example. You need a second procesing on it(regex, deserializers etc) to get the final data.

See what you get with webClient.DownloadString("https://graph.facebook.com/HavelVaclav?access_token=????") without any encoding

{"id":"100000042150992",
    "name":"Havel V\u00e1clav",
    "first_name":"Havel",
    "last_name":"V\u00e1clav",
    "link":"http:\/\/www.facebook.com\/havel.vaclav",
    "username":"havel.vaclav",
    "gender":"male",
    "locale":"cs_CZ"
}

Would your encoding change \/ to /?

So, the problem is not in your StreamReader.

edited Nov 30 '11 at 19:51

answered Nov 30 '11 at 19:10

L.B

114,136
19
178
224

But you accepted another answer? Where your best solution would be Json.Net, JavaScriptSerializer etc. – L.B Nov 30 '11 at 20:15
both answers were good. it was a tossup. your answer was what i implemented but @Guffa's answer was the correct answer to my exact question. I want to give you both credit. – Barka Nov 30 '11 at 20:37

converting \u0040 to @ in C#

2 Answers2