Characters in string changed after downloading HTML from the internet

Using the following code, I can download the HTML of a file from the internet:

WebClient wc = new WebClient();

// ....

string downloadedFile = wc.DownloadString("http://www.myurl.com/");

However, sometimes the file contains "interesting" characters like é to é, to ↠and フシギダネ to フシギダãƒ.

I think it may be something to do with different unicode types or something, as each character gets changed into 2 new ones, perhaps each character being split in half but I have very little knowledge in this area. What do you think is wrong?


Solution 1:

Here's a wrapped download class which supports gzip and checks encoding header and meta tags in order to decode it correctly.

Instantiate the class, and call GetPage().

public class HttpDownloader
{
    private readonly string _referer;
    private readonly string _userAgent;

    public Encoding Encoding { get; set; }
    public WebHeaderCollection Headers { get; set; }
    public Uri Url { get; set; }

    public HttpDownloader(string url, string referer, string userAgent)
    {
        Encoding = Encoding.GetEncoding("ISO-8859-1");
        Url = new Uri(url); // verify the uri
        _userAgent = userAgent;
        _referer = referer;
    }

    public string GetPage()
    {
        HttpWebRequest request = (HttpWebRequest)WebRequest.Create(Url);
        if (!string.IsNullOrEmpty(_referer))
            request.Referer = _referer;
        if (!string.IsNullOrEmpty(_userAgent))
            request.UserAgent = _userAgent;

        request.Headers.Add(HttpRequestHeader.AcceptEncoding, "gzip,deflate");

        using (HttpWebResponse response = (HttpWebResponse)request.GetResponse())
        {
            Headers = response.Headers;
            Url = response.ResponseUri;
            return ProcessContent(response);
        }

    }

    private string ProcessContent(HttpWebResponse response)
    {
        SetEncodingFromHeader(response);

        Stream s = response.GetResponseStream();
        if (response.ContentEncoding.ToLower().Contains("gzip"))
            s = new GZipStream(s, CompressionMode.Decompress);
        else if (response.ContentEncoding.ToLower().Contains("deflate"))
            s = new DeflateStream(s, CompressionMode.Decompress);  

        MemoryStream memStream = new MemoryStream();
        int bytesRead;
        byte[] buffer = new byte[0x1000];
        for (bytesRead = s.Read(buffer, 0, buffer.Length); bytesRead > 0; bytesRead = s.Read(buffer, 0, buffer.Length))
        {
            memStream.Write(buffer, 0, bytesRead);
        }
        s.Close();
        string html;
        memStream.Position = 0;
        using (StreamReader r = new StreamReader(memStream, Encoding))
        {
            html = r.ReadToEnd().Trim();
            html = CheckMetaCharSetAndReEncode(memStream, html);
        }            

        return html;
    }

    private void SetEncodingFromHeader(HttpWebResponse response)
    {
        string charset = null;
        if (string.IsNullOrEmpty(response.CharacterSet))
        {
            Match m = Regex.Match(response.ContentType, @";\s*charset\s*=\s*(?<charset>.*)", RegexOptions.IgnoreCase);
            if (m.Success)
            {
                charset = m.Groups["charset"].Value.Trim(new[] { '\'', '"' });
            }
        }
        else
        {
            charset = response.CharacterSet;
        }
        if (!string.IsNullOrEmpty(charset))
        {
            try
            {
                Encoding = Encoding.GetEncoding(charset);
            }
            catch (ArgumentException)
            {
            }
        }
    }

    private string CheckMetaCharSetAndReEncode(Stream memStream, string html)
    {
        Match m = new Regex(@"<meta\s+.*?charset\s*=\s*""?(?<charset>[A-Za-z0-9_-]+)""?", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(html);            
        if (m.Success)
        {
            string charset = m.Groups["charset"].Value.ToLower() ?? "iso-8859-1";
            if ((charset == "unicode") || (charset == "utf-16"))
            {
                charset = "utf-8";
            }

            try
            {
                Encoding metaEncoding = Encoding.GetEncoding(charset);
                if (Encoding != metaEncoding)
                {
                    memStream.Position = 0L;
                    StreamReader recodeReader = new StreamReader(memStream, metaEncoding);
                    html = recodeReader.ReadToEnd().Trim();
                    recodeReader.Close();
                }
            }
            catch (ArgumentException)
            {
            }
        }

        return html;
    }
}

Solution 2:

Since I am not allowed to comment (insufficient reputation), I'll have to post an additional answer. I am using Mikael's great class routinely, but I encountered a practical problem with the regex that tries to find the charset meta-info. This

Match m = new Regex(@"<meta\s+.*?charset\s*=\s*(?<charset>[A-Za-z0-9_-]+)", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(html); 

fails on this

<meta charset="UTF-8"/>

whereas this

Match m = new Regex(@"<meta\s+.*?charset\s*=\s*""?(?<charset>[A-Za-z0-9_-]+)""?", RegexOptions.Singleline | RegexOptions.IgnoreCase).Match(html);

does not.

Thanks, Mikael.