Hi

in my application I get a lot of strings which I have to "clean up"
before I pass them to a third-party library. The strings I have contain
characters which are invalid for the third-party library, so I have to
either remove them or replace them with reasonable alternatives.

What is a good method of doing this?

At the moment I have the following:

string Clean(string element)
{
  element = element.Replace(",", "");
  element = element.Replace("-", "");
  element = element.Replace("!", "");
  element = element.Replace("/", "");
  element = element.Replace("\\", "");

  element = element.Replace("æ", "ae");
  element = element.Replace("Æ", "AE");
  element = element.Replace("ä", "ae");
  element = element.Replace("Ä", "AE");
  element = element.Replace("ø", "oe");
  element = element.Replace("Ø", "OE");
  element = element.Replace("ö", "oe");
  element = element.Replace("Ö", "OE");
  element = element.Replace("å", "aa");
  element = element.Replace("Å", "AA");

  element = element.Trim(' ', '.');

  return element;
}


Thanks,
Peter

It will work but you are scanning the whole string for each replace.  I have
no experience of it but Regex.Replace is likely to only scan the string once
and call a delegate each time it finds a match against one of many patterns
you specify...


http://msdn.microsoft.com/en-us/library/ms149475.aspx

Show quote Hide quote You are creating a new String for every replace.  Using a
StringBuilder instead avoids this and may run faster:

  string Clean(string element) {
    StringBuilder sb = new StringBuilder(element);
    sb.Replace(",", "");
    // Other replaces

    return sb.ToString();
  }

rossum

Any chance to get a new version of the 3rd party library?
Some of these replacements are locale sensitive.
And even for the locales where they are valid, they affect (negatively)
the quality of the text.
So that is not "clean up", that is "crap"

Imagine someone whould do this to English strings:
   element = element.Replace("w", "vv");
because some stupid library does not support 'w'.

Show quote Hide quote My take: build a "conversion matrix" and then run every character in that
string through the matrix, outputting a clean string in the end. Something
like this (air code!):

private Dictionary<char, string> _conversions;

// Constructor
public <your class name>
{
    // Ideally you would read these from a database or settings file so
    // that you wouldn't have to recompile if you find new things to replace
    _conversions.Add(',', "");
    _conversions.Add('-', "");
    _conversions.Add('!', "");
    _conversions.Add('/', "");
    _conversions.Add('\\', "");
    _conversions.Add('æ', "ae");
    _conversions.Add('Æ', "AE");
    _conversions.Add('ä', "ae");
    _conversions.Add('Ä', "AE");
    _conversions.Add('ø', "oe");
    _conversions.Add('Ø', "OE");
    _conversions.Add('ö', "oe");
    _conversions.Add('Ö', "OE");
    _conversions.Add('å', "aa");
    _conversions.Add('Å', "AA");
}

private string Clean(string element)
{
    StringBuilder sb = new StringBuilder();

    foreach(char c in element)
    {
        // NOTE: The following line may not compile since one option returns
        // a string and the other a char. In that case, make it a full blown
        // if/else clause.
        sb.Append(_conversions.Contains(c) ? _conversions[c] : c);
    }

    return sb.ToString().Trim(' ', '.');
}

Oh, and for what it's worth, it sounds like your third-party library
sucks....

Thanks for all the comments.

With regards to the 3rd-party library, it is a content management
system, and it imposes rules on the names that can be used for path
elements and the "items" or "nodes" which make up the hierarchical
content structure. Some things I do accept, like / or \ in a name (much
the same as in windows) - but I don't really know why one can't use [
or ) or "international" letters like æ or ø. I don't have an exhaustive
list of all the invalid characters.

The data I receive comes from a database, and I have to then insert it
in the CMS - which gives problems if I read "invalid" strings from the
database, so I have to make some sort of "conversion".


/Peter

Is the result visible somewhere "as is", or it will always go thru some
"conversion layer"?

Maybe you can come with some kind of escaping system?

For instance have the string as utf-8, then escape all bytes > 127
When you get them back, you unescape and get the original utf-8 strings,
not characters damaged.

Mihai N. wrote:

Hi - I'm not sure I completely follow you. What I am doing is reading
company data from a database, and putting them into the hierarchical
structure of the CMS (as items/nodes in the CMS) - as well as some
accompanying data (like contact info, address, images etc).

This is to make it easy for site editors to access and change
information which is shown on some of the website's pages.

Eg.

IT companies
  microsoft
  yahoo

And some of the companies might have "illegal" characters in their
names (eg ! in Yahoo!).


/Peter

How to Verify if EventHandler for ValueChanged has been set?
Procedure xxx expects parameter '@ID' which was not supplied
Linq Style
Can C# 3.0 not be used in websites?
Design of a suer interface similar to visual studio
Cross Platform encryption/decryption
SQLException - message property was empty
Bogus control property values
Import a form
Repository and Services