A more extensive listing of problematic word characters
if (buffer.IndexOf('\u2013') > -1) buffer = buffer.Replace('\u2013', '-');
if (buffer.IndexOf('\u2014') > -1) buffer = buffer.Replace('\u2014', '-');
if (buffer.IndexOf('\u2015') > -1) buffer = buffer.Replace('\u2015', '-');
if (buffer.IndexOf('\u2017') > -1) buffer = buffer.Replace('\u2017', '_');
if (buffer.IndexOf('\u2018') > -1) buffer = buffer.Replace('\u2018', '\'');
if (buffer.IndexOf('\u2019') > -1) buffer = buffer.Replace('\u2019', '\'');
if (buffer.IndexOf('\u201a') > -1) buffer = buffer.Replace('\u201a', ',');
if (buffer.IndexOf('\u201b') > -1) buffer = buffer.Replace('\u201b', '\'');
if (buffer.IndexOf('\u201c') > -1) buffer = buffer.Replace('\u201c', '\"');
if (buffer.IndexOf('\u201d') > -1) buffer = buffer.Replace('\u201d', '\"');
if (buffer.IndexOf('\u201e') > -1) buffer = buffer.Replace('\u201e', '\"');
if (buffer.IndexOf('\u2026') > -1) buffer = buffer.Replace("\u2026", "...");
if (buffer.IndexOf('\u2032') > -1) buffer = buffer.Replace('\u2032', '\'');
if (buffer.IndexOf('\u2033') > -1) buffer = buffer.Replace('\u2033', '\"');
When I encountered this problem I wrote an extension method to the String class in C#.
public static class StringExtensions
{
public static string StripIncompatableQuotes(this string s)
{
if (!string.IsNullOrEmpty(s))
return s.Replace('\u2018', '\'').Replace('\u2019', '\'').Replace('\u201c', '\"').Replace('\u201d', '\"');
else
return s;
}
}
This simply replaces the silly 'smart quotes' with normal quotes.
[EDIT] Fixed to also support replacement of 'double smart quotes'.
To extend on Nick van Esch's popular answer, here is the code with the names of the characters in the comments.
if (buffer.IndexOf('\u2013') > -1) buffer = buffer.Replace('\u2013', '-'); // en dash
if (buffer.IndexOf('\u2014') > -1) buffer = buffer.Replace('\u2014', '-'); // em dash
if (buffer.IndexOf('\u2015') > -1) buffer = buffer.Replace('\u2015', '-'); // horizontal bar
if (buffer.IndexOf('\u2017') > -1) buffer = buffer.Replace('\u2017', '_'); // double low line
if (buffer.IndexOf('\u2018') > -1) buffer = buffer.Replace('\u2018', '\''); // left single quotation mark
if (buffer.IndexOf('\u2019') > -1) buffer = buffer.Replace('\u2019', '\''); // right single quotation mark
if (buffer.IndexOf('\u201a') > -1) buffer = buffer.Replace('\u201a', ','); // single low-9 quotation mark
if (buffer.IndexOf('\u201b') > -1) buffer = buffer.Replace('\u201b', '\''); // single high-reversed-9 quotation mark
if (buffer.IndexOf('\u201c') > -1) buffer = buffer.Replace('\u201c', '\"'); // left double quotation mark
if (buffer.IndexOf('\u201d') > -1) buffer = buffer.Replace('\u201d', '\"'); // right double quotation mark
if (buffer.IndexOf('\u201e') > -1) buffer = buffer.Replace('\u201e', '\"'); // double low-9 quotation mark
if (buffer.IndexOf('\u2026') > -1) buffer = buffer.Replace("\u2026", "..."); // horizontal ellipsis
if (buffer.IndexOf('\u2032') > -1) buffer = buffer.Replace('\u2032', '\''); // prime
if (buffer.IndexOf('\u2033') > -1) buffer = buffer.Replace('\u2033', '\"'); // double prime
Note that what you have is inherently a corrupt CSV file. Indiscriminately replacing all typographer's quotes with straight quotes won't necessarily fix your file. For all you know, some of the typographer's quotes were supposed to be there, as part of a field's value. Replacing them with straight quotes might not leave you with a valid CSV file, either.
I don't think there is an algorithmic way to fix a file that is corrupt in the way you describe. Your time might be better spent investigating how you come to have such invalid files in the first place, and then putting a stop to it. Is someone using Word to edit your data files, for instance?
According to the Character Map application that comes with Windows, the Unicode values for the curly quotes are 0x201c and 0x201d. Replace those values with the straight quote 0x0022, and you should be good to go.
String.Replace(0x201c, '"');
String.Replace(0x201d, '"');
The VB equivalent of what @Matthew wrote:
Public Module StringExtensions
<Extension()>
Public Function StripIncompatableQuotes(BadString As String) As String
If Not String.IsNullOrEmpty(BadString) Then
Return BadString.Replace(ChrW(&H2018), "'").Replace(ChrW(&H2019), "'").Replace(ChrW(&H201C), """").Replace(ChrW(&H201D), """")
Else
Return BadString
End If
End Function
End Module
I have a whole great big... program... that does precisely this. You can rip out the script and use it at your leasure. It does all sorts of replacements, and is located at http://bitbucket.org/nesteruk/typografix
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With