用转义字符反转字符串

Question

I have a string that might contain escape characters. Let's assume this is '\\'. I follow the MSDN Escape Sequences definition

I want to reverse this string, but keep the escape sequences.

Example:

string input = @"Hello\_World";
string reversed = @"dlroW\_elloH";

Note that in my input string the backslashes are separate characters. The reversed string is meant to be used in a SQL LIKE statement where the underscore is not meant as a wild card, but literally as an underscore. The backslash in the SQL LIKE functions as an escape character

The problem is, that if a character in my original string is preceded by a backslash, then in my reversed string this backslash should still precede the character: @"_" (two separate characters) should in reverse still be @"_".

Bonus points: Reverse escape sequences with numbers '\\x0128'

I've tried it as extension functions:

public static string EscapedReverse(this string txt, char escapeChar)
{
    IList<char> charList = txt.ToList();
    return new string(EscapedReverse(charList, escapeChar).ToArray());
}

public static IEnumerable<char> EscapedReverse(this IList<char> text, char escapeChar)
{
    int i = text.Count-1;
    // Text[i] is the last character of the sequence;
    // text[i] is the next character to return, except if text[i-1] is escapeChar
    while (i > 0)
    {
        if(text[i-1] == escapeChar)
        {
            yield return text[i-1];
            yield return text[i];
            i -= 2;
        }
        else
        {
            yield return text[i];
            i -= 1;
        }
    }
    // return the last character
    if (i == 0)
        yield return text[i];
}

This works. However, my string is converted to array / list twice. I wondered if there would be a smarter method where the elements don't have to be accessed so often?

Addition: what is my problem anyway?

Comments suggested to add more information about my problem.

There is a requirement to show a list of matching elements while an operator is typing in a text box. Most elements he can see start with a similar prefix. The difference the operator searches for is in the end of the name.

Therefore we want to show a list of names ending with the typed character. So if the operator types "World" he will see a list with all names ending with "World".

The already existing database (change is out of the question) has a table with a NAME and a REVERSEDNAME. Software takes care that if a name is inserted or updated the correct reversed name is inserted / updated. REVERSEDNAME is indexed, so using a WHERE with reversed name is fast.

So if I need to return all names ending with "World", I need to return the names of all records where the REVERSEDNAME starts with the reverse of "WORLD":

SELECT TOP 30 [MYTABLE].[NAME] as Name
FROM [MYTABLE]
WHERE [MYTABLE].REVERSEDNAME LIKE 'dlroW%'

This works fine as long as no wild cards (like underscore) are used. This was solved by the software by escaping the underscore character (I know, bad design, the fact that SQL LIKE uses underscore as wild card should not seep through, but I have to live with this existing software)

So the operator types @"My_World" My software received @"My_World", the backslash is a separate character I have to reverse to @"dlrow_yM", note that the backslash is still before the underscore

My Dapper code:

IEnumerable<string> FetchNamesEndingWith(string nameEnd)

// here is my reversal procedure:
string reversedNameEnd = nameEnd.EscapedReverse() = '%';

using (var dbConnection = this.CreateOpenDbConnection())
{
    return dbConnection.Query<string>(@"
        SELECT TOP 30 [MYTABLE].[NAME] as Name
        FROM [MYTABLE]
        WHERE [MYTABLE].REVERSEDNAME LIKE @param ESCAPE '\'",
        new {param = reversedNameEnd});
}

MSDN about using escape characters in SQL LIKE

Changing the escape character to a different character doesn't help. The problem is not that the escape character is a backslash, but that reversing my string should keep the escape character in front of the escaped character.

My code works, I only wondered if there would be a better algorithm that doesn't copy the string twice. Not only for this specific problem, but also if in future problems I need to reverse strings and keep certain characters in place.

Answer 1

You can use regular expressions:

var pattern =  @"\\x[1-9a-fA-F]{4}|\\x[1-9a-fA-F]{2}|\\[0-7]{3}|\\.|.";
var rgx = new Regex(pattern);
return new string(
          rgx.Matches(txt)
          .Cast<Match>()
          .OrderByDescending(x => x.Index)
          .SelectMany(x => x.Value)
          .ToArray());

pattern covers single characters and escape sequences in formats:

\x????
\x??
\???
\?