从前面带有分隔符的文本中删除单词（使用正则表达式）

Question

I need to remove words from the text with separators next to them. I already removed words but I don't know how I can remove separators at the same time. Any suggestions?
At the moment I have:

static void Main(string[] args)
        {
            Program p = new Program();
            string text = "";
            text = p.ReadText("Duomenys.txt", text);
            string[] wordsToDelete = { "Hello", "Thanks", "kinda" };
            char[] separators = { ' ', '.', ',', '!', '?', ':', ';', '(', ')', '\t' };
            p.DeleteWordsFromText(text, wordsToDelete, separators);
        }

        public string ReadText(string file, string text)
        {     
            text = File.ReadAllText(file);           
            return text;
        }

        public void DeleteWordsFromText(string text, string[] wordsToDelete, char[] separators)
        {
            Console.WriteLine(text);
            for (int i = 0; i < wordsToDelete.Length; i++)
            {
                text = Regex.Replace(text, wordsToDelete[i], String.Empty);
            }
            Console.WriteLine("-------------------------------------------");
            Console.WriteLine(text);
        }

The results should be:

how are you?
I am  good.

I have:

, how are you?
, I am . good.

Duomenys.txt

Hello, how are you? 
Thanks, I am kinda. good. 

Answer 1

You can build the regex like follows:

var regex = new Regex(@"\b(" 
    + string.Join("|", wordsToDelete.Select(Regex.Escape)) + ")(" 
    + string.Join("|", separators.Select(c => Regex.Escape(new string(c, 1)))) + ")?");

Explanation:

• the \\b at the start matches a word boundary. Just in case you get "XYZThanks"
• the next part builds a regex construct matching any of the wordsToDelete
• the last part builds a regex construct matching any of the separators; the trailing "?" is there because you said you want to replace the word also if no separator follows

Answer 2

You may build a regex like

\b(?:Hello|Thanks|kinda)\b[ .,!?:;()    ]*

where \\b(?:Hello|Thanks|kinda)\\b will match any words to delete as whole words and [ .,!?:;() ]* will match all your separators 0 or more times after the words to delete.

The C# solution :

char[] separators = { ' ', '.', ',', '!', '?', ':', ';', '(', ')', '\t' };
string[] wordsToDelete = { "Hello", "Thanks", "kinda" };
string SepPattern = new String(separators).Replace(@"\", @"\\").Replace("^", @"\^").Replace("-", @"\-").Replace("]", @"\]");
var pattern = $@"\b(?:{string.Join("|", wordsToDelete.Select(Regex.Escape))})\b[{SepPattern}]*";
// => \b(?:Hello|Thanks|kinda)\b[ .,!?:;()  ]*
Regex rx = new Regex(pattern, RegexOptions.Compiled);
// RegexOptions.IgnoreCase can be added to the above flags for case insensitive matching: RegexOptions.IgnoreCase | RegexOptions.Compiled
DeleteWordsFromText("Hello, how are you?", rx);
DeleteWordsFromText("Thanks, I am kinda. good.", rx);

Here is the DeleteWordsFromText method:

public static void DeleteWordsFromText(string text, Regex p)
{
    Console.WriteLine($"---- {text} ----");
    Console.WriteLine(p.Replace(text, ""));
}

Output:

---- Hello, how are you? ----
how are you?
---- Thanks, I am kinda. good. ----
I am good.

Notes :

• string SepPattern = new String(separators).Replace(@"\\", @"\\\\").Replace("^", @"\\^").Replace("-", @"\\-").Replace("]", @"\\]"); - it is a separator pattern that will be used inside a character class, and since only ^ , - , \\ , ] chars require escaping inside a character class, only these chars are escaped
• $@"\\b(?:{string.Join("|", wordsToDelete.Select(Regex.Escape))})\\b" - this will build the alternation from the words to delete and will only match them as whole words.

Pattern details

• \\b - word boundary
• (?: - start of a non-capturing group:
  • Hello - Hello word
  • | - or
  • Thanks - Thanls word
  • | - or
  • kinda - kinda word
• ) - end of the group
• \\b - word boundary
• [ .,!?:;() ]* - any 0+ chars inside the character class.

See the regex demo .

Answer 3

I would not use Regex. In 3 months from now, you'll not understand the Regex any more and fixing bugs is a hard thing then.

I would use simple loops. Everyone will understand:

public void DeleteWordsFromText(string text, string[] wordsToDelete, char[] separators)
{
    Console.WriteLine(text);
    foreach (string word in wordsToDelete)
    {
        foreach(char separator in separators)
        {
            text = text.Replace(word + separator, String.Empty);
        }
    }
    Console.WriteLine("-------------------------------------------");
    Console.WriteLine(text);
}