how to extract a whole sentence by a single word match in a string?

Question

So I have got a whole string (about 10k chars) and then searching for a word(or many words) in that string. With regex(word).Matches(scrappedstring).

But how to do so to extract the whole sentence, that contains that word. I was thinking of taking a substring after the searched word until the first dot/exclamation mark/question mark/etc. But how to take the part of the sentence before the searched word ?

Or maybe there's a better logic ?

Answer 1

If your boundaries are e.g. ., !, ? and ;, match all sentences across [^.!?;]*(wordmatch)[^.!?;]* expression. It will give all sentences with desired wordmatch inside.

Example:

var s = "First sentence. Second with wordmatch ? Third one; The last wordmatch, EOM!";
var r = new Regex("[^.!?;]*(wordmatch)[^.!?;]*");
var m = r.Matches(s);

var result = Enumerable.Range(0, m.Count).Select(index => m[index].Value).ToList();

Answer 2

You can get substrings between sentence finishers (dot/exclamation mark/qustion mark/etc) and search for the word in each sentence inside a loop.

Then return the substring when you find the matching word.

Answer 3

Once you have a position, you would then read up to the next ., or end of the file.. but you also need to read backwards from the beginning of the word to a . or the beginning of the file. Those two positions mean you can then extract the sentence.

Note, it's not fool-proof... in its simplest form as outlined above e.g. would mean the sentence started after the g. which is not probably the case.

Answer 4

Extract the sentances from the input. Then search for the specified word(s) within each sentance. Return the sentances where the word(s) is present.

    public List<string> GetMatchedString(string match, string input)
    {
        var sentanceList = input.Split(new char[] { '.', '?', '!' });
        var regex = new Regex(match);
        return sentanceList.Where(sentance => regex.Matches(sentance,0).Count > 0).ToList();
    }

Answer 5

You can do that using a process in 2 steps.

1st you fragment the phrases and then filter each one has the word.

something like this:

var input = "A large text with many sentences. Many chars in a string!. A sentence without the pattern word.";

//Step 1: fragment phrase.
var patternPhrase = @"(?<=(^|[.!?]\s*))[^ .!?][^.!?]+[.!?]";

//Step 2: filter out only the phrases containing the word.
var patternWord = @"many";

var result = Regex
    .Matches(input, patternPhrase) // step 1
    .Cast<Match>()
    .Select(s => s.Value)
    .Where(w => Regex.IsMatch(w, patternWord, RegexOptions.IgnoreCase)); // step 2

foreach (var item in result)
{
    //do something with any phrase.
}