提取包含特定单词的句子

Question

I want to get the sentences in a textfile which contain a particular keyword. I tried a lot but not able to get the proper sentences that contain the keyword....I have more that one set of keywords if any of this match with the paragraph then it should be taken. For eg :if my text file contains words like robbery,robbed etc then that sentence shold be extracted.. Below is the code which I tried. Is there anyway to solve this using regex. Any help will be appreciated.

  BufferedReader br1 = new BufferedReader(new FileReader("/home/pgrms/Documents/test/one.txt"));
    String str="";

    while(br1 .ready()) 
    {
        str+=br1 .readLine() +"\n";

    }
Pattern re = Pattern.compile("[^.!?\\s][^.!?]*(?:[.!?](?!['\"]?\\s|$)[^.!?]*)*[.!?]?['\"]?(?=\\s|$)", Pattern.MULTILINE | Pattern.COMMENTS);
Matcher match = re.matcher(str);
String sentenceString="";
while (match .find())
{
    sentenceString=match.group(0);
    System.out.println(sentenceString);
}

Answer 1

Here is an example for when you have a list of predefined keywords:

import java.io.BufferedReader;
import java.io.FileReader;
import java.util.regex.*;
public class Tester {

    public static void main(String [] args){
        try {
            BufferedReader br1 = new BufferedReader(new FileReader("input"));
            String[] words = {"robbery","robbed", "robbers"};
            String word_re = words[0];   
            String str="";

            for (int i = 1; i < words.length; i++)
                word_re += "|" + words[i];
            word_re = "[^.]*\\b(" + word_re + ")\\b[^.]*[.]";
            while(br1.ready()) { str += br1.readLine(); }
            Pattern re = Pattern.compile(word_re, 
                    Pattern.MULTILINE | Pattern.COMMENTS | 
                    Pattern.CASE_INSENSITIVE);
            Matcher match = re.matcher(str);
            String sentenceString="";
            while (match .find()) {
                sentenceString = match.group(0);
                System.out.println(sentenceString);
            }
        } catch (Exception e) {}
    }

}

This creates a regex of the form:

[^.]*\b(robbery|robbed|robbers)\b[^.]*[.]

Answer 2

In general, to check if a sentence contains rob or robbery or robbed , you can add a lookehead after the beginning of string anchor, before the rest of your regex pattern:

(?=.*(?:rob|robbery|robbed))

In this case, it is more efficient to group the rob then check for potential suffixes:

(?=.*(?:rob(?:ery|ed)?))

In your Java code, we can (for instance) modify your loop like this:

while (match.find())
{
    sentenceString=match.group(0);
    if (sentenceString.matches("(?=.*(?:rob(?:ery|ed)?))")) {
        System.out.println(sentenceString);
    }
}

Explain Regex

(?=                      # look ahead to see if there is:
  .*                     #   any character except \n (0 or more times
                         #   (matching the most amount possible))
  (?:                    #   group, but do not capture:
    rob                  #     'rob'
    (?:                  #     group, but do not capture (optional
                         #     (matching the most amount possible)):
      ery                #       'ery'
     |                   #      OR
      ed                 #       'ed'
    )?                   #     end of grouping
  )                      #   end of grouping
)                        # end of look-ahead

Answer 3

Take a look at the ICU Project and icu4j. It does boundary analysis, so it splits sentences and words for you, and will do it for different languages.

For the rest, you can either match the words against a Pattern (as others have suggested), or check it against a Set of the words you're interested in.