prog-intro-2025/java/wspp/WordScanner.java

package wspp;

import java.io.*;
import java.nio.charset.StandardCharsets;

public class WordScanner {
    private BufferedReader br;
    private String line = null;
    private int pos = 0;

    public WordScanner(String fileName) throws IOException {
        br = new BufferedReader(new InputStreamReader(
            new FileInputStream(fileName), StandardCharsets.UTF_8));
    }

    private boolean hasNextLine() throws IOException {
        if (line != null && pos < line.length()) return true;
        line = br.readLine();
        pos = 0;
        return line != null;
    }

    private boolean isWordChar(char c) {
        return Character.isLetter(c) || c == '\'' || 
               Character.getType(c) == Character.DASH_PUNCTUATION;
    }

    public boolean hasNextWord() throws IOException {
        while (hasNextLine()) {
            while (pos < line.length() && !isWordChar(line.charAt(pos))) {
                pos++;
            }
            if (pos < line.length()) {
                return true;
            }
        }
        return false;
    }

    public String nextWord() {
        while (pos < line.length() && !isWordChar(line.charAt(pos))) {
            pos++;
        }

        int start = pos;
        while (pos < line.length() && isWordChar(line.charAt(pos))) {
            pos++;
        }

        return line.substring(start, pos).toLowerCase();
    }

    public void close() throws IOException {
        br.close();
    }
}