1. import java.io.*;
  2. import java.util.ArrayList;
  3. import java.util.List;
  4. import java.util.Random;
  6. public class DataCleaner {
  8. private static final int MAX_RETRIES = 3;
  9. private static final int CLEANING_ATTEMPTS = 3;
  10. private static final Random random = new Random();
  12. public static List<String> cleanData(String filePath) throws IOException {
  13. List<String> cleanedData = new ArrayList<>();
  14. int retryCount = 0;
  16. while (retryCount < MAX_RETRIES) {
  17. try {
  18. cleanedData = cleanFile(filePath);
  19. break; // Exit loop if cleaning is successful
  20. } catch (IOException e) {
  21. retryCount++;
  22. if (retryCount < MAX_RETRIES) {
  23. System.err.println("Error cleaning file. Retry attempt: " + retryCount);
  24. try {
  25. Thread.sleep(1000 * (random.nextDouble() * 3 + 1)); // Random delay between 1-3 seconds
  26. } catch (InterruptedException ie) {
  27. Thread.currentThread().interrupt();
  28. }
  29. } else {
  30. System.err.println("Failed to clean file after multiple retries.");
  31. throw e; // Re-throw the exception after max retries
  32. }
  33. }
  34. }
  36. return cleanedData;
  37. }
  39. private static List<String> cleanFile(String filePath) throws IOException {
  40. List<String> lines = new ArrayList<>();
  41. try (BufferedReader reader = new BufferedReader(new FileReader(filePath))) {
  42. String line;
  43. while ((line = reader.readLine()) != null) {
  44. line = cleanLine(line);
  45. if (!line.isEmpty()) { //Only add non-empty lines
  46. lines.add(line);
  47. }
  48. }
  49. }
  50. return lines;
  51. }
  53. private static String cleanLine(String line) {
  54. // Remove leading/trailing whitespace
  55. line = line.trim();
  57. // Remove multiple spaces
  58. line = line.replaceAll("\\s+", " ");
  60. //Remove special characters (keep only alphanumeric and spaces)
  61. line = line.replaceAll("[^a-zA-Z0-9\\s]", "");
  63. //Remove extra spaces
  64. line = line.trim();
  66. //Remove empty strings
  67. if(line.isEmpty()){
  68. return "";
  69. }
  71. return line;
  72. }
  75. public static void main(String[] args) {
  76. try {
  77. // Example usage
  78. List<String> data = cleanData("input.txt"); // Replace with your file path
  79. for (String cleanedLine : data) {
  80. System.out.println(cleanedLine);
  81. }
  82. } catch (IOException e) {
  83. System.err.println("Error: " + e.getMessage());
  84. }
  85. }
  86. }

Add your comment