net.sourceforge.phpeclipse/src/net/sourceforge/phpeclipse/wizards/html/StringDivider.java

   1 /*
   2  * $Id: StringDivider.java,v 1.2 2006-10-21 23:18:43 pombredanne Exp $
   3  * Copyright Narushima Hironori. All rights reserved.
   4  */
   5 package net.sourceforge.phpeclipse.wizards.html;
   6
   7 import java.util.Arrays;
   8 import java.util.regex.Pattern;
   9
  10 public class StringDivider {
  11
  12         static Pattern tagNameChoosePattern = Pattern
  13                         .compile("<[\\s/]*(\\w+)\\s*.*>");
  14
  15         String[] splitRegexpCandidates = { "\t", ",", "\\s", "\\s+", };
  16
  17         public StringDivider() {
  18         }
  19
  20         public String[][] divide(String content) {
  21                 return divide(content, getDivideSuitedRegexp(content));
  22         }
  23
  24         public String[][] divide(String content, String regexp) {
  25                 String[] lines = content.split("\n");
  26                 int len = lines.length;
  27                 String[][] dist = new String[len][];
  28
  29                 int max = Integer.MIN_VALUE;
  30                 for (int i = 0; i < len; i++) {
  31                         String line = lines[i];
  32                         String[] cells = line.split(regexp);
  33                         dist[i] = cells;
  34                         if (max < cells.length) {
  35                                 max = cells.length;
  36                         }
  37                 }
  38                 for (int i = 0; i < len; i++) {
  39                         String[] newArray = new String[max];
  40                         Arrays.fill(newArray, "");
  41                         System.arraycopy(dist[i], 0, newArray, 0, dist[i].length);
  42                         dist[i] = newArray;
  43                 }
  44                 return dist;
  45         }
  46
  47         public String getDivideSuitedRegexp(String content) {
  48                 String[] lines = content.split("\n");
  49
  50                 String resultRegexp = null;
  51                 int score = Integer.MAX_VALUE, cellCount = Integer.MIN_VALUE;
  52
  53                 for (int i = 0; i < splitRegexpCandidates.length; i++) {
  54                         String regexp = splitRegexpCandidates[i];
  55                         int max = Integer.MIN_VALUE, min = Integer.MAX_VALUE;
  56                         for (int j = 0; j < lines.length; j++) {
  57                                 String[] vals = lines[j].split(regexp);
  58                                 if (max < vals.length) {
  59                                         max = vals.length;
  60                                 }
  61                                 if (min > vals.length) {
  62                                         min = vals.length;
  63                                 }
  64                         }
  65                         int s = max - min;
  66                         if (score > s || (score == s && max > cellCount)) {
  67                                 cellCount = max;
  68                                 score = s;
  69                                 resultRegexp = regexp;
  70                         }
  71                 }
  72                 return resultRegexp;
  73         }
  74
  75 }