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

   1 /*
   2  * $Id: StringDivider.java,v 1.1 2004-10-05 20:51:57 jsurfer 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.compile("<[\\s/]*(\\w+)\\s*.*>");
  13
  14         String[] splitRegexpCandidates = {
  15                 "\t",
  16                 ",",
  17                 "\\s",
  18                 "\\s+",
  19         };
  20
  21         public StringDivider() {
  22         }
  23
  24         public String[][] divide(String content){
  25                 return divide(content, getDivideSuitedRegexp(content));
  26         }
  27
  28         public String[][] divide(String content, String regexp){
  29                 String[] lines = content.split("\n");
  30                 int len = lines.length;
  31                 String[][] dist = new String[len][];
  32
  33                 int max = Integer.MIN_VALUE;
  34                 for (int i = 0; i < len; i++) {
  35                         String line = lines[i];
  36                         String[] cells = line.split(regexp);
  37                         dist[i] = cells;
  38                         if(max < cells.length){
  39                                 max = cells.length;
  40                         }
  41                 }
  42                 for (int i = 0; i < len; i++) {
  43                         String[] newArray = new String[max];
  44                         Arrays.fill(newArray, "");
  45                         System.arraycopy(dist[i], 0, newArray, 0, dist[i].length);
  46                         dist[i] = newArray;
  47                 }
  48                 return dist;
  49         }
  50
  51         public String getDivideSuitedRegexp(String content){
  52                 String[] lines = content.split("\n");
  53
  54                 String resultRegexp = null;
  55                 int score = Integer.MAX_VALUE, cellCount = Integer.MIN_VALUE;
  56
  57                 for (int i = 0; i < splitRegexpCandidates.length; i++) {
  58                         String regexp = splitRegexpCandidates[i];
  59                         int max = Integer.MIN_VALUE, min = Integer.MAX_VALUE;
  60                         for (int j = 0; j < lines.length; j++) {
  61                                 String[] vals = lines[j].split(regexp);
  62                                 if(max < vals.length){
  63                                         max = vals.length;
  64                                 }
  65                                 if(min > vals.length){
  66                                         min = vals.length;
  67                                 }
  68                         }
  69                         int s = max - min;
  70                         if( score > s || (score == s && max > cellCount)){
  71                                 cellCount = max;
  72                                 score = s;
  73                                 resultRegexp = regexp;
  74                         }
  75                 }
  76                 return resultRegexp;
  77         }
  78
  79 }