Compare_en


1   package com.icl.saxon.sort;
2   
3   // Copyright � International Computers Limited 1998
4   // See conditions of use
5   
6   /**
7    * A Comparer used with lang="en". Note this only does anything intelligent with characters
8    * in ISO 8859/1, which are mapped to their unaccented equivalents
9    *
10   * @author Michael H. Kay (mhkay@iclway.co.uk)
11   *
12   */
13  
14  public class Compare_en extends TextComparer {
15  
16      // Following string maps Latin-1 characters in the range C0-FF to equivalent unaccented letter
17  
18      private static String   supp =
19      "AAAAAAACEEEEIIII[NOOOOO*OUUUUY]Saaaaaaaceeeeiiii{nooooo*ouuuuy}y";
20  
21      int caseOrder = UPPERCASE_FIRST;
22  
23      /**
24      * Compare two string objects, in three phases:
25      * (a) ignoring accents and case
26      * (b) if still equal, ignoring case
27      * (c) if still equal, taking case into account
28      * @return <0 if a<b, 0 if a=b, >0 if a>b
29      */
30  
31      public int compare(Object   a, Object   b) {
32  
33          char[] a1 = ((String  )a).toCharArray();
34          char[] b1 = ((String  )b).toCharArray();
35          int alen = a1.length;
36          int blen = b1.length;
37  
38          // strip off the accents
39          for (int k=0; k<alen; k++) {
40              int code = (int)a1[k];
41              if (code>=192 && code<=255) {
42                  a1[k] = supp.charAt(code-192);
43              }
44          }
45          for (int k=0; k<blen; k++) {
46              int code = (int)b1[k];
47              if (code>=192 && code<=255) {
48                  b1[k] = supp.charAt(code-192);
49              }
50          }
51          int i = 0;
52          int j = 0;
53          
54          // do an accent-blind comparison
55          while (true) {
56              if (i==alen && j==blen) break;
57              if (i==alen) return -1;
58              if (j==blen) return +1;
59              int diff = Character.toLowerCase(a1[i]) -
60                               Character.toLowerCase(b1[j]);
61              i++;
62              j++;
63              if (diff!=0) return diff;
64          }
65  
66          // still equal - try again with accents, but still case-blind
67          
68          a1 = ((String  )a).toCharArray();
69          b1 = ((String  )b).toCharArray();
70          i = 0;
71          j = 0;
72          while (true) {
73              if (i==alen && j==blen) break;
74              if (i==alen) return -1;
75              if (j==blen) return +1;
76              int diff = Character.toLowerCase(a1[i]) -
77                               Character.toLowerCase(b1[j]);
78              i++;
79              j++;
80              if (diff!=0) return diff;
81          }
82  
83          // still equal: take case into account
84          
85          i = 0;
86          j = 0;
87          while (true) {
88              if (i==alen) return 0;
89              int diff = a1[i++] - b1[j++]; 
90              if (diff!=0) {
91                  if (caseOrder==LOWERCASE_FIRST) {
92                      return (Character.isLowerCase(a1[i-1]) ? -1 : +1);
93                  } else {
94                      return (Character.isUpperCase(a1[i-1]) ? -1 : +1);
95                  }
96              }
97          }        
98              
99      }
100 
101     public Comparer setCaseOrder(int caseOrder) {
102         this.caseOrder = caseOrder;
103         return this;
104     }
105         
106 }
107
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags