StringMatcher


1   /*******************************************************************************
2    * Copyright (c) 2000, 2004 IBM Corporation and others.
3    * All rights reserved. This program and the accompanying materials 
4    * are made available under the terms of the Common Public License v1.0
5    * which accompanies this distribution, and is available at
6    * http://www.eclipse.org/legal/cpl-v10.html
7    * 
8    * Contributors:
9    *     IBM Corporation - initial API and implementation
10   *******************************************************************************/
11  package org.eclipse.ui.externaltools.internal.model;
12   
13  import java.util.Vector  ;
14  
15  /**
16   * A string pattern matcher, suppporting * and ? wildcards.
17   */
18  public class StringMatcher {
19      protected String   fPattern;
20      protected int fLength; // pattern length
21      protected boolean fIgnoreWildCards;
22      protected boolean fIgnoreCase;
23      protected boolean fHasLeadingStar;
24      protected boolean fHasTrailingStar;
25      protected String   fSegments[]; //the given pattern is split into * separated segments
26  
27      /* boundary value beyond which we don't need to search in the text */
28      protected int fBound= 0;
29      
30  
31      protected static final char fSingleWildCard= '\u0000';
32      
33      public static class Position {
34          int start; //inclusive
35          int end; //exclusive
36          public Position(int start, int end) {
37              this.start= start;
38              this.end= end;
39          }
40          public int getStart() {
41              return start;
42          }
43          public int getEnd() {
44              return end;
45          }
46      }
47      /**
48       * StringMatcher constructor takes in a String object that is a simple 
49       * pattern which may contain '*' for 0 and many characters and
50       * '?' for exactly one character.  
51       *
52       * Literal '*' and '?' characters must be escaped in the pattern 
53       * e.g., "\*" means literal "*", etc.
54       *
55       * Escaping any other character (including the escape character itself), 
56       * just results in that character in the pattern.
57       * e.g., "\a" means "a" and "\\" means "\"
58       *
59       * If invoking the StringMatcher with string literals in Java, don't forget
60       * escape characters are represented by "\\".
61       *
62       * @param pattern the pattern to match text against
63       * @param ignoreCase if true, case is ignored
64       * @param ignoreWildCards if true, wild cards and their escape sequences are ignored
65       *        (everything is taken literally).
66       */
67      public StringMatcher(String   pattern, boolean ignoreCase, boolean ignoreWildCards) {
68          if (pattern == null)
69              throw new IllegalArgumentException  ();
70          fIgnoreCase= ignoreCase;
71          fIgnoreWildCards= ignoreWildCards;
72          fPattern= pattern;
73          fLength= pattern.length();
74          
75          if (fIgnoreWildCards) {
76              parseNoWildCards();
77          } else {
78              parseWildCards();
79          }
80      }
81      /**
82       * Find the first occurrence of the pattern between <code>start</code>(inclusive) 
83       * and <code>end</code>(exclusive).  
84       * @param text the String object to search in 
85       * @param start the starting index of the search range, inclusive
86       * @param end the ending index of the search range, exclusive
87       * @return an StringMatcher.Position</code> object that keeps the starting 
88       * (inclusive) and ending positions (exclusive) of the first occurrence of the 
89       * pattern in the specified range of the text; return null if not found or subtext
90       * is empty (start==end). A pair of zeros is returned if pattern is empty string
91       * Note that for pattern like "*abc*" with leading and trailing stars, position of "abc"
92       * is returned. For a pattern like"*??*" in text "abcdf", (1,3) is returned
93       */
94      public StringMatcher.Position find(String   text, int start, int end) {
95          if (text == null)
96              throw new IllegalArgumentException  ();
97              
98          int tlen= text.length();
99          if (start < 0)
100             start= 0;
101         if (end > tlen)
102             end= tlen;
103         if (end < 0 ||start >= end )
104             return null;
105         if (fLength == 0)
106             return new Position(start, start);
107         if (fIgnoreWildCards) {
108             int x= posIn(text, start, end);
109             if (x < 0)
110                 return null;
111             return new Position(x, x+fLength);
112         }
113 
114         int segCount= fSegments.length;
115         if (segCount == 0)//pattern contains only '*'(s)
116             return new Position (start, end);
117                     
118         int curPos= start;
119         int matchStart= -1;
120         int i;
121         for (i= 0; i < segCount && curPos < end; ++i) {
122             String   current= fSegments[i];
123             int nextMatch= regExpPosIn(text, curPos, end, current);
124             if (nextMatch < 0 )
125                 return null;
126             if(i == 0)
127                 matchStart= nextMatch;
128             curPos= nextMatch + current.length();
129         }
130         if (i < segCount)
131             return null;
132         return new Position(matchStart, curPos);
133     }
134     /**
135      * match the given <code>text</code> with the pattern 
136      * @return true if matched eitherwise false
137      * @param text a String object 
138      */
139     public boolean match(String   text) {
140         return match(text, 0, text.length());
141     }
142     /**
143      * Given the starting (inclusive) and the ending (exclusive) positions in the   
144      * <code>text</code>, determine if the given substring matches with aPattern  
145      * @return true if the specified portion of the text matches the pattern
146      * @param text a String object that contains the substring to match 
147      * @param start marks the starting position (inclusive) of the substring
148      * @param end marks the ending index (exclusive) of the substring 
149      */
150     public boolean match(String   text, int start, int end) {
151         if (null == text)
152             throw new IllegalArgumentException  ();
153 
154         if (start > end)
155             return false;
156 
157         if (fIgnoreWildCards)
158             return (end - start == fLength) && fPattern.regionMatches(fIgnoreCase, 0, text, start, fLength);
159         int segCount= fSegments.length;
160         if (segCount == 0 && (fHasLeadingStar || fHasTrailingStar))  // pattern contains only '*'(s)
161             return true;
162         if (start == end)
163             return fLength == 0;
164         if (fLength == 0)
165             return start == end;
166 
167         int tlen= text.length();
168         if (start < 0)
169             start= 0;
170         if (end > tlen)
171             end= tlen;
172 
173         int tCurPos= start;
174         int bound= end - fBound;
175         if ( bound < 0)
176             return false;
177         int i=0;
178         String   current= fSegments[i];
179         int segLength= current.length();
180 
181         /* process first segment */
182         if (!fHasLeadingStar){
183             if(!regExpRegionMatches(text, start, current, 0, segLength)) {
184                 return false;
185             } 
186             ++i;
187             tCurPos= tCurPos + segLength;
188         }
189         if ((fSegments.length == 1) && (!fHasLeadingStar) && (!fHasTrailingStar)) {
190             // only one segment to match, no wildcards specified
191             return tCurPos == end;
192         }
193         /* process middle segments */
194         while (i < segCount) {
195             current= fSegments[i];
196             int currentMatch;
197             int k= current.indexOf(fSingleWildCard);
198             if (k < 0) {
199                 currentMatch= textPosIn(text, tCurPos, end, current);
200                 if (currentMatch < 0)
201                     return false;
202             } else {
203                 currentMatch= regExpPosIn(text, tCurPos, end, current);
204                 if (currentMatch < 0)
205                     return false;
206             }
207             tCurPos= currentMatch + current.length();
208             i++;
209         }
210 
211         /* process final segment */
212         if (!fHasTrailingStar && tCurPos != end) {
213             int clen= current.length();
214             return regExpRegionMatches(text, end - clen, current, 0, clen);
215         }
216         return i == segCount ;
217     }
218 
219     /**
220      * This method parses the given pattern into segments seperated by wildcard '*' characters.
221      * Since wildcards are not being used in this case, the pattern consists of a single segment.
222      */
223     private void parseNoWildCards() {
224         fSegments= new String  [1];
225         fSegments[0]= fPattern;
226         fBound= fLength;
227     }
228     /**
229      * Parses the given pattern into segments seperated by wildcard '*' characters.
230      * @param p, a String object that is a simple regular expression with '*' and/or '?'
231      */
232     private void parseWildCards() {
233         if(fPattern.startsWith("*"))//$NON-NLS-1$
234             fHasLeadingStar= true;
235         if(fPattern.endsWith("*")) {//$NON-NLS-1$
236             /* make sure it's not an escaped wildcard */
237             if (fLength > 1 && fPattern.charAt(fLength - 2) != '\\') {
238                 fHasTrailingStar= true;
239             }
240         }
241 
242         Vector   temp= new Vector  ();
243 
244         int pos= 0;
245         StringBuffer   buf= new StringBuffer  ();
246         while (pos < fLength) {
247             char c= fPattern.charAt(pos++);
248             switch (c) {
249                 case '\\':
250                     if (pos >= fLength) {
251                         buf.append(c);
252                     } else {
253                         char next= fPattern.charAt(pos++);
254                         /* if it's an escape sequence */
255                         if (next == '*' || next == '?' || next == '\\') {
256                             buf.append(next);
257                         } else {
258                             /* not an escape sequence, just insert literally */
259                             buf.append(c);
260                             buf.append(next);
261                         }
262                     }
263                 break;
264                 case '*':
265                     if (buf.length() > 0) {
266                         /* new segment */
267                         temp.addElement(buf.toString());
268                         fBound += buf.length();
269                         buf.setLength(0);
270                     }
271                 break;
272                 case '?':
273                     /* append special character representing single match wildcard */
274                     buf.append(fSingleWildCard);
275                 break;
276                 default:
277                     buf.append(c);
278             }
279         }
280 
281         /* add last buffer to segment list */
282         if (buf.length() > 0) {
283             temp.addElement(buf.toString());
284             fBound += buf.length();
285         }
286             
287         fSegments= new String  [temp.size()];
288         temp.copyInto(fSegments);
289     }
290     /** 
291      * @param text a string which contains no wildcard
292      * @param start the starting index in the text for search, inclusive
293      * @param end the stopping point of search, exclusive
294      * @return the starting index in the text of the pattern , or -1 if not found 
295      */
296     protected int posIn(String   text, int start, int end) {//no wild card in pattern
297         int max= end - fLength;
298         
299         if (!fIgnoreCase) {
300             int i= text.indexOf(fPattern, start);
301             if (i == -1 || i > max)
302                 return -1;
303             return i;
304         }
305         
306         for (int i= start; i <= max; ++i) {
307             if (text.regionMatches(true, i, fPattern, 0, fLength))
308                 return i;
309         }
310         
311         return -1;
312     }
313     /** 
314      * @param text a simple regular expression that may only contain '?'(s)
315      * @param start the starting index in the text for search, inclusive
316      * @param end the stopping point of search, exclusive
317      * @param p a simple regular expression that may contains '?'
318      * @param caseIgnored whether the pattern is not casesensitive
319      * @return the starting index in the text of the pattern , or -1 if not found 
320      */
321     protected int regExpPosIn(String   text, int start, int end, String   p) {
322         int plen= p.length();
323         
324         int max= end - plen;
325         for (int i= start; i <= max; ++i) {
326             if (regExpRegionMatches(text, i, p, 0, plen))
327                 return i;
328         }
329         return -1;
330     }
331     /**
332      * 
333      * @return boolean
334      * @param text a String to match
335      * @param start int that indicates the starting index of match, inclusive
336      * @param end int that indicates the ending index of match, exclusive
337      * @param p a simple regular expression that may contain '?'
338      * @param ignoreCase boolean indicating wether code>p</code> is case sensitive
339      */
340     protected boolean regExpRegionMatches(String   text, int tStart, String   p, int pStart, int plen) {
341         while (plen-- > 0) {
342             char tchar= text.charAt(tStart++);
343             char pchar= p.charAt(pStart++);
344 
345             /* process wild cards */
346             if (!fIgnoreWildCards) {
347                 /* skip single wild cards */
348                 if (pchar == fSingleWildCard) {
349                     continue;
350                 }
351             }
352             if (pchar == tchar)
353                 continue;
354             if (fIgnoreCase) {
355                 if (Character.toUpperCase(tchar) == Character.toUpperCase(pchar))
356                     continue;
357                 // comparing after converting to upper case doesn't handle all cases;
358                 // also compare after converting to lower case
359                 if (Character.toLowerCase(tchar) == Character.toLowerCase(pchar))
360                     continue;
361             }
362             return false;
363         }
364         return true;
365     }
366     /** 
367      * @param text the string to match
368      * @param start the starting index in the text for search, inclusive
369      * @param end the stopping point of search, exclusive
370      * @param p a string that has no wildcard
371      * @param ignoreCase boolean indicating wether code>p</code> is case sensitive
372      * @return the starting index in the text of the pattern , or -1 if not found 
373      */
374     protected int textPosIn(String   text, int start, int end, String   p) { 
375         
376         int plen= p.length();
377         int max= end - plen;
378         
379         if (!fIgnoreCase) {
380             int i= text.indexOf(p, start);
381             if (i == -1 || i > max)
382                 return -1;
383             return i;
384         }
385         
386         for (int i= start; i <= max; ++i) {
387             if (text.regionMatches(true, i, p, 0, plen))
388                 return i;
389         }
390         
391         return -1;
392     }
393 }
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags