Source code

001/*
002 *  gnu/regexp/RESyntax.java
003 *  Copyright (C) 1998-2001 Wes Biggs
004 *
005 *  This library is free software; you can redistribute it and/or modify
006 *  it under the terms of the GNU Lesser General Public License as published
007 *  by the Free Software Foundation; either version 2.1 of the License, or
008 *  (at your option) any later version.
009 *
010 *  This library is distributed in the hope that it will be useful,
011 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
012 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
013 *  GNU Lesser General Public License for more details.
014 *
015 *  You should have received a copy of the GNU Lesser General Public License
016 *  along with this program; if not, write to the Free Software
017 *  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
018 */
019
020package gnu.regexp;
021import java.io.Serializable;
022import java.util.BitSet;
023
024/**
025 * An RESyntax specifies the way a regular expression will be compiled.
026 * This class provides a number of predefined useful constants for
027 * emulating popular regular expression syntaxes.  Additionally the
028 * user may construct his or her own syntax, using any combination of the
029 * syntax bit constants.  The syntax is an optional argument to any of the
030 * matching methods on class RE.
031 *
032 * @author <A HREF="mailto:wes@cacas.org">Wes Biggs</A>
033 */
034
035public final class RESyntax implements Serializable {
036    static final String DEFAULT_LINE_SEPARATOR = System.getProperty("line.separator");
037
038    private static final String SYNTAX_IS_FINAL = RE.getLocalizedMessage("syntax.final");
039
040    private BitSet bits;
041
042    // true for the constant defined syntaxes
043    private boolean isFinal = false;
044
045    private String lineSeparator = DEFAULT_LINE_SEPARATOR;
046
047  // Values for constants are bit indexes
048
049  /**
050   * Syntax bit. Backslash is an escape character in lists.
051   */
052  public static final int RE_BACKSLASH_ESCAPE_IN_LISTS =  0;
053
054  /**
055   * Syntax bit. Use \? instead of ? and \+ instead of +.
056   */
057  public static final int RE_BK_PLUS_QM                =  1;
058
059  /**
060   * Syntax bit. POSIX character classes ([:...:]) in lists are allowed.
061   */
062  public static final int RE_CHAR_CLASSES              =  2;
063
064  /**
065   * Syntax bit. ^ and $ are special everywhere.
066   * <B>Not implemented.</B>
067   */
068  public static final int RE_CONTEXT_INDEP_ANCHORS     =  3; 
069
070  /**
071   * Syntax bit. Repetition operators are only special in valid positions.
072   * <B>Not implemented.</B>
073   */
074  public static final int RE_CONTEXT_INDEP_OPS         =  4; 
075
076  /**
077   * Syntax bit. Repetition and alternation operators are invalid
078   * at start and end of pattern and other places. 
079   * <B>Not implemented</B>.
080   */
081  public static final int RE_CONTEXT_INVALID_OPS       =  5; 
082
083  /**
084   * Syntax bit. Match-any-character operator (.) matches a newline.
085   */
086  public static final int RE_DOT_NEWLINE               =  6;
087
088  /**
089   * Syntax bit. Match-any-character operator (.) does not match a null.
090   */
091  public static final int RE_DOT_NOT_NULL              =  7;
092
093  /**
094   * Syntax bit. Intervals ({x}, {x,}, {x,y}) are allowed.
095   */
096  public static final int RE_INTERVALS                 =  8;
097
098  /**
099   * Syntax bit. No alternation (|), match one-or-more (+), or 
100   * match zero-or-one (?) operators.
101   */
102  public static final int RE_LIMITED_OPS               =  9;
103
104  /**
105   * Syntax bit. Newline is an alternation operator.
106   */
107  public static final int RE_NEWLINE_ALT               = 10; // impl.
108
109  /**
110   * Syntax bit. Intervals use { } instead of \{ \}
111   */
112  public static final int RE_NO_BK_BRACES              = 11; 
113
114  /**
115   * Syntax bit. Grouping uses ( ) instead of \( \).
116   */
117  public static final int RE_NO_BK_PARENS              = 12;
118
119  /**
120   * Syntax bit. Backreferences not allowed.
121   */
122  public static final int RE_NO_BK_REFS                = 13;
123
124  /**
125   * Syntax bit. Alternation uses | instead of \|
126   */
127  public static final int RE_NO_BK_VBAR                = 14;
128
129  /**
130   * Syntax bit. <B>Not implemented</B>.
131   */
132  public static final int RE_NO_EMPTY_RANGES           = 15;
133
134  /**
135   * Syntax bit. An unmatched right parenthesis (')' or '\)', depending
136   * on RE_NO_BK_PARENS) will throw an exception when compiling.
137   */
138  public static final int RE_UNMATCHED_RIGHT_PAREN_ORD = 16;
139
140  /**
141   * Syntax bit. <B>Not implemented.</B>
142   */
143  public static final int RE_HAT_LISTS_NOT_NEWLINE     = 17;
144
145  /**
146   * Syntax bit.  Stingy matching is allowed (+?, *?, ??, {x,y}?).
147   */
148  public static final int RE_STINGY_OPS                = 18;
149
150  /**
151   * Syntax bit. Allow character class escapes (\d, \D, \s, \S, \w, \W).
152   */
153  public static final int RE_CHAR_CLASS_ESCAPES        = 19;
154
155  /**
156   * Syntax bit. Allow use of (?:xxx) grouping (subexpression is not saved).
157   */
158  public static final int RE_PURE_GROUPING             = 20;
159
160  /**
161   * Syntax bit. Allow use of (?=xxx) and (?!xxx) apply the subexpression
162   * to the text following the current position without consuming that text.
163   */
164  public static final int RE_LOOKAHEAD                 = 21;
165
166  /**
167   * Syntax bit. Allow beginning- and end-of-string anchors (\A, \Z).
168   */
169  public static final int RE_STRING_ANCHORS            = 22;
170
171  /**
172   * Syntax bit. Allow embedded comments, (?#comment), as in Perl5.
173   */
174  public static final int RE_COMMENTS                  = 23;
175
176  /**
177   * Syntax bit. Allow character class escapes within lists, as in Perl5.
178   */
179  public static final int RE_CHAR_CLASS_ESC_IN_LISTS   = 24;
180
181  private static final int BIT_TOTAL                   = 25;
182
183  /**
184   * Predefined syntax.
185   * Emulates regular expression support in the awk utility.
186   */
187  public static final RESyntax RE_SYNTAX_AWK;
188
189  /**
190   * Predefined syntax.
191   * Emulates regular expression support in the ed utility.
192   */
193  public static final RESyntax RE_SYNTAX_ED;
194
195  /**
196   * Predefined syntax.
197   * Emulates regular expression support in the egrep utility.
198   */
199  public static final RESyntax RE_SYNTAX_EGREP;
200
201  /**
202   * Predefined syntax.
203   * Emulates regular expression support in the GNU Emacs editor.
204   */
205  public static final RESyntax RE_SYNTAX_EMACS;
206
207  /**
208   * Predefined syntax.
209   * Emulates regular expression support in the grep utility.
210   */
211  public static final RESyntax RE_SYNTAX_GREP;
212
213  /**
214   * Predefined syntax.
215   * Emulates regular expression support in the POSIX awk specification.
216   */
217  public static final RESyntax RE_SYNTAX_POSIX_AWK;
218
219  /**
220   * Predefined syntax.
221   * Emulates POSIX basic regular expression support.
222   */
223  public static final RESyntax RE_SYNTAX_POSIX_BASIC;
224
225  /**
226   * Predefined syntax.
227   * Emulates regular expression support in the POSIX egrep specification.
228   */
229  public static final RESyntax RE_SYNTAX_POSIX_EGREP;
230
231  /**
232   * Predefined syntax.
233   * Emulates POSIX extended regular expression support.
234   */
235  public static final RESyntax RE_SYNTAX_POSIX_EXTENDED;
236
237  /**
238   * Predefined syntax.
239   * Emulates POSIX basic minimal regular expressions.
240   */
241  public static final RESyntax RE_SYNTAX_POSIX_MINIMAL_BASIC;
242
243  /**
244   * Predefined syntax.
245   * Emulates POSIX extended minimal regular expressions.
246   */
247  public static final RESyntax RE_SYNTAX_POSIX_MINIMAL_EXTENDED;
248
249  /**
250   * Predefined syntax.
251   * Emulates regular expression support in the sed utility.
252   */
253  public static final RESyntax RE_SYNTAX_SED;
254
255  /**
256   * Predefined syntax.
257   * Emulates regular expression support in Larry Wall's perl, version 4,
258   */
259  public static final RESyntax RE_SYNTAX_PERL4;
260
261  /**
262   * Predefined syntax.
263   * Emulates regular expression support in Larry Wall's perl, version 4,
264   * using single line mode (/s modifier).
265   */
266  public static final RESyntax RE_SYNTAX_PERL4_S; // single line mode (/s)
267
268  /**
269   * Predefined syntax.
270   * Emulates regular expression support in Larry Wall's perl, version 5.
271   */
272  public static final RESyntax RE_SYNTAX_PERL5;  
273
274  /**
275   * Predefined syntax.
276   * Emulates regular expression support in Larry Wall's perl, version 5,
277   * using single line mode (/s modifier).
278   */
279  public static final RESyntax RE_SYNTAX_PERL5_S;
280  
281  static {
282      // Define syntaxes
283      
284      RE_SYNTAX_EMACS = new RESyntax().makeFinal();
285      
286      RESyntax RE_SYNTAX_POSIX_COMMON = new RESyntax()
287          .set(RE_CHAR_CLASSES)
288          .set(RE_DOT_NEWLINE)
289          .set(RE_DOT_NOT_NULL)
290          .set(RE_INTERVALS)
291          .set(RE_NO_EMPTY_RANGES)
292          .makeFinal();
293      
294      RE_SYNTAX_POSIX_BASIC = new RESyntax(RE_SYNTAX_POSIX_COMMON)
295          .set(RE_BK_PLUS_QM)
296          .makeFinal();
297      
298      RE_SYNTAX_POSIX_EXTENDED = new RESyntax(RE_SYNTAX_POSIX_COMMON)
299          .set(RE_CONTEXT_INDEP_ANCHORS)
300          .set(RE_CONTEXT_INDEP_OPS)
301          .set(RE_NO_BK_BRACES)
302          .set(RE_NO_BK_PARENS)
303          .set(RE_NO_BK_VBAR)
304          .set(RE_UNMATCHED_RIGHT_PAREN_ORD)
305          .makeFinal();
306
307      RE_SYNTAX_AWK = new RESyntax()
308          .set(RE_BACKSLASH_ESCAPE_IN_LISTS)
309          .set(RE_DOT_NOT_NULL)
310          .set(RE_NO_BK_PARENS)
311          .set(RE_NO_BK_REFS)
312          .set(RE_NO_BK_VBAR)
313          .set(RE_NO_EMPTY_RANGES)
314          .set(RE_UNMATCHED_RIGHT_PAREN_ORD)
315          .makeFinal();
316      
317      RE_SYNTAX_POSIX_AWK = new RESyntax(RE_SYNTAX_POSIX_EXTENDED)
318          .set(RE_BACKSLASH_ESCAPE_IN_LISTS)
319          .makeFinal();
320      
321      RE_SYNTAX_GREP = new RESyntax()
322          .set(RE_BK_PLUS_QM)
323          .set(RE_CHAR_CLASSES)
324          .set(RE_HAT_LISTS_NOT_NEWLINE)
325          .set(RE_INTERVALS)
326          .set(RE_NEWLINE_ALT)
327          .makeFinal();
328      
329      RE_SYNTAX_EGREP = new RESyntax()
330          .set(RE_CHAR_CLASSES)
331          .set(RE_CONTEXT_INDEP_ANCHORS)
332          .set(RE_CONTEXT_INDEP_OPS)
333          .set(RE_HAT_LISTS_NOT_NEWLINE)
334          .set(RE_NEWLINE_ALT)
335          .set(RE_NO_BK_PARENS)
336          .set(RE_NO_BK_VBAR)
337          .makeFinal();
338    
339      RE_SYNTAX_POSIX_EGREP = new RESyntax(RE_SYNTAX_EGREP)
340          .set(RE_INTERVALS)
341          .set(RE_NO_BK_BRACES)
342          .makeFinal();
343    
344      /* P1003.2/D11.2, section 4.20.7.1, lines 5078ff.  */
345    
346      RE_SYNTAX_ED = new RESyntax(RE_SYNTAX_POSIX_BASIC)
347          .makeFinal();
348    
349      RE_SYNTAX_SED = new RESyntax(RE_SYNTAX_POSIX_BASIC)
350          .makeFinal();
351      
352      RE_SYNTAX_POSIX_MINIMAL_BASIC = new RESyntax(RE_SYNTAX_POSIX_COMMON)
353          .set(RE_LIMITED_OPS)
354          .makeFinal();
355      
356      /* Differs from RE_SYNTAX_POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS
357         replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */
358      
359      RE_SYNTAX_POSIX_MINIMAL_EXTENDED = new RESyntax(RE_SYNTAX_POSIX_COMMON)
360          .set(RE_CONTEXT_INDEP_ANCHORS)
361          .set(RE_CONTEXT_INVALID_OPS)
362          .set(RE_NO_BK_BRACES)
363          .set(RE_NO_BK_PARENS)
364          .set(RE_NO_BK_REFS)
365          .set(RE_NO_BK_VBAR)
366          .set(RE_UNMATCHED_RIGHT_PAREN_ORD)
367          .makeFinal();
368      
369      /* There is no official Perl spec, but here's a "best guess" */
370      
371      RE_SYNTAX_PERL4 = new RESyntax()
372          .set(RE_BACKSLASH_ESCAPE_IN_LISTS)
373          .set(RE_CONTEXT_INDEP_ANCHORS)
374          .set(RE_CONTEXT_INDEP_OPS)          // except for '{', apparently
375          .set(RE_INTERVALS)
376          .set(RE_NO_BK_BRACES)
377          .set(RE_NO_BK_PARENS)
378          .set(RE_NO_BK_VBAR)
379          .set(RE_NO_EMPTY_RANGES)
380          .set(RE_CHAR_CLASS_ESCAPES)    // \d,\D,\w,\W,\s,\S
381          .makeFinal();
382      
383      RE_SYNTAX_PERL4_S = new RESyntax(RE_SYNTAX_PERL4)
384          .set(RE_DOT_NEWLINE)
385          .makeFinal();
386      
387      RE_SYNTAX_PERL5 = new RESyntax(RE_SYNTAX_PERL4)
388          .set(RE_PURE_GROUPING)          // (?:)
389          .set(RE_STINGY_OPS)             // *?,??,+?,{}?
390          .set(RE_LOOKAHEAD)              // (?=)(?!)
391          .set(RE_STRING_ANCHORS)         // \A,\Z
392          .set(RE_CHAR_CLASS_ESC_IN_LISTS)// \d,\D,\w,\W,\s,\S within []
393          .set(RE_COMMENTS)              // (?#)
394          .makeFinal();
395      
396      RE_SYNTAX_PERL5_S = new RESyntax(RE_SYNTAX_PERL5)
397          .set(RE_DOT_NEWLINE)
398          .makeFinal();
399  }
400
401  /**
402   * Construct a new syntax object with all bits turned off.
403   * This is equivalent to RE_SYNTAX_EMACS.
404   */
405  public RESyntax() {
406    bits = new BitSet(BIT_TOTAL);
407  }
408
409    /**
410     * Called internally when constructing predefined syntaxes
411     * so their interpretation cannot vary.  Conceivably useful
412     * for your syntaxes as well.  Causes IllegalAccessError to
413     * be thrown if any attempt to modify the syntax is made.
414     *
415     * @return this object for convenient chaining
416     */
417    public RESyntax makeFinal() {
418        isFinal = true;
419        return this;
420    }
421
422  /**
423   * Construct a new syntax object with all bits set the same 
424   * as the other syntax.
425   */
426  public RESyntax(RESyntax other) {
427    bits = (BitSet) other.bits.clone();
428  }
429
430  /**
431   * Check if a given bit is set in this syntax.
432   */
433  public boolean get(int index) {
434    return bits.get(index);
435  }
436
437  /**
438   * Set a given bit in this syntax. 
439   *
440   * @param index the constant (RESyntax.RE_xxx) bit to set.
441   * @return a reference to this object for easy chaining.
442   */
443  public RESyntax set(int index) {
444      if (isFinal) throw new IllegalAccessError(SYNTAX_IS_FINAL);
445    bits.set(index);
446    return this;
447  }
448
449  /**
450   * Clear a given bit in this syntax. 
451   *
452   * @param index the constant (RESyntax.RE_xxx) bit to clear.
453   * @return a reference to this object for easy chaining.
454   */
455  public RESyntax clear(int index) {
456      if (isFinal) throw new IllegalAccessError(SYNTAX_IS_FINAL);
457      bits.clear(index);
458      return this;
459  }
460
461    /**
462     * Changes the line separator string for regular expressions
463     * created using this RESyntax.  The default separator is the
464     * value returned by the system property "line.separator", which
465     * should be correct when reading platform-specific files from a
466     * filesystem.  However, many programs may collect input from
467     * sources where the line separator is differently specified (for
468     * example, in the applet environment, the text box widget
469     * interprets line breaks as single-character newlines,
470     * regardless of the host platform.
471     *
472     * Note that setting the line separator to a character or
473     * characters that have specific meaning within the current syntax
474     * can cause unexpected chronosynclastic infundibula.
475     *
476     * @return this object for convenient chaining 
477     */
478    public RESyntax setLineSeparator(String aSeparator) {
479        if (isFinal) throw new IllegalAccessError(SYNTAX_IS_FINAL);
480        lineSeparator = aSeparator;
481        return this;
482    }
483
484    /**
485     * Returns the currently active line separator string.  The default
486     * is the platform-dependent system property "line.separator".
487     */
488    public String getLineSeparator() {
489        return lineSeparator;
490    }
491}