[apple/icu.git] / icuSources / i18n / regexst.cpp

//
//  regexst.h
//
//  Copyright (C) 2004, International Business Machines Corporation and others.
//  All Rights Reserved.
//
//  This file contains class RegexStaticSets
//
//  This class is internal to the regular expression implementation.
//  For the public Regular Expression API, see the file "unicode/regex.h"
//
//  RegexStaticSets groups together the common UnicodeSets that are needed
//   for compiling or executing RegularExpressions.  This grouping simplifies
//   the thread safe lazy creation and sharing of these sets across
//   all instances of regular expressions.
//
#include "unicode/utypes.h"

#if !UCONFIG_NO_REGULAR_EXPRESSIONS

#include "unicode/unistr.h"
#include "unicode/uniset.h"
#include "unicode/uchar.h"
#include "unicode/regex.h"
#include "uprops.h"
#include "cmemory.h"
#include "cstring.h"
#include "uassert.h"
#include "ucln_in.h"
#include "umutex.h"

#include "regexcst.h"   // Contains state table for the regex pattern parser.
                        //   generated by a Perl script.
#include "regexst.h"


U_NAMESPACE_BEGIN


//----------------------------------------------------------------------------------------
//
// Unicode Set pattern strings for all of the required constant sets.
//               Initialized with hex values for portability to EBCDIC based machines.
//                Really ugly, but there's no good way to avoid it.
//
//----------------------------------------------------------------------------------------

// "Rule Char" Characters are those with no special meaning, and therefore do not
//    need to be escaped to appear as literals in a regexp.  Expressed
//    as the inverse of those needing escaping --  [^\*\?\+\[\(\)\{\}\^\$\|\\\.]
static const UChar gRuleSet_rule_char_pattern[]       = {
 //   [    ^      \     *     \     ?     \     +     \     [     \     (     /     )
    0x5b, 0x5e, 0x5c, 0x2a, 0x5c, 0x3f, 0x5c, 0x2b, 0x5c, 0x5b, 0x5c, 0x28, 0x5c, 0x29,
 //   \     {    \     }     \     ^     \     $     \     |     \     \     \     .     ]
    0x5c, 0x7b,0x5c, 0x7d, 0x5c, 0x5e, 0x5c, 0x24, 0x5c, 0x7c, 0x5c, 0x5c, 0x5c, 0x2e, 0x5d, 0};


static const UChar gRuleSet_digit_char_pattern[] = {
//    [    0      -    9     ]
    0x5b, 0x30, 0x2d, 0x39, 0x5d, 0};
//static const UnicodeSet *gRuleDigits = NULL;


//
//   Here are the backslash escape characters that ICU's unescape() function
//    will handle.
//
static const UChar gUnescapeCharPattern[] = {
//    [     a     c     e     f     n     r     t     u     U     x    ]
    0x5b, 0x61, 0x63, 0x65, 0x66, 0x6e, 0x72, 0x74, 0x75, 0x55, 0x78, 0x5d, 0};


//
//  White space characters that may appear within a pattern in free-form mode
//
static const UChar gRuleWhiteSpacePattern[] = {
    /* "[[:Cf:][:WSpace:]]" */
    91, 91, 58, 67, 102, 58, 93, 91, 58, 87,
        83, 112, 97, 99, 101, 58, 93, 93, 0 };


//
//  Unicode Set Definitions for Regular Expression  \w
//
static const UChar gIsWordPattern[] = {
//    [     \     p     {    A     l     p     h     a     b     e     t     i      c    }
    0x5b, 0x5c, 0x70, 0x7b, 0x61, 0x6c, 0x70, 0x68, 0x61, 0x62, 0x65, 0x74, 0x69, 0x63, 0x7d,
//          \     p     {    M     }                               Mark
          0x5c, 0x70, 0x7b, 0x4d, 0x7d, 
//          \     p     {    N     d     }                         Digit_Numeric
          0x5c, 0x70, 0x7b, 0x4e, 0x64, 0x7d,
//          \     p     {    P     c     }      ]                  Connector_Punctuation
          0x5c, 0x70, 0x7b, 0x50, 0x63, 0x7d, 0x5d, 0};


//
//  Unicode Set Definitions for Regular Expression  \s
//
    static const UChar gIsSpacePattern[] = {
//        [     \     p     {     W     h     i     t     e     S     p     a     c     e     }     ]
        0x5b, 0x5c, 0x70, 0x7b, 0x57, 0x68, 0x69, 0x74, 0x65, 0x53, 0x70, 0x61, 0x63, 0x65, 0x7d, 0x5d, 0};


//
//  UnicodeSets used in implementation of Grapheme Cluster detection, \X
//
    static const UChar gGC_ControlPattern[] = {
//    [     [     :     Z     l     :     ]     [     :     Z     p     :     ]    
    0x5b, 0x5b, 0x3a, 0x5A, 0x6c, 0x3a, 0x5d, 0x5b, 0x3a, 0x5A, 0x70, 0x3a, 0x5d, 
//    [     :     C     c     :     ]     [     :     C     f     :     ]     -
    0x5b, 0x3a, 0x43, 0x63, 0x3a, 0x5d, 0x5b, 0x3a, 0x43, 0x66, 0x3a, 0x5d, 0x2d,
//    [     :     G     r     a     p     h     e     m     e     _
    0x5b, 0x3a, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
//    E     x     t     e     n     d     :     ]     ]
    0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x3a, 0x5d, 0x5d, 0};

    static const UChar gGC_ExtendPattern[] = {
//    [     \     p     {     G     r     a     p     h     e     m     e     _
    0x5b, 0x5c, 0x70, 0x7b, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
//    E     x     t     e     n     d     }     ]
    0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x7d, 0x5d, 0};

    static const UChar gGC_LPattern[] = {
//    [     \     p     {     H     a     n     g     u     l     _     S     y     l    
    0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
//    l     a     b     l     e     _     T     y     p     e     =     L     }     ]
    0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x7d,  0x5d, 0}; 

    static const UChar gGC_VPattern[] = {
//    [     \     p     {     H     a     n     g     u     l     _     S     y     l    
    0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
//    l     a     b     l     e     _     T     y     p     e     =     V     }     ]
    0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x56, 0x7d,  0x5d, 0}; 

    static const UChar gGC_TPattern[] = {
//    [     \     p     {     H     a     n     g     u     l     _     S     y     l    
    0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
//    l     a     b     l     e     _     T     y     p     e     =     T     }    ]
    0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x54, 0x7d, 0x5d, 0}; 

    static const UChar gGC_LVPattern[] = {
//    [     \     p     {     H     a     n     g     u     l     _     S     y     l    
    0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
//    l     a     b     l     e     _     T     y     p     e     =     L     V     }     ]
    0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x7d, 0x5d, 0}; 

    static const UChar gGC_LVTPattern[] = {
//    [     \     p     {     H     a     n     g     u     l     _     S     y     l    
    0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
//    l     a     b     l     e     _     T     y     p     e     =     L     V     T     }     ]
    0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x54, 0x7d, 0x5d, 0}; 

RegexStaticSets *RegexStaticSets::gStaticSets = NULL;

RegexStaticSets::RegexStaticSets(UErrorCode *status) {
    // First zero out everything  
    int i;
    for (i=0; i<URX_LAST_SET; i++) {
        fPropSets[i] = NULL;
    }
    for (i=0; i<10; i++) {
        fRuleSets[i] = NULL;
    }
    fUnescapeCharSet = NULL;
    fRuleDigits      = NULL;
    fEmptyString     = NULL;

    // Then init the sets to their correct values.
    fPropSets[URX_ISWORD_SET]  = new UnicodeSet(gIsWordPattern,     *status);
    fPropSets[URX_ISSPACE_SET] = new UnicodeSet(gIsSpacePattern,    *status);    
    fPropSets[URX_GC_EXTEND]   = new UnicodeSet(gGC_ExtendPattern,  *status);
    fPropSets[URX_GC_CONTROL]  = new UnicodeSet(gGC_ControlPattern, *status);
    fPropSets[URX_GC_L]        = new UnicodeSet(gGC_LPattern,       *status);
    fPropSets[URX_GC_V]        = new UnicodeSet(gGC_VPattern,       *status);
    fPropSets[URX_GC_T]        = new UnicodeSet(gGC_TPattern,       *status);
    fPropSets[URX_GC_LV]       = new UnicodeSet(gGC_LVPattern,      *status);
    fPropSets[URX_GC_LVT]      = new UnicodeSet(gGC_LVTPattern,     *status);
    if (U_FAILURE(*status)) {
        // Bail out if we were unable to create the above sets.
        // The rest of the initialization needs them, so we cannot proceed.
        return;
    }
    
    
    //
    // The following sets  are dynamically constructed, because their
    //   intialization strings would be unreasonable.
    //
    
    
    //
    //  "Normal" is the set of characters that don't need special handling
    //            when finding grapheme cluster boundaries.
    //
    fPropSets[URX_GC_NORMAL] = new UnicodeSet;
    fPropSets[URX_GC_NORMAL]->complement();
    fPropSets[URX_GC_NORMAL]->remove(0xac00, 0xd7a4);
    fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_CONTROL]);
    fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_L]);
    fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_V]);
    fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_T]);
    
    // Initialize the 8-bit fast bit sets from the parallel full
    //   UnicodeSets.
    for (i=0; i<URX_LAST_SET; i++) {
        fPropSets8[i].init(fPropSets[i]);
    }

    // Sets used while parsing rules, but not referenced from the parse state table
    fRuleSets[kRuleSet_rule_char-128]   = new UnicodeSet(gRuleSet_rule_char_pattern,  *status);
    fRuleSets[kRuleSet_white_space-128] = new UnicodeSet(gRuleWhiteSpacePattern,      *status);
    fRuleSets[kRuleSet_digit_char-128]  = new UnicodeSet(gRuleSet_digit_char_pattern, *status);
    fRuleDigits                         = new UnicodeSet(gRuleSet_digit_char_pattern, *status);
    fUnescapeCharSet                    = new UnicodeSet(gUnescapeCharPattern,        *status);

    // Empty UnicodeString, for use by matchers with NULL input.
    fEmptyString = new UnicodeString;
}


RegexStaticSets::~RegexStaticSets() {
    int i;

    for (i=0; i<URX_LAST_SET; i++) {
        delete fPropSets[i];
        fPropSets[i] = NULL;
    }
    for (i=0; i<10; i++) {
        delete fRuleSets[i];
        fRuleSets[i] = NULL;
    }
    delete fUnescapeCharSet;
    fUnescapeCharSet = NULL;
    delete fRuleDigits;
    fRuleDigits = NULL;
    delete fEmptyString;
    fEmptyString = NULL;
}


//----------------------------------------------------------------------------------
//
//   regex_cleanup      Memory cleanup function, free/delete all
//                      cached memory.  Called by ICU's u_cleanup() function.
//
//----------------------------------------------------------------------------------
UBool
RegexStaticSets::cleanup(void) {
    delete RegexStaticSets::gStaticSets;
    RegexStaticSets::gStaticSets = NULL;
    return TRUE;
}

U_CDECL_BEGIN
static UBool U_CALLCONV
regex_cleanup(void) {
    return RegexStaticSets::cleanup();
}
U_CDECL_END

void RegexStaticSets::initGlobals(UErrorCode *status) {
    umtx_lock(NULL);
    RegexStaticSets *p = gStaticSets;
    umtx_unlock(NULL);
    if (p == NULL) {
        p = new RegexStaticSets(status);
        if (U_FAILURE(*status)) {
            delete p;
            return;
        }
        umtx_lock(NULL);
        if (gStaticSets == NULL) {
            gStaticSets = p;
            p = NULL;
        }
        umtx_unlock(NULL);
        if (p) {
            delete p;
        }
        ucln_i18n_registerCleanup(UCLN_I18N_REGEX, regex_cleanup);
    }
}
    

U_NAMESPACE_END
#endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS
Commit	Line	Data
b75a7d8f A	1	//
	2	// regexst.h
	3	//
374ca955	4	// Copyright (C) 2004, International Business Machines Corporation and others.
b75a7d8f A	5	// All Rights Reserved.
	6	//
	7	// This file contains class RegexStaticSets
	8	//
	9	// This class is internal to the regular expression implementation.
	10	// For the public Regular Expression API, see the file "unicode/regex.h"
	11	//
	12	// RegexStaticSets groups together the common UnicodeSets that are needed
	13	// for compiling or executing RegularExpressions. This grouping simplifies
	14	// the thread safe lazy creation and sharing of these sets across
	15	// all instances of regular expressions.
	16	//
	17	#include "unicode/utypes.h"
	18
	19	#if !UCONFIG_NO_REGULAR_EXPRESSIONS
	20
	21	#include "unicode/unistr.h"
	22	#include "unicode/uniset.h"
	23	#include "unicode/uchar.h"
	24	#include "unicode/regex.h"
	25	#include "uprops.h"
	26	#include "cmemory.h"
	27	#include "cstring.h"
	28	#include "uassert.h"
	29	#include "ucln_in.h"
	30	#include "umutex.h"
	31
	32	#include "regexcst.h" // Contains state table for the regex pattern parser.
	33	// generated by a Perl script.
	34	#include "regexst.h"
	35
	36
	37
	38	U_NAMESPACE_BEGIN
	39
	40
	41	//----------------------------------------------------------------------------------------
	42	//
	43	// Unicode Set pattern strings for all of the required constant sets.
	44	// Initialized with hex values for portability to EBCDIC based machines.
	45	// Really ugly, but there's no good way to avoid it.
	46	//
	47	//----------------------------------------------------------------------------------------
	48
	49	// "Rule Char" Characters are those with no special meaning, and therefore do not
	50	// need to be escaped to appear as literals in a regexp. Expressed
	51	// as the inverse of those needing escaping -- [^\*\?\+\[\(\)\{\}\^\$\\|\\\.]
	52	static const UChar gRuleSet_rule_char_pattern[] = {
	53	// [ ^ \ * \ ? \ + \ [ \ ( / )
	54	0x5b, 0x5e, 0x5c, 0x2a, 0x5c, 0x3f, 0x5c, 0x2b, 0x5c, 0x5b, 0x5c, 0x28, 0x5c, 0x29,
	55	// \ { \ } \ ^ \ $ \ \| \ \ \ . ]
	56	0x5c, 0x7b,0x5c, 0x7d, 0x5c, 0x5e, 0x5c, 0x24, 0x5c, 0x7c, 0x5c, 0x5c, 0x5c, 0x2e, 0x5d, 0};
	57
	58
	59	static const UChar gRuleSet_digit_char_pattern[] = {
	60	// [ 0 - 9 ]
	61	0x5b, 0x30, 0x2d, 0x39, 0x5d, 0};
	62	//static const UnicodeSet *gRuleDigits = NULL;
	63
	64
	65
	66	//
	67	// Here are the backslash escape characters that ICU's unescape() function
	68	// will handle.
69	//
70	static const UChar gUnescapeCharPattern[] = {
71	// [ a c e f n r t u U x ]
72	0x5b, 0x61, 0x63, 0x65, 0x66, 0x6e, 0x72, 0x74, 0x75, 0x55, 0x78, 0x5d, 0};
73
74
75	//
76	// White space characters that may appear within a pattern in free-form mode
77	//
78	static const UChar gRuleWhiteSpacePattern[] = {
79	/* "[[:Cf:][:WSpace:]]" */
80	91, 91, 58, 67, 102, 58, 93, 91, 58, 87,
81	83, 112, 97, 99, 101, 58, 93, 93, 0 };
82
83
84
85	//
86	// Unicode Set Definitions for Regular Expression \w
87	//
88	static const UChar gIsWordPattern[] = {
374ca955 A	89	// [ \ p { A l p h a b e t i c }
	90	0x5b, 0x5c, 0x70, 0x7b, 0x61, 0x6c, 0x70, 0x68, 0x61, 0x62, 0x65, 0x74, 0x69, 0x63, 0x7d,
	91	// \ p { M } Mark
	92	0x5c, 0x70, 0x7b, 0x4d, 0x7d,
	93	// \ p { N d } Digit_Numeric
	94	0x5c, 0x70, 0x7b, 0x4e, 0x64, 0x7d,
	95	// \ p { P c } ] Connector_Punctuation
	96	0x5c, 0x70, 0x7b, 0x50, 0x63, 0x7d, 0x5d, 0};
b75a7d8f A	97
	98
	99	//
	100	// Unicode Set Definitions for Regular Expression \s
	101	//
	102	static const UChar gIsSpacePattern[] = {
374ca955 A	103	// [ \ p { W h i t e S p a c e } ]
374ca955 A	104	0x5b, 0x5c, 0x70, 0x7b, 0x57, 0x68, 0x69, 0x74, 0x65, 0x53, 0x70, 0x61, 0x63, 0x65, 0x7d, 0x5d, 0};
b75a7d8f A	105
	106
	107	//
	108	// UnicodeSets used in implementation of Grapheme Cluster detection, \X
	109	//
	110	static const UChar gGC_ControlPattern[] = {
	111	// [ [ : Z l : ] [ : Z p : ]
	112	0x5b, 0x5b, 0x3a, 0x5A, 0x6c, 0x3a, 0x5d, 0x5b, 0x3a, 0x5A, 0x70, 0x3a, 0x5d,
374ca955 A	113	// [ : C c : ] [ : C f : ] -
	114	0x5b, 0x3a, 0x43, 0x63, 0x3a, 0x5d, 0x5b, 0x3a, 0x43, 0x66, 0x3a, 0x5d, 0x2d,
	115	// [ : G r a p h e m e _
	116	0x5b, 0x3a, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
	117	// E x t e n d : ] ]
	118	0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x3a, 0x5d, 0x5d, 0};
b75a7d8f A	119
	120	static const UChar gGC_ExtendPattern[] = {
	121	// [ \ p { G r a p h e m e _
	122	0x5b, 0x5c, 0x70, 0x7b, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
	123	// E x t e n d } ]
	124	0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x7d, 0x5d, 0};
	125
	126	static const UChar gGC_LPattern[] = {
	127	// [ \ p { H a n g u l _ S y l
	128	0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
	129	// l a b l e _ T y p e = L } ]
	130	0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x7d, 0x5d, 0};
	131
	132	static const UChar gGC_VPattern[] = {
	133	// [ \ p { H a n g u l _ S y l
	134	0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
	135	// l a b l e _ T y p e = V } ]
	136	0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x56, 0x7d, 0x5d, 0};
	137
	138	static const UChar gGC_TPattern[] = {
	139	// [ \ p { H a n g u l _ S y l
	140	0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
	141	// l a b l e _ T y p e = T } ]
	142	0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x54, 0x7d, 0x5d, 0};
	143
	144	static const UChar gGC_LVPattern[] = {
	145	// [ \ p { H a n g u l _ S y l
	146	0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
	147	// l a b l e _ T y p e = L V } ]
	148	0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x7d, 0x5d, 0};
	149
	150	static const UChar gGC_LVTPattern[] = {
	151	// [ \ p { H a n g u l _ S y l
	152	0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
	153	// l a b l e _ T y p e = L V T } ]
	154	0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x54, 0x7d, 0x5d, 0};
	155
	156	RegexStaticSets *RegexStaticSets::gStaticSets = NULL;
	157
	158	RegexStaticSets::RegexStaticSets(UErrorCode *status) {
	159	// First zero out everything
	160	int i;
	161	for (i=0; i<URX_LAST_SET; i++) {
	162	fPropSets[i] = NULL;
	163	}
	164	for (i=0; i<10; i++) {
	165	fRuleSets[i] = NULL;
	166	}
	167	fUnescapeCharSet = NULL;
	168	fRuleDigits = NULL;
	169	fEmptyString = NULL;
	170
	171	// Then init the sets to their correct values.
	172	fPropSets[URX_ISWORD_SET] = new UnicodeSet(gIsWordPattern, *status);
	173	fPropSets[URX_ISSPACE_SET] = new UnicodeSet(gIsSpacePattern, *status);
	174	fPropSets[URX_GC_EXTEND] = new UnicodeSet(gGC_ExtendPattern, *status);
	175	fPropSets[URX_GC_CONTROL] = new UnicodeSet(gGC_ControlPattern, *status);
	176	fPropSets[URX_GC_L] = new UnicodeSet(gGC_LPattern, *status);
	177	fPropSets[URX_GC_V] = new UnicodeSet(gGC_VPattern, *status);
	178	fPropSets[URX_GC_T] = new UnicodeSet(gGC_TPattern, *status);
	179	fPropSets[URX_GC_LV] = new UnicodeSet(gGC_LVPattern, *status);
	180	fPropSets[URX_GC_LVT] = new UnicodeSet(gGC_LVTPattern, *status);
	181	if (U_FAILURE(*status)) {
	182	// Bail out if we were unable to create the above sets.
183	// The rest of the initialization needs them, so we cannot proceed.
184	return;
185	}
186
187
188	//
189	// The following sets are dynamically constructed, because their
190	// intialization strings would be unreasonable.
191	//
192
193
194	//
195	// "Normal" is the set of characters that don't need special handling
196	// when finding grapheme cluster boundaries.
197	//
198	fPropSets[URX_GC_NORMAL] = new UnicodeSet;
199	fPropSets[URX_GC_NORMAL]->complement();
200	fPropSets[URX_GC_NORMAL]->remove(0xac00, 0xd7a4);
201	fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_CONTROL]);
202	fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_L]);
203	fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_V]);
204	fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_T]);
205
206	// Initialize the 8-bit fast bit sets from the parallel full
207	// UnicodeSets.
208	for (i=0; i<URX_LAST_SET; i++) {
209	fPropSets8[i].init(fPropSets[i]);
210	}
211
212	// Sets used while parsing rules, but not referenced from the parse state table
213	fRuleSets[kRuleSet_rule_char-128] = new UnicodeSet(gRuleSet_rule_char_pattern, *status);
214	fRuleSets[kRuleSet_white_space-128] = new UnicodeSet(gRuleWhiteSpacePattern, *status);
215	fRuleSets[kRuleSet_digit_char-128] = new UnicodeSet(gRuleSet_digit_char_pattern, *status);
216	fRuleDigits = new UnicodeSet(gRuleSet_digit_char_pattern, *status);
217	fUnescapeCharSet = new UnicodeSet(gUnescapeCharPattern, *status);
218
219	// Empty UnicodeString, for use by matchers with NULL input.
220	fEmptyString = new UnicodeString;
374ca955	221	}
b75a7d8f A	222
	223
	224	RegexStaticSets::~RegexStaticSets() {
	225	int i;
	226
	227	for (i=0; i<URX_LAST_SET; i++) {
	228	delete fPropSets[i];
	229	fPropSets[i] = NULL;
	230	}
	231	for (i=0; i<10; i++) {
	232	delete fRuleSets[i];
	233	fRuleSets[i] = NULL;
	234	}
	235	delete fUnescapeCharSet;
	236	fUnescapeCharSet = NULL;
	237	delete fRuleDigits;
	238	fRuleDigits = NULL;
	239	delete fEmptyString;
	240	fEmptyString = NULL;
374ca955 A	241	}
	242
	243
	244	//----------------------------------------------------------------------------------
	245	//
	246	// regex_cleanup Memory cleanup function, free/delete all
	247	// cached memory. Called by ICU's u_cleanup() function.
	248	//
	249	//----------------------------------------------------------------------------------
	250	UBool
	251	RegexStaticSets::cleanup(void) {
	252	delete RegexStaticSets::gStaticSets;
	253	RegexStaticSets::gStaticSets = NULL;
	254	return TRUE;
	255	}
b75a7d8f	256
374ca955 A	257	U_CDECL_BEGIN
	258	static UBool U_CALLCONV
	259	regex_cleanup(void) {
	260	return RegexStaticSets::cleanup();
	261	}
	262	U_CDECL_END
b75a7d8f A	263
	264	void RegexStaticSets::initGlobals(UErrorCode *status) {
	265	umtx_lock(NULL);
	266	RegexStaticSets *p = gStaticSets;
	267	umtx_unlock(NULL);
	268	if (p == NULL) {
	269	p = new RegexStaticSets(status);
	270	if (U_FAILURE(*status)) {
	271	delete p;
	272	return;
	273	}
	274	umtx_lock(NULL);
	275	if (gStaticSets == NULL) {
	276	gStaticSets = p;
	277	p = NULL;
	278	}
	279	umtx_unlock(NULL);
	280	if (p) {
	281	delete p;
	282	}
374ca955	283	ucln_i18n_registerCleanup(UCLN_I18N_REGEX, regex_cleanup);
b75a7d8f A	284	}
	285	}
	286
	287
	288	U_NAMESPACE_END
	289	#endif // !UCONFIG_NO_REGULAR_EXPRESSIONS