michael@0: //
michael@0: //  regexst.h
michael@0: //
michael@0: //  Copyright (C) 2004-2013, International Business Machines Corporation and others.
michael@0: //  All Rights Reserved.
michael@0: //
michael@0: //  This file contains class RegexStaticSets
michael@0: //
michael@0: //  This class is internal to the regular expression implementation.
michael@0: //  For the public Regular Expression API, see the file "unicode/regex.h"
michael@0: //
michael@0: //  RegexStaticSets groups together the common UnicodeSets that are needed
michael@0: //   for compiling or executing RegularExpressions.  This grouping simplifies
michael@0: //   the thread safe lazy creation and sharing of these sets across
michael@0: //   all instances of regular expressions.
michael@0: //
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: #if !UCONFIG_NO_REGULAR_EXPRESSIONS
michael@0: 
michael@0: #include "unicode/unistr.h"
michael@0: #include "unicode/uniset.h"
michael@0: #include "unicode/uchar.h"
michael@0: #include "unicode/regex.h"
michael@0: #include "uprops.h"
michael@0: #include "cmemory.h"
michael@0: #include "cstring.h"
michael@0: #include "uassert.h"
michael@0: #include "ucln_in.h"
michael@0: #include "umutex.h"
michael@0: 
michael@0: #include "regexcst.h"   // Contains state table for the regex pattern parser.
michael@0:                         //   generated by a Perl script.
michael@0: #include "regexst.h"
michael@0: 
michael@0: 
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: 
michael@0: //------------------------------------------------------------------------------
michael@0: //
michael@0: // Unicode Set pattern strings for all of the required constant sets.
michael@0: //               Initialized with hex values for portability to EBCDIC based machines.
michael@0: //                Really ugly, but there's no good way to avoid it.
michael@0: //
michael@0: //------------------------------------------------------------------------------
michael@0: 
michael@0: // "Rule Char" Characters are those with no special meaning, and therefore do not
michael@0: //    need to be escaped to appear as literals in a regexp.  Expressed
michael@0: //    as the inverse of those needing escaping --  [^\*\?\+\[\(\)\{\}\^\$\|\\\.]
michael@0: static const UChar gRuleSet_rule_char_pattern[]       = {
michael@0:  //   [    ^      \     *     \     ?     \     +     \     [     \     (     /     )
michael@0:     0x5b, 0x5e, 0x5c, 0x2a, 0x5c, 0x3f, 0x5c, 0x2b, 0x5c, 0x5b, 0x5c, 0x28, 0x5c, 0x29,
michael@0:  //   \     {    \     }     \     ^     \     $     \     |     \     \     \     .     ]
michael@0:     0x5c, 0x7b,0x5c, 0x7d, 0x5c, 0x5e, 0x5c, 0x24, 0x5c, 0x7c, 0x5c, 0x5c, 0x5c, 0x2e, 0x5d, 0};
michael@0: 
michael@0: 
michael@0: static const UChar gRuleSet_digit_char_pattern[] = {
michael@0: //    [    0      -    9     ]
michael@0:     0x5b, 0x30, 0x2d, 0x39, 0x5d, 0};
michael@0: 
michael@0: //
michael@0: //   Here are the backslash escape characters that ICU's unescape() function
michael@0: //    will handle.
michael@0: //
michael@0: static const UChar gUnescapeCharPattern[] = {
michael@0: //    [     a     c     e     f     n     r     t     u     U     x    ]
michael@0:     0x5b, 0x61, 0x63, 0x65, 0x66, 0x6e, 0x72, 0x74, 0x75, 0x55, 0x78, 0x5d, 0};
michael@0: 
michael@0: 
michael@0: //
michael@0: //  Unicode Set Definitions for Regular Expression  \w
michael@0: //
michael@0: static const UChar gIsWordPattern[] = {
michael@0: //    [     \     p     {    A     l     p     h     a     b     e     t     i      c    }
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x61, 0x6c, 0x70, 0x68, 0x61, 0x62, 0x65, 0x74, 0x69, 0x63, 0x7d,
michael@0: //          \     p     {    M     }                               Mark
michael@0:           0x5c, 0x70, 0x7b, 0x4d, 0x7d,
michael@0: //          \     p     {    N     d     }                         Digit_Numeric
michael@0:           0x5c, 0x70, 0x7b, 0x4e, 0x64, 0x7d,
michael@0: //          \     p     {    P     c     }                         Connector_Punctuation
michael@0:           0x5c, 0x70, 0x7b, 0x50, 0x63, 0x7d,
michael@0: //          \     u     2    0     0     c      \     u     2    0     0     d     ]
michael@0:           0x5c, 0x75, 0x32, 0x30, 0x30, 0x63, 0x5c, 0x75, 0x32, 0x30, 0x30, 0x64, 0x5d, 0};
michael@0: 
michael@0: 
michael@0: //
michael@0: //  Unicode Set Definitions for Regular Expression  \s
michael@0: //
michael@0: static const UChar gIsSpacePattern[] = {
michael@0: //        [     \     p     {     W     h     i     t     e     S     p     a     c     e     }     ]
michael@0:         0x5b, 0x5c, 0x70, 0x7b, 0x57, 0x68, 0x69, 0x74, 0x65, 0x53, 0x70, 0x61, 0x63, 0x65, 0x7d, 0x5d, 0};
michael@0: 
michael@0: 
michael@0: //
michael@0: //  UnicodeSets used in implementation of Grapheme Cluster detection, \X
michael@0: //
michael@0: static const UChar gGC_ControlPattern[] = {
michael@0: //    [     [     :     Z     l     :     ]     [     :     Z     p     :     ]
michael@0:     0x5b, 0x5b, 0x3a, 0x5A, 0x6c, 0x3a, 0x5d, 0x5b, 0x3a, 0x5A, 0x70, 0x3a, 0x5d,
michael@0: //    [     :     C     c     :     ]     [     :     C     f     :     ]     -
michael@0:     0x5b, 0x3a, 0x43, 0x63, 0x3a, 0x5d, 0x5b, 0x3a, 0x43, 0x66, 0x3a, 0x5d, 0x2d,
michael@0: //    [     :     G     r     a     p     h     e     m     e     _
michael@0:     0x5b, 0x3a, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
michael@0: //    E     x     t     e     n     d     :     ]     ]
michael@0:     0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x3a, 0x5d, 0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_ExtendPattern[] = {
michael@0: //    [     \     p     {     G     r     a     p     h     e     m     e     _
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x47, 0x72, 0x61, 0x70, 0x68, 0x65, 0x6d, 0x65, 0x5f,
michael@0: //    E     x     t     e     n     d     }     ]
michael@0:     0x45, 0x78, 0x74, 0x65, 0x6e, 0x64, 0x7d, 0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_LPattern[] = {
michael@0: //    [     \     p     {     H     a     n     g     u     l     _     S     y     l
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
michael@0: //    l     a     b     l     e     _     T     y     p     e     =     L     }     ]
michael@0:     0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x7d,  0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_VPattern[] = {
michael@0: //    [     \     p     {     H     a     n     g     u     l     _     S     y     l
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
michael@0: //    l     a     b     l     e     _     T     y     p     e     =     V     }     ]
michael@0:     0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x56, 0x7d,  0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_TPattern[] = {
michael@0: //    [     \     p     {     H     a     n     g     u     l     _     S     y     l
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
michael@0: //    l     a     b     l     e     _     T     y     p     e     =     T     }    ]
michael@0:     0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x54, 0x7d, 0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_LVPattern[] = {
michael@0: //    [     \     p     {     H     a     n     g     u     l     _     S     y     l
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
michael@0: //    l     a     b     l     e     _     T     y     p     e     =     L     V     }     ]
michael@0:     0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x7d, 0x5d, 0};
michael@0: 
michael@0: static const UChar gGC_LVTPattern[] = {
michael@0: //    [     \     p     {     H     a     n     g     u     l     _     S     y     l
michael@0:     0x5b, 0x5c, 0x70, 0x7b, 0x48, 0x61, 0x6e, 0x67, 0x75, 0x6c, 0x5f, 0x53, 0x79, 0x6c,
michael@0: //    l     a     b     l     e     _     T     y     p     e     =     L     V     T     }     ]
michael@0:     0x6c, 0x61, 0x62, 0x6c, 0x65, 0x5f, 0x54, 0x79, 0x70, 0x65, 0x3d, 0x4c, 0x56, 0x54, 0x7d, 0x5d, 0};
michael@0: 
michael@0: 
michael@0: RegexStaticSets *RegexStaticSets::gStaticSets = NULL;
michael@0: UInitOnce gStaticSetsInitOnce = U_INITONCE_INITIALIZER;
michael@0: 
michael@0: RegexStaticSets::RegexStaticSets(UErrorCode *status)
michael@0: :
michael@0: fUnescapeCharSet(UnicodeString(TRUE, gUnescapeCharPattern, -1), *status),
michael@0: fRuleDigitsAlias(NULL),
michael@0: fEmptyText(NULL)
michael@0: {
michael@0:     // First zero out everything
michael@0:     int i;
michael@0:     for (i=0; i<URX_LAST_SET; i++) {
michael@0:         fPropSets[i] = NULL;
michael@0:     }
michael@0:     // Then init the sets to their correct values.
michael@0:     fPropSets[URX_ISWORD_SET]  = new UnicodeSet(UnicodeString(TRUE, gIsWordPattern, -1),     *status);
michael@0:     fPropSets[URX_ISSPACE_SET] = new UnicodeSet(UnicodeString(TRUE, gIsSpacePattern, -1),    *status);
michael@0:     fPropSets[URX_GC_EXTEND]   = new UnicodeSet(UnicodeString(TRUE, gGC_ExtendPattern, -1),  *status);
michael@0:     fPropSets[URX_GC_CONTROL]  = new UnicodeSet(UnicodeString(TRUE, gGC_ControlPattern, -1), *status);
michael@0:     fPropSets[URX_GC_L]        = new UnicodeSet(UnicodeString(TRUE, gGC_LPattern, -1),       *status);
michael@0:     fPropSets[URX_GC_V]        = new UnicodeSet(UnicodeString(TRUE, gGC_VPattern, -1),       *status);
michael@0:     fPropSets[URX_GC_T]        = new UnicodeSet(UnicodeString(TRUE, gGC_TPattern, -1),       *status);
michael@0:     fPropSets[URX_GC_LV]       = new UnicodeSet(UnicodeString(TRUE, gGC_LVPattern, -1),      *status);
michael@0:     fPropSets[URX_GC_LVT]      = new UnicodeSet(UnicodeString(TRUE, gGC_LVTPattern, -1),     *status);
michael@0:     
michael@0:     // Check for null pointers
michael@0:     if (fPropSets[URX_ISWORD_SET] == NULL || fPropSets[URX_ISSPACE_SET] == NULL || fPropSets[URX_GC_EXTEND] == NULL || 
michael@0:         fPropSets[URX_GC_CONTROL] == NULL || fPropSets[URX_GC_L] == NULL || fPropSets[URX_GC_V] == NULL || 
michael@0:         fPropSets[URX_GC_T] == NULL || fPropSets[URX_GC_LV] == NULL || fPropSets[URX_GC_LVT] == NULL) {
michael@0:         goto ExitConstrDeleteAll;
michael@0:     }
michael@0:     if (U_FAILURE(*status)) {
michael@0:         // Bail out if we were unable to create the above sets.
michael@0:         // The rest of the initialization needs them, so we cannot proceed.
michael@0:         return;
michael@0:     }
michael@0: 
michael@0: 
michael@0:     //
michael@0:     // The following sets  are dynamically constructed, because their
michael@0:     //   initialization strings would be unreasonable.
michael@0:     //
michael@0: 
michael@0: 
michael@0:     //
michael@0:     //  "Normal" is the set of characters that don't need special handling
michael@0:     //            when finding grapheme cluster boundaries.
michael@0:     //
michael@0:     fPropSets[URX_GC_NORMAL] = new UnicodeSet(0, UnicodeSet::MAX_VALUE);
michael@0:     // Null pointer check
michael@0:     if (fPropSets[URX_GC_NORMAL] == NULL) {
michael@0:     	goto ExitConstrDeleteAll;
michael@0:     }
michael@0:     fPropSets[URX_GC_NORMAL]->remove(0xac00, 0xd7a4);
michael@0:     fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_CONTROL]);
michael@0:     fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_L]);
michael@0:     fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_V]);
michael@0:     fPropSets[URX_GC_NORMAL]->removeAll(*fPropSets[URX_GC_T]);
michael@0: 
michael@0:     // Initialize the 8-bit fast bit sets from the parallel full
michael@0:     //   UnicodeSets.
michael@0:     for (i=0; i<URX_LAST_SET; i++) {
michael@0:         if (fPropSets[i]) {
michael@0:             fPropSets[i]->compact();
michael@0:             fPropSets8[i].init(fPropSets[i]);
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     // Sets used while parsing rules, but not referenced from the parse state table
michael@0:     fRuleSets[kRuleSet_rule_char-128]   = UnicodeSet(UnicodeString(TRUE, gRuleSet_rule_char_pattern, -1),   *status);
michael@0:     fRuleSets[kRuleSet_digit_char-128]  = UnicodeSet(UnicodeString(TRUE, gRuleSet_digit_char_pattern, -1),  *status);
michael@0:     fRuleDigitsAlias = &fRuleSets[kRuleSet_digit_char-128];
michael@0:     for (i=0; i<(int32_t)(sizeof(fRuleSets)/sizeof(fRuleSets[0])); i++) {
michael@0:         fRuleSets[i].compact();
michael@0:     }
michael@0:     
michael@0:     // Finally, initialize an empty string for utility purposes
michael@0:     fEmptyText = utext_openUChars(NULL, NULL, 0, status);
michael@0:     
michael@0:     return; // If we reached this point, everything is fine so just exit
michael@0: 
michael@0: ExitConstrDeleteAll: // Remove fPropSets and fRuleSets and return error
michael@0:     for (i=0; i<URX_LAST_SET; i++) {
michael@0:         delete fPropSets[i];
michael@0:         fPropSets[i] = NULL;
michael@0:     }
michael@0:     *status = U_MEMORY_ALLOCATION_ERROR;
michael@0: }
michael@0: 
michael@0: 
michael@0: RegexStaticSets::~RegexStaticSets() {
michael@0:     int32_t i;
michael@0: 
michael@0:     for (i=0; i<URX_LAST_SET; i++) {
michael@0:         delete fPropSets[i];
michael@0:         fPropSets[i] = NULL;
michael@0:     }
michael@0:     fRuleDigitsAlias = NULL;
michael@0:     
michael@0:     utext_close(fEmptyText);
michael@0: }
michael@0: 
michael@0: 
michael@0: //------------------------------------------------------------------------------
michael@0: //
michael@0: //   regex_cleanup      Memory cleanup function, free/delete all
michael@0: //                      cached memory.  Called by ICU's u_cleanup() function.
michael@0: //
michael@0: //------------------------------------------------------------------------------
michael@0: UBool
michael@0: RegexStaticSets::cleanup(void) {
michael@0:     delete RegexStaticSets::gStaticSets;
michael@0:     RegexStaticSets::gStaticSets = NULL;
michael@0:     gStaticSetsInitOnce.reset();
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static UBool U_CALLCONV
michael@0: regex_cleanup(void) {
michael@0:     return RegexStaticSets::cleanup();
michael@0: }
michael@0: 
michael@0: static void U_CALLCONV initStaticSets(UErrorCode &status) {
michael@0:     U_ASSERT(RegexStaticSets::gStaticSets == NULL);
michael@0:     ucln_i18n_registerCleanup(UCLN_I18N_REGEX, regex_cleanup);
michael@0:     RegexStaticSets::gStaticSets = new RegexStaticSets(&status);
michael@0:     if (U_FAILURE(status)) {
michael@0:         delete RegexStaticSets::gStaticSets;
michael@0:         RegexStaticSets::gStaticSets = NULL;
michael@0:     }
michael@0:     if (RegexStaticSets::gStaticSets == NULL && U_SUCCESS(status)) {
michael@0:         status = U_MEMORY_ALLOCATION_ERROR;
michael@0:     }
michael@0: }
michael@0: U_CDECL_END
michael@0: 
michael@0: void RegexStaticSets::initGlobals(UErrorCode *status) {
michael@0:     umtx_initOnce(gStaticSetsInitOnce, &initStaticSets, *status);
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS