scuffed-code/icu4c/source/i18n/regexst.cpp
Andy Heninger 54a60fe6f4 ICU-11548 Improve regex static UnicodeSets handling
Compiled regular expression patterns make use of several shared common
UnicodeSets. This change simplifies the creation and use of these
static UnicodeSets.

- Pointer fields to the static sets are removed from the compiled patterns,
  and the static variables are accessed directly. The deleted pointers
  were a hold-over from earlier code that did not use shared statics.

- The UnicodeSet pattern literals are changed from hex constants to
  u"string literals".

- The size of fRuleSets (from regexst.h) is changed from a hard-coded 10
  to the number of UnicodeSets actually required. Doing this required
  a change to regexcst.pl to export the required size. Changing and
  rerunning this perl code resulted in massive but benign changes to
  the generated file regexcst.h, the result of perl having changed its
  order of enumeration of hashes since the file was last regenerated.

- UnicodeSets are frozen when possible. Should result in faster matching.
2020-01-30 15:13:07 -08:00

173 lines
6.6 KiB
C++

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
//
// regexst.h
//
// Copyright (C) 2004-2015, 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
// "Rule Char" Characters are those with special meaning, and therefore
// need to be escaped to appear as literals in a regexp.
constexpr char16_t const *gRuleSet_rule_chars = u"*?+[(){}^$|\\.";
//
// The backslash escape characters that ICU's unescape() function will handle.
//
constexpr char16_t const *gUnescapeChars = u"acefnrtuUx";
//
// Unicode Set pattern for Regular Expression \w
//
constexpr char16_t const *gIsWordPattern = u"[\\p{Alphabetic}\\p{M}\\p{Nd}\\p{Pc}\\u200c\\u200d]";
//
// Unicode Set Definitions for Regular Expression \s
//
constexpr char16_t const *gIsSpacePattern = u"[\\p{WhiteSpace}]";
//
// UnicodeSets used in implementation of Grapheme Cluster detection, \X
//
constexpr char16_t const *gGC_ControlPattern = u"[[:Zl:][:Zp:][:Cc:][:Cf:]-[:Grapheme_Extend:]]";
constexpr char16_t const *gGC_ExtendPattern = u"[\\p{Grapheme_Extend}]";
constexpr char16_t const *gGC_LPattern = u"[\\p{Hangul_Syllable_Type=L}]";
constexpr char16_t const *gGC_VPattern = u"[\\p{Hangul_Syllable_Type=V}]";
constexpr char16_t const *gGC_TPattern = u"[\\p{Hangul_Syllable_Type=T}]";
constexpr char16_t const *gGC_LVPattern = u"[\\p{Hangul_Syllable_Type=LV}]";
constexpr char16_t const *gGC_LVTPattern = u"[\\p{Hangul_Syllable_Type=LVT}]";
RegexStaticSets *RegexStaticSets::gStaticSets = nullptr;
UInitOnce gStaticSetsInitOnce = U_INITONCE_INITIALIZER;
RegexStaticSets::RegexStaticSets(UErrorCode *status) {
// Initialize the shared static sets to their correct values.
fUnescapeCharSet.addAll(UnicodeString(true, gUnescapeChars, -1)).freeze();
fPropSets[URX_ISWORD_SET].applyPattern(UnicodeString(true, gIsWordPattern, -1), *status).freeze();
fPropSets[URX_ISSPACE_SET].applyPattern(UnicodeString(true, gIsSpacePattern, -1), *status).freeze();
fPropSets[URX_GC_EXTEND].applyPattern(UnicodeString(TRUE, gGC_ExtendPattern, -1), *status).freeze();
fPropSets[URX_GC_CONTROL].applyPattern(UnicodeString(TRUE, gGC_ControlPattern, -1), *status).freeze();
fPropSets[URX_GC_L].applyPattern(UnicodeString(TRUE, gGC_LPattern, -1), *status).freeze();
fPropSets[URX_GC_V].applyPattern(UnicodeString(TRUE, gGC_VPattern, -1), *status).freeze();
fPropSets[URX_GC_T].applyPattern(UnicodeString(TRUE, gGC_TPattern, -1), *status).freeze();
fPropSets[URX_GC_LV].applyPattern(UnicodeString(TRUE, gGC_LVPattern, -1), *status).freeze();
fPropSets[URX_GC_LVT].applyPattern(UnicodeString(TRUE, gGC_LVTPattern, -1), *status).freeze();
//
// "Normal" is the set of characters that don't need special handling
// when finding grapheme cluster boundaries.
//
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]);
fPropSets[URX_GC_NORMAL].freeze();
// Initialize the 8-bit fast bit sets from the parallel full
// UnicodeSets.
//
// TODO: 25 Oct 2019 are these fast 8-bit sets worth keeping?
// Measured 3.5% gain on (non) matching with the pattern "x(?:\\S+)+x"
// This runs in exponential time, making it easy to adjust the time for
// convenient measuring.
//
// This 8 bit optimization dates from the early days of ICU,
// with a less optimized UnicodeSet. At the time, the difference
// was substantial.
for (int32_t 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]
.addAll(UnicodeString(gRuleSet_rule_chars)).complement().freeze();
fRuleSets[kRuleSet_digit_char-128].add(u'0', u'9').freeze();
fRuleSets[kRuleSet_ascii_letter-128].add(u'A', u'Z').add(u'a', u'z').freeze();
fRuleDigitsAlias = &fRuleSets[kRuleSet_digit_char-128];
// Finally, initialize an empty UText string for utility purposes
fEmptyText = utext_openUChars(nullptr, nullptr, 0, status);
}
RegexStaticSets::~RegexStaticSets() {
fRuleDigitsAlias = nullptr;
utext_close(fEmptyText);
}
//------------------------------------------------------------------------------
//
// regex_cleanup Memory cleanup function, free/delete all
// cached memory. Called by ICU's u_cleanup() function.
//
//------------------------------------------------------------------------------
U_CDECL_BEGIN
static UBool U_CALLCONV
regex_cleanup(void) {
delete RegexStaticSets::gStaticSets;
RegexStaticSets::gStaticSets = nullptr;
gStaticSetsInitOnce.reset();
return TRUE;
}
static void U_CALLCONV initStaticSets(UErrorCode &status) {
U_ASSERT(RegexStaticSets::gStaticSets == nullptr);
ucln_i18n_registerCleanup(UCLN_I18N_REGEX, regex_cleanup);
RegexStaticSets::gStaticSets = new RegexStaticSets(&status);
if (U_FAILURE(status)) {
delete RegexStaticSets::gStaticSets;
RegexStaticSets::gStaticSets = nullptr;
}
if (RegexStaticSets::gStaticSets == nullptr && U_SUCCESS(status)) {
status = U_MEMORY_ALLOCATION_ERROR;
}
}
U_CDECL_END
void RegexStaticSets::initGlobals(UErrorCode *status) {
umtx_initOnce(gStaticSetsInitOnce, &initStaticSets, *status);
}
U_NAMESPACE_END
#endif // !UCONFIG_NO_REGULAR_EXPRESSIONS