blob: 6ffe61058e71e05ef01c8e888d834e12b9c9814c [file] [log] [blame]
/*
**************************************************************************
* Copyright (C) 2002-2013 International Business Machines Corporation *
* and others. All rights reserved. *
**************************************************************************
*/
//
// file: rematch.cpp
//
// Contains the implementation of class RegexMatcher,
// which is one of the main API classes for the ICU regular expression package.
//
#include "unicode/utypes.h"
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
#include "unicode/regex.h"
#include "unicode/uniset.h"
#include "unicode/uchar.h"
#include "unicode/ustring.h"
#include "unicode/rbbi.h"
#include "unicode/utf.h"
#include "unicode/utf16.h"
#include "uassert.h"
#include "cmemory.h"
#include "uvector.h"
#include "uvectr32.h"
#include "uvectr64.h"
#include "regeximp.h"
#include "regexst.h"
#include "regextxt.h"
#include "ucase.h"
// #include <malloc.h> // Needed for heapcheck testing
// Find progress callback
// ----------------------
// Macro to inline test & call to ReportFindProgress(). Eliminates unnecessary function call.
//
#define REGEXFINDPROGRESS_INTERRUPT(pos, status) \
(fFindProgressCallbackFn != NULL) && (ReportFindProgress(pos, status) == FALSE)
// Smart Backtracking
// ------------------
// When a failure would go back to a LOOP_C instruction,
// strings, characters, and setrefs scan backwards for a valid start
// character themselves, pop the stack, and save state, emulating the
// LOOP_C's effect but assured that the next character of input is a
// possible matching character.
//
// Good idea in theory; unfortunately it only helps out a few specific
// cases and slows the engine down a little in the rest.
U_NAMESPACE_BEGIN
// Default limit for the size of the back track stack, to avoid system
// failures causedby heap exhaustion. Units are in 32 bit words, not bytes.
// This value puts ICU's limits higher than most other regexp implementations,
// which use recursion rather than the heap, and take more storage per
// backtrack point.
//
static const int32_t DEFAULT_BACKTRACK_STACK_CAPACITY = 8000000;
// Time limit counter constant.
// Time limits for expression evaluation are in terms of quanta of work by
// the engine, each of which is 10,000 state saves.
// This constant determines that state saves per tick number.
static const int32_t TIMER_INITIAL_VALUE = 10000;
//-----------------------------------------------------------------------------
//
// Constructor and Destructor
//
//-----------------------------------------------------------------------------
RegexMatcher::RegexMatcher(const RegexPattern *pat) {
fDeferredStatus = U_ZERO_ERROR;
init(fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return;
}
if (pat==NULL) {
fDeferredStatus = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
fPattern = pat;
init2(RegexStaticSets::gStaticSets->fEmptyText, fDeferredStatus);
}
RegexMatcher::RegexMatcher(const UnicodeString &regexp, const UnicodeString &input,
uint32_t flags, UErrorCode &status) {
init(status);
if (U_FAILURE(status)) {
return;
}
UParseError pe;
fPatternOwned = RegexPattern::compile(regexp, flags, pe, status);
fPattern = fPatternOwned;
UText inputText = UTEXT_INITIALIZER;
utext_openConstUnicodeString(&inputText, &input, &status);
init2(&inputText, status);
utext_close(&inputText);
fInputUniStrMaybeMutable = TRUE;
}
RegexMatcher::RegexMatcher(UText *regexp, UText *input,
uint32_t flags, UErrorCode &status) {
init(status);
if (U_FAILURE(status)) {
return;
}
UParseError pe;
fPatternOwned = RegexPattern::compile(regexp, flags, pe, status);
if (U_FAILURE(status)) {
return;
}
fPattern = fPatternOwned;
init2(input, status);
}
RegexMatcher::RegexMatcher(const UnicodeString &regexp,
uint32_t flags, UErrorCode &status) {
init(status);
if (U_FAILURE(status)) {
return;
}
UParseError pe;
fPatternOwned = RegexPattern::compile(regexp, flags, pe, status);
if (U_FAILURE(status)) {
return;
}
fPattern = fPatternOwned;
init2(RegexStaticSets::gStaticSets->fEmptyText, status);
}
RegexMatcher::RegexMatcher(UText *regexp,
uint32_t flags, UErrorCode &status) {
init(status);
if (U_FAILURE(status)) {
return;
}
UParseError pe;
fPatternOwned = RegexPattern::compile(regexp, flags, pe, status);
if (U_FAILURE(status)) {
return;
}
fPattern = fPatternOwned;
init2(RegexStaticSets::gStaticSets->fEmptyText, status);
}
RegexMatcher::~RegexMatcher() {
delete fStack;
if (fData != fSmallData) {
uprv_free(fData);
fData = NULL;
}
if (fPatternOwned) {
delete fPatternOwned;
fPatternOwned = NULL;
fPattern = NULL;
}
if (fInput) {
delete fInput;
}
if (fInputText) {
utext_close(fInputText);
}
if (fAltInputText) {
utext_close(fAltInputText);
}
#if UCONFIG_NO_BREAK_ITERATION==0
delete fWordBreakItr;
#endif
}
//
// init() common initialization for use by all constructors.
// Initialize all fields, get the object into a consistent state.
// This must be done even when the initial status shows an error,
// so that the object is initialized sufficiently well for the destructor
// to run safely.
//
void RegexMatcher::init(UErrorCode &status) {
fPattern = NULL;
fPatternOwned = NULL;
fFrameSize = 0;
fRegionStart = 0;
fRegionLimit = 0;
fAnchorStart = 0;
fAnchorLimit = 0;
fLookStart = 0;
fLookLimit = 0;
fActiveStart = 0;
fActiveLimit = 0;
fTransparentBounds = FALSE;
fAnchoringBounds = TRUE;
fMatch = FALSE;
fMatchStart = 0;
fMatchEnd = 0;
fLastMatchEnd = -1;
fAppendPosition = 0;
fHitEnd = FALSE;
fRequireEnd = FALSE;
fStack = NULL;
fFrame = NULL;
fTimeLimit = 0;
fTime = 0;
fTickCounter = 0;
fStackLimit = DEFAULT_BACKTRACK_STACK_CAPACITY;
fCallbackFn = NULL;
fCallbackContext = NULL;
fFindProgressCallbackFn = NULL;
fFindProgressCallbackContext = NULL;
fTraceDebug = FALSE;
fDeferredStatus = status;
fData = fSmallData;
fWordBreakItr = NULL;
fStack = NULL;
fInputText = NULL;
fAltInputText = NULL;
fInput = NULL;
fInputLength = 0;
fInputUniStrMaybeMutable = FALSE;
if (U_FAILURE(status)) {
fDeferredStatus = status;
}
}
//
// init2() Common initialization for use by RegexMatcher constructors, part 2.
// This handles the common setup to be done after the Pattern is available.
//
void RegexMatcher::init2(UText *input, UErrorCode &status) {
if (U_FAILURE(status)) {
fDeferredStatus = status;
return;
}
if (fPattern->fDataSize > (int32_t)(sizeof(fSmallData)/sizeof(fSmallData[0]))) {
fData = (int64_t *)uprv_malloc(fPattern->fDataSize * sizeof(int64_t));
if (fData == NULL) {
status = fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
return;
}
}
fStack = new UVector64(status);
if (fStack == NULL) {
status = fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
return;
}
reset(input);
setStackLimit(DEFAULT_BACKTRACK_STACK_CAPACITY, status);
if (U_FAILURE(status)) {
fDeferredStatus = status;
return;
}
}
static const UChar BACKSLASH = 0x5c;
static const UChar DOLLARSIGN = 0x24;
//--------------------------------------------------------------------------------
//
// appendReplacement
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::appendReplacement(UnicodeString &dest,
const UnicodeString &replacement,
UErrorCode &status) {
UText replacementText = UTEXT_INITIALIZER;
utext_openConstUnicodeString(&replacementText, &replacement, &status);
if (U_SUCCESS(status)) {
UText resultText = UTEXT_INITIALIZER;
utext_openUnicodeString(&resultText, &dest, &status);
if (U_SUCCESS(status)) {
appendReplacement(&resultText, &replacementText, status);
utext_close(&resultText);
}
utext_close(&replacementText);
}
return *this;
}
//
// appendReplacement, UText mode
//
RegexMatcher &RegexMatcher::appendReplacement(UText *dest,
UText *replacement,
UErrorCode &status) {
if (U_FAILURE(status)) {
return *this;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return *this;
}
if (fMatch == FALSE) {
status = U_REGEX_INVALID_STATE;
return *this;
}
// Copy input string from the end of previous match to start of current match
int64_t destLen = utext_nativeLength(dest);
if (fMatchStart > fAppendPosition) {
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
destLen += utext_replace(dest, destLen, destLen, fInputText->chunkContents+fAppendPosition,
(int32_t)(fMatchStart-fAppendPosition), &status);
} else {
int32_t len16;
if (UTEXT_USES_U16(fInputText)) {
len16 = (int32_t)(fMatchStart-fAppendPosition);
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
len16 = utext_extract(fInputText, fAppendPosition, fMatchStart, NULL, 0, &lengthStatus);
}
UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16+1));
if (inputChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return *this;
}
utext_extract(fInputText, fAppendPosition, fMatchStart, inputChars, len16+1, &status);
destLen += utext_replace(dest, destLen, destLen, inputChars, len16, &status);
uprv_free(inputChars);
}
}
fAppendPosition = fMatchEnd;
// scan the replacement text, looking for substitutions ($n) and \escapes.
// TODO: optimize this loop by efficiently scanning for '$' or '\',
// move entire ranges not containing substitutions.
UTEXT_SETNATIVEINDEX(replacement, 0);
UChar32 c = UTEXT_NEXT32(replacement);
while (c != U_SENTINEL) {
if (c == BACKSLASH) {
// Backslash Escape. Copy the following char out without further checks.
// Note: Surrogate pairs don't need any special handling
// The second half wont be a '$' or a '\', and
// will move to the dest normally on the next
// loop iteration.
c = UTEXT_CURRENT32(replacement);
if (c == U_SENTINEL) {
break;
}
if (c==0x55/*U*/ || c==0x75/*u*/) {
// We have a \udddd or \Udddddddd escape sequence.
int32_t offset = 0;
struct URegexUTextUnescapeCharContext context = U_REGEX_UTEXT_UNESCAPE_CONTEXT(replacement);
UChar32 escapedChar = u_unescapeAt(uregex_utext_unescape_charAt, &offset, INT32_MAX, &context);
if (escapedChar != (UChar32)0xFFFFFFFF) {
if (U_IS_BMP(escapedChar)) {
UChar c16 = (UChar)escapedChar;
destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status);
} else {
UChar surrogate[2];
surrogate[0] = U16_LEAD(escapedChar);
surrogate[1] = U16_TRAIL(escapedChar);
if (U_SUCCESS(status)) {
destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status);
}
}
// TODO: Report errors for mal-formed \u escapes?
// As this is, the original sequence is output, which may be OK.
if (context.lastOffset == offset) {
(void)UTEXT_PREVIOUS32(replacement);
} else if (context.lastOffset != offset-1) {
utext_moveIndex32(replacement, offset - context.lastOffset - 1);
}
}
} else {
(void)UTEXT_NEXT32(replacement);
// Plain backslash escape. Just put out the escaped character.
if (U_IS_BMP(c)) {
UChar c16 = (UChar)c;
destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status);
} else {
UChar surrogate[2];
surrogate[0] = U16_LEAD(c);
surrogate[1] = U16_TRAIL(c);
if (U_SUCCESS(status)) {
destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status);
}
}
}
} else if (c != DOLLARSIGN) {
// Normal char, not a $. Copy it out without further checks.
if (U_IS_BMP(c)) {
UChar c16 = (UChar)c;
destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status);
} else {
UChar surrogate[2];
surrogate[0] = U16_LEAD(c);
surrogate[1] = U16_TRAIL(c);
if (U_SUCCESS(status)) {
destLen += utext_replace(dest, destLen, destLen, surrogate, 2, &status);
}
}
} else {
// We've got a $. Pick up a capture group number if one follows.
// Consume at most the number of digits necessary for the largest capture
// number that is valid for this pattern.
int32_t numDigits = 0;
int32_t groupNum = 0;
UChar32 digitC;
for (;;) {
digitC = UTEXT_CURRENT32(replacement);
if (digitC == U_SENTINEL) {
break;
}
if (u_isdigit(digitC) == FALSE) {
break;
}
(void)UTEXT_NEXT32(replacement);
groupNum=groupNum*10 + u_charDigitValue(digitC);
numDigits++;
if (numDigits >= fPattern->fMaxCaptureDigits) {
break;
}
}
if (numDigits == 0) {
// The $ didn't introduce a group number at all.
// Treat it as just part of the substitution text.
UChar c16 = DOLLARSIGN;
destLen += utext_replace(dest, destLen, destLen, &c16, 1, &status);
} else {
// Finally, append the capture group data to the destination.
destLen += appendGroup(groupNum, dest, status);
if (U_FAILURE(status)) {
// Can fail if group number is out of range.
break;
}
}
}
if (U_FAILURE(status)) {
break;
} else {
c = UTEXT_NEXT32(replacement);
}
}
return *this;
}
//--------------------------------------------------------------------------------
//
// appendTail Intended to be used in conjunction with appendReplacement()
// To the destination string, append everything following
// the last match position from the input string.
//
// Note: Match ranges do not affect appendTail or appendReplacement
//
//--------------------------------------------------------------------------------
UnicodeString &RegexMatcher::appendTail(UnicodeString &dest) {
UErrorCode status = U_ZERO_ERROR;
UText resultText = UTEXT_INITIALIZER;
utext_openUnicodeString(&resultText, &dest, &status);
if (U_SUCCESS(status)) {
appendTail(&resultText, status);
utext_close(&resultText);
}
return dest;
}
//
// appendTail, UText mode
//
UText *RegexMatcher::appendTail(UText *dest, UErrorCode &status) {
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return dest;
}
if (fInputLength > fAppendPosition) {
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
int64_t destLen = utext_nativeLength(dest);
utext_replace(dest, destLen, destLen, fInputText->chunkContents+fAppendPosition,
(int32_t)(fInputLength-fAppendPosition), &status);
} else {
int32_t len16;
if (UTEXT_USES_U16(fInputText)) {
len16 = (int32_t)(fInputLength-fAppendPosition);
} else {
len16 = utext_extract(fInputText, fAppendPosition, fInputLength, NULL, 0, &status);
status = U_ZERO_ERROR; // buffer overflow
}
UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16));
if (inputChars == NULL) {
fDeferredStatus = U_MEMORY_ALLOCATION_ERROR;
} else {
utext_extract(fInputText, fAppendPosition, fInputLength, inputChars, len16, &status); // unterminated
int64_t destLen = utext_nativeLength(dest);
utext_replace(dest, destLen, destLen, inputChars, len16, &status);
uprv_free(inputChars);
}
}
}
return dest;
}
//--------------------------------------------------------------------------------
//
// end
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::end(UErrorCode &err) const {
return end(0, err);
}
int64_t RegexMatcher::end64(UErrorCode &err) const {
return end64(0, err);
}
int64_t RegexMatcher::end64(int32_t group, UErrorCode &err) const {
if (U_FAILURE(err)) {
return -1;
}
if (fMatch == FALSE) {
err = U_REGEX_INVALID_STATE;
return -1;
}
if (group < 0 || group > fPattern->fGroupMap->size()) {
err = U_INDEX_OUTOFBOUNDS_ERROR;
return -1;
}
int64_t e = -1;
if (group == 0) {
e = fMatchEnd;
} else {
// Get the position within the stack frame of the variables for
// this capture group.
int32_t groupOffset = fPattern->fGroupMap->elementAti(group-1);
U_ASSERT(groupOffset < fPattern->fFrameSize);
U_ASSERT(groupOffset >= 0);
e = fFrame->fExtra[groupOffset + 1];
}
return e;
}
int32_t RegexMatcher::end(int32_t group, UErrorCode &err) const {
return (int32_t)end64(group, err);
}
//--------------------------------------------------------------------------------
//
// find()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::find() {
// Start at the position of the last match end. (Will be zero if the
// matcher has been reset.)
//
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
return findUsingChunk();
}
int64_t startPos = fMatchEnd;
if (startPos==0) {
startPos = fActiveStart;
}
if (fMatch) {
// Save the position of any previous successful match.
fLastMatchEnd = fMatchEnd;
if (fMatchStart == fMatchEnd) {
// Previous match had zero length. Move start position up one position
// to avoid sending find() into a loop on zero-length matches.
if (startPos >= fActiveLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
UTEXT_SETNATIVEINDEX(fInputText, startPos);
(void)UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
}
} else {
if (fLastMatchEnd >= 0) {
// A previous find() failed to match. Don't try again.
// (without this test, a pattern with a zero-length match
// could match again at the end of an input string.)
fHitEnd = TRUE;
return FALSE;
}
}
// Compute the position in the input string beyond which a match can not begin, because
// the minimum length match would extend past the end of the input.
// Note: some patterns that cannot match anything will have fMinMatchLength==Max Int.
// Be aware of possible overflows if making changes here.
int64_t testStartLimit;
if (UTEXT_USES_U16(fInputText)) {
testStartLimit = fActiveLimit - fPattern->fMinMatchLen;
if (startPos > testStartLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
} else {
// For now, let the matcher discover that it can't match on its own
// We don't know how long the match len is in native characters
testStartLimit = fActiveLimit;
}
UChar32 c;
U_ASSERT(startPos >= 0);
switch (fPattern->fStartType) {
case START_NO_INFO:
// No optimization was found.
// Try a match at each input position.
for (;;) {
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
if (startPos >= testStartLimit) {
fHitEnd = TRUE;
return FALSE;
}
UTEXT_SETNATIVEINDEX(fInputText, startPos);
(void)UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testStartLimit the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
U_ASSERT(FALSE);
case START_START:
// Matches are only possible at the start of the input string
// (pattern begins with ^ or \A)
if (startPos > fActiveStart) {
fMatch = FALSE;
return FALSE;
}
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
return fMatch;
case START_SET:
{
// Match may start on any char from a pre-computed set.
U_ASSERT(fPattern->fMinMatchLen > 0);
int64_t pos;
UTEXT_SETNATIVEINDEX(fInputText, startPos);
for (;;) {
c = UTEXT_NEXT32(fInputText);
pos = UTEXT_GETNATIVEINDEX(fInputText);
// c will be -1 (U_SENTINEL) at end of text, in which case we
// skip this next block (so we don't have a negative array index)
// and handle end of text in the following block.
if (c >= 0 && ((c<256 && fPattern->fInitialChars8->contains(c)) ||
(c>=256 && fPattern->fInitialChars->contains(c)))) {
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
UTEXT_SETNATIVEINDEX(fInputText, pos);
}
if (startPos >= testStartLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
startPos = pos;
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
U_ASSERT(FALSE);
case START_STRING:
case START_CHAR:
{
// Match starts on exactly one char.
U_ASSERT(fPattern->fMinMatchLen > 0);
UChar32 theChar = fPattern->fInitialChar;
int64_t pos;
UTEXT_SETNATIVEINDEX(fInputText, startPos);
for (;;) {
c = UTEXT_NEXT32(fInputText);
pos = UTEXT_GETNATIVEINDEX(fInputText);
if (c == theChar) {
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
UTEXT_SETNATIVEINDEX(fInputText, pos);
}
if (startPos >= testStartLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
startPos = pos;
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
U_ASSERT(FALSE);
case START_LINE:
{
UChar32 c;
if (startPos == fAnchorStart) {
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
UTEXT_SETNATIVEINDEX(fInputText, startPos);
c = UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
} else {
UTEXT_SETNATIVEINDEX(fInputText, startPos);
c = UTEXT_PREVIOUS32(fInputText);
UTEXT_SETNATIVEINDEX(fInputText, startPos);
}
if (fPattern->fFlags & UREGEX_UNIX_LINES) {
for (;;) {
if (c == 0x0a) {
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
UTEXT_SETNATIVEINDEX(fInputText, startPos);
}
if (startPos >= testStartLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
c = UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testStartLimit the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
} else {
for (;;) {
if (((c & 0x7f) <= 0x29) && // First quickly bypass as many chars as possible
((c<=0x0d && c>=0x0a) || c==0x85 ||c==0x2028 || c==0x2029 )) {
if (c == 0x0d && startPos < fActiveLimit && UTEXT_CURRENT32(fInputText) == 0x0a) {
(void)UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
}
MatchAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
UTEXT_SETNATIVEINDEX(fInputText, startPos);
}
if (startPos >= testStartLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
c = UTEXT_NEXT32(fInputText);
startPos = UTEXT_GETNATIVEINDEX(fInputText);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testStartLimit the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
}
default:
U_ASSERT(FALSE);
}
U_ASSERT(FALSE);
return FALSE;
}
UBool RegexMatcher::find(int64_t start, UErrorCode &status) {
if (U_FAILURE(status)) {
return FALSE;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return FALSE;
}
this->reset(); // Note: Reset() is specified by Java Matcher documentation.
// This will reset the region to be the full input length.
if (start < 0) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
int64_t nativeStart = start;
if (nativeStart < fActiveStart || nativeStart > fActiveLimit) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
fMatchEnd = nativeStart;
return find();
}
//--------------------------------------------------------------------------------
//
// findUsingChunk() -- like find(), but with the advance knowledge that the
// entire string is available in the UText's chunk buffer.
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::findUsingChunk() {
// Start at the position of the last match end. (Will be zero if the
// matcher has been reset.
//
int32_t startPos = (int32_t)fMatchEnd;
if (startPos==0) {
startPos = (int32_t)fActiveStart;
}
const UChar *inputBuf = fInputText->chunkContents;
if (fMatch) {
// Save the position of any previous successful match.
fLastMatchEnd = fMatchEnd;
if (fMatchStart == fMatchEnd) {
// Previous match had zero length. Move start position up one position
// to avoid sending find() into a loop on zero-length matches.
if (startPos >= fActiveLimit) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
U16_FWD_1(inputBuf, startPos, fInputLength);
}
} else {
if (fLastMatchEnd >= 0) {
// A previous find() failed to match. Don't try again.
// (without this test, a pattern with a zero-length match
// could match again at the end of an input string.)
fHitEnd = TRUE;
return FALSE;
}
}
// Compute the position in the input string beyond which a match can not begin, because
// the minimum length match would extend past the end of the input.
// Note: some patterns that cannot match anything will have fMinMatchLength==Max Int.
// Be aware of possible overflows if making changes here.
int32_t testLen = (int32_t)(fActiveLimit - fPattern->fMinMatchLen);
if (startPos > testLen) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
UChar32 c;
U_ASSERT(startPos >= 0);
switch (fPattern->fStartType) {
case START_NO_INFO:
// No optimization was found.
// Try a match at each input position.
for (;;) {
MatchChunkAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
if (startPos >= testLen) {
fHitEnd = TRUE;
return FALSE;
}
U16_FWD_1(inputBuf, startPos, fActiveLimit);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testLen the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
U_ASSERT(FALSE);
case START_START:
// Matches are only possible at the start of the input string
// (pattern begins with ^ or \A)
if (startPos > fActiveStart) {
fMatch = FALSE;
return FALSE;
}
MatchChunkAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
return fMatch;
case START_SET:
{
// Match may start on any char from a pre-computed set.
U_ASSERT(fPattern->fMinMatchLen > 0);
for (;;) {
int32_t pos = startPos;
U16_NEXT(inputBuf, startPos, fActiveLimit, c); // like c = inputBuf[startPos++];
if ((c<256 && fPattern->fInitialChars8->contains(c)) ||
(c>=256 && fPattern->fInitialChars->contains(c))) {
MatchChunkAt(pos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
}
if (pos >= testLen) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
U_ASSERT(FALSE);
case START_STRING:
case START_CHAR:
{
// Match starts on exactly one char.
U_ASSERT(fPattern->fMinMatchLen > 0);
UChar32 theChar = fPattern->fInitialChar;
for (;;) {
int32_t pos = startPos;
U16_NEXT(inputBuf, startPos, fActiveLimit, c); // like c = inputBuf[startPos++];
if (c == theChar) {
MatchChunkAt(pos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
}
if (pos >= testLen) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
U_ASSERT(FALSE);
case START_LINE:
{
UChar32 c;
if (startPos == fAnchorStart) {
MatchChunkAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
U16_FWD_1(inputBuf, startPos, fActiveLimit);
}
if (fPattern->fFlags & UREGEX_UNIX_LINES) {
for (;;) {
c = inputBuf[startPos-1];
if (c == 0x0a) {
MatchChunkAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
}
if (startPos >= testLen) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
U16_FWD_1(inputBuf, startPos, fActiveLimit);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testLen the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
} else {
for (;;) {
c = inputBuf[startPos-1];
if (((c & 0x7f) <= 0x29) && // First quickly bypass as many chars as possible
((c<=0x0d && c>=0x0a) || c==0x85 ||c==0x2028 || c==0x2029 )) {
if (c == 0x0d && startPos < fActiveLimit && inputBuf[startPos] == 0x0a) {
startPos++;
}
MatchChunkAt(startPos, FALSE, fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
if (fMatch) {
return TRUE;
}
}
if (startPos >= testLen) {
fMatch = FALSE;
fHitEnd = TRUE;
return FALSE;
}
U16_FWD_1(inputBuf, startPos, fActiveLimit);
// Note that it's perfectly OK for a pattern to have a zero-length
// match at the end of a string, so we must make sure that the loop
// runs with startPos == testLen the last time through.
if (REGEXFINDPROGRESS_INTERRUPT(startPos, fDeferredStatus))
return FALSE;
}
}
}
default:
U_ASSERT(FALSE);
}
U_ASSERT(FALSE);
return FALSE;
}
//--------------------------------------------------------------------------------
//
// group()
//
//--------------------------------------------------------------------------------
UnicodeString RegexMatcher::group(UErrorCode &status) const {
return group(0, status);
}
// Return immutable shallow clone
UText *RegexMatcher::group(UText *dest, int64_t &group_len, UErrorCode &status) const {
return group(0, dest, group_len, status);
}
// Return immutable shallow clone
UText *RegexMatcher::group(int32_t groupNum, UText *dest, int64_t &group_len, UErrorCode &status) const {
group_len = 0;
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
} else if (fMatch == FALSE) {
status = U_REGEX_INVALID_STATE;
} else if (groupNum < 0 || groupNum > fPattern->fGroupMap->size()) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
}
if (U_FAILURE(status)) {
return dest;
}
int64_t s, e;
if (groupNum == 0) {
s = fMatchStart;
e = fMatchEnd;
} else {
int32_t groupOffset = fPattern->fGroupMap->elementAti(groupNum-1);
U_ASSERT(groupOffset < fPattern->fFrameSize);
U_ASSERT(groupOffset >= 0);
s = fFrame->fExtra[groupOffset];
e = fFrame->fExtra[groupOffset+1];
}
if (s < 0) {
// A capture group wasn't part of the match
return utext_clone(dest, fInputText, FALSE, TRUE, &status);
}
U_ASSERT(s <= e);
group_len = e - s;
dest = utext_clone(dest, fInputText, FALSE, TRUE, &status);
if (dest)
UTEXT_SETNATIVEINDEX(dest, s);
return dest;
}
UnicodeString RegexMatcher::group(int32_t groupNum, UErrorCode &status) const {
UnicodeString result;
if (U_FAILURE(status)) {
return result;
}
UText resultText = UTEXT_INITIALIZER;
utext_openUnicodeString(&resultText, &result, &status);
group(groupNum, &resultText, status);
utext_close(&resultText);
return result;
}
// Return deep (mutable) clone
// Technology Preview (as an API), but note that the UnicodeString API is implemented
// using this function.
UText *RegexMatcher::group(int32_t groupNum, UText *dest, UErrorCode &status) const {
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
} else if (fMatch == FALSE) {
status = U_REGEX_INVALID_STATE;
} else if (groupNum < 0 || groupNum > fPattern->fGroupMap->size()) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
}
if (U_FAILURE(status)) {
return dest;
}
int64_t s, e;
if (groupNum == 0) {
s = fMatchStart;
e = fMatchEnd;
} else {
int32_t groupOffset = fPattern->fGroupMap->elementAti(groupNum-1);
U_ASSERT(groupOffset < fPattern->fFrameSize);
U_ASSERT(groupOffset >= 0);
s = fFrame->fExtra[groupOffset];
e = fFrame->fExtra[groupOffset+1];
}
if (s < 0) {
// A capture group wasn't part of the match
if (dest) {
utext_replace(dest, 0, utext_nativeLength(dest), NULL, 0, &status);
return dest;
} else {
return utext_openUChars(NULL, NULL, 0, &status);
}
}
U_ASSERT(s <= e);
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
U_ASSERT(e <= fInputLength);
if (dest) {
utext_replace(dest, 0, utext_nativeLength(dest), fInputText->chunkContents+s, (int32_t)(e-s), &status);
} else {
UText groupText = UTEXT_INITIALIZER;
utext_openUChars(&groupText, fInputText->chunkContents+s, e-s, &status);
dest = utext_clone(NULL, &groupText, TRUE, FALSE, &status);
utext_close(&groupText);
}
} else {
int32_t len16;
if (UTEXT_USES_U16(fInputText)) {
len16 = (int32_t)(e-s);
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
len16 = utext_extract(fInputText, s, e, NULL, 0, &lengthStatus);
}
UChar *groupChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16+1));
if (groupChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return dest;
}
utext_extract(fInputText, s, e, groupChars, len16+1, &status);
if (dest) {
utext_replace(dest, 0, utext_nativeLength(dest), groupChars, len16, &status);
} else {
UText groupText = UTEXT_INITIALIZER;
utext_openUChars(&groupText, groupChars, len16, &status);
dest = utext_clone(NULL, &groupText, TRUE, FALSE, &status);
utext_close(&groupText);
}
uprv_free(groupChars);
}
return dest;
}
//--------------------------------------------------------------------------------
//
// appendGroup() -- currently internal only, appends a group to a UText rather
// than replacing its contents
//
//--------------------------------------------------------------------------------
int64_t RegexMatcher::appendGroup(int32_t groupNum, UText *dest, UErrorCode &status) const {
if (U_FAILURE(status)) {
return 0;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return 0;
}
int64_t destLen = utext_nativeLength(dest);
if (fMatch == FALSE) {
status = U_REGEX_INVALID_STATE;
return utext_replace(dest, destLen, destLen, NULL, 0, &status);
}
if (groupNum < 0 || groupNum > fPattern->fGroupMap->size()) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return utext_replace(dest, destLen, destLen, NULL, 0, &status);
}
int64_t s, e;
if (groupNum == 0) {
s = fMatchStart;
e = fMatchEnd;
} else {
int32_t groupOffset = fPattern->fGroupMap->elementAti(groupNum-1);
U_ASSERT(groupOffset < fPattern->fFrameSize);
U_ASSERT(groupOffset >= 0);
s = fFrame->fExtra[groupOffset];
e = fFrame->fExtra[groupOffset+1];
}
if (s < 0) {
// A capture group wasn't part of the match
return utext_replace(dest, destLen, destLen, NULL, 0, &status);
}
U_ASSERT(s <= e);
int64_t deltaLen;
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
U_ASSERT(e <= fInputLength);
deltaLen = utext_replace(dest, destLen, destLen, fInputText->chunkContents+s, (int32_t)(e-s), &status);
} else {
int32_t len16;
if (UTEXT_USES_U16(fInputText)) {
len16 = (int32_t)(e-s);
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
len16 = utext_extract(fInputText, s, e, NULL, 0, &lengthStatus);
}
UChar *groupChars = (UChar *)uprv_malloc(sizeof(UChar)*(len16+1));
if (groupChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
utext_extract(fInputText, s, e, groupChars, len16+1, &status);
deltaLen = utext_replace(dest, destLen, destLen, groupChars, len16, &status);
uprv_free(groupChars);
}
return deltaLen;
}
//--------------------------------------------------------------------------------
//
// groupCount()
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::groupCount() const {
return fPattern->fGroupMap->size();
}
//--------------------------------------------------------------------------------
//
// hasAnchoringBounds()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::hasAnchoringBounds() const {
return fAnchoringBounds;
}
//--------------------------------------------------------------------------------
//
// hasTransparentBounds()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::hasTransparentBounds() const {
return fTransparentBounds;
}
//--------------------------------------------------------------------------------
//
// hitEnd()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::hitEnd() const {
return fHitEnd;
}
//--------------------------------------------------------------------------------
//
// input()
//
//--------------------------------------------------------------------------------
const UnicodeString &RegexMatcher::input() const {
if (!fInput) {
UErrorCode status = U_ZERO_ERROR;
int32_t len16;
if (UTEXT_USES_U16(fInputText)) {
len16 = (int32_t)fInputLength;
} else {
len16 = utext_extract(fInputText, 0, fInputLength, NULL, 0, &status);
status = U_ZERO_ERROR; // overflow, length status
}
UnicodeString *result = new UnicodeString(len16, 0, 0);
UChar *inputChars = result->getBuffer(len16);
utext_extract(fInputText, 0, fInputLength, inputChars, len16, &status); // unterminated warning
result->releaseBuffer(len16);
(*(const UnicodeString **)&fInput) = result; // pointer assignment, rather than operator=
}
return *fInput;
}
//--------------------------------------------------------------------------------
//
// inputText()
//
//--------------------------------------------------------------------------------
UText *RegexMatcher::inputText() const {
return fInputText;
}
//--------------------------------------------------------------------------------
//
// getInput() -- like inputText(), but makes a clone or copies into another UText
//
//--------------------------------------------------------------------------------
UText *RegexMatcher::getInput (UText *dest, UErrorCode &status) const {
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return dest;
}
if (dest) {
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
utext_replace(dest, 0, utext_nativeLength(dest), fInputText->chunkContents, (int32_t)fInputLength, &status);
} else {
int32_t input16Len;
if (UTEXT_USES_U16(fInputText)) {
input16Len = (int32_t)fInputLength;
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
input16Len = utext_extract(fInputText, 0, fInputLength, NULL, 0, &lengthStatus); // buffer overflow error
}
UChar *inputChars = (UChar *)uprv_malloc(sizeof(UChar)*(input16Len));
if (inputChars == NULL) {
return dest;
}
status = U_ZERO_ERROR;
utext_extract(fInputText, 0, fInputLength, inputChars, input16Len, &status); // not terminated warning
status = U_ZERO_ERROR;
utext_replace(dest, 0, utext_nativeLength(dest), inputChars, input16Len, &status);
uprv_free(inputChars);
}
return dest;
} else {
return utext_clone(NULL, fInputText, FALSE, TRUE, &status);
}
}
static UBool compat_SyncMutableUTextContents(UText *ut);
static UBool compat_SyncMutableUTextContents(UText *ut) {
UBool retVal = FALSE;
// In the following test, we're really only interested in whether the UText should switch
// between heap and stack allocation. If length hasn't changed, we won't, so the chunkContents
// will still point to the correct data.
if (utext_nativeLength(ut) != ut->nativeIndexingLimit) {
UnicodeString *us=(UnicodeString *)ut->context;
// Update to the latest length.
// For example, (utext_nativeLength(ut) != ut->nativeIndexingLimit).
int32_t newLength = us->length();
// Update the chunk description.
// The buffer may have switched between stack- and heap-based.
ut->chunkContents = us->getBuffer();
ut->chunkLength = newLength;
ut->chunkNativeLimit = newLength;
ut->nativeIndexingLimit = newLength;
retVal = TRUE;
}
return retVal;
}
//--------------------------------------------------------------------------------
//
// lookingAt()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::lookingAt(UErrorCode &status) {
if (U_FAILURE(status)) {
return FALSE;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return FALSE;
}
if (fInputUniStrMaybeMutable) {
if (compat_SyncMutableUTextContents(fInputText)) {
fInputLength = utext_nativeLength(fInputText);
reset();
}
}
else {
resetPreserveRegion();
}
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
MatchChunkAt((int32_t)fActiveStart, FALSE, status);
} else {
MatchAt(fActiveStart, FALSE, status);
}
return fMatch;
}
UBool RegexMatcher::lookingAt(int64_t start, UErrorCode &status) {
if (U_FAILURE(status)) {
return FALSE;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return FALSE;
}
reset();
if (start < 0) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
if (fInputUniStrMaybeMutable) {
if (compat_SyncMutableUTextContents(fInputText)) {
fInputLength = utext_nativeLength(fInputText);
reset();
}
}
int64_t nativeStart;
nativeStart = start;
if (nativeStart < fActiveStart || nativeStart > fActiveLimit) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
MatchChunkAt((int32_t)nativeStart, FALSE, status);
} else {
MatchAt(nativeStart, FALSE, status);
}
return fMatch;
}
//--------------------------------------------------------------------------------
//
// matches()
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::matches(UErrorCode &status) {
if (U_FAILURE(status)) {
return FALSE;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return FALSE;
}
if (fInputUniStrMaybeMutable) {
if (compat_SyncMutableUTextContents(fInputText)) {
fInputLength = utext_nativeLength(fInputText);
reset();
}
}
else {
resetPreserveRegion();
}
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
MatchChunkAt((int32_t)fActiveStart, TRUE, status);
} else {
MatchAt(fActiveStart, TRUE, status);
}
return fMatch;
}
UBool RegexMatcher::matches(int64_t start, UErrorCode &status) {
if (U_FAILURE(status)) {
return FALSE;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return FALSE;
}
reset();
if (start < 0) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
if (fInputUniStrMaybeMutable) {
if (compat_SyncMutableUTextContents(fInputText)) {
fInputLength = utext_nativeLength(fInputText);
reset();
}
}
int64_t nativeStart;
nativeStart = start;
if (nativeStart < fActiveStart || nativeStart > fActiveLimit) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return FALSE;
}
if (UTEXT_FULL_TEXT_IN_CHUNK(fInputText, fInputLength)) {
MatchChunkAt((int32_t)nativeStart, TRUE, status);
} else {
MatchAt(nativeStart, TRUE, status);
}
return fMatch;
}
//--------------------------------------------------------------------------------
//
// pattern
//
//--------------------------------------------------------------------------------
const RegexPattern &RegexMatcher::pattern() const {
return *fPattern;
}
//--------------------------------------------------------------------------------
//
// region
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::region(int64_t regionStart, int64_t regionLimit, int64_t startIndex, UErrorCode &status) {
if (U_FAILURE(status)) {
return *this;
}
if (regionStart>regionLimit || regionStart<0 || regionLimit<0) {
status = U_ILLEGAL_ARGUMENT_ERROR;
}
int64_t nativeStart = regionStart;
int64_t nativeLimit = regionLimit;
if (nativeStart > fInputLength || nativeLimit > fInputLength) {
status = U_ILLEGAL_ARGUMENT_ERROR;
}
if (startIndex == -1)
this->reset();
else
resetPreserveRegion();
fRegionStart = nativeStart;
fRegionLimit = nativeLimit;
fActiveStart = nativeStart;
fActiveLimit = nativeLimit;
if (startIndex != -1) {
if (startIndex < fActiveStart || startIndex > fActiveLimit) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
}
fMatchEnd = startIndex;
}
if (!fTransparentBounds) {
fLookStart = nativeStart;
fLookLimit = nativeLimit;
}
if (fAnchoringBounds) {
fAnchorStart = nativeStart;
fAnchorLimit = nativeLimit;
}
return *this;
}
RegexMatcher &RegexMatcher::region(int64_t start, int64_t limit, UErrorCode &status) {
return region(start, limit, -1, status);
}
//--------------------------------------------------------------------------------
//
// regionEnd
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::regionEnd() const {
return (int32_t)fRegionLimit;
}
int64_t RegexMatcher::regionEnd64() const {
return fRegionLimit;
}
//--------------------------------------------------------------------------------
//
// regionStart
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::regionStart() const {
return (int32_t)fRegionStart;
}
int64_t RegexMatcher::regionStart64() const {
return fRegionStart;
}
//--------------------------------------------------------------------------------
//
// replaceAll
//
//--------------------------------------------------------------------------------
UnicodeString RegexMatcher::replaceAll(const UnicodeString &replacement, UErrorCode &status) {
UText replacementText = UTEXT_INITIALIZER;
UText resultText = UTEXT_INITIALIZER;
UnicodeString resultString;
if (U_FAILURE(status)) {
return resultString;
}
utext_openConstUnicodeString(&replacementText, &replacement, &status);
utext_openUnicodeString(&resultText, &resultString, &status);
replaceAll(&replacementText, &resultText, status);
utext_close(&resultText);
utext_close(&replacementText);
return resultString;
}
//
// replaceAll, UText mode
//
UText *RegexMatcher::replaceAll(UText *replacement, UText *dest, UErrorCode &status) {
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return dest;
}
if (dest == NULL) {
UnicodeString emptyString;
UText empty = UTEXT_INITIALIZER;
utext_openUnicodeString(&empty, &emptyString, &status);
dest = utext_clone(NULL, &empty, TRUE, FALSE, &status);
utext_close(&empty);
}
if (U_SUCCESS(status)) {
reset();
while (find()) {
appendReplacement(dest, replacement, status);
if (U_FAILURE(status)) {
break;
}
}
appendTail(dest, status);
}
return dest;
}
//--------------------------------------------------------------------------------
//
// replaceFirst
//
//--------------------------------------------------------------------------------
UnicodeString RegexMatcher::replaceFirst(const UnicodeString &replacement, UErrorCode &status) {
UText replacementText = UTEXT_INITIALIZER;
UText resultText = UTEXT_INITIALIZER;
UnicodeString resultString;
utext_openConstUnicodeString(&replacementText, &replacement, &status);
utext_openUnicodeString(&resultText, &resultString, &status);
replaceFirst(&replacementText, &resultText, status);
utext_close(&resultText);
utext_close(&replacementText);
return resultString;
}
//
// replaceFirst, UText mode
//
UText *RegexMatcher::replaceFirst(UText *replacement, UText *dest, UErrorCode &status) {
if (U_FAILURE(status)) {
return dest;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return dest;
}
reset();
if (!find()) {
return getInput(dest, status);
}
if (dest == NULL) {
UnicodeString emptyString;
UText empty = UTEXT_INITIALIZER;
utext_openUnicodeString(&empty, &emptyString, &status);
dest = utext_clone(NULL, &empty, TRUE, FALSE, &status);
utext_close(&empty);
}
appendReplacement(dest, replacement, status);
appendTail(dest, status);
return dest;
}
//--------------------------------------------------------------------------------
//
// requireEnd
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::requireEnd() const {
return fRequireEnd;
}
//--------------------------------------------------------------------------------
//
// reset
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::reset() {
fRegionStart = 0;
fRegionLimit = fInputLength;
fActiveStart = 0;
fActiveLimit = fInputLength;
fAnchorStart = 0;
fAnchorLimit = fInputLength;
fLookStart = 0;
fLookLimit = fInputLength;
resetPreserveRegion();
return *this;
}
void RegexMatcher::resetPreserveRegion() {
fMatchStart = 0;
fMatchEnd = 0;
fLastMatchEnd = -1;
fAppendPosition = 0;
fMatch = FALSE;
fHitEnd = FALSE;
fRequireEnd = FALSE;
fTime = 0;
fTickCounter = TIMER_INITIAL_VALUE;
//resetStack(); // more expensive than it looks...
}
RegexMatcher &RegexMatcher::reset(const UnicodeString &input) {
fInputText = utext_openConstUnicodeString(fInputText, &input, &fDeferredStatus);
if (fPattern->fNeedsAltInput) {
fAltInputText = utext_clone(fAltInputText, fInputText, FALSE, TRUE, &fDeferredStatus);
}
fInputLength = utext_nativeLength(fInputText);
reset();
delete fInput;
fInput = NULL;
// Do the following for any UnicodeString.
// This is for compatibility for those clients who modify the input string "live" during regex operations.
fInputUniStrMaybeMutable = TRUE;
if (fWordBreakItr != NULL) {
#if UCONFIG_NO_BREAK_ITERATION==0
UErrorCode status = U_ZERO_ERROR;
fWordBreakItr->setText(fInputText, status);
#endif
}
return *this;
}
RegexMatcher &RegexMatcher::reset(UText *input) {
if (fInputText != input) {
fInputText = utext_clone(fInputText, input, FALSE, TRUE, &fDeferredStatus);
if (fPattern->fNeedsAltInput) fAltInputText = utext_clone(fAltInputText, fInputText, FALSE, TRUE, &fDeferredStatus);
fInputLength = utext_nativeLength(fInputText);
delete fInput;
fInput = NULL;
if (fWordBreakItr != NULL) {
#if UCONFIG_NO_BREAK_ITERATION==0
UErrorCode status = U_ZERO_ERROR;
fWordBreakItr->setText(input, status);
#endif
}
}
reset();
fInputUniStrMaybeMutable = FALSE;
return *this;
}
/*RegexMatcher &RegexMatcher::reset(const UChar *) {
fDeferredStatus = U_INTERNAL_PROGRAM_ERROR;
return *this;
}*/
RegexMatcher &RegexMatcher::reset(int64_t position, UErrorCode &status) {
if (U_FAILURE(status)) {
return *this;
}
reset(); // Reset also resets the region to be the entire string.
if (position < 0 || position > fActiveLimit) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return *this;
}
fMatchEnd = position;
return *this;
}
//--------------------------------------------------------------------------------
//
// refresh
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::refreshInputText(UText *input, UErrorCode &status) {
if (U_FAILURE(status)) {
return *this;
}
if (input == NULL) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return *this;
}
if (utext_nativeLength(fInputText) != utext_nativeLength(input)) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return *this;
}
int64_t pos = utext_getNativeIndex(fInputText);
// Shallow read-only clone of the new UText into the existing input UText
fInputText = utext_clone(fInputText, input, FALSE, TRUE, &status);
if (U_FAILURE(status)) {
return *this;
}
utext_setNativeIndex(fInputText, pos);
if (fAltInputText != NULL) {
pos = utext_getNativeIndex(fAltInputText);
fAltInputText = utext_clone(fAltInputText, input, FALSE, TRUE, &status);
if (U_FAILURE(status)) {
return *this;
}
utext_setNativeIndex(fAltInputText, pos);
}
return *this;
}
//--------------------------------------------------------------------------------
//
// setTrace
//
//--------------------------------------------------------------------------------
void RegexMatcher::setTrace(UBool state) {
fTraceDebug = state;
}
//---------------------------------------------------------------------
//
// split
//
//---------------------------------------------------------------------
int32_t RegexMatcher::split(const UnicodeString &input,
UnicodeString dest[],
int32_t destCapacity,
UErrorCode &status)
{
UText inputText = UTEXT_INITIALIZER;
utext_openConstUnicodeString(&inputText, &input, &status);
if (U_FAILURE(status)) {
return 0;
}
UText **destText = (UText **)uprv_malloc(sizeof(UText*)*destCapacity);
if (destText == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
return 0;
}
int32_t i;
for (i = 0; i < destCapacity; i++) {
destText[i] = utext_openUnicodeString(NULL, &dest[i], &status);
}
int32_t fieldCount = split(&inputText, destText, destCapacity, status);
for (i = 0; i < destCapacity; i++) {
utext_close(destText[i]);
}
uprv_free(destText);
utext_close(&inputText);
return fieldCount;
}
//
// split, UText mode
//
int32_t RegexMatcher::split(UText *input,
UText *dest[],
int32_t destCapacity,
UErrorCode &status)
{
//
// Check arguements for validity
//
if (U_FAILURE(status)) {
return 0;
};
if (destCapacity < 1) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return 0;
}
//
// Reset for the input text
//
reset(input);
int64_t nextOutputStringStart = 0;
if (fActiveLimit == 0) {
return 0;
}
//
// Loop through the input text, searching for the delimiter pattern
//
int32_t i;
int32_t numCaptureGroups = fPattern->fGroupMap->size();
for (i=0; ; i++) {
if (i>=destCapacity-1) {
// There is one or zero output string left.
// Fill the last output string with whatever is left from the input, then exit the loop.
// ( i will be == destCapacity if we filled the output array while processing
// capture groups of the delimiter expression, in which case we will discard the
// last capture group saved in favor of the unprocessed remainder of the
// input string.)
i = destCapacity-1;
if (fActiveLimit > nextOutputStringStart) {
if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) {
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]),
input->chunkContents+nextOutputStringStart,
(int32_t)(fActiveLimit-nextOutputStringStart), &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart,
fActiveLimit-nextOutputStringStart, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
int32_t remaining16Length =
utext_extract(input, nextOutputStringStart, fActiveLimit, NULL, 0, &lengthStatus);
UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1));
if (remainingChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
break;
}
utext_extract(input, nextOutputStringStart, fActiveLimit, remainingChars, remaining16Length+1, &status);
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, remainingChars, remaining16Length, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
uprv_free(remainingChars);
}
}
break;
}
if (find()) {
// We found another delimiter. Move everything from where we started looking
// up until the start of the delimiter into the next output string.
if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) {
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]),
input->chunkContents+nextOutputStringStart,
(int32_t)(fMatchStart-nextOutputStringStart), &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart,
fMatchStart-nextOutputStringStart, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
int32_t remaining16Length = utext_extract(input, nextOutputStringStart, fMatchStart, NULL, 0, &lengthStatus);
UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1));
if (remainingChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
break;
}
utext_extract(input, nextOutputStringStart, fMatchStart, remainingChars, remaining16Length+1, &status);
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, remainingChars, remaining16Length, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
uprv_free(remainingChars);
}
nextOutputStringStart = fMatchEnd;
// If the delimiter pattern has capturing parentheses, the captured
// text goes out into the next n destination strings.
int32_t groupNum;
for (groupNum=1; groupNum<=numCaptureGroups; groupNum++) {
if (i >= destCapacity-2) {
// Never fill the last available output string with capture group text.
// It will filled with the last field, the remainder of the
// unsplit input text.
break;
}
i++;
dest[i] = group(groupNum, dest[i], status);
}
if (nextOutputStringStart == fActiveLimit) {
// The delimiter was at the end of the string. We're done, but first
// we output one last empty string, for the empty field following
// the delimiter at the end of input.
if (i+1 < destCapacity) {
++i;
if (dest[i] == NULL) {
dest[i] = utext_openUChars(NULL, NULL, 0, &status);
} else {
static UChar emptyString[] = {(UChar)0};
utext_replace(dest[i], 0, utext_nativeLength(dest[i]), emptyString, 0, &status);
}
}
break;
}
}
else
{
// We ran off the end of the input while looking for the next delimiter.
// All the remaining text goes into the current output string.
if (UTEXT_FULL_TEXT_IN_CHUNK(input, fInputLength)) {
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]),
input->chunkContents+nextOutputStringStart,
(int32_t)(fActiveLimit-nextOutputStringStart), &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, input->chunkContents+nextOutputStringStart,
fActiveLimit-nextOutputStringStart, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
} else {
UErrorCode lengthStatus = U_ZERO_ERROR;
int32_t remaining16Length = utext_extract(input, nextOutputStringStart, fActiveLimit, NULL, 0, &lengthStatus);
UChar *remainingChars = (UChar *)uprv_malloc(sizeof(UChar)*(remaining16Length+1));
if (remainingChars == NULL) {
status = U_MEMORY_ALLOCATION_ERROR;
break;
}
utext_extract(input, nextOutputStringStart, fActiveLimit, remainingChars, remaining16Length+1, &status);
if (dest[i]) {
utext_replace(dest[i], 0, utext_nativeLength(dest[i]), remainingChars, remaining16Length, &status);
} else {
UText remainingText = UTEXT_INITIALIZER;
utext_openUChars(&remainingText, remainingChars, remaining16Length, &status);
dest[i] = utext_clone(NULL, &remainingText, TRUE, FALSE, &status);
utext_close(&remainingText);
}
uprv_free(remainingChars);
}
break;
}
if (U_FAILURE(status)) {
break;
}
} // end of for loop
return i+1;
}
//--------------------------------------------------------------------------------
//
// start
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::start(UErrorCode &status) const {
return start(0, status);
}
int64_t RegexMatcher::start64(UErrorCode &status) const {
return start64(0, status);
}
//--------------------------------------------------------------------------------
//
// start(int32_t group, UErrorCode &status)
//
//--------------------------------------------------------------------------------
int64_t RegexMatcher::start64(int32_t group, UErrorCode &status) const {
if (U_FAILURE(status)) {
return -1;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return -1;
}
if (fMatch == FALSE) {
status = U_REGEX_INVALID_STATE;
return -1;
}
if (group < 0 || group > fPattern->fGroupMap->size()) {
status = U_INDEX_OUTOFBOUNDS_ERROR;
return -1;
}
int64_t s;
if (group == 0) {
s = fMatchStart;
} else {
int32_t groupOffset = fPattern->fGroupMap->elementAti(group-1);
U_ASSERT(groupOffset < fPattern->fFrameSize);
U_ASSERT(groupOffset >= 0);
s = fFrame->fExtra[groupOffset];
}
return s;
}
int32_t RegexMatcher::start(int32_t group, UErrorCode &status) const {
return (int32_t)start64(group, status);
}
//--------------------------------------------------------------------------------
//
// useAnchoringBounds
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::useAnchoringBounds(UBool b) {
fAnchoringBounds = b;
fAnchorStart = (fAnchoringBounds ? fRegionStart : 0);
fAnchorLimit = (fAnchoringBounds ? fRegionLimit : fInputLength);
return *this;
}
//--------------------------------------------------------------------------------
//
// useTransparentBounds
//
//--------------------------------------------------------------------------------
RegexMatcher &RegexMatcher::useTransparentBounds(UBool b) {
fTransparentBounds = b;
fLookStart = (fTransparentBounds ? 0 : fRegionStart);
fLookLimit = (fTransparentBounds ? fInputLength : fRegionLimit);
return *this;
}
//--------------------------------------------------------------------------------
//
// setTimeLimit
//
//--------------------------------------------------------------------------------
void RegexMatcher::setTimeLimit(int32_t limit, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return;
}
if (limit < 0) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
fTimeLimit = limit;
}
//--------------------------------------------------------------------------------
//
// getTimeLimit
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::getTimeLimit() const {
return fTimeLimit;
}
//--------------------------------------------------------------------------------
//
// setStackLimit
//
//--------------------------------------------------------------------------------
void RegexMatcher::setStackLimit(int32_t limit, UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
if (U_FAILURE(fDeferredStatus)) {
status = fDeferredStatus;
return;
}
if (limit < 0) {
status = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
// Reset the matcher. This is needed here in case there is a current match
// whose final stack frame (containing the match results, pointed to by fFrame)
// would be lost by resizing to a smaller stack size.
reset();
if (limit == 0) {
// Unlimited stack expansion
fStack->setMaxCapacity(0);
} else {
// Change the units of the limit from bytes to ints, and bump the size up
// to be big enough to hold at least one stack frame for the pattern,
// if it isn't there already.
int32_t adjustedLimit = limit / sizeof(int32_t);
if (adjustedLimit < fPattern->fFrameSize) {
adjustedLimit = fPattern->fFrameSize;
}
fStack->setMaxCapacity(adjustedLimit);
}
fStackLimit = limit;
}
//--------------------------------------------------------------------------------
//
// getStackLimit
//
//--------------------------------------------------------------------------------
int32_t RegexMatcher::getStackLimit() const {
return fStackLimit;
}
//--------------------------------------------------------------------------------
//
// setMatchCallback
//
//--------------------------------------------------------------------------------
void RegexMatcher::setMatchCallback(URegexMatchCallback *callback,
const void *context,
UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fCallbackFn = callback;
fCallbackContext = context;
}
//--------------------------------------------------------------------------------
//
// getMatchCallback
//
//--------------------------------------------------------------------------------
void RegexMatcher::getMatchCallback(URegexMatchCallback *&callback,
const void *&context,
UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
callback = fCallbackFn;
context = fCallbackContext;
}
//--------------------------------------------------------------------------------
//
// setMatchCallback
//
//--------------------------------------------------------------------------------
void RegexMatcher::setFindProgressCallback(URegexFindProgressCallback *callback,
const void *context,
UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
fFindProgressCallbackFn = callback;
fFindProgressCallbackContext = context;
}
//--------------------------------------------------------------------------------
//
// getMatchCallback
//
//--------------------------------------------------------------------------------
void RegexMatcher::getFindProgressCallback(URegexFindProgressCallback *&callback,
const void *&context,
UErrorCode &status) {
if (U_FAILURE(status)) {
return;
}
callback = fFindProgressCallbackFn;
context = fFindProgressCallbackContext;
}
//================================================================================
//
// Code following this point in this file is the internal
// Match Engine Implementation.
//
//================================================================================
//--------------------------------------------------------------------------------
//
// resetStack
// Discard any previous contents of the state save stack, and initialize a
// new stack frame to all -1. The -1s are needed for capture group limits,
// where they indicate that a group has not yet matched anything.
//--------------------------------------------------------------------------------
REStackFrame *RegexMatcher::resetStack() {
// Discard any previous contents of the state save stack, and initialize a
// new stack frame with all -1 data. The -1s are needed for capture group limits,
// where they indicate that a group has not yet matched anything.
fStack->removeAllElements();
REStackFrame *iFrame = (REStackFrame *)fStack->reserveBlock(fPattern->fFrameSize, fDeferredStatus);
int32_t i;
for (i=0; i<fPattern->fFrameSize-RESTACKFRAME_HDRCOUNT; i++) {
iFrame->fExtra[i] = -1;
}
return iFrame;
}
//--------------------------------------------------------------------------------
//
// isWordBoundary
// in perl, "xab..cd..", \b is true at positions 0,3,5,7
// For us,
// If the current char is a combining mark,
// \b is FALSE.
// Else Scan backwards to the first non-combining char.
// We are at a boundary if the this char and the original chars are
// opposite in membership in \w set
//
// parameters: pos - the current position in the input buffer
//
// TODO: double-check edge cases at region boundaries.
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::isWordBoundary(int64_t pos) {
UBool isBoundary = FALSE;
UBool cIsWord = FALSE;
if (pos >= fLookLimit) {
fHitEnd = TRUE;
} else {
// Determine whether char c at current position is a member of the word set of chars.
// If we're off the end of the string, behave as though we're not at a word char.
UTEXT_SETNATIVEINDEX(fInputText, pos);
UChar32 c = UTEXT_CURRENT32(fInputText);
if (u_hasBinaryProperty(c, UCHAR_GRAPHEME_EXTEND) || u_charType(c) == U_FORMAT_CHAR) {
// Current char is a combining one. Not a boundary.
return FALSE;
}
cIsWord = fPattern->fStaticSets[URX_ISWORD_SET]->contains(c);
}
// Back up until we come to a non-combining char, determine whether
// that char is a word char.
UBool prevCIsWord = FALSE;
for (;;) {
if (UTEXT_GETNATIVEINDEX(fInputText) <= fLookStart) {
break;
}
UChar32 prevChar = UTEXT_PREVIOUS32(fInputText);
if (!(u_hasBinaryProperty(prevChar, UCHAR_GRAPHEME_EXTEND)
|| u_charType(prevChar) == U_FORMAT_CHAR)) {
prevCIsWord = fPattern->fStaticSets[URX_ISWORD_SET]->contains(prevChar);
break;
}
}
isBoundary = cIsWord ^ prevCIsWord;
return isBoundary;
}
UBool RegexMatcher::isChunkWordBoundary(int32_t pos) {
UBool isBoundary = FALSE;
UBool cIsWord = FALSE;
const UChar *inputBuf = fInputText->chunkContents;
if (pos >= fLookLimit) {
fHitEnd = TRUE;
} else {
// Determine whether char c at current position is a member of the word set of chars.
// If we're off the end of the string, behave as though we're not at a word char.
UChar32 c;
U16_GET(inputBuf, fLookStart, pos, fLookLimit, c);
if (u_hasBinaryProperty(c, UCHAR_GRAPHEME_EXTEND) || u_charType(c) == U_FORMAT_CHAR) {
// Current char is a combining one. Not a boundary.
return FALSE;
}
cIsWord = fPattern->fStaticSets[URX_ISWORD_SET]->contains(c);
}
// Back up until we come to a non-combining char, determine whether
// that char is a word char.
UBool prevCIsWord = FALSE;
for (;;) {
if (pos <= fLookStart) {
break;
}
UChar32 prevChar;
U16_PREV(inputBuf, fLookStart, pos, prevChar);
if (!(u_hasBinaryProperty(prevChar, UCHAR_GRAPHEME_EXTEND)
|| u_charType(prevChar) == U_FORMAT_CHAR)) {
prevCIsWord = fPattern->fStaticSets[URX_ISWORD_SET]->contains(prevChar);
break;
}
}
isBoundary = cIsWord ^ prevCIsWord;
return isBoundary;
}
//--------------------------------------------------------------------------------
//
// isUWordBoundary
//
// Test for a word boundary using RBBI word break.
//
// parameters: pos - the current position in the input buffer
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::isUWordBoundary(int64_t pos) {
UBool returnVal = FALSE;
#if UCONFIG_NO_BREAK_ITERATION==0
// If we haven't yet created a break iterator for this matcher, do it now.
if (fWordBreakItr == NULL) {
fWordBreakItr =
(RuleBasedBreakIterator *)BreakIterator::createWordInstance(Locale::getEnglish(), fDeferredStatus);
if (U_FAILURE(fDeferredStatus)) {
return FALSE;
}
fWordBreakItr->setText(fInputText, fDeferredStatus);
}
if (pos >= fLookLimit) {
fHitEnd = TRUE;
returnVal = TRUE; // With Unicode word rules, only positions within the interior of "real"
// words are not boundaries. All non-word chars stand by themselves,
// with word boundaries on both sides.
} else {
if (!UTEXT_USES_U16(fInputText)) {
// !!!: Would like a better way to do this!
UErrorCode status = U_ZERO_ERROR;
pos = utext_extract(fInputText, 0, pos, NULL, 0, &status);
}
returnVal = fWordBreakItr->isBoundary((int32_t)pos);
}
#endif
return returnVal;
}
//--------------------------------------------------------------------------------
//
// IncrementTime This function is called once each TIMER_INITIAL_VALUE state
// saves. Increment the "time" counter, and call the
// user callback function if there is one installed.
//
// If the match operation needs to be aborted, either for a time-out
// or because the user callback asked for it, just set an error status.
// The engine will pick that up and stop in its outer loop.
//
//--------------------------------------------------------------------------------
void RegexMatcher::IncrementTime(UErrorCode &status) {
fTickCounter = TIMER_INITIAL_VALUE;
fTime++;
if (fCallbackFn != NULL) {
if ((*fCallbackFn)(fCallbackContext, fTime) == FALSE) {
status = U_REGEX_STOPPED_BY_CALLER;
return;
}
}
if (fTimeLimit > 0 && fTime >= fTimeLimit) {
status = U_REGEX_TIME_OUT;
}
}
//--------------------------------------------------------------------------------
//
// ReportFindProgress This function is called once for each advance in the target
// string from the find() function, and calls the user progress callback
// function if there is one installed.
//
// NOTE:
//
// If the match operation needs to be aborted because the user
// callback asked for it, just set an error status.
// The engine will pick that up and stop in its outer loop.
//
//--------------------------------------------------------------------------------
UBool RegexMatcher::ReportFindProgress(int64_t matchIndex, UErrorCode &status) {
if (fFindProgressCallbackFn != NULL) {
if ((*fFindProgressCallbackFn)(fFindProgressCallbackContext, matchIndex) == FALSE) {
status = U_ZERO_ERROR /*U_REGEX_STOPPED_BY_CALLER*/;
return FALSE;
}
}
return TRUE;
}
//--------------------------------------------------------------------------------
//
// StateSave
// Make a new stack frame, initialized as a copy of the current stack frame.
// Set the pattern index in the original stack frame from the operand value
// in the opcode. Execution of the engine continues with the state in
// the newly created stack frame
//
// Note that reserveBlock() may grow the stack, resulting in the
// whole thing being relocated in memory.
//
// Parameters:
// fp The top frame pointer when called. At return, a new
// fame will be present
// savePatIdx An index into the compiled pattern. Goes into the original
// (not new) frame. If execution ever back-tracks out of the
// new frame, this will be where we continue from in the pattern.
// Return
// The new frame pointer.
//
//--------------------------------------------------------------------------------
inline REStackFrame *RegexMatcher::StateSave(REStackFrame *fp, int64_t savePatIdx, UErrorCode &status) {
// push storage for a new frame.
int64_t *newFP = fStack->reserveBlock(fFrameSize, status);
if (newFP == NULL) {
// Failure on attempted stack expansion.
// Stack function set some other error code, change it to a more
// specific one for regular expressions.
status = U_REGEX_STACK_OVERFLOW;
// We need to return a writable stack frame, so just return the
// previous frame. The match operation will stop quickly
// because of the error status, after which the frame will never
// be looked at again.
return fp;
}
fp = (REStackFrame *)(newFP - fFrameSize); // in case of realloc of stack.
// New stack frame = copy of old top frame.
int64_t *source = (int64_t *)fp;
int64_t *dest = newFP;
for (;;) {
*dest++ = *source++;
if (source == newFP) {
break;
}
}
fTickCounter--;
if (fTickCounter <= 0) {
IncrementTime(status); // Re-initializes fTickCounter
}
fp->fPatIdx = savePatIdx;
return (REStackFrame *)newFP;
}
//--------------------------------------------------------------------------------
//
// MatchAt This is the actual matching engine.
//
// startIdx: begin matching a this index.
// toEnd: if true, match must extend to end of the input region
//
//--------------------------------------------------------------------------------
void RegexMatcher::MatchAt(int64_t startIdx, UBool toEnd, UErrorCode &status) {
UBool isMatch = FALSE; // True if the we have a match.
int64_t backSearchIndex = U_INT64_MAX; // used after greedy single-character matches for searching backwards
int32_t op; // Operation from the compiled pattern, split into
int32_t opType; // the opcode
int32_t opValue; // and the operand value.
#ifdef REGEX_RUN_DEBUG
if (fTraceDebug)
{
printf("MatchAt(startIdx=%ld)\n", startIdx);
printf("Original Pattern: ");
UChar32 c = utext_next32From(fPattern->fPattern, 0);
while (c != U_SENTINEL) {
if (