chromium/third_party/icu/source/common/filteredbrk.cpp

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
* Copyright (C) 2014-2015, International Business Machines Corporation and
* others. All Rights Reserved.
*******************************************************************************
*/

#include "unicode/utypes.h"
#if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION

#include "cmemory.h"

#include "unicode/filteredbrk.h"
#include "unicode/ucharstriebuilder.h"
#include "unicode/ures.h"

#include "uresimp.h" // ures_getByKeyWithFallback
#include "ubrkimpl.h" // U_ICUDATA_BRKITR
#include "uvector.h"
#include "cmemory.h"
#include "umutex.h"

U_NAMESPACE_BEGIN

#ifndef FB_DEBUG
#define FB_DEBUG …
#endif

#if FB_DEBUG
#include <stdio.h>
static void _fb_trace(const char *m, const UnicodeString *s, UBool b, int32_t d, const char *f, int l) {
  char buf[2048];
  if(s) {
    s->extract(0,s->length(),buf,2048);
  } else {
    strcpy(buf,"nullptr");
  }
  fprintf(stderr,"%s:%d: %s. s='%s'(%p), b=%c, d=%d\n",
          f, l, m, buf, (const void*)s, b?'T':'F',(int)d);
}

#define FB_TRACE …
#else
#define FB_TRACE(m,s,b,d) …
#endif

/**
 * Used with sortedInsert()
 */
static int32_t U_CALLCONV compareUnicodeString(UElement t1, UElement t2) { … }

/**
 * A UVector which implements a set of strings.
 */
class UStringSet : public UVector { … };

/**
 * Virtual, won't be inlined
 */
UStringSet::~UStringSet() { … }

/* ----------------------------------------------------------- */


/* Filtered Break constants */
static const int32_t kPARTIAL = …; //< partial - need to run through forward trie
static const int32_t kMATCH   = …; //< exact match - skip this one.
static const int32_t kSuppressInReverse = …;
static const int32_t kAddToForward = …;
static const char16_t kFULLSTOP = …; // '.'

/**
 * Shared data for SimpleFilteredSentenceBreakIterator
 */
class SimpleFilteredSentenceBreakData : public UMemory { … };

SimpleFilteredSentenceBreakData::~SimpleFilteredSentenceBreakData() { … }

/**
 * Concrete implementation
 */
class SimpleFilteredSentenceBreakIterator : public BreakIterator { … };

SimpleFilteredSentenceBreakIterator::SimpleFilteredSentenceBreakIterator(const SimpleFilteredSentenceBreakIterator& other)
  : … { … }


SimpleFilteredSentenceBreakIterator::SimpleFilteredSentenceBreakIterator(BreakIterator *adopt, UCharsTrie *forwards, UCharsTrie *backwards, UErrorCode &status) : … { … }

SimpleFilteredSentenceBreakIterator::~SimpleFilteredSentenceBreakIterator() { … }

void SimpleFilteredSentenceBreakIterator::resetState(UErrorCode &status) { … }

SimpleFilteredSentenceBreakIterator::EFBMatchResult
SimpleFilteredSentenceBreakIterator::breakExceptionAt(int32_t n) { … }

// the workhorse single next.
int32_t
SimpleFilteredSentenceBreakIterator::internalNext(int32_t n) { … }

int32_t
SimpleFilteredSentenceBreakIterator::internalPrev(int32_t n) { … }


int32_t
SimpleFilteredSentenceBreakIterator::next() { … }

int32_t
SimpleFilteredSentenceBreakIterator::first() { … }

int32_t
SimpleFilteredSentenceBreakIterator::preceding(int32_t offset) { … }

int32_t
SimpleFilteredSentenceBreakIterator::previous() { … }

UBool SimpleFilteredSentenceBreakIterator::isBoundary(int32_t offset) { … }
 
int32_t
SimpleFilteredSentenceBreakIterator::next(int32_t offset) { … }

int32_t
SimpleFilteredSentenceBreakIterator::following(int32_t offset) { … }

int32_t
SimpleFilteredSentenceBreakIterator::last() { … }


/**
 * Concrete implementation of builder class.
 */
class SimpleFilteredBreakIteratorBuilder : public FilteredBreakIteratorBuilder { … };

SimpleFilteredBreakIteratorBuilder::~SimpleFilteredBreakIteratorBuilder()
{ … }

SimpleFilteredBreakIteratorBuilder::SimpleFilteredBreakIteratorBuilder(UErrorCode &status) 
  : … { … }

SimpleFilteredBreakIteratorBuilder::SimpleFilteredBreakIteratorBuilder(const Locale &fromLocale, UErrorCode &status)
  : … { … }

UBool
SimpleFilteredBreakIteratorBuilder::suppressBreakAfter(const UnicodeString& exception, UErrorCode& status)
{ … }

UBool
SimpleFilteredBreakIteratorBuilder::unsuppressBreakAfter(const UnicodeString& exception, UErrorCode& status)
{ … }

/**
 * Jitterbug 2974: MSVC has a bug whereby new X[0] behaves badly.
 * Work around this.
 *
 * Note: "new UnicodeString[subCount]" ends up calling global operator new
 * on MSVC2012 for some reason.
 */
static inline UnicodeString* newUnicodeStringArray(size_t count) { … }

BreakIterator *
SimpleFilteredBreakIteratorBuilder::build(BreakIterator* adoptBreakIterator, UErrorCode& status) { … }


// ----------- Base class implementation

FilteredBreakIteratorBuilder::FilteredBreakIteratorBuilder() { … }

FilteredBreakIteratorBuilder::~FilteredBreakIteratorBuilder() { … }

FilteredBreakIteratorBuilder *
FilteredBreakIteratorBuilder::createInstance(const Locale& where, UErrorCode& status) { … }

FilteredBreakIteratorBuilder *
FilteredBreakIteratorBuilder::createInstance(UErrorCode &status) { … }

FilteredBreakIteratorBuilder *
FilteredBreakIteratorBuilder::createEmptyInstance(UErrorCode& status) { … }

U_NAMESPACE_END

#endif //#if !UCONFIG_NO_BREAK_ITERATION && !UCONFIG_NO_FILTERED_BREAK_ITERATION