uspoof_conf.cpp | Explore in Territory

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
******************************************************************************
*
*   Copyright (C) 2008-2015, International Business Machines
*   Corporation and others.  All Rights Reserved.
*
******************************************************************************
*   file name:  uspoof_conf.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2009Jan05  (refactoring earlier files)
*   created by: Andy Heninger
*
*   Internal classes for compiling confusable data into its binary (runtime) form.
*/

#include "unicode/utypes.h"
#include "unicode/uspoof.h"
#if !UCONFIG_NO_REGULAR_EXPRESSIONS
#if !UCONFIG_NO_NORMALIZATION

#include "unicode/unorm.h"
#include "unicode/uregex.h"
#include "unicode/ustring.h"
#include "cmemory.h"
#include "uspoof_impl.h"
#include "uhash.h"
#include "uvector.h"
#include "uassert.h"
#include "uarrsort.h"
#include "uspoof_conf.h"

U_NAMESPACE_USE


//---------------------------------------------------------------------
//
//  buildConfusableData   Compile the source confusable data, as defined by
//                        the Unicode data file confusables.txt, into the binary
//                        structures used by the confusable detector.
//
//                        The binary structures are described in uspoof_impl.h
//
//     1.  Parse the data, making a hash table mapping from a UChar32 to a String.
//
//     2.  Sort all of the strings encountered by length, since they will need to
//         be stored in that order in the final string table.
//         TODO: Sorting these strings by length is no longer needed since the removal of
//         the string lengths table.  This logic can be removed to save processing time
//         when building confusables data.
//
//     3.  Build a list of keys (UChar32s) from the four mapping tables.  Sort the
//         list because that will be the ordering of our runtime table.
//
//     4.  Generate the run time string table.  This is generated before the key & value
//         tables because we need the string indexes when building those tables.
//
//     5.  Build the run-time key and value tables.  These are parallel tables, and are built
//         at the same time
//

SPUString::SPUString(LocalPointer<UnicodeString> s) { … }


SPUString::~SPUString() { … }


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


SPUStringPool::~SPUStringPool() { … }


int32_t SPUStringPool::size() { … }

SPUString *SPUStringPool::getByIndex(int32_t index) { … }


// Comparison function for ordering strings in the string pool.
// Compare by length first, then, within a group of the same length,
// by code point order.
// Conforms to the type signature for a USortComparator in uvector.h

static int32_t U_CALLCONV SPUStringCompare(UHashTok left, UHashTok right) { … }

void SPUStringPool::sort(UErrorCode &status) { … }


SPUString *SPUStringPool::addString(UnicodeString *src, UErrorCode &status) { … }



ConfusabledataBuilder::ConfusabledataBuilder(SpoofImpl *spImpl, UErrorCode &status) : … { … }


ConfusabledataBuilder::~ConfusabledataBuilder() { … }


void ConfusabledataBuilder::buildConfusableData(SpoofImpl * spImpl, const char * confusables,
    int32_t confusablesLen, int32_t *errorType, UParseError *pe, UErrorCode &status) { … }


void ConfusabledataBuilder::build(const char * confusables, int32_t confusablesLen,
               UErrorCode &status) { … }

//
// outputData     The confusable data has been compiled and stored in intermediate
//                collections and strings.  Copy it from there to the final flat
//                binary array.
//
//                Note that as each section is added to the output data, the
//                expand (reserveSpace() function will likely relocate it in memory.
//                Be careful with pointers.
//
void ConfusabledataBuilder::outputData(UErrorCode &status) { … }

#endif
#endif // !UCONFIG_NO_REGULAR_EXPRESSIONS
chromium/third_party/icu/source/i18n/uspoof_conf.cpp
