uspoof_impl.h | Explore in Territory

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
***************************************************************************
* Copyright (C) 2008-2013, International Business Machines Corporation
* and others. All Rights Reserved.
***************************************************************************
*
*  uspoof_impl.h
*
*    Implementation header for spoof detection
*
*/

#ifndef USPOOFIM_H
#define USPOOFIM_H

#include "uassert.h"
#include "unicode/utypes.h"
#include "unicode/uspoof.h"
#include "unicode/uscript.h"
#include "unicode/udata.h"
#include "udataswp.h"
#include "utrie2.h"

#if !UCONFIG_NO_NORMALIZATION

#ifdef __cplusplus

#include "capi_helper.h"
#include "umutex.h"

U_NAMESPACE_BEGIN

// The maximum length (in UTF-16 UChars) of the skeleton replacement string resulting from
//   a single input code point.  This is function of the unicode.org data.
#define USPOOF_MAX_SKELETON_EXPANSION …

// The default stack buffer size for copies or conversions or normalizations
// of input strings being checked.  (Used in multiple places.)
#define USPOOF_STACK_BUFFER_SIZE …

// Magic number for sanity checking spoof data.
#define USPOOF_MAGIC …

// Magic number for sanity checking spoof checkers.
#define USPOOF_CHECK_MAGIC …

class ScriptSet;
class SpoofData;
struct SpoofDataHeader;
class ConfusableDataUtils;

/**
  *  Class SpoofImpl corresponds directly to the plain C API opaque type
  *  USpoofChecker.  One can be cast to the other.
  */
class SpoofImpl : public UObject,
        public IcuCApiHelper<USpoofChecker, SpoofImpl, USPOOF_MAGIC> { … };

/**
 *  Class CheckResult corresponds directly to the plain C API opaque type
 *  USpoofCheckResult.  One can be cast to the other.
 */
class CheckResult : public UObject,
        public IcuCApiHelper<USpoofCheckResult, CheckResult, USPOOF_CHECK_MAGIC> { … };


//
//  Confusable Mappings Data Structures, version 2.0
//
//    For the confusable data, we are essentially implementing a map,
//       key:    a code point
//       value:  a string.  Most commonly one char in length, but can be more.
//
//    The keys are stored as a sorted array of 32 bit ints.
//             bits 0-23    a code point value
//             bits 24-31   length of value string, in UChars (between 1 and 256 UChars).
//        The key table is sorted in ascending code point order.  (not on the
//        32 bit int value, the flag bits do not participate in the sorting.)
//
//        Lookup is done by means of a binary search in the key table.
//
//    The corresponding values are kept in a parallel array of 16 bit ints.
//        If the value string is of length 1, it is literally in the value array.
//        For longer strings, the value array contains an index into the strings table.
//
//    String Table:
//       The strings table contains all of the value strings (those of length two or greater)
//       concatenated together into one long char16_t (UTF-16) array.
//
//       There is no nul character or other mark between adjacent strings.
//
//----------------------------------------------------------------------------
//
//  Changes from format version 1 to format version 2:
//      1) Removal of the whole-script confusable data tables.
//      2) Removal of the SL/SA/ML/MA and multi-table flags in the key bitmask.
//      3) Expansion of string length value in the key bitmask from 2 bits to 8 bits.
//      4) Removal of the string lengths table since 8 bits is sufficient for the
//         lengths of all entries in confusables.txt.



// Internal functions for manipulating confusable data table keys
#define USPOOF_CONFUSABLE_DATA_FORMAT_VERSION …
class ConfusableDataUtils { … };


//-------------------------------------------------------------------------------------
//
//  SpoofData
//
//    A small class that wraps the raw (usually memory mapped) spoof data.
//    Serves two primary functions:
//      1.  Convenience.  Contains real pointers to the data, to avoid dealing with
//          the offsets in the raw data.
//      2.  Reference counting.  When a spoof checker is cloned, the raw data is shared
//          and must be retained until all checkers using the data are closed.
//    Nothing in this struct includes state that is specific to any particular
//    USpoofDetector object.
//
//---------------------------------------------------------------------------------------
class SpoofData: public UMemory { … };

//---------------------------------------------------------------------------------------
//
//  Raw Binary Data Formats, as loaded from the ICU data file,
//    or as built by the builder.
//
//---------------------------------------------------------------------------------------
struct SpoofDataHeader { … };



U_NAMESPACE_END
#endif /* __cplusplus */

/**
  * Endianness swap function for binary spoof data.
  * @internal
  */
U_CAPI int32_t U_EXPORT2
uspoof_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData,
            UErrorCode *status);


#endif

#endif  /* USPOOFIM_H */
godot/thirdparty/icu4c/i18n/uspoof_impl.h
