chromium/third_party/icu/source/common/dictionarydata.h

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
* Copyright (C) 2014, International Business Machines
* Corporation and others.  All Rights Reserved.
*******************************************************************************
* dictionarydata.h
*
* created on: 2012may31
* created by: Markus W. Scherer & Maxime Serrano
*/

#ifndef __DICTIONARYDATA_H__
#define __DICTIONARYDATA_H__

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/utext.h"
#include "unicode/udata.h"
#include "udataswp.h"
#include "unicode/uobject.h"
#include "unicode/ustringtrie.h"

U_NAMESPACE_BEGIN

class UCharsTrie;
class BytesTrie;

class U_COMMON_API DictionaryData : public UMemory { … };

/**
 * Wrapper class around generic dictionaries, implementing matches().
 * getType() should return a TRIE_TYPE_??? constant from DictionaryData.
 * 
 * All implementations of this interface must be thread-safe if they are to be used inside of the
 * dictionary-based break iteration code.
 */
class U_COMMON_API DictionaryMatcher : public UMemory { … };

// Implementation of the DictionaryMatcher interface for a UCharsTrie dictionary
class U_COMMON_API UCharsDictionaryMatcher : public DictionaryMatcher { … };

// Implementation of the DictionaryMatcher interface for a BytesTrie dictionary
class U_COMMON_API BytesDictionaryMatcher : public DictionaryMatcher { … };

U_NAMESPACE_END

U_CAPI int32_t U_EXPORT2
udict_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData, UErrorCode *pErrorCode);

/**
 * Format of dictionary .dict data files.
 * Format version 1.0.
 *
 * A dictionary .dict data file contains a byte-serialized BytesTrie or
 * a UChars-serialized UCharsTrie.
 * Such files are used in dictionary-based break iteration (DBBI).
 *
 * For a BytesTrie, a transformation type is specified for
 * transforming Unicode strings into byte sequences.
 *
 * A .dict file begins with a standard ICU data file header
 * (DataHeader, see ucmndata.h and unicode/udata.h).
 * The UDataInfo.dataVersion field is currently unused (set to 0.0.0.0).
 *
 * After the header, the file contains the following parts.
 * Constants are defined in the DictionaryData class.
 *
 * For the data structure of BytesTrie & UCharsTrie see
 * https://icu.unicode.org/design/struct/tries
 * and the bytestrie.h and ucharstrie.h header files.
 *
 * int32_t indexes[indexesLength]; -- indexesLength=indexes[IX_STRING_TRIE_OFFSET]/4;
 *
 *      The first four indexes are byte offsets in ascending order.
 *      Each byte offset marks the start of the next part in the data file,
 *      and the end of the previous one.
 *      When two consecutive byte offsets are the same, then the corresponding part is empty.
 *      Byte offsets are offsets from after the header,
 *      that is, from the beginning of the indexes[].
 *      Each part starts at an offset with proper alignment for its data.
 *      If necessary, the previous part may include padding bytes to achieve this alignment.
 *
 *      trieType=indexes[IX_TRIE_TYPE] defines the trie type.
 *      transform=indexes[IX_TRANSFORM] defines the Unicode-to-bytes transformation.
 *          If the transformation type is TRANSFORM_TYPE_OFFSET,
 *          then the lower 21 bits contain the offset code point.
 *          Each code point c is mapped to byte b = (c - offset).
 *          Code points outside the range offset..(offset+0xff) cannot be mapped
 *          and do not occur in the dictionary.
 *
 * stringTrie; -- a serialized BytesTrie or UCharsTrie
 *
 *      The dictionary maps strings to specific values (TRIE_HAS_VALUES bit set in trieType),
 *      or it maps all strings to 0 (TRIE_HAS_VALUES bit not set).
 */

#endif  /* !UCONFIG_NO_BREAK_ITERATION */
#endif  /* __DICTIONARYDATA_H__ */