1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/propname.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,210 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (c) 2002-2011, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +* Author: Alan Liu
1.10 +* Created: October 30 2002
1.11 +* Since: ICU 2.4
1.12 +* 2010nov19 Markus Scherer Rewrite for formatVersion 2.
1.13 +**********************************************************************
1.14 +*/
1.15 +#ifndef PROPNAME_H
1.16 +#define PROPNAME_H
1.17 +
1.18 +#include "unicode/utypes.h"
1.19 +#include "unicode/bytestrie.h"
1.20 +#include "unicode/uchar.h"
1.21 +#include "udataswp.h"
1.22 +#include "uprops.h"
1.23 +
1.24 +/*
1.25 + * This header defines the in-memory layout of the property names data
1.26 + * structure representing the UCD data files PropertyAliases.txt and
1.27 + * PropertyValueAliases.txt. It is used by:
1.28 + * propname.cpp - reads data
1.29 + * genpname - creates data
1.30 + */
1.31 +
1.32 +/* low-level char * property name comparison -------------------------------- */
1.33 +
1.34 +U_CDECL_BEGIN
1.35 +
1.36 +/**
1.37 + * \var uprv_comparePropertyNames
1.38 + * Unicode property names and property value names are compared "loosely".
1.39 + *
1.40 + * UCD.html 4.0.1 says:
1.41 + * For all property names, property value names, and for property values for
1.42 + * Enumerated, Binary, or Catalog properties, use the following
1.43 + * loose matching rule:
1.44 + *
1.45 + * LM3. Ignore case, whitespace, underscore ('_'), and hyphens.
1.46 + *
1.47 + * This function does just that, for (char *) name strings.
1.48 + * It is almost identical to ucnv_compareNames() but also ignores
1.49 + * C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC).
1.50 + *
1.51 + * @internal
1.52 + */
1.53 +
1.54 +U_CAPI int32_t U_EXPORT2
1.55 +uprv_compareASCIIPropertyNames(const char *name1, const char *name2);
1.56 +
1.57 +U_CAPI int32_t U_EXPORT2
1.58 +uprv_compareEBCDICPropertyNames(const char *name1, const char *name2);
1.59 +
1.60 +#if U_CHARSET_FAMILY==U_ASCII_FAMILY
1.61 +# define uprv_comparePropertyNames uprv_compareASCIIPropertyNames
1.62 +#elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY
1.63 +# define uprv_comparePropertyNames uprv_compareEBCDICPropertyNames
1.64 +#else
1.65 +# error U_CHARSET_FAMILY is not valid
1.66 +#endif
1.67 +
1.68 +U_CDECL_END
1.69 +
1.70 +/* UDataMemory structure and signatures ------------------------------------- */
1.71 +
1.72 +#define PNAME_DATA_NAME "pnames"
1.73 +#define PNAME_DATA_TYPE "icu"
1.74 +
1.75 +/* Fields in UDataInfo: */
1.76 +
1.77 +/* PNAME_SIG[] is encoded as numeric literals for compatibility with the HP compiler */
1.78 +#define PNAME_SIG_0 ((uint8_t)0x70) /* p */
1.79 +#define PNAME_SIG_1 ((uint8_t)0x6E) /* n */
1.80 +#define PNAME_SIG_2 ((uint8_t)0x61) /* a */
1.81 +#define PNAME_SIG_3 ((uint8_t)0x6D) /* m */
1.82 +
1.83 +U_NAMESPACE_BEGIN
1.84 +
1.85 +class PropNameData {
1.86 +public:
1.87 + enum {
1.88 + // Byte offsets from the start of the data, after the generic header.
1.89 + IX_VALUE_MAPS_OFFSET,
1.90 + IX_BYTE_TRIES_OFFSET,
1.91 + IX_NAME_GROUPS_OFFSET,
1.92 + IX_RESERVED3_OFFSET,
1.93 + IX_RESERVED4_OFFSET,
1.94 + IX_TOTAL_SIZE,
1.95 +
1.96 + // Other values.
1.97 + IX_MAX_NAME_LENGTH,
1.98 + IX_RESERVED7,
1.99 + IX_COUNT
1.100 + };
1.101 +
1.102 + static const char *getPropertyName(int32_t property, int32_t nameChoice);
1.103 + static const char *getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice);
1.104 +
1.105 + static int32_t getPropertyEnum(const char *alias);
1.106 + static int32_t getPropertyValueEnum(int32_t property, const char *alias);
1.107 +
1.108 +private:
1.109 + static int32_t findProperty(int32_t property);
1.110 + static int32_t findPropertyValueNameGroup(int32_t valueMapIndex, int32_t value);
1.111 + static const char *getName(const char *nameGroup, int32_t nameIndex);
1.112 + static UBool containsName(BytesTrie &trie, const char *name);
1.113 +
1.114 + static int32_t getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias);
1.115 +
1.116 + static const int32_t indexes[];
1.117 + static const int32_t valueMaps[];
1.118 + static const uint8_t bytesTries[];
1.119 + static const char nameGroups[];
1.120 +};
1.121 +
1.122 +/*
1.123 + * pnames.icu formatVersion 2
1.124 + *
1.125 + * formatVersion 2 is new in ICU 4.8.
1.126 + * In ICU 4.8, the pnames.icu data file is used only in ICU4J.
1.127 + * ICU4C 4.8 has the same data structures hardcoded in source/common/propname_data.h.
1.128 + *
1.129 + * For documentation of pnames.icu formatVersion 1 see ICU4C 4.6 (2010-dec-01)
1.130 + * or earlier versions of this header file (source/common/propname.h).
1.131 + *
1.132 + * The pnames.icu begins with the standard ICU DataHeader/UDataInfo.
1.133 + * After that:
1.134 + *
1.135 + * int32_t indexes[8];
1.136 + *
1.137 + * (See the PropNameData::IX_... constants.)
1.138 + *
1.139 + * The first 6 indexes are byte offsets from the beginning of the data
1.140 + * (beginning of indexes[]) to following structures.
1.141 + * The length of each structure is the difference between its offset
1.142 + * and the next one.
1.143 + * All offsets are filled in: Where there is no data between two offsets,
1.144 + * those two offsets are the same.
1.145 + * The last offset (indexes[PropNameData::IX_TOTAL_SIZE]) indicates the
1.146 + * total number of bytes in the file. (Not counting the standard headers.)
1.147 + *
1.148 + * The sixth index (indexes[PropNameData::IX_MAX_NAME_LENGTH]) has the
1.149 + * maximum length of any Unicode property (or property value) alias.
1.150 + * (Without normalization, that is, including underscores etc.)
1.151 + *
1.152 + * int32_t valueMaps[];
1.153 + *
1.154 + * The valueMaps[] begins with a map from UProperty enums to properties,
1.155 + * followed by the per-property value maps from property values to names,
1.156 + * for those properties that have named values.
1.157 + * (Binary & enumerated, plus General_Category_Mask.)
1.158 + *
1.159 + * valueMaps[0] contains the number of UProperty enum ranges.
1.160 + * For each range:
1.161 + * int32_t start, limit -- first and last+1 UProperty enum of a dense range
1.162 + * Followed by (limit-start) pairs of
1.163 + * int32_t nameGroupOffset;
1.164 + * Offset into nameGroups[] for the property's names/aliases.
1.165 + * int32_t valueMapIndex;
1.166 + * Offset of the property's value map in the valueMaps[] array.
1.167 + * If the valueMapIndex is 0, then the property does not have named values.
1.168 + *
1.169 + * For each property's value map:
1.170 + * int32_t bytesTrieOffset; -- Offset into bytesTries[] for name->value mapping.
1.171 + * int32_t numRanges;
1.172 + * If numRanges is in the range 1..15, then that many ranges of values follow.
1.173 + * Per range:
1.174 + * int32_t start, limit -- first and last+1 UProperty enum of a range
1.175 + * Followed by (limit-start) entries of
1.176 + * int32_t nameGroupOffset;
1.177 + * Offset into nameGroups[] for the property value's names/aliases.
1.178 + * If the nameGroupOffset is 0, then this is not a named value for this property.
1.179 + * (That is, the ranges need not be dense.)
1.180 + * If numRanges is >=0x10, then (numRanges-0x10) sorted values
1.181 + * and then (numRanges-0x10) corresponding nameGroupOffsets follow.
1.182 + * Values are sorted as signed integers.
1.183 + * In this case, the set of values is dense; no nameGroupOffset will be 0.
1.184 + *
1.185 + * For both properties and property values, ranges are sorted by their start/limit values.
1.186 + *
1.187 + * uint8_t bytesTries[];
1.188 + *
1.189 + * This is a sequence of BytesTrie structures, byte-serialized tries for
1.190 + * mapping from names/aliases to values.
1.191 + * The first one maps from property names/aliases to UProperty enum constants.
1.192 + * The following ones are indexed by property value map bytesTrieOffsets
1.193 + * for mapping each property's names/aliases to their property values.
1.194 + *
1.195 + * char nameGroups[];
1.196 + *
1.197 + * This is a sequence of property name groups.
1.198 + * Each group is a list of names/aliases (invariant-character strings) for
1.199 + * one property or property value, in the order of UCharNameChoice.
1.200 + * The first byte of each group is the number of names in the group.
1.201 + * It is followed by that many NUL-terminated strings.
1.202 + * The first string is for the short name; if there is no short name,
1.203 + * then the first string is empty.
1.204 + * The second string is the long name. Further strings are additional aliases.
1.205 + *
1.206 + * The first name group is for a property rather than a property value,
1.207 + * so that a nameGroupOffset of 0 can be used to indicate "no value"
1.208 + * in a property's sparse value ranges.
1.209 + */
1.210 +
1.211 +U_NAMESPACE_END
1.212 +
1.213 +#endif
