icuSources/common/propname.h

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 **********************************************************************
   5 * Copyright (c) 2002-2011, International Business Machines
   6 * Corporation and others.  All Rights Reserved.
   7 **********************************************************************
   8 * Author: Alan Liu
   9 * Created: October 30 2002
  10 * Since: ICU 2.4
  11 * 2010nov19 Markus Scherer  Rewrite for formatVersion 2.
  12 **********************************************************************
  13 */
  14 #ifndef PROPNAME_H
  15 #define PROPNAME_H
  16
  17 #include "unicode/utypes.h"
  18 #include "unicode/bytestrie.h"
  19 #include "unicode/uchar.h"
  20 #include "udataswp.h"
  21 #include "uprops.h"
  22
  23 /*
  24  * This header defines the in-memory layout of the property names data
  25  * structure representing the UCD data files PropertyAliases.txt and
  26  * PropertyValueAliases.txt.  It is used by:
  27  *   propname.cpp - reads data
  28  *   genpname     - creates data
  29  */
  30
  31 /* low-level char * property name comparison -------------------------------- */
  32
  33 U_CDECL_BEGIN
  34
  35 /**
  36  * \var uprv_comparePropertyNames
  37  * Unicode property names and property value names are compared "loosely".
  38  *
  39  * UCD.html 4.0.1 says:
  40  *   For all property names, property value names, and for property values for
  41  *   Enumerated, Binary, or Catalog properties, use the following
  42  *   loose matching rule:
  43  *
  44  *   LM3. Ignore case, whitespace, underscore ('_'), and hyphens.
  45  *
  46  * This function does just that, for (char *) name strings.
  47  * It is almost identical to ucnv_compareNames() but also ignores
  48  * C0 White_Space characters (U+0009..U+000d, and U+0085 on EBCDIC).
  49  *
  50  * @internal
  51  */
  52
  53 U_CAPI int32_t U_EXPORT2
  54 uprv_compareASCIIPropertyNames(const char *name1, const char *name2);
  55
  56 U_CAPI int32_t U_EXPORT2
  57 uprv_compareEBCDICPropertyNames(const char *name1, const char *name2);
  58
  59 #if U_CHARSET_FAMILY==U_ASCII_FAMILY
  60 #   define uprv_comparePropertyNames uprv_compareASCIIPropertyNames
  61 #elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY
  62 #   define uprv_comparePropertyNames uprv_compareEBCDICPropertyNames
  63 #else
  64 #   error U_CHARSET_FAMILY is not valid
  65 #endif
  66
  67 U_CDECL_END
  68
  69 /* UDataMemory structure and signatures ------------------------------------- */
  70
  71 #define PNAME_DATA_NAME "pnames"
  72 #define PNAME_DATA_TYPE "icu"
  73
  74 /* Fields in UDataInfo: */
  75
  76 /* PNAME_SIG[] is encoded as numeric literals for compatibility with the HP compiler */
  77 #define PNAME_SIG_0 ((uint8_t)0x70) /* p */
  78 #define PNAME_SIG_1 ((uint8_t)0x6E) /* n */
  79 #define PNAME_SIG_2 ((uint8_t)0x61) /* a */
  80 #define PNAME_SIG_3 ((uint8_t)0x6D) /* m */
  81
  82 U_NAMESPACE_BEGIN
  83
  84 class PropNameData {
  85 public:
  86     enum {
  87         // Byte offsets from the start of the data, after the generic header.
  88         IX_VALUE_MAPS_OFFSET,
  89         IX_BYTE_TRIES_OFFSET,
  90         IX_NAME_GROUPS_OFFSET,
  91         IX_RESERVED3_OFFSET,
  92         IX_RESERVED4_OFFSET,
  93         IX_TOTAL_SIZE,
  94
  95         // Other values.
  96         IX_MAX_NAME_LENGTH,
  97         IX_RESERVED7,
  98         IX_COUNT
  99     };
 100
 101     static const char *getPropertyName(int32_t property, int32_t nameChoice);
 102     static const char *getPropertyValueName(int32_t property, int32_t value, int32_t nameChoice);
 103
 104     static int32_t getPropertyEnum(const char *alias);
 105     static int32_t getPropertyValueEnum(int32_t property, const char *alias);
 106
 107 private:
 108     static int32_t findProperty(int32_t property);
 109     static int32_t findPropertyValueNameGroup(int32_t valueMapIndex, int32_t value);
 110     static const char *getName(const char *nameGroup, int32_t nameIndex);
 111     static UBool containsName(BytesTrie &trie, const char *name);
 112
 113     static int32_t getPropertyOrValueEnum(int32_t bytesTrieOffset, const char *alias);
 114
 115     static const int32_t indexes[];
 116     static const int32_t valueMaps[];
 117     static const uint8_t bytesTries[];
 118     static const char nameGroups[];
 119 };
 120
 121 /*
 122  * pnames.icu formatVersion 2
 123  *
 124  * formatVersion 2 is new in ICU 4.8.
 125  * In ICU 4.8, the pnames.icu data file is used only in ICU4J.
 126  * ICU4C 4.8 has the same data structures hardcoded in source/common/propname_data.h.
 127  *
 128  * For documentation of pnames.icu formatVersion 1 see ICU4C 4.6 (2010-dec-01)
 129  * or earlier versions of this header file (source/common/propname.h).
 130  *
 131  * The pnames.icu begins with the standard ICU DataHeader/UDataInfo.
 132  * After that:
 133  *
 134  * int32_t indexes[8];
 135  *
 136  *      (See the PropNameData::IX_... constants.)
 137  *
 138  *      The first 6 indexes are byte offsets from the beginning of the data
 139  *      (beginning of indexes[]) to following structures.
 140  *      The length of each structure is the difference between its offset
 141  *      and the next one.
 142  *      All offsets are filled in: Where there is no data between two offsets,
 143  *      those two offsets are the same.
 144  *      The last offset (indexes[PropNameData::IX_TOTAL_SIZE]) indicates the
 145  *      total number of bytes in the file. (Not counting the standard headers.)
 146  *
 147  *      The sixth index (indexes[PropNameData::IX_MAX_NAME_LENGTH]) has the
 148  *      maximum length of any Unicode property (or property value) alias.
 149  *      (Without normalization, that is, including underscores etc.)
 150  *
 151  * int32_t valueMaps[];
 152  *
 153  *      The valueMaps[] begins with a map from UProperty enums to properties,
 154  *      followed by the per-property value maps from property values to names,
 155  *      for those properties that have named values.
 156  *      (Binary & enumerated, plus General_Category_Mask.)
 157  *
 158  *      valueMaps[0] contains the number of UProperty enum ranges.
 159  *      For each range:
 160  *        int32_t start, limit -- first and last+1 UProperty enum of a dense range
 161  *        Followed by (limit-start) pairs of
 162  *          int32_t nameGroupOffset;
 163  *            Offset into nameGroups[] for the property's names/aliases.
 164  *          int32_t valueMapIndex;
 165  *            Offset of the property's value map in the valueMaps[] array.
 166  *            If the valueMapIndex is 0, then the property does not have named values.
 167  *
 168  *      For each property's value map:
 169  *      int32_t bytesTrieOffset; -- Offset into bytesTries[] for name->value mapping.
 170  *      int32_t numRanges;
 171  *        If numRanges is in the range 1..15, then that many ranges of values follow.
 172  *        Per range:
 173  *          int32_t start, limit -- first and last+1 UProperty enum of a range
 174  *          Followed by (limit-start) entries of
 175  *            int32_t nameGroupOffset;
 176  *              Offset into nameGroups[] for the property value's names/aliases.
 177  *              If the nameGroupOffset is 0, then this is not a named value for this property.
 178  *              (That is, the ranges need not be dense.)
 179  *        If numRanges is >=0x10, then (numRanges-0x10) sorted values
 180  *        and then (numRanges-0x10) corresponding nameGroupOffsets follow.
 181  *        Values are sorted as signed integers.
 182  *        In this case, the set of values is dense; no nameGroupOffset will be 0.
 183  *
 184  *      For both properties and property values, ranges are sorted by their start/limit values.
 185  *
 186  * uint8_t bytesTries[];
 187  *
 188  *      This is a sequence of BytesTrie structures, byte-serialized tries for
 189  *      mapping from names/aliases to values.
 190  *      The first one maps from property names/aliases to UProperty enum constants.
 191  *      The following ones are indexed by property value map bytesTrieOffsets
 192  *      for mapping each property's names/aliases to their property values.
 193  *
 194  * char nameGroups[];
 195  *
 196  *      This is a sequence of property name groups.
 197  *      Each group is a list of names/aliases (invariant-character strings) for
 198  *      one property or property value, in the order of UCharNameChoice.
 199  *      The first byte of each group is the number of names in the group.
 200  *      It is followed by that many NUL-terminated strings.
 201  *      The first string is for the short name; if there is no short name,
 202  *      then the first string is empty.
 203  *      The second string is the long name. Further strings are additional aliases.
 204  *
 205  *      The first name group is for a property rather than a property value,
 206  *      so that a nameGroupOffset of 0 can be used to indicate "no value"
 207  *      in a property's sparse value ranges.
 208  */
 209
 210 U_NAMESPACE_END
 211
 212 #endif