icuSources/common/propname.h

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