icuSources/i18n/unicode/region.h

   1 /*
   2  *******************************************************************************
   3  * Copyright (C) 2014-2016, International Business Machines Corporation and others.
   4  * All Rights Reserved.
   5  *******************************************************************************
   6  */
   7
   8 #ifndef REGION_H
   9 #define REGION_H
  10
  11 /**
  12  * \file
  13  * \brief C++ API: Region classes (territory containment)
  14  */
  15
  16 #include "unicode/utypes.h"
  17 #include "unicode/uregion.h"
  18
  19 #if !UCONFIG_NO_FORMATTING
  20
  21 #include "unicode/uobject.h"
  22 #include "unicode/uniset.h"
  23 #include "unicode/unistr.h"
  24 #include "unicode/strenum.h"
  25
  26 U_NAMESPACE_BEGIN
  27
  28 /**
  29  * <code>Region</code> is the class representing a Unicode Region Code, also known as a
  30  * Unicode Region Subtag, which is defined based upon the BCP 47 standard. We often think of
  31  * "regions" as "countries" when defining the characteristics of a locale.  Region codes There are different
  32  * types of region codes that are important to distinguish.
  33  * <p>
  34  *  Macroregion - A code for a "macro geographical (continental) region, geographical sub-region, or
  35  *  selected economic and other grouping" as defined in
  36  *  UN M.49 (http://unstats.un.org/unsd/methods/m49/m49regin.htm).
  37  *  These are typically 3-digit codes, but contain some 2-letter codes, such as the LDML code QO
  38  *  added for Outlying Oceania.  Not all UNM.49 codes are defined in LDML, but most of them are.
  39  *  Macroregions are represented in ICU by one of three region types: WORLD ( region code 001 ),
  40  *  CONTINENTS ( regions contained directly by WORLD ), and SUBCONTINENTS ( things contained directly
  41  *  by a continent ).
  42  *  <p>
  43  *  TERRITORY - A Region that is not a Macroregion. These are typically codes for countries, but also
  44  *  include areas that are not separate countries, such as the code "AQ" for Antarctica or the code
  45  *  "HK" for Hong Kong (SAR China). Overseas dependencies of countries may or may not have separate
  46  *  codes. The codes are typically 2-letter codes aligned with the ISO 3166 standard, but BCP47 allows
  47  *  for the use of 3-digit codes in the future.
  48  *  <p>
  49  *  UNKNOWN - The code ZZ is defined by Unicode LDML for use to indicate that the Region is unknown,
  50  *  or that the value supplied as a region was invalid.
  51  *  <p>
  52  *  DEPRECATED - Region codes that have been defined in the past but are no longer in modern usage,
  53  *  usually due to a country splitting into multiple territories or changing its name.
  54  *  <p>
  55  *  GROUPING - A widely understood grouping of territories that has a well defined membership such
  56  *  that a region code has been assigned for it.  Some of these are UNM.49 codes that do't fall into
  57  *  the world/continent/sub-continent hierarchy, while others are just well known groupings that have
  58  *  their own region code. Region "EU" (European Union) is one such region code that is a grouping.
  59  *  Groupings will never be returned by the getContainingRegion() API, since a different type of region
  60  *  ( WORLD, CONTINENT, or SUBCONTINENT ) will always be the containing region instead.
  61  *
  62  * The Region class is not intended for public subclassing.
  63  *
  64  * @author       John Emmons
  65  * @stable ICU 51
  66  */
  67
  68 class U_I18N_API Region : public UObject {
  69 public:
  70     /**
  71      * Destructor.
  72      * @stable ICU 51
  73      */
  74     virtual ~Region();
  75
  76     /**
  77      * Returns true if the two regions are equal.
  78      * @stable ICU 51
  79      */
  80     UBool operator==(const Region &that) const;
  81
  82     /**
  83      * Returns true if the two regions are NOT equal; that is, if operator ==() returns false.
  84      * @stable ICU 51
  85      */
  86     UBool operator!=(const Region &that) const;
  87
  88     /**
  89      * Returns a pointer to a Region using the given region code.  The region code can be either 2-letter ISO code,
  90      * 3-letter ISO code,  UNM.49 numeric code, or other valid Unicode Region Code as defined by the LDML specification.
  91      * The identifier will be canonicalized internally using the supplemental metadata as defined in the CLDR.
  92      * If the region code is NULL or not recognized, the appropriate error code will be set ( U_ILLEGAL_ARGUMENT_ERROR )
  93      * @stable ICU 51
  94      */
  95     static const Region* U_EXPORT2 getInstance(const char *region_code, UErrorCode &status);
  96
  97     /**
  98      * Returns a pointer to a Region using the given numeric region code. If the numeric region code is not recognized,
  99      * the appropriate error code will be set ( U_ILLEGAL_ARGUMENT_ERROR ).
 100      * @stable ICU 51
 101      */
 102     static const Region* U_EXPORT2 getInstance (int32_t code, UErrorCode &status);
 103
 104     /**
 105      * Returns an enumeration over the IDs of all known regions that match the given type.
 106      * @stable ICU 55
 107      */
 108     static StringEnumeration* U_EXPORT2 getAvailable(URegionType type, UErrorCode &status);
 109
 110     /**
 111      * Returns a pointer to the region that contains this region.  Returns NULL if this region is code "001" (World)
 112      * or "ZZ" (Unknown region). For example, calling this method with region "IT" (Italy) returns the
 113      * region "039" (Southern Europe).
 114      * @stable ICU 51
 115      */
 116     const Region* getContainingRegion() const;
 117
 118     /**
 119      * Return a pointer to the region that geographically contains this region and matches the given type,
 120      * moving multiple steps up the containment chain if necessary.  Returns NULL if no containing region can be found
 121      * that matches the given type. Note: The URegionTypes = "URGN_GROUPING", "URGN_DEPRECATED", or "URGN_UNKNOWN"
 122      * are not appropriate for use in this API. NULL will be returned in this case. For example, calling this method
 123      * with region "IT" (Italy) for type "URGN_CONTINENT" returns the region "150" ( Europe ).
 124      * @stable ICU 51
 125      */
 126     const Region* getContainingRegion(URegionType type) const;
 127
 128     /**
 129      * Return an enumeration over the IDs of all the regions that are immediate children of this region in the
 130      * region hierarchy. These returned regions could be either macro regions, territories, or a mixture of the two,
 131      * depending on the containment data as defined in CLDR.  This API may return NULL if this region doesn't have
 132      * any sub-regions. For example, calling this method with region "150" (Europe) returns an enumeration containing
 133      * the various sub regions of Europe - "039" (Southern Europe) - "151" (Eastern Europe) - "154" (Northern Europe)
 134      * and "155" (Western Europe).
 135      * @stable ICU 55
 136      */
 137     StringEnumeration* getContainedRegions(UErrorCode &status) const;
 138
 139     /**
 140      * Returns an enumeration over the IDs of all the regions that are children of this region anywhere in the region
 141      * hierarchy and match the given type.  This API may return an empty enumeration if this region doesn't have any
 142      * sub-regions that match the given type. For example, calling this method with region "150" (Europe) and type
 143      * "URGN_TERRITORY" returns a set containing all the territories in Europe ( "FR" (France) - "IT" (Italy) - "DE" (Germany) etc. )
 144      * @stable ICU 55
 145      */
 146     StringEnumeration* getContainedRegions( URegionType type, UErrorCode &status ) const;
 147
 148     /**
 149      * Returns true if this region contains the supplied other region anywhere in the region hierarchy.
 150      * @stable ICU 51
 151      */
 152     UBool contains(const Region &other) const;
 153
 154     /**
 155      * For deprecated regions, return an enumeration over the IDs of the regions that are the preferred replacement
 156      * regions for this region.  Returns null for a non-deprecated region.  For example, calling this method with region
 157      * "SU" (Soviet Union) would return a list of the regions containing "RU" (Russia), "AM" (Armenia), "AZ" (Azerbaijan), etc...
 158      * @stable ICU 55
 159      */
 160     StringEnumeration* getPreferredValues(UErrorCode &status) const;
 161
 162     /**
 163      * Return this region's canonical region code.
 164      * @stable ICU 51
 165      */
 166     const char* getRegionCode() const;
 167
 168     /**
 169      * Return this region's numeric code.
 170      * Returns a negative value if the given region does not have a numeric code assigned to it.
 171      * @stable ICU 51
 172      */
 173     int32_t getNumericCode() const;
 174
 175     /**
 176      * Returns the region type of this region.
 177      * @stable ICU 51
 178      */
 179     URegionType getType() const;
 180
 181 #ifndef U_HIDE_INTERNAL_API
 182     /**
 183      * Cleans up statically allocated memory.
 184      * @internal
 185      */
 186     static void cleanupRegionData();
 187 #endif  /* U_HIDE_INTERNAL_API */
 188
 189 private:
 190     char id[4];
 191     UnicodeString idStr;
 192     int32_t code;
 193     URegionType type;
 194     Region *containingRegion;
 195     UVector *containedRegions;
 196     UVector *preferredValues;
 197
 198     /**
 199      * Default Constructor. Internal - use factory methods only.
 200      */
 201     Region();
 202
 203
 204     /*
 205      * Initializes the region data from the ICU resource bundles.  The region data
 206      * contains the basic relationships such as which regions are known, what the numeric
 207      * codes are, any known aliases, and the territory containment data.
 208      *
 209      * If the region data has already loaded, then this method simply returns without doing
 210      * anything meaningful.
 211      */
 212
 213     static void loadRegionData(UErrorCode &status);
 214
 215 };
 216
 217 U_NAMESPACE_END
 218
 219 #endif /* #if !UCONFIG_NO_FORMATTING */
 220 #endif // REGION_H
 221
 222 //eof