1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/identifier_info.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,204 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (C) 2013, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +*
1.10 +* indentifier_info.h
1.11 +*
1.12 +* created on: 2013 Jan 7
1.13 +* created by: Andy Heninger
1.14 +*/
1.15 +
1.16 +#ifndef __IDENTIFIER_INFO_H__
1.17 +#define __IDENTIFIER_INFO_H__
1.18 +
1.19 +#include "unicode/utypes.h"
1.20 +
1.21 +#include "unicode/uniset.h"
1.22 +#include "unicode/uspoof.h"
1.23 +#include "uhash.h"
1.24 +
1.25 +U_NAMESPACE_BEGIN
1.26 +
1.27 +class ScriptSet;
1.28 +
1.29 +// TODO(andy): review consistency of reference vs pointer arguments to the funcions.
1.30 +
1.31 +/**
1.32 + * This class analyzes a possible identifier for script and identifier status. Use it by calling setIdentifierProfile
1.33 + * then setIdentifier. Available methods include:
1.34 + * <ol>
1.35 + * <li>call getScripts for the specific scripts in the identifier. The identifier contains at least one character in
1.36 + * each of these.
1.37 + * <li>call getAlternates to get cases where a character is not limited to a single script. For example, it could be
1.38 + * either Katakana or Hiragana.
1.39 + * <li>call getCommonAmongAlternates to find out if any scripts are common to all the alternates.
1.40 + * <li>call getNumerics to get a representative character (with value zero) for each of the decimal number systems in
1.41 + * the identifier.
1.42 + * <li>call getRestrictionLevel to see what the UTS36 restriction level is.
1.43 + * </ol>
1.44 + *
1.45 + * This is a port from ICU4J of class com.ibm.icu.text.IdentifierInfo
1.46 + */
1.47 +class U_I18N_API IdentifierInfo : public UMemory {
1.48 +
1.49 + public:
1.50 + /**
1.51 + * Create an identifier info object. Subsequently, call setIdentifier(), etc.
1.52 + * @internal
1.53 + */
1.54 + IdentifierInfo(UErrorCode &status);
1.55 +
1.56 + /**
1.57 + * Destructor
1.58 + */
1.59 + virtual ~IdentifierInfo();
1.60 +
1.61 + private:
1.62 + /* Disallow copying for now. Can be added if there's a need. */
1.63 + IdentifierInfo(const IdentifierInfo &other);
1.64 +
1.65 + public:
1.66 +
1.67 + /**
1.68 + * Set the identifier profile: the characters that are to be allowed in the identifier.
1.69 + *
1.70 + * @param identifierProfile the characters that are to be allowed in the identifier
1.71 + * @return this
1.72 + * @internal
1.73 + */
1.74 + IdentifierInfo &setIdentifierProfile(const UnicodeSet &identifierProfile);
1.75 +
1.76 + /**
1.77 + * Get the identifier profile: the characters that are to be allowed in the identifier.
1.78 + *
1.79 + * @return The characters that are to be allowed in the identifier.
1.80 + * @internal
1.81 + */
1.82 + const UnicodeSet &getIdentifierProfile() const;
1.83 +
1.84 +
1.85 + /**
1.86 + * Set an identifier to analyze. Afterwards, call methods like getScripts()
1.87 + *
1.88 + * @param identifier the identifier to analyze
1.89 + * @param status Errorcode, set if errors occur.
1.90 + * @return this
1.91 + * @internal
1.92 + */
1.93 + IdentifierInfo &setIdentifier(const UnicodeString &identifier, UErrorCode &status);
1.94 +
1.95 +
1.96 + /**
1.97 + * Get the identifier that was analyzed. The returned string is owned by the ICU library,
1.98 + * and must not be deleted by the caller.
1.99 + *
1.100 + * @return the identifier that was analyzed.
1.101 + * @internal
1.102 + */
1.103 + const UnicodeString *getIdentifier() const;
1.104 +
1.105 +
1.106 + /**
1.107 + * Get the scripts found in the identifiers.
1.108 + *
1.109 + * @return the set of explicit scripts.
1.110 + * @internal
1.111 + */
1.112 + const ScriptSet *getScripts() const;
1.113 +
1.114 + /**
1.115 + * Get the set of alternate scripts found in the identifiers. That is, when a character can be in two scripts, then
1.116 + * the set consisting of those scripts will be returned.
1.117 + *
1.118 + * @return a uhash, with each key being of type (ScriptSet *).
1.119 + * This is a set, not a map, so the value stored in the uhash is not relevant.
1.120 + * (It is, in fact, 1).
1.121 + * Ownership of the uhash and its contents remains with the IndetifierInfo object,
1.122 + * and remains valid until a new identifer is set or until the object is deleted.
1.123 + * @internal
1.124 + */
1.125 + const UHashtable *getAlternates() const;
1.126 +
1.127 + /**
1.128 + * Get the representative characters (zeros) for the numerics found in the identifier.
1.129 + *
1.130 + * @return the set of explicit scripts.
1.131 + * @internal
1.132 + */
1.133 + const UnicodeSet *getNumerics() const;
1.134 +
1.135 + /**
1.136 + * Find out which scripts are in common among the alternates.
1.137 + *
1.138 + * @return the set of scripts that are in common among the alternates.
1.139 + * @internal
1.140 + */
1.141 + const ScriptSet *getCommonAmongAlternates() const;
1.142 +
1.143 + /**
1.144 + * Get the number of scripts appearing in the identifier.
1.145 + * Note: Common and Inherited scripts are omitted from the count.
1.146 + * Note: Result may be high when the identifier contains characters
1.147 + * with alternate scripts. The distinction between
1.148 + * 0, 1 and > 1 will remain valid, however.
1.149 + * @return the number of scripts.
1.150 + */
1.151 + int32_t getScriptCount() const;
1.152 +
1.153 +#if !UCONFIG_NO_NORMALIZATION
1.154 +
1.155 + /**
1.156 + * Find the "tightest" restriction level that the identifier satisfies.
1.157 + *
1.158 + * @return the restriction level.
1.159 + * @internal
1.160 + */
1.161 + URestrictionLevel getRestrictionLevel(UErrorCode &status) const;
1.162 +
1.163 +#endif /*!UCONFIG_NO_NORMALIZATION */
1.164 +
1.165 + UnicodeString toString() const;
1.166 +
1.167 + /**
1.168 + * Produce a readable string of alternates.
1.169 + *
1.170 + * @param alternates a UHashtable of UScriptSets.
1.171 + * Keys only, no meaningful values in the UHash.
1.172 + * @return display form
1.173 + * @internal
1.174 + */
1.175 + static UnicodeString &displayAlternates(UnicodeString &dest, const UHashtable *alternates, UErrorCode &status);
1.176 +
1.177 + /**
1.178 + * Static memory cleanup function.
1.179 + * @internal
1.180 + */
1.181 + static UBool cleanup();
1.182 + private:
1.183 +
1.184 + IdentifierInfo & clear();
1.185 + UBool containsWithAlternates(const ScriptSet &container, const ScriptSet &containee) const;
1.186 +
1.187 + UnicodeString *fIdentifier;
1.188 + ScriptSet *fRequiredScripts;
1.189 + UHashtable *fScriptSetSet;
1.190 + ScriptSet *fCommonAmongAlternates;
1.191 + UnicodeSet *fNumerics;
1.192 + UnicodeSet *fIdentifierProfile;
1.193 +
1.194 + static UnicodeSet *ASCII;
1.195 + static ScriptSet *JAPANESE;
1.196 + static ScriptSet *CHINESE;
1.197 + static ScriptSet *KOREAN;
1.198 + static ScriptSet *CONFUSABLE_WITH_LATIN;
1.199 +
1.200 +
1.201 +
1.202 +};
1.203 +
1.204 +U_NAMESPACE_END
1.205 +
1.206 +#endif // __IDENTIFIER_INFO_H__
1.207 +
