michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (C) 2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *
michael@0: * indentifier_info.h
michael@0: * 
michael@0: * created on: 2013 Jan 7
michael@0: * created by: Andy Heninger
michael@0: */
michael@0: 
michael@0: #ifndef __IDENTIFIER_INFO_H__
michael@0: #define __IDENTIFIER_INFO_H__
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: #include "unicode/uniset.h"
michael@0: #include "unicode/uspoof.h"
michael@0: #include "uhash.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: class ScriptSet;
michael@0: 
michael@0: // TODO(andy): review consistency of reference vs pointer arguments to the funcions.
michael@0: 
michael@0: /**
michael@0:  * This class analyzes a possible identifier for script and identifier status. Use it by calling setIdentifierProfile
michael@0:  * then setIdentifier. Available methods include:
michael@0:  * <ol>
michael@0:  * <li>call getScripts for the specific scripts in the identifier. The identifier contains at least one character in
michael@0:  * each of these.
michael@0:  * <li>call getAlternates to get cases where a character is not limited to a single script. For example, it could be
michael@0:  * either Katakana or Hiragana.
michael@0:  * <li>call getCommonAmongAlternates to find out if any scripts are common to all the alternates.
michael@0:  * <li>call getNumerics to get a representative character (with value zero) for each of the decimal number systems in
michael@0:  * the identifier.
michael@0:  * <li>call getRestrictionLevel to see what the UTS36 restriction level is.
michael@0:  * </ol>
michael@0:  * 
michael@0:  * This is a port from ICU4J of class com.ibm.icu.text.IdentifierInfo
michael@0:  */
michael@0: class U_I18N_API IdentifierInfo : public UMemory {
michael@0: 
michael@0:   public:
michael@0:     /**
michael@0:      * Create an identifier info object. Subsequently, call setIdentifier(), etc.
michael@0:      * @internal
michael@0:      */
michael@0:     IdentifierInfo(UErrorCode &status);
michael@0: 
michael@0:     /**
michael@0:       * Destructor
michael@0:       */
michael@0:     virtual ~IdentifierInfo();
michael@0: 
michael@0:   private:
michael@0:     /* Disallow copying for now. Can be added if there's a need. */
michael@0:     IdentifierInfo(const IdentifierInfo &other);
michael@0: 
michael@0:   public:
michael@0:      
michael@0:     /**
michael@0:      * Set the identifier profile: the characters that are to be allowed in the identifier.
michael@0:      * 
michael@0:      * @param identifierProfile the characters that are to be allowed in the identifier
michael@0:      * @return this
michael@0:      * @internal
michael@0:      */
michael@0:     IdentifierInfo &setIdentifierProfile(const UnicodeSet &identifierProfile);
michael@0: 
michael@0:     /**
michael@0:      * Get the identifier profile: the characters that are to be allowed in the identifier.
michael@0:      * 
michael@0:      * @return The characters that are to be allowed in the identifier.
michael@0:      * @internal
michael@0:      */
michael@0:     const UnicodeSet &getIdentifierProfile() const;
michael@0: 
michael@0: 
michael@0:     /**
michael@0:      * Set an identifier to analyze. Afterwards, call methods like getScripts()
michael@0:      * 
michael@0:      * @param identifier the identifier to analyze
michael@0:      * @param status Errorcode, set if errors occur.
michael@0:      * @return this
michael@0:      * @internal
michael@0:      */
michael@0:     IdentifierInfo &setIdentifier(const UnicodeString &identifier, UErrorCode &status);
michael@0: 
michael@0: 
michael@0:     /**
michael@0:      * Get the identifier that was analyzed. The returned string is owned by the ICU library,
michael@0:      * and must not be deleted by the caller.
michael@0:      * 
michael@0:      * @return the identifier that was analyzed.
michael@0:      * @internal
michael@0:      */
michael@0:     const UnicodeString *getIdentifier() const;
michael@0:     
michael@0: 
michael@0:     /**
michael@0:      * Get the scripts found in the identifiers.
michael@0:      * 
michael@0:      * @return the set of explicit scripts.
michael@0:      * @internal
michael@0:      */
michael@0:     const ScriptSet *getScripts() const;
michael@0: 
michael@0:     /**
michael@0:      * Get the set of alternate scripts found in the identifiers. That is, when a character can be in two scripts, then
michael@0:      * the set consisting of those scripts will be returned.
michael@0:      * 
michael@0:      * @return a uhash, with each key being of type (ScriptSet *). 
michael@0:      *         This is a set, not a map, so the value stored in the uhash is not relevant.
michael@0:      *         (It is, in fact, 1).
michael@0:      *         Ownership of the uhash and its contents remains with the IndetifierInfo object, 
michael@0:      *         and remains valid until a new identifer is set or until the object is deleted.
michael@0:      * @internal
michael@0:      */
michael@0:     const UHashtable *getAlternates() const;
michael@0: 
michael@0:     /**
michael@0:      * Get the representative characters (zeros) for the numerics found in the identifier.
michael@0:      * 
michael@0:      * @return the set of explicit scripts.
michael@0:      * @internal
michael@0:      */
michael@0:     const UnicodeSet *getNumerics() const;
michael@0: 
michael@0:     /**
michael@0:      * Find out which scripts are in common among the alternates.
michael@0:      * 
michael@0:      * @return the set of scripts that are in common among the alternates.
michael@0:      * @internal
michael@0:      */
michael@0:     const ScriptSet *getCommonAmongAlternates() const;
michael@0: 
michael@0:     /**
michael@0:       * Get the number of scripts appearing in the identifier.
michael@0:       *   Note: Common and Inherited scripts are omitted from the count.
michael@0:       *   Note: Result may be high when the identifier contains characters
michael@0:       *         with alternate scripts. The distinction between
michael@0:       *         0, 1 and > 1 will remain valid, however.
michael@0:       * @return the number of scripts.
michael@0:       */
michael@0:     int32_t getScriptCount() const;
michael@0: 
michael@0: #if !UCONFIG_NO_NORMALIZATION
michael@0: 
michael@0:     /**
michael@0:      * Find the "tightest" restriction level that the identifier satisfies.
michael@0:      * 
michael@0:      * @return the restriction level.
michael@0:      * @internal
michael@0:      */
michael@0:     URestrictionLevel getRestrictionLevel(UErrorCode &status) const;
michael@0: 
michael@0: #endif /*!UCONFIG_NO_NORMALIZATION */
michael@0: 
michael@0:     UnicodeString toString() const;
michael@0: 
michael@0:     /**
michael@0:      * Produce a readable string of alternates.
michael@0:      * 
michael@0:      * @param alternates a UHashtable of UScriptSets.
michael@0:      *        Keys only, no meaningful values in the UHash.
michael@0:      * @return display form
michael@0:      * @internal
michael@0:      */
michael@0:     static UnicodeString &displayAlternates(UnicodeString &dest, const UHashtable *alternates, UErrorCode &status);
michael@0: 
michael@0:     /**
michael@0:      * Static memory cleanup function.
michael@0:      * @internal
michael@0:      */
michael@0:     static UBool      cleanup();
michael@0:   private:
michael@0: 
michael@0:     IdentifierInfo  & clear();
michael@0:     UBool             containsWithAlternates(const ScriptSet &container, const ScriptSet &containee) const;
michael@0: 
michael@0:     UnicodeString     *fIdentifier;
michael@0:     ScriptSet         *fRequiredScripts;
michael@0:     UHashtable        *fScriptSetSet;
michael@0:     ScriptSet         *fCommonAmongAlternates;
michael@0:     UnicodeSet        *fNumerics;
michael@0:     UnicodeSet        *fIdentifierProfile;
michael@0: 
michael@0:     static UnicodeSet *ASCII;
michael@0:     static ScriptSet  *JAPANESE;
michael@0:     static ScriptSet  *CHINESE;
michael@0:     static ScriptSet  *KOREAN;
michael@0:     static ScriptSet  *CONFUSABLE_WITH_LATIN;
michael@0: 
michael@0: 
michael@0: 
michael@0: };
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: #endif // __IDENTIFIER_INFO_H__
michael@0: