Home | History | Annotate | Download | only in unicode
      1 /*
      2 ***************************************************************************
      3 *   Copyright (C) 1999-2011 International Business Machines Corporation   *
      4 *   and others. All rights reserved.                                      *
      5 ***************************************************************************
      6 
      7 **********************************************************************
      8 *   Date        Name        Description
      9 *   10/22/99    alan        Creation.
     10 *   11/11/99    rgillam     Complete port from Java.
     11 **********************************************************************
     12 */
     13 
     14 #ifndef RBBI_H
     15 #define RBBI_H
     16 
     17 #include "unicode/utypes.h"
     18 
     19 /**
     20  * \file
     21  * \brief C++ API: Rule Based Break Iterator
     22  */
     23 
     24 #if !UCONFIG_NO_BREAK_ITERATION
     25 
     26 #include "unicode/brkiter.h"
     27 #include "unicode/udata.h"
     28 #include "unicode/parseerr.h"
     29 #include "unicode/schriter.h"
     30 #include "unicode/uchriter.h"
     31 
     32 
     33 struct UTrie;
     34 
     35 U_NAMESPACE_BEGIN
     36 
     37 /** @internal */
     38 struct RBBIDataHeader;
     39 class  RuleBasedBreakIteratorTables;
     40 class  BreakIterator;
     41 class  RBBIDataWrapper;
     42 class  UStack;
     43 class  LanguageBreakEngine;
     44 class  UnhandledEngine;
     45 struct RBBIStateTable;
     46 
     47 
     48 
     49 
     50 /**
     51  *
     52  * A subclass of BreakIterator whose behavior is specified using a list of rules.
     53  * <p>Instances of this class are most commonly created by the factory methods of
     54  *  BreakIterator::createWordInstance(), BreakIterator::createLineInstance(), etc.,
     55  *  and then used via the abstract API in class BreakIterator</p>
     56  *
     57  * <p>See the ICU User Guide for information on Break Iterator Rules.</p>
     58  *
     59  * <p>This class is not intended to be subclassed.  (Class DictionaryBasedBreakIterator
     60  *    is a subclass, but that relationship is effectively internal to the ICU
     61  *    implementation.  The subclassing interface to RulesBasedBreakIterator is
     62  *    not part of the ICU API, and may not remain stable.</p>
     63  *
     64  */
     65 class U_COMMON_API RuleBasedBreakIterator : public BreakIterator {
     66 
     67 protected:
     68     /**
     69      * The UText through which this BreakIterator accesses the text
     70      * @internal
     71      */
     72     UText  *fText;
     73 
     74     /**
     75      *   A character iterator that refers to the same text as the UText, above.
     76      *   Only included for compatibility with old API, which was based on CharacterIterators.
     77      *   Value may be adopted from outside, or one of fSCharIter or fDCharIter, below.
     78      */
     79     CharacterIterator  *fCharIter;
     80 
     81     /**
     82      *   When the input text is provided by a UnicodeString, this will point to
     83      *    a characterIterator that wraps that data.  Needed only for the
     84      *    implementation of getText(), a backwards compatibility issue.
     85      */
     86     StringCharacterIterator *fSCharIter;
     87 
     88     /**
     89      *  When the input text is provided by a UText, this
     90      *    dummy CharacterIterator over an empty string will
     91      *    be returned from getText()
     92      */
     93     UCharCharacterIterator *fDCharIter;
     94 
     95     /**
     96      * The rule data for this BreakIterator instance
     97      * @internal
     98      */
     99     RBBIDataWrapper    *fData;
    100 
    101     /** Index of the Rule {tag} values for the most recent match.
    102      *  @internal
    103     */
    104     int32_t             fLastRuleStatusIndex;
    105 
    106     /**
    107      * Rule tag value valid flag.
    108      * Some iterator operations don't intrinsically set the correct tag value.
    109      * This flag lets us lazily compute the value if we are ever asked for it.
    110      * @internal
    111      */
    112     UBool               fLastStatusIndexValid;
    113 
    114     /**
    115      * Counter for the number of characters encountered with the "dictionary"
    116      *   flag set.
    117      * @internal
    118      */
    119     uint32_t            fDictionaryCharCount;
    120 
    121     /**
    122      * When a range of characters is divided up using the dictionary, the break
    123      * positions that are discovered are stored here, preventing us from having
    124      * to use either the dictionary or the state table again until the iterator
    125      * leaves this range of text. Has the most impact for line breaking.
    126      * @internal
    127      */
    128     int32_t*            fCachedBreakPositions;
    129 
    130     /**
    131      * The number of elements in fCachedBreakPositions
    132      * @internal
    133      */
    134     int32_t             fNumCachedBreakPositions;
    135 
    136     /**
    137      * if fCachedBreakPositions is not null, this indicates which item in the
    138      * cache the current iteration position refers to
    139      * @internal
    140      */
    141     int32_t             fPositionInCache;
    142 
    143     /**
    144      *
    145      * If present, UStack of LanguageBreakEngine objects that might handle
    146      * dictionary characters. Searched from top to bottom to find an object to
    147      * handle a given character.
    148      * @internal
    149      */
    150     UStack              *fLanguageBreakEngines;
    151 
    152     /**
    153      *
    154      * If present, the special LanguageBreakEngine used for handling
    155      * characters that are in the dictionary set, but not handled by any
    156      * LangugageBreakEngine.
    157      * @internal
    158      */
    159     UnhandledEngine     *fUnhandledBreakEngine;
    160 
    161     /**
    162      *
    163      * The type of the break iterator, or -1 if it has not been set.
    164      * @internal
    165      */
    166     int32_t             fBreakType;
    167 
    168 protected:
    169     //=======================================================================
    170     // constructors
    171     //=======================================================================
    172 
    173     /**
    174      * Constant to be used in the constructor
    175      * RuleBasedBreakIterator(RBBIDataHeader*, EDontAdopt, UErrorCode &);
    176      * which does not adopt the memory indicated by the RBBIDataHeader*
    177      * parameter.
    178      *
    179      * @internal
    180      */
    181     enum EDontAdopt {
    182         kDontAdopt
    183     };
    184 
    185     /**
    186      * Constructor from a flattened set of RBBI data in malloced memory.
    187      *             RulesBasedBreakIterators built from a custom set of rules
    188      *             are created via this constructor; the rules are compiled
    189      *             into memory, then the break iterator is constructed here.
    190      *
    191      *             The break iterator adopts the memory, and will
    192      *             free it when done.
    193      * @internal
    194      */
    195     RuleBasedBreakIterator(RBBIDataHeader* data, UErrorCode &status);
    196 
    197     /**
    198      * Constructor from a flattened set of RBBI data in memory which need not
    199      *             be malloced (e.g. it may be a memory-mapped file, etc.).
    200      *
    201      *             This version does not adopt the memory, and does not
    202      *             free it when done.
    203      * @internal
    204      */
    205     RuleBasedBreakIterator(const RBBIDataHeader* data, enum EDontAdopt dontAdopt, UErrorCode &status);
    206 
    207 
    208     friend class RBBIRuleBuilder;
    209     /** @internal */
    210     friend class BreakIterator;
    211 
    212 
    213 
    214 public:
    215 
    216     /** Default constructor.  Creates an empty shell of an iterator, with no
    217      *  rules or text to iterate over.   Object can subsequently be assigned to.
    218      *  @stable ICU 2.2
    219      */
    220     RuleBasedBreakIterator();
    221 
    222     /**
    223      * Copy constructor.  Will produce a break iterator with the same behavior,
    224      * and which iterates over the same text, as the one passed in.
    225      * @param that The RuleBasedBreakIterator passed to be copied
    226      * @stable ICU 2.0
    227      */
    228     RuleBasedBreakIterator(const RuleBasedBreakIterator& that);
    229 
    230     /**
    231      * Construct a RuleBasedBreakIterator from a set of rules supplied as a string.
    232      * @param rules The break rules to be used.
    233      * @param parseError  In the event of a syntax error in the rules, provides the location
    234      *                    within the rules of the problem.
    235      * @param status Information on any errors encountered.
    236      * @stable ICU 2.2
    237      */
    238     RuleBasedBreakIterator( const UnicodeString    &rules,
    239                              UParseError           &parseError,
    240                              UErrorCode            &status);
    241 
    242 
    243 
    244 
    245     /**
    246      * Contruct a RuleBasedBreakIterator from a set of precompiled binary rules.
    247      * Binary rules are obtained from RulesBasedBreakIterator::getBinaryRules().
    248      * Construction of a break iterator in this way is substantially faster than
    249      * constuction from source rules.
    250      *
    251      * Ownership of the storage containing the compiled rules remains with the
    252      * caller of this function.  The compiled rules must not be  modified or
    253      * deleted during the life of the break iterator.
    254      *
    255      * The compiled rules are not compatible across different major versions of ICU.
    256      * The compiled rules are comaptible only between machines with the same
    257      * byte ordering (little or big endian) and the same base character set family
    258      * (ASCII or EBCDIC).
    259      *
    260      * @see #getBinaryRules
    261      * @param compiledRules A pointer to the compiled break rules to be used.
    262      * @param ruleLength The length of the compiled break rules, in bytes.  This
    263      *   corresponds to the length value produced by getBinaryRules().
    264      * @param status Information on any errors encountered, including invalid
    265      *   binary rules.
    266      * @draft ICU 4.8
    267      */
    268     RuleBasedBreakIterator(const uint8_t *compiledRules,
    269                            uint32_t       ruleLength,
    270                            UErrorCode    &status);
    271 
    272 
    273     /**
    274      * This constructor uses the udata interface to create a BreakIterator
    275      * whose internal tables live in a memory-mapped file.  "image" is an
    276      * ICU UDataMemory handle for the pre-compiled break iterator tables.
    277      * @param image handle to the memory image for the break iterator data.
    278      *        Ownership of the UDataMemory handle passes to the Break Iterator,
    279      *        which will be responsible for closing it when it is no longer needed.
    280      * @param status Information on any errors encountered.
    281      * @see udata_open
    282      * @see #getBinaryRules
    283      * @stable ICU 2.8
    284      */
    285     RuleBasedBreakIterator(UDataMemory* image, UErrorCode &status);
    286 
    287     /**
    288      * Destructor
    289      *  @stable ICU 2.0
    290      */
    291     virtual ~RuleBasedBreakIterator();
    292 
    293     /**
    294      * Assignment operator.  Sets this iterator to have the same behavior,
    295      * and iterate over the same text, as the one passed in.
    296      * @param that The RuleBasedBreakItertor passed in
    297      * @return the newly created RuleBasedBreakIterator
    298      *  @stable ICU 2.0
    299      */
    300     RuleBasedBreakIterator& operator=(const RuleBasedBreakIterator& that);
    301 
    302     /**
    303      * Equality operator.  Returns TRUE if both BreakIterators are of the
    304      * same class, have the same behavior, and iterate over the same text.
    305      * @param that The BreakIterator to be compared for equality
    306      * @return TRUE if both BreakIterators are of the
    307      * same class, have the same behavior, and iterate over the same text.
    308      *  @stable ICU 2.0
    309      */
    310     virtual UBool operator==(const BreakIterator& that) const;
    311 
    312     /**
    313      * Not-equal operator.  If operator== returns TRUE, this returns FALSE,
    314      * and vice versa.
    315      * @param that The BreakIterator to be compared for inequality
    316      * @return TRUE if both BreakIterators are not same.
    317      *  @stable ICU 2.0
    318      */
    319     UBool operator!=(const BreakIterator& that) const;
    320 
    321     /**
    322      * Returns a newly-constructed RuleBasedBreakIterator with the same
    323      * behavior, and iterating over the same text, as this one.
    324      * Differs from the copy constructor in that it is polymorphic, and
    325      * will correctly clone (copy) a derived class.
    326      * clone() is thread safe.  Multiple threads may simultaeneously
    327      * clone the same source break iterator.
    328      * @return a newly-constructed RuleBasedBreakIterator
    329      * @stable ICU 2.0
    330      */
    331     virtual BreakIterator* clone() const;
    332 
    333     /**
    334      * Compute a hash code for this BreakIterator
    335      * @return A hash code
    336      *  @stable ICU 2.0
    337      */
    338     virtual int32_t hashCode(void) const;
    339 
    340     /**
    341      * Returns the description used to create this iterator
    342      * @return the description used to create this iterator
    343      *  @stable ICU 2.0
    344      */
    345     virtual const UnicodeString& getRules(void) const;
    346 
    347     //=======================================================================
    348     // BreakIterator overrides
    349     //=======================================================================
    350 
    351     /**
    352      * <p>
    353      * Return a CharacterIterator over the text being analyzed.
    354      * The returned character iterator is owned by the break iterator, and must
    355      * not be deleted by the caller.  Repeated calls to this function may
    356      * return the same CharacterIterator.
    357      * </p>
    358      * <p>
    359      * The returned character iterator must not be used concurrently with
    360      * the break iterator.  If concurrent operation is needed, clone the
    361      * returned character iterator first and operate on the clone.
    362      * </p>
    363      * <p>
    364      * When the break iterator is operating on text supplied via a UText,
    365      * this function will fail.  Lacking any way to signal failures, it
    366      * returns an CharacterIterator containing no text.
    367      * The function getUText() provides similar functionality,
    368      * is reliable, and is more efficient.
    369      * </p>
    370      *
    371      * TODO:  deprecate this function?
    372      *
    373      * @return An iterator over the text being analyzed.
    374      * @stable ICU 2.0
    375      */
    376     virtual  CharacterIterator& getText(void) const;
    377 
    378 
    379     /**
    380       *  Get a UText for the text being analyzed.
    381       *  The returned UText is a shallow clone of the UText used internally
    382       *  by the break iterator implementation.  It can safely be used to
    383       *  access the text without impacting any break iterator operations,
    384       *  but the underlying text itself must not be altered.
    385       *
    386       * @param fillIn A UText to be filled in.  If NULL, a new UText will be
    387       *           allocated to hold the result.
    388       * @param status receives any error codes.
    389       * @return   The current UText for this break iterator.  If an input
    390       *           UText was provided, it will always be returned.
    391       * @stable ICU 3.4
    392       */
    393      virtual UText *getUText(UText *fillIn, UErrorCode &status) const;
    394 
    395     /**
    396      * Set the iterator to analyze a new piece of text.  This function resets
    397      * the current iteration position to the beginning of the text.
    398      * @param newText An iterator over the text to analyze.  The BreakIterator
    399      * takes ownership of the character iterator.  The caller MUST NOT delete it!
    400      *  @stable ICU 2.0
    401      */
    402     virtual void adoptText(CharacterIterator* newText);
    403 
    404     /**
    405      * Set the iterator to analyze a new piece of text.  This function resets
    406      * the current iteration position to the beginning of the text.
    407      * @param newText The text to analyze.
    408      *  @stable ICU 2.0
    409      */
    410     virtual void setText(const UnicodeString& newText);
    411 
    412     /**
    413      * Reset the break iterator to operate over the text represented by
    414      * the UText.  The iterator position is reset to the start.
    415      *
    416      * This function makes a shallow clone of the supplied UText.  This means
    417      * that the caller is free to immediately close or otherwise reuse the
    418      * Utext that was passed as a parameter, but that the underlying text itself
    419      * must not be altered while being referenced by the break iterator.
    420      *
    421      * @param text    The UText used to change the text.
    422      * @param status  Receives any error codes.
    423      * @stable ICU 3.4
    424      */
    425     virtual void  setText(UText *text, UErrorCode &status);
    426 
    427     /**
    428      * Sets the current iteration position to the beginning of the text.
    429      * @return The offset of the beginning of the text.
    430      *  @stable ICU 2.0
    431      */
    432     virtual int32_t first(void);
    433 
    434     /**
    435      * Sets the current iteration position to the end of the text.
    436      * @return The text's past-the-end offset.
    437      *  @stable ICU 2.0
    438      */
    439     virtual int32_t last(void);
    440 
    441     /**
    442      * Advances the iterator either forward or backward the specified number of steps.
    443      * Negative values move backward, and positive values move forward.  This is
    444      * equivalent to repeatedly calling next() or previous().
    445      * @param n The number of steps to move.  The sign indicates the direction
    446      * (negative is backwards, and positive is forwards).
    447      * @return The character offset of the boundary position n boundaries away from
    448      * the current one.
    449      *  @stable ICU 2.0
    450      */
    451     virtual int32_t next(int32_t n);
    452 
    453     /**
    454      * Advances the iterator to the next boundary position.
    455      * @return The position of the first boundary after this one.
    456      *  @stable ICU 2.0
    457      */
    458     virtual int32_t next(void);
    459 
    460     /**
    461      * Moves the iterator backwards, to the last boundary preceding this one.
    462      * @return The position of the last boundary position preceding this one.
    463      *  @stable ICU 2.0
    464      */
    465     virtual int32_t previous(void);
    466 
    467     /**
    468      * Sets the iterator to refer to the first boundary position following
    469      * the specified position.
    470      * @param offset The position from which to begin searching for a break position.
    471      * @return The position of the first break after the current position.
    472      *  @stable ICU 2.0
    473      */
    474     virtual int32_t following(int32_t offset);
    475 
    476     /**
    477      * Sets the iterator to refer to the last boundary position before the
    478      * specified position.
    479      * @param offset The position to begin searching for a break from.
    480      * @return The position of the last boundary before the starting position.
    481      *  @stable ICU 2.0
    482      */
    483     virtual int32_t preceding(int32_t offset);
    484 
    485     /**
    486      * Returns true if the specfied position is a boundary position.  As a side
    487      * effect, leaves the iterator pointing to the first boundary position at
    488      * or after "offset".
    489      * @param offset the offset to check.
    490      * @return True if "offset" is a boundary position.
    491      *  @stable ICU 2.0
    492      */
    493     virtual UBool isBoundary(int32_t offset);
    494 
    495     /**
    496      * Returns the current iteration position.
    497      * @return The current iteration position.
    498      * @stable ICU 2.0
    499      */
    500     virtual int32_t current(void) const;
    501 
    502 
    503     /**
    504      * Return the status tag from the break rule that determined the most recently
    505      * returned break position.  For break rules that do not specify a
    506      * status, a default value of 0 is returned.  If more than one break rule
    507      * would cause a boundary to be located at some position in the text,
    508      * the numerically largest of the applicable status values is returned.
    509      * <p>
    510      * Of the standard types of ICU break iterators, only word break and
    511      * line break provide status values.  The values are defined in
    512      * the header file ubrk.h.  For Word breaks, the status allows distinguishing between words
    513      * that contain alphabetic letters, "words" that appear to be numbers,
    514      * punctuation and spaces, words containing ideographic characters, and
    515      * more.  For Line Break, the status distinguishes between hard (mandatory) breaks
    516      * and soft (potential) break positions.
    517      * <p>
    518      * <code>getRuleStatus()</code> can be called after obtaining a boundary
    519      * position from <code>next()</code>, <code>previous()</code>, or
    520      * any other break iterator functions that returns a boundary position.
    521      * <p>
    522      * When creating custom break rules, one is free to define whatever
    523      * status values may be convenient for the application.
    524      * <p>
    525      * Note: this function is not thread safe.  It should not have been
    526      *       declared const, and the const remains only for compatibility
    527      *       reasons.  (The function is logically const, but not bit-wise const).
    528      * <p>
    529      * @return the status from the break rule that determined the most recently
    530      * returned break position.
    531      *
    532      * @see UWordBreak
    533      * @stable ICU 2.2
    534      */
    535     virtual int32_t getRuleStatus() const;
    536 
    537    /**
    538     * Get the status (tag) values from the break rule(s) that determined the most
    539     * recently returned break position.
    540     * <p>
    541     * The returned status value(s) are stored into an array provided by the caller.
    542     * The values are stored in sorted (ascending) order.
    543     * If the capacity of the output array is insufficient to hold the data,
    544     *  the output will be truncated to the available length, and a
    545     *  U_BUFFER_OVERFLOW_ERROR will be signaled.
    546     *
    547     * @param fillInVec an array to be filled in with the status values.
    548     * @param capacity  the length of the supplied vector.  A length of zero causes
    549     *                  the function to return the number of status values, in the
    550     *                  normal way, without attemtping to store any values.
    551     * @param status    receives error codes.
    552     * @return          The number of rule status values from rules that determined
    553     *                  the most recent boundary returned by the break iterator.
    554     *                  In the event of a U_BUFFER_OVERFLOW_ERROR, the return value
    555     *                  is the total number of status values that were available,
    556     *                  not the reduced number that were actually returned.
    557     * @see getRuleStatus
    558     * @stable ICU 3.0
    559     */
    560     virtual int32_t getRuleStatusVec(int32_t *fillInVec, int32_t capacity, UErrorCode &status);
    561 
    562     /**
    563      * Returns a unique class ID POLYMORPHICALLY.  Pure virtual override.
    564      * This method is to implement a simple version of RTTI, since not all
    565      * C++ compilers support genuine RTTI.  Polymorphic operator==() and
    566      * clone() methods call this method.
    567      *
    568      * @return          The class ID for this object. All objects of a
    569      *                  given class have the same class ID.  Objects of
    570      *                  other classes have different class IDs.
    571      * @stable ICU 2.0
    572      */
    573     virtual UClassID getDynamicClassID(void) const;
    574 
    575     /**
    576      * Returns the class ID for this class.  This is useful only for
    577      * comparing to a return value from getDynamicClassID().  For example:
    578      *
    579      *      Base* polymorphic_pointer = createPolymorphicObject();
    580      *      if (polymorphic_pointer->getDynamicClassID() ==
    581      *          Derived::getStaticClassID()) ...
    582      *
    583      * @return          The class ID for all objects of this class.
    584      * @stable ICU 2.0
    585      */
    586     static UClassID U_EXPORT2 getStaticClassID(void);
    587 
    588     /*
    589      * Create a clone (copy) of this break iterator in memory provided
    590      *  by the caller.  The idea is to increase performance by avoiding
    591      *  a storage allocation.  Use of this functoin is NOT RECOMMENDED.
    592      *  Performance gains are minimal, and correct buffer management is
    593      *  tricky.  Use clone() instead.
    594      *
    595      * @param stackBuffer  The pointer to the memory into which the cloned object
    596      *                     should be placed.  If NULL,  allocate heap memory
    597      *                     for the cloned object.
    598      * @param BufferSize   The size of the buffer.  If zero, return the required
    599      *                     buffer size, but do not clone the object.  If the
    600      *                     size was too small (but not zero), allocate heap
    601      *                     storage for the cloned object.
    602      *
    603      * @param status       Error status.  U_SAFECLONE_ALLOCATED_WARNING will be
    604      *                     returned if the the provided buffer was too small, and
    605      *                     the clone was therefore put on the heap.
    606      *
    607      * @return  Pointer to the clone object.  This may differ from the stackBuffer
    608      *          address if the byte alignment of the stack buffer was not suitable
    609      *          or if the stackBuffer was too small to hold the clone.
    610      * @stable ICU 2.0
    611      */
    612     virtual BreakIterator *  createBufferClone(void *stackBuffer,
    613                                                int32_t &BufferSize,
    614                                                UErrorCode &status);
    615 
    616 
    617     /**
    618      * Return the binary form of compiled break rules,
    619      * which can then be used to create a new break iterator at some
    620      * time in the future.  Creating a break iterator from pre-compiled rules
    621      * is much faster than building one from the source form of the
    622      * break rules.
    623      *
    624      * The binary data can only be used with the same version of ICU
    625      *  and on the same platform type (processor endian-ness)
    626      *
    627      * @param length Returns the length of the binary data.  (Out paramter.)
    628      *
    629      * @return   A pointer to the binary (compiled) rule data.  The storage
    630      *           belongs to the RulesBasedBreakIterator object, not the
    631      *           caller, and must not be modified or deleted.
    632      * @draft ICU 4.8
    633      */
    634     virtual const uint8_t *getBinaryRules(uint32_t &length);
    635 
    636 
    637 protected:
    638     //=======================================================================
    639     // implementation
    640     //=======================================================================
    641     /**
    642      * Dumps caches and performs other actions associated with a complete change
    643      * in text or iteration position.
    644      * @internal
    645      */
    646     virtual void reset(void);
    647 
    648 #if 0
    649     /**
    650       * Return true if the category lookup for this char
    651       * indicates that it is in the set of dictionary lookup chars.
    652       * This function is intended for use by dictionary based break iterators.
    653       * @return true if the category lookup for this char
    654       * indicates that it is in the set of dictionary lookup chars.
    655       * @internal
    656       */
    657     virtual UBool isDictionaryChar(UChar32);
    658 
    659     /**
    660       * Get the type of the break iterator.
    661       * @internal
    662       */
    663     virtual int32_t getBreakType() const;
    664 #endif
    665 
    666     /**
    667       * Set the type of the break iterator.
    668       * @internal
    669       */
    670     virtual void setBreakType(int32_t type);
    671 
    672     /**
    673       * Common initialization function, used by constructors and bufferClone.
    674       *   (Also used by DictionaryBasedBreakIterator::createBufferClone().)
    675       * @internal
    676       */
    677     void init();
    678 
    679 private:
    680 
    681     /**
    682      * This method backs the iterator back up to a "safe position" in the text.
    683      * This is a position that we know, without any context, must be a break position.
    684      * The various calling methods then iterate forward from this safe position to
    685      * the appropriate position to return.  (For more information, see the description
    686      * of buildBackwardsStateTable() in RuleBasedBreakIterator.Builder.)
    687      * @param statetable state table used of moving backwards
    688      * @internal
    689      */
    690     int32_t handlePrevious(const RBBIStateTable *statetable);
    691 
    692     /**
    693      * This method is the actual implementation of the next() method.  All iteration
    694      * vectors through here.  This method initializes the state machine to state 1
    695      * and advances through the text character by character until we reach the end
    696      * of the text or the state machine transitions to state 0.  We update our return
    697      * value every time the state machine passes through a possible end state.
    698      * @param statetable state table used of moving forwards
    699      * @internal
    700      */
    701     int32_t handleNext(const RBBIStateTable *statetable);
    702 
    703 protected:
    704 
    705     /**
    706      * This is the function that actually implements dictionary-based
    707      * breaking.  Covering at least the range from startPos to endPos,
    708      * it checks for dictionary characters, and if it finds them determines
    709      * the appropriate object to deal with them. It may cache found breaks in
    710      * fCachedBreakPositions as it goes. It may well also look at text outside
    711      * the range startPos to endPos.
    712      * If going forward, endPos is the normal Unicode break result, and
    713      * if goind in reverse, startPos is the normal Unicode break result
    714      * @param startPos  The start position of a range of text
    715      * @param endPos    The end position of a range of text
    716      * @param reverse   The call is for the reverse direction
    717      * @internal
    718      */
    719     int32_t checkDictionary(int32_t startPos, int32_t endPos, UBool reverse);
    720 
    721 private:
    722 
    723     /**
    724      * This function returns the appropriate LanguageBreakEngine for a
    725      * given character c.
    726      * @param c         A character in the dictionary set
    727      * @internal
    728      */
    729     const LanguageBreakEngine *getLanguageBreakEngine(UChar32 c);
    730 
    731     /**
    732      *  @internal
    733      */
    734     void makeRuleStatusValid();
    735 
    736 };
    737 
    738 //------------------------------------------------------------------------------
    739 //
    740 //   Inline Functions Definitions ...
    741 //
    742 //------------------------------------------------------------------------------
    743 
    744 inline UBool RuleBasedBreakIterator::operator!=(const BreakIterator& that) const {
    745     return !operator==(that);
    746 }
    747 
    748 U_NAMESPACE_END
    749 
    750 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
    751 
    752 #endif
    753