sl@0: /*
sl@0: *******************************************************************************
sl@0: *
sl@0: *   Copyright (C) 2002-2005, International Business Machines
sl@0: *   Corporation and others.  All Rights Reserved.
sl@0: *
sl@0: *******************************************************************************
sl@0: *   file name:  uiter.h
sl@0: *   encoding:   US-ASCII
sl@0: *   tab size:   8 (not used)
sl@0: *   indentation:4
sl@0: *
sl@0: *   created on: 2002jan18
sl@0: *   created by: Markus W. Scherer
sl@0: */
sl@0: 
sl@0: #ifndef __UITER_H__
sl@0: #define __UITER_H__
sl@0: 
sl@0: /**
sl@0:  * \file
sl@0:  * \brief C API: Unicode Character Iteration
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  */
sl@0: 
sl@0: #include "unicode/utypes.h"
sl@0: 
sl@0: #ifdef XP_CPLUSPLUS
sl@0:     U_NAMESPACE_BEGIN
sl@0: 
sl@0:     class CharacterIterator;
sl@0:     class Replaceable;
sl@0: 
sl@0:     U_NAMESPACE_END
sl@0: #endif
sl@0: 
sl@0: U_CDECL_BEGIN
sl@0: 
sl@0: struct UCharIterator;
sl@0: typedef struct UCharIterator UCharIterator; /**< C typedef for struct UCharIterator. @stable ICU 2.1 */
sl@0: 
sl@0: /**
sl@0:  * Origin constants for UCharIterator.getIndex() and UCharIterator.move().
sl@0:  * @see UCharIteratorMove
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef enum UCharIteratorOrigin {
sl@0:     UITER_START, UITER_CURRENT, UITER_LIMIT, UITER_ZERO, UITER_LENGTH
sl@0: } UCharIteratorOrigin;
sl@0: 
sl@0: /** Constants for UCharIterator. @stable ICU 2.6 */
sl@0: enum {
sl@0:     /**
sl@0:      * Constant value that may be returned by UCharIteratorMove
sl@0:      * indicating that the final UTF-16 index is not known, but that the move succeeded.
sl@0:      * This can occur when moving relative to limit or length, or
sl@0:      * when moving relative to the current index after a setState()
sl@0:      * when the current UTF-16 index is not known.
sl@0:      *
sl@0:      * It would be very inefficient to have to count from the beginning of the text
sl@0:      * just to get the current/limit/length index after moving relative to it.
sl@0:      * The actual index can be determined with getIndex(UITER_CURRENT)
sl@0:      * which will count the UChars if necessary.
sl@0:      *
sl@0:      * @stable ICU 2.6
sl@0:      */
sl@0:     UITER_UNKNOWN_INDEX=-2
sl@0: };
sl@0: 
sl@0: 
sl@0: /**
sl@0:  * Constant for UCharIterator getState() indicating an error or
sl@0:  * an unknown state.
sl@0:  * Returned by uiter_getState()/UCharIteratorGetState
sl@0:  * when an error occurs.
sl@0:  * Also, some UCharIterator implementations may not be able to return
sl@0:  * a valid state for each position. This will be clearly documented
sl@0:  * for each such iterator (none of the public ones here).
sl@0:  *
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: #define UITER_NO_STATE ((uint32_t)0xffffffff)
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.getIndex().
sl@0:  *
sl@0:  * Gets the current position, or the start or limit of the
sl@0:  * iteration range.
sl@0:  *
sl@0:  * This function may perform slowly for UITER_CURRENT after setState() was called,
sl@0:  * or for UITER_LENGTH, because an iterator implementation may have to count
sl@0:  * UChars if the underlying storage is not UTF-16.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @param origin get the 0, start, limit, length, or current index
sl@0:  * @return the requested index, or U_SENTINEL in an error condition
sl@0:  *
sl@0:  * @see UCharIteratorOrigin
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef int32_t U_CALLCONV
sl@0: UCharIteratorGetIndex(UCharIterator *iter, UCharIteratorOrigin origin);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.move().
sl@0:  *
sl@0:  * Use iter->move(iter, index, UITER_ZERO) like CharacterIterator::setIndex(index).
sl@0:  *
sl@0:  * Moves the current position relative to the start or limit of the
sl@0:  * iteration range, or relative to the current position itself.
sl@0:  * The movement is expressed in numbers of code units forward
sl@0:  * or backward by specifying a positive or negative delta.
sl@0:  * Out of bounds movement will be pinned to the start or limit.
sl@0:  *
sl@0:  * This function may perform slowly for moving relative to UITER_LENGTH
sl@0:  * because an iterator implementation may have to count the rest of the
sl@0:  * UChars if the native storage is not UTF-16.
sl@0:  *
sl@0:  * When moving relative to the limit or length, or
sl@0:  * relative to the current position after setState() was called,
sl@0:  * move() may return UITER_UNKNOWN_INDEX (-2) to avoid an inefficient
sl@0:  * determination of the actual UTF-16 index.
sl@0:  * The actual index can be determined with getIndex(UITER_CURRENT)
sl@0:  * which will count the UChars if necessary.
sl@0:  * See UITER_UNKNOWN_INDEX for details.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @param delta can be positive, zero, or negative
sl@0:  * @param origin move relative to the 0, start, limit, length, or current index
sl@0:  * @return the new index, or U_SENTINEL on an error condition,
sl@0:  *         or UITER_UNKNOWN_INDEX when the index is not known.
sl@0:  *
sl@0:  * @see UCharIteratorOrigin
sl@0:  * @see UCharIterator
sl@0:  * @see UITER_UNKNOWN_INDEX
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef int32_t U_CALLCONV
sl@0: UCharIteratorMove(UCharIterator *iter, int32_t delta, UCharIteratorOrigin origin);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.hasNext().
sl@0:  *
sl@0:  * Check if current() and next() can still
sl@0:  * return another code unit.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return boolean value for whether current() and next() can still return another code unit
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef UBool U_CALLCONV
sl@0: UCharIteratorHasNext(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.hasPrevious().
sl@0:  *
sl@0:  * Check if previous() can still return another code unit.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return boolean value for whether previous() can still return another code unit
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef UBool U_CALLCONV
sl@0: UCharIteratorHasPrevious(UCharIterator *iter);
sl@0:  
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.current().
sl@0:  *
sl@0:  * Return the code unit at the current position,
sl@0:  * or U_SENTINEL if there is none (index is at the limit).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the current code unit
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef UChar32 U_CALLCONV
sl@0: UCharIteratorCurrent(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.next().
sl@0:  *
sl@0:  * Return the code unit at the current index and increment
sl@0:  * the index (post-increment, like s[i++]),
sl@0:  * or return U_SENTINEL if there is none (index is at the limit).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the current code unit (and post-increment the current index)
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef UChar32 U_CALLCONV
sl@0: UCharIteratorNext(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.previous().
sl@0:  *
sl@0:  * Decrement the index and return the code unit from there
sl@0:  * (pre-decrement, like s[--i]),
sl@0:  * or return U_SENTINEL if there is none (index is at the start).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the previous code unit (after pre-decrementing the current index)
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef UChar32 U_CALLCONV
sl@0: UCharIteratorPrevious(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.reservedFn().
sl@0:  * Reserved for future use.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @param something some integer argument
sl@0:  * @return some integer
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: typedef int32_t U_CALLCONV
sl@0: UCharIteratorReserved(UCharIterator *iter, int32_t something);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.getState().
sl@0:  *
sl@0:  * Get the "state" of the iterator in the form of a single 32-bit word.
sl@0:  * It is recommended that the state value be calculated to be as small as
sl@0:  * is feasible. For strings with limited lengths, fewer than 32 bits may
sl@0:  * be sufficient.
sl@0:  *
sl@0:  * This is used together with setState()/UCharIteratorSetState
sl@0:  * to save and restore the iterator position more efficiently than with
sl@0:  * getIndex()/move().
sl@0:  *
sl@0:  * The iterator state is defined as a uint32_t value because it is designed
sl@0:  * for use in ucol_nextSortKeyPart() which provides 32 bits to store the state
sl@0:  * of the character iterator.
sl@0:  *
sl@0:  * With some UCharIterator implementations (e.g., UTF-8),
sl@0:  * getting and setting the UTF-16 index with existing functions
sl@0:  * (getIndex(UITER_CURRENT) followed by move(pos, UITER_ZERO)) is possible but
sl@0:  * relatively slow because the iterator has to "walk" from a known index
sl@0:  * to the requested one.
sl@0:  * This takes more time the farther it needs to go.
sl@0:  *
sl@0:  * An opaque state value allows an iterator implementation to provide
sl@0:  * an internal index (UTF-8: the source byte array index) for
sl@0:  * fast, constant-time restoration.
sl@0:  *
sl@0:  * After calling setState(), a getIndex(UITER_CURRENT) may be slow because
sl@0:  * the UTF-16 index may not be restored as well, but the iterator can deliver
sl@0:  * the correct text contents and move relative to the current position
sl@0:  * without performance degradation.
sl@0:  *
sl@0:  * Some UCharIterator implementations may not be able to return
sl@0:  * a valid state for each position, in which case they return UITER_NO_STATE instead.
sl@0:  * This will be clearly documented for each such iterator (none of the public ones here).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the state word
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see UCharIteratorSetState
sl@0:  * @see UITER_NO_STATE
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: typedef uint32_t U_CALLCONV
sl@0: UCharIteratorGetState(const UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Function type declaration for UCharIterator.setState().
sl@0:  *
sl@0:  * Restore the "state" of the iterator using a state word from a getState() call.
sl@0:  * The iterator object need not be the same one as for which getState() was called,
sl@0:  * but it must be of the same type (set up using the same uiter_setXYZ function)
sl@0:  * and it must iterate over the same string
sl@0:  * (binary identical regardless of memory address).
sl@0:  * For more about the state word see UCharIteratorGetState.
sl@0:  *
sl@0:  * After calling setState(), a getIndex(UITER_CURRENT) may be slow because
sl@0:  * the UTF-16 index may not be restored as well, but the iterator can deliver
sl@0:  * the correct text contents and move relative to the current position
sl@0:  * without performance degradation.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @param state the state word from a getState() call
sl@0:  *              on a same-type, same-string iterator
sl@0:  * @param pErrorCode Must be a valid pointer to an error code value,
sl@0:  *                   which must not indicate a failure before the function call.
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see UCharIteratorGetState
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: typedef void U_CALLCONV
sl@0: UCharIteratorSetState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode);
sl@0: 
sl@0: 
sl@0: /**
sl@0:  * C API for code unit iteration.
sl@0:  * This can be used as a C wrapper around
sl@0:  * CharacterIterator, Replaceable, or implemented using simple strings, etc.
sl@0:  *
sl@0:  * There are two roles for using UCharIterator:
sl@0:  *
sl@0:  * A "provider" sets the necessary function pointers and controls the "protected"
sl@0:  * fields of the UCharIterator structure. A "provider" passes a UCharIterator
sl@0:  * into C APIs that need a UCharIterator as an abstract, flexible string interface.
sl@0:  *
sl@0:  * Implementations of such C APIs are "callers" of UCharIterator functions;
sl@0:  * they only use the "public" function pointers and never access the "protected"
sl@0:  * fields directly.
sl@0:  *
sl@0:  * The current() and next() functions only check the current index against the
sl@0:  * limit, and previous() only checks the current index against the start,
sl@0:  * to see if the iterator already reached the end of the iteration range.
sl@0:  *
sl@0:  * The assumption - in all iterators - is that the index is moved via the API,
sl@0:  * which means it won't go out of bounds, or the index is modified by
sl@0:  * user code that knows enough about the iterator implementation to set valid
sl@0:  * index values.
sl@0:  *
sl@0:  * UCharIterator functions return code unit values 0..0xffff,
sl@0:  * or U_SENTINEL if the iteration bounds are reached.
sl@0:  *
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: struct UCharIterator {
sl@0:     /**
sl@0:      * (protected) Pointer to string or wrapped object or similar.
sl@0:      * Not used by caller.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     const void *context;
sl@0: 
sl@0:     /**
sl@0:      * (protected) Length of string or similar.
sl@0:      * Not used by caller.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     int32_t length;
sl@0: 
sl@0:     /**
sl@0:      * (protected) Start index or similar.
sl@0:      * Not used by caller.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     int32_t start;
sl@0: 
sl@0:     /**
sl@0:      * (protected) Current index or similar.
sl@0:      * Not used by caller.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     int32_t index;
sl@0: 
sl@0:     /**
sl@0:      * (protected) Limit index or similar.
sl@0:      * Not used by caller.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     int32_t limit;
sl@0: 
sl@0:     /**
sl@0:      * (protected) Used by UTF-8 iterators and possibly others.
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     int32_t reservedField;
sl@0: 
sl@0:     /**
sl@0:      * (public) Returns the current position or the
sl@0:      * start or limit index of the iteration range.
sl@0:      *
sl@0:      * @see UCharIteratorGetIndex
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorGetIndex *getIndex;
sl@0: 
sl@0:     /**
sl@0:      * (public) Moves the current position relative to the start or limit of the
sl@0:      * iteration range, or relative to the current position itself.
sl@0:      * The movement is expressed in numbers of code units forward
sl@0:      * or backward by specifying a positive or negative delta.
sl@0:      *
sl@0:      * @see UCharIteratorMove
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorMove *move;
sl@0: 
sl@0:     /**
sl@0:      * (public) Check if current() and next() can still
sl@0:      * return another code unit.
sl@0:      *
sl@0:      * @see UCharIteratorHasNext
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorHasNext *hasNext;
sl@0: 
sl@0:     /**
sl@0:      * (public) Check if previous() can still return another code unit.
sl@0:      *
sl@0:      * @see UCharIteratorHasPrevious
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorHasPrevious *hasPrevious;
sl@0: 
sl@0:     /**
sl@0:      * (public) Return the code unit at the current position,
sl@0:      * or U_SENTINEL if there is none (index is at the limit).
sl@0:      *
sl@0:      * @see UCharIteratorCurrent
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorCurrent *current;
sl@0: 
sl@0:     /**
sl@0:      * (public) Return the code unit at the current index and increment
sl@0:      * the index (post-increment, like s[i++]),
sl@0:      * or return U_SENTINEL if there is none (index is at the limit).
sl@0:      *
sl@0:      * @see UCharIteratorNext
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorNext *next;
sl@0: 
sl@0:     /**
sl@0:      * (public) Decrement the index and return the code unit from there
sl@0:      * (pre-decrement, like s[--i]),
sl@0:      * or return U_SENTINEL if there is none (index is at the start).
sl@0:      *
sl@0:      * @see UCharIteratorPrevious
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorPrevious *previous;
sl@0: 
sl@0:     /**
sl@0:      * (public) Reserved for future use. Currently NULL.
sl@0:      *
sl@0:      * @see UCharIteratorReserved
sl@0:      * @stable ICU 2.1
sl@0:      */
sl@0:     UCharIteratorReserved *reservedFn;
sl@0: 
sl@0:     /**
sl@0:      * (public) Return the state of the iterator, to be restored later with setState().
sl@0:      * This function pointer is NULL if the iterator does not implement it.
sl@0:      *
sl@0:      * @see UCharIteratorGet
sl@0:      * @stable ICU 2.6
sl@0:      */
sl@0:     UCharIteratorGetState *getState;
sl@0: 
sl@0:     /**
sl@0:      * (public) Restore the iterator state from the state word from a call
sl@0:      * to getState().
sl@0:      * This function pointer is NULL if the iterator does not implement it.
sl@0:      *
sl@0:      * @see UCharIteratorSet
sl@0:      * @stable ICU 2.6
sl@0:      */
sl@0:     UCharIteratorSetState *setState;
sl@0: };
sl@0: 
sl@0: /**
sl@0:  * Helper function for UCharIterator to get the code point
sl@0:  * at the current index.
sl@0:  *
sl@0:  * Return the code point that includes the code unit at the current position,
sl@0:  * or U_SENTINEL if there is none (index is at the limit).
sl@0:  * If the current code unit is a lead or trail surrogate,
sl@0:  * then the following or preceding surrogate is used to form
sl@0:  * the code point value.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the current code point
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see U16_GET
sl@0:  * @see UnicodeString::char32At()
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE UChar32 U_EXPORT2
sl@0: uiter_current32(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Helper function for UCharIterator to get the next code point.
sl@0:  *
sl@0:  * Return the code point at the current index and increment
sl@0:  * the index (post-increment, like s[i++]),
sl@0:  * or return U_SENTINEL if there is none (index is at the limit).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the current code point (and post-increment the current index)
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see U16_NEXT
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE UChar32 U_EXPORT2
sl@0: uiter_next32(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Helper function for UCharIterator to get the previous code point.
sl@0:  *
sl@0:  * Decrement the index and return the code point from there
sl@0:  * (pre-decrement, like s[--i]),
sl@0:  * or return U_SENTINEL if there is none (index is at the start).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the previous code point (after pre-decrementing the current index)
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see U16_PREV
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE UChar32 U_EXPORT2
sl@0: uiter_previous32(UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Get the "state" of the iterator in the form of a single 32-bit word.
sl@0:  * This is a convenience function that calls iter->getState(iter)
sl@0:  * if iter->getState is not NULL;
sl@0:  * if it is NULL or any other error occurs, then UITER_NO_STATE is returned.
sl@0:  *
sl@0:  * Some UCharIterator implementations may not be able to return
sl@0:  * a valid state for each position, in which case they return UITER_NO_STATE instead.
sl@0:  * This will be clearly documented for each such iterator (none of the public ones here).
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @return the state word
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see UCharIteratorGetState
sl@0:  * @see UITER_NO_STATE
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: U_STABLE uint32_t U_EXPORT2
sl@0: uiter_getState(const UCharIterator *iter);
sl@0: 
sl@0: /**
sl@0:  * Restore the "state" of the iterator using a state word from a getState() call.
sl@0:  * This is a convenience function that calls iter->setState(iter, state, pErrorCode)
sl@0:  * if iter->setState is not NULL; if it is NULL, then U_UNSUPPORTED_ERROR is set.
sl@0:  *
sl@0:  * @param iter the UCharIterator structure ("this pointer")
sl@0:  * @param state the state word from a getState() call
sl@0:  *              on a same-type, same-string iterator
sl@0:  * @param pErrorCode Must be a valid pointer to an error code value,
sl@0:  *                   which must not indicate a failure before the function call.
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see UCharIteratorSetState
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setState(UCharIterator *iter, uint32_t state, UErrorCode *pErrorCode);
sl@0: 
sl@0: /**
sl@0:  * Set up a UCharIterator to iterate over a string.
sl@0:  *
sl@0:  * Sets the UCharIterator function pointers for iteration over the string s
sl@0:  * with iteration boundaries start=index=0 and length=limit=string length.
sl@0:  * The "provider" may set the start, index, and limit values at any time
sl@0:  * within the range 0..length.
sl@0:  * The length field will be ignored.
sl@0:  *
sl@0:  * The string pointer s is set into UCharIterator.context without copying
sl@0:  * or reallocating the string contents.
sl@0:  *
sl@0:  * getState() simply returns the current index.
sl@0:  * move() will always return the final index.
sl@0:  *
sl@0:  * @param iter UCharIterator structure to be set for iteration
sl@0:  * @param s String to iterate over
sl@0:  * @param length Length of s, or -1 if NUL-terminated
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setString(UCharIterator *iter, const UChar *s, int32_t length);
sl@0: 
sl@0: /**
sl@0:  * Set up a UCharIterator to iterate over a UTF-16BE string
sl@0:  * (byte vector with a big-endian pair of bytes per UChar).
sl@0:  *
sl@0:  * Everything works just like with a normal UChar iterator (uiter_setString),
sl@0:  * except that UChars are assembled from byte pairs,
sl@0:  * and that the length argument here indicates an even number of bytes.
sl@0:  *
sl@0:  * getState() simply returns the current index.
sl@0:  * move() will always return the final index.
sl@0:  *
sl@0:  * @param iter UCharIterator structure to be set for iteration
sl@0:  * @param s UTF-16BE string to iterate over
sl@0:  * @param length Length of s as an even number of bytes, or -1 if NUL-terminated
sl@0:  *               (NUL means pair of 0 bytes at even index from s)
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @see uiter_setString
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setUTF16BE(UCharIterator *iter, const char *s, int32_t length);
sl@0: 
sl@0: /**
sl@0:  * Set up a UCharIterator to iterate over a UTF-8 string.
sl@0:  *
sl@0:  * Sets the UCharIterator function pointers for iteration over the UTF-8 string s
sl@0:  * with UTF-8 iteration boundaries 0 and length.
sl@0:  * The implementation counts the UTF-16 index on the fly and
sl@0:  * lazily evaluates the UTF-16 length of the text.
sl@0:  *
sl@0:  * The start field is used as the UTF-8 offset, the limit field as the UTF-8 length.
sl@0:  * When the reservedField is not 0, then it contains a supplementary code point
sl@0:  * and the UTF-16 index is between the two corresponding surrogates.
sl@0:  * At that point, the UTF-8 index is behind that code point.
sl@0:  *
sl@0:  * The UTF-8 string pointer s is set into UCharIterator.context without copying
sl@0:  * or reallocating the string contents.
sl@0:  *
sl@0:  * getState() returns a state value consisting of
sl@0:  * - the current UTF-8 source byte index (bits 31..1)
sl@0:  * - a flag (bit 0) that indicates whether the UChar position is in the middle
sl@0:  *   of a surrogate pair
sl@0:  *   (from a 4-byte UTF-8 sequence for the corresponding supplementary code point)
sl@0:  *
sl@0:  * getState() cannot also encode the UTF-16 index in the state value.
sl@0:  * move(relative to limit or length), or
sl@0:  * move(relative to current) after setState(), may return UITER_UNKNOWN_INDEX.
sl@0:  *
sl@0:  * @param iter UCharIterator structure to be set for iteration
sl@0:  * @param s UTF-8 string to iterate over
sl@0:  * @param length Length of s in bytes, or -1 if NUL-terminated
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.6
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setUTF8(UCharIterator *iter, const char *s, int32_t length);
sl@0: 
sl@0: #ifdef XP_CPLUSPLUS
sl@0: 
sl@0: /**
sl@0:  * Set up a UCharIterator to wrap around a C++ CharacterIterator.
sl@0:  *
sl@0:  * Sets the UCharIterator function pointers for iteration using the
sl@0:  * CharacterIterator charIter.
sl@0:  *
sl@0:  * The CharacterIterator pointer charIter is set into UCharIterator.context
sl@0:  * without copying or cloning the CharacterIterator object.
sl@0:  * The other "protected" UCharIterator fields are set to 0 and will be ignored.
sl@0:  * The iteration index and boundaries are controlled by the CharacterIterator.
sl@0:  *
sl@0:  * getState() simply returns the current index.
sl@0:  * move() will always return the final index.
sl@0:  *
sl@0:  * @param iter UCharIterator structure to be set for iteration
sl@0:  * @param charIter CharacterIterator to wrap
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setCharacterIterator(UCharIterator *iter, CharacterIterator *charIter);
sl@0: 
sl@0: /**
sl@0:  * Set up a UCharIterator to iterate over a C++ Replaceable.
sl@0:  *
sl@0:  * Sets the UCharIterator function pointers for iteration over the
sl@0:  * Replaceable rep with iteration boundaries start=index=0 and
sl@0:  * length=limit=rep->length().
sl@0:  * The "provider" may set the start, index, and limit values at any time
sl@0:  * within the range 0..length=rep->length().
sl@0:  * The length field will be ignored.
sl@0:  *
sl@0:  * The Replaceable pointer rep is set into UCharIterator.context without copying
sl@0:  * or cloning/reallocating the Replaceable object.
sl@0:  *
sl@0:  * getState() simply returns the current index.
sl@0:  * move() will always return the final index.
sl@0:  *
sl@0:  * @param iter UCharIterator structure to be set for iteration
sl@0:  * @param rep Replaceable to iterate over
sl@0:  *
sl@0:  * @see UCharIterator
sl@0:  * @stable ICU 2.1
sl@0:  */
sl@0: U_STABLE void U_EXPORT2
sl@0: uiter_setReplaceable(UCharIterator *iter, const Replaceable *rep);
sl@0: 
sl@0: #endif
sl@0: 
sl@0: U_CDECL_END
sl@0: 
sl@0: #endif