1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unorm_it.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,148 @@
1.4 +/*
1.5 +*******************************************************************************
1.6 +*
1.7 +* Copyright (C) 2003, International Business Machines
1.8 +* Corporation and others. All Rights Reserved.
1.9 +*
1.10 +*******************************************************************************
1.11 +* file name: unorm_it.h
1.12 +* encoding: US-ASCII
1.13 +* tab size: 8 (not used)
1.14 +* indentation:4
1.15 +*
1.16 +* created on: 2003jan21
1.17 +* created by: Markus W. Scherer
1.18 +*/
1.19 +
1.20 +#ifndef __UNORM_IT_H__
1.21 +#define __UNORM_IT_H__
1.22 +
1.23 +#include "unicode/utypes.h"
1.24 +
1.25 +#if !UCONFIG_NO_COLLATION && !UCONFIG_NO_NORMALIZATION
1.26 +
1.27 +#include "unicode/uiter.h"
1.28 +#include "unicode/unorm.h"
1.29 +
1.30 +/**
1.31 + * Normalizing UCharIterator wrapper.
1.32 + * This internal API basically duplicates the functionality of the C++ Normalizer
1.33 + * but
1.34 + * - it actually implements a character iterator (UCharIterator)
1.35 + * with few restrictions (see unorm_setIter())
1.36 + * - it supports UCharIterator getState()/setState()
1.37 + * - it uses lower-level APIs and buffers more text and states,
1.38 + * hopefully resulting in higher performance
1.39 + *
1.40 + * Usage example:
1.41 + * \code
1.42 + * function(UCharIterator *srcIter) {
1.43 + * UNormIterator *uni;
1.44 + * UCharIterator *iter;
1.45 + * UErrorCode errorCode;
1.46 + *
1.47 + * errorCode=U_ZERO_ERROR;
1.48 + * uni=unorm_openIter(&errorCode);
1.49 + * if(U_FAILURE(errorCode)) {
1.50 + * // report error
1.51 + * return;
1.52 + * }
1.53 + *
1.54 + * iter=unorm_setIter(uni, srcIter, UNORM_FCD, &errorCode);
1.55 + * if(U_FAILURE(errorCode)) {
1.56 + * // report error
1.57 + * } else {
1.58 + * // use iter to iterate over the canonically ordered
1.59 + * // version of srcIter's text
1.60 + * uint32_t state;
1.61 + *
1.62 + * ...
1.63 + *
1.64 + * state=uiter_getState(iter);
1.65 + * if(state!=UITER_NO_STATE) {
1.66 + * // use valid state, store it, use iter some more
1.67 + * ...
1.68 + *
1.69 + * // later restore iter to the saved state:
1.70 + * uiter_setState(iter, state, &errorCode);
1.71 + *
1.72 + * ...
1.73 + * }
1.74 + *
1.75 + * ...
1.76 + * }
1.77 + * unorm_closeIter(uni);
1.78 + * }
1.79 + * \endcode
1.80 + *
1.81 + * See also the ICU test suites.
1.82 + *
1.83 + * @internal
1.84 + */
1.85 +struct UNormIterator;
1.86 +typedef struct UNormIterator UNormIterator;
1.87 +
1.88 +/**
1.89 + * Size of a stack buffer to hold a UNormIterator, see the stackMem parameter
1.90 + * of unorm_openIter().
1.91 + *
1.92 + * @internal
1.93 + */
1.94 +#define UNORM_ITER_SIZE 1024
1.95 +
1.96 +/**
1.97 + * Open a normalizing iterator. Must be closed later.
1.98 + * Use unorm_setIter().
1.99 + *
1.100 + * @param stackMem Pointer to preallocated (stack-allocated) buffer to hold
1.101 + * the UNormIterator if possible; can be NULL.
1.102 + * @param stackMemSize Number of bytes at stackMem; can be 0,
1.103 + * or should be >= UNORM_ITER_SIZE for a non-NULL stackMem.
1.104 + * @param pErrorCode ICU error code
1.105 + * @return an allocated and pre-initialized UNormIterator
1.106 + * @internal
1.107 + */
1.108 +U_CAPI UNormIterator * U_EXPORT2
1.109 +unorm_openIter(void *stackMem, int32_t stackMemSize, UErrorCode *pErrorCode);
1.110 +
1.111 +/**
1.112 + * Close a normalizing iterator.
1.113 + *
1.114 + * @param uni UNormIterator from unorm_openIter()
1.115 + * @internal
1.116 + */
1.117 +U_CAPI void U_EXPORT2
1.118 +unorm_closeIter(UNormIterator *uni);
1.119 +
1.120 +/**
1.121 + * Set a UCharIterator and a normalization mode for the normalizing iterator
1.122 + * to wrap. The normalizing iterator will read from the character iterator,
1.123 + * normalize the text, and in turn deliver it with its own wrapper UCharIterator
1.124 + * interface which it returns.
1.125 + *
1.126 + * The source iterator remains at its current position through the unorm_setIter()
1.127 + * call but will be used and moved as soon as the
1.128 + * the returned normalizing iterator is.
1.129 + *
1.130 + * The returned interface pointer is valid for as long as the normalizing iterator
1.131 + * is open and until another unorm_setIter() call is made on it.
1.132 + *
1.133 + * The normalizing iterator's UCharIterator interface has the following properties:
1.134 + * - getIndex() and move() will almost always return UITER_UNKNOWN_INDEX
1.135 + * - getState() will return UITER_NO_STATE for unknown states for positions
1.136 + * that are not at normalization boundaries
1.137 + *
1.138 + * @param uni UNormIterator from unorm_openIter()
1.139 + * @param iter The source text UCharIterator to be wrapped. It is aliases into the normalizing iterator.
1.140 + * Must support getState() and setState().
1.141 + * @param mode The normalization mode.
1.142 + * @param pErrorCode ICU error code
1.143 + * @return an alias to the normalizing iterator's UCharIterator interface
1.144 + * @internal
1.145 + */
1.146 +U_CAPI UCharIterator * U_EXPORT2
1.147 +unorm_setIter(UNormIterator *uni, UCharIterator *iter, UNormalizationMode mode, UErrorCode *pErrorCode);
1.148 +
1.149 +#endif /* uconfig.h switches */
1.150 +
1.151 +#endif
