1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/common/unicode/usetiter.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,318 @@
1.4 +/*
1.5 +**********************************************************************
1.6 +* Copyright (c) 2002-2008, International Business Machines
1.7 +* Corporation and others. All Rights Reserved.
1.8 +**********************************************************************
1.9 +*/
1.10 +#ifndef USETITER_H
1.11 +#define USETITER_H
1.12 +
1.13 +#include "unicode/utypes.h"
1.14 +#include "unicode/uobject.h"
1.15 +#include "unicode/unistr.h"
1.16 +
1.17 +/**
1.18 + * \file
1.19 + * \brief C++ API: UnicodeSetIterator iterates over the contents of a UnicodeSet.
1.20 + */
1.21 +
1.22 +U_NAMESPACE_BEGIN
1.23 +
1.24 +class UnicodeSet;
1.25 +class UnicodeString;
1.26 +
1.27 +/**
1.28 + *
1.29 + * UnicodeSetIterator iterates over the contents of a UnicodeSet. It
1.30 + * iterates over either code points or code point ranges. After all
1.31 + * code points or ranges have been returned, it returns the
1.32 + * multicharacter strings of the UnicodeSet, if any.
1.33 + *
1.34 + * This class is not intended to be subclassed. Consider any fields
1.35 + * or methods declared as "protected" to be private. The use of
1.36 + * protected in this class is an artifact of history.
1.37 + *
1.38 + * <p>To iterate over code points and strings, use a loop like this:
1.39 + * <pre>
1.40 + * UnicodeSetIterator it(set);
1.41 + * while (it.next()) {
1.42 + * processItem(it.getString());
1.43 + * }
1.44 + * </pre>
1.45 + * <p>Each item in the set is accessed as a string. Set elements
1.46 + * consisting of single code points are returned as strings containing
1.47 + * just the one code point.
1.48 + *
1.49 + * <p>To iterate over code point ranges, instead of individual code points,
1.50 + * use a loop like this:
1.51 + * <pre>
1.52 + * UnicodeSetIterator it(set);
1.53 + * while (it.nextRange()) {
1.54 + * if (it.isString()) {
1.55 + * processString(it.getString());
1.56 + * } else {
1.57 + * processCodepointRange(it.getCodepoint(), it.getCodepointEnd());
1.58 + * }
1.59 + * }
1.60 + * </pre>
1.61 + * @author M. Davis
1.62 + * @stable ICU 2.4
1.63 + */
1.64 +class U_COMMON_API UnicodeSetIterator : public UObject {
1.65 +
1.66 + protected:
1.67 +
1.68 + /**
1.69 + * Value of <tt>codepoint</tt> if the iterator points to a string.
1.70 + * If <tt>codepoint == IS_STRING</tt>, then examine
1.71 + * <tt>string</tt> for the current iteration result.
1.72 + * @stable ICU 2.4
1.73 + */
1.74 + enum { IS_STRING = -1 };
1.75 +
1.76 + /**
1.77 + * Current code point, or the special value <tt>IS_STRING</tt>, if
1.78 + * the iterator points to a string.
1.79 + * @stable ICU 2.4
1.80 + */
1.81 + UChar32 codepoint;
1.82 +
1.83 + /**
1.84 + * When iterating over ranges using <tt>nextRange()</tt>,
1.85 + * <tt>codepointEnd</tt> contains the inclusive end of the
1.86 + * iteration range, if <tt>codepoint != IS_STRING</tt>. If
1.87 + * iterating over code points using <tt>next()</tt>, or if
1.88 + * <tt>codepoint == IS_STRING</tt>, then the value of
1.89 + * <tt>codepointEnd</tt> is undefined.
1.90 + * @stable ICU 2.4
1.91 + */
1.92 + UChar32 codepointEnd;
1.93 +
1.94 + /**
1.95 + * If <tt>codepoint == IS_STRING</tt>, then <tt>string</tt> points
1.96 + * to the current string. If <tt>codepoint != IS_STRING</tt>, the
1.97 + * value of <tt>string</tt> is undefined.
1.98 + * @stable ICU 2.4
1.99 + */
1.100 + const UnicodeString* string;
1.101 +
1.102 + public:
1.103 +
1.104 + /**
1.105 + * Create an iterator over the given set. The iterator is valid
1.106 + * only so long as <tt>set</tt> is valid.
1.107 + * @param set set to iterate over
1.108 + * @stable ICU 2.4
1.109 + */
1.110 + UnicodeSetIterator(const UnicodeSet& set);
1.111 +
1.112 + /**
1.113 + * Create an iterator over nothing. <tt>next()</tt> and
1.114 + * <tt>nextRange()</tt> return false. This is a convenience
1.115 + * constructor allowing the target to be set later.
1.116 + * @stable ICU 2.4
1.117 + */
1.118 + UnicodeSetIterator();
1.119 +
1.120 + /**
1.121 + * Destructor.
1.122 + * @stable ICU 2.4
1.123 + */
1.124 + virtual ~UnicodeSetIterator();
1.125 +
1.126 + /**
1.127 + * Returns true if the current element is a string. If so, the
1.128 + * caller can retrieve it with <tt>getString()</tt>. If this
1.129 + * method returns false, the current element is a code point or
1.130 + * code point range, depending on whether <tt>next()</tt> or
1.131 + * <tt>nextRange()</tt> was called.
1.132 + * Elements of types string and codepoint can both be retrieved
1.133 + * with the function <tt>getString()</tt>.
1.134 + * Elements of type codepoint can also be retrieved with
1.135 + * <tt>getCodepoint()</tt>.
1.136 + * For ranges, <tt>getCodepoint()</tt> returns the starting codepoint
1.137 + * of the range, and <tt>getCodepointEnd()</tt> returns the end
1.138 + * of the range.
1.139 + * @stable ICU 2.4
1.140 + */
1.141 + inline UBool isString() const;
1.142 +
1.143 + /**
1.144 + * Returns the current code point, if <tt>isString()</tt> returned
1.145 + * false. Otherwise returns an undefined result.
1.146 + * @stable ICU 2.4
1.147 + */
1.148 + inline UChar32 getCodepoint() const;
1.149 +
1.150 + /**
1.151 + * Returns the end of the current code point range, if
1.152 + * <tt>isString()</tt> returned false and <tt>nextRange()</tt> was
1.153 + * called. Otherwise returns an undefined result.
1.154 + * @stable ICU 2.4
1.155 + */
1.156 + inline UChar32 getCodepointEnd() const;
1.157 +
1.158 + /**
1.159 + * Returns the current string, if <tt>isString()</tt> returned
1.160 + * true. If the current iteration item is a code point, a UnicodeString
1.161 + * containing that single code point is returned.
1.162 + *
1.163 + * Ownership of the returned string remains with the iterator.
1.164 + * The string is guaranteed to remain valid only until the iterator is
1.165 + * advanced to the next item, or until the iterator is deleted.
1.166 + *
1.167 + * @stable ICU 2.4
1.168 + */
1.169 + const UnicodeString& getString();
1.170 +
1.171 + /**
1.172 + * Advances the iteration position to the next element in the set,
1.173 + * which can be either a single code point or a string.
1.174 + * If there are no more elements in the set, return false.
1.175 + *
1.176 + * <p>
1.177 + * If <tt>isString() == TRUE</tt>, the value is a
1.178 + * string, otherwise the value is a
1.179 + * single code point. Elements of either type can be retrieved
1.180 + * with the function <tt>getString()</tt>, while elements of
1.181 + * consisting of a single code point can be retrieved with
1.182 + * <tt>getCodepoint()</tt>
1.183 + *
1.184 + * <p>The order of iteration is all code points in sorted order,
1.185 + * followed by all strings sorted order. Do not mix
1.186 + * calls to <tt>next()</tt> and <tt>nextRange()</tt> without
1.187 + * calling <tt>reset()</tt> between them. The results of doing so
1.188 + * are undefined.
1.189 + *
1.190 + * @return true if there was another element in the set.
1.191 + * @stable ICU 2.4
1.192 + */
1.193 + UBool next();
1.194 +
1.195 + /**
1.196 + * Returns the next element in the set, either a code point range
1.197 + * or a string. If there are no more elements in the set, return
1.198 + * false. If <tt>isString() == TRUE</tt>, the value is a
1.199 + * string and can be accessed with <tt>getString()</tt>. Otherwise the value is a
1.200 + * range of one or more code points from <tt>getCodepoint()</tt> to
1.201 + * <tt>getCodepointeEnd()</tt> inclusive.
1.202 + *
1.203 + * <p>The order of iteration is all code points ranges in sorted
1.204 + * order, followed by all strings sorted order. Ranges are
1.205 + * disjoint and non-contiguous. The value returned from <tt>getString()</tt>
1.206 + * is undefined unless <tt>isString() == TRUE</tt>. Do not mix calls to
1.207 + * <tt>next()</tt> and <tt>nextRange()</tt> without calling
1.208 + * <tt>reset()</tt> between them. The results of doing so are
1.209 + * undefined.
1.210 + *
1.211 + * @return true if there was another element in the set.
1.212 + * @stable ICU 2.4
1.213 + */
1.214 + UBool nextRange();
1.215 +
1.216 + /**
1.217 + * Sets this iterator to visit the elements of the given set and
1.218 + * resets it to the start of that set. The iterator is valid only
1.219 + * so long as <tt>set</tt> is valid.
1.220 + * @param set the set to iterate over.
1.221 + * @stable ICU 2.4
1.222 + */
1.223 + void reset(const UnicodeSet& set);
1.224 +
1.225 + /**
1.226 + * Resets this iterator to the start of the set.
1.227 + * @stable ICU 2.4
1.228 + */
1.229 + void reset();
1.230 +
1.231 + /**
1.232 + * ICU "poor man's RTTI", returns a UClassID for this class.
1.233 + *
1.234 + * @stable ICU 2.4
1.235 + */
1.236 + static UClassID U_EXPORT2 getStaticClassID();
1.237 +
1.238 + /**
1.239 + * ICU "poor man's RTTI", returns a UClassID for the actual class.
1.240 + *
1.241 + * @stable ICU 2.4
1.242 + */
1.243 + virtual UClassID getDynamicClassID() const;
1.244 +
1.245 + // ======================= PRIVATES ===========================
1.246 +
1.247 + protected:
1.248 +
1.249 + // endElement and nextElements are really UChar32's, but we keep
1.250 + // them as signed int32_t's so we can do comparisons with
1.251 + // endElement set to -1. Leave them as int32_t's.
1.252 + /** The set
1.253 + * @stable ICU 2.4
1.254 + */
1.255 + const UnicodeSet* set;
1.256 + /** End range
1.257 + * @stable ICU 2.4
1.258 + */
1.259 + int32_t endRange;
1.260 + /** Range
1.261 + * @stable ICU 2.4
1.262 + */
1.263 + int32_t range;
1.264 + /** End element
1.265 + * @stable ICU 2.4
1.266 + */
1.267 + int32_t endElement;
1.268 + /** Next element
1.269 + * @stable ICU 2.4
1.270 + */
1.271 + int32_t nextElement;
1.272 + //UBool abbreviated;
1.273 + /** Next string
1.274 + * @stable ICU 2.4
1.275 + */
1.276 + int32_t nextString;
1.277 + /** String count
1.278 + * @stable ICU 2.4
1.279 + */
1.280 + int32_t stringCount;
1.281 +
1.282 + /**
1.283 + * Points to the string to use when the caller asks for a
1.284 + * string and the current iteration item is a code point, not a string.
1.285 + * @internal
1.286 + */
1.287 + UnicodeString *cpString;
1.288 +
1.289 + /** Copy constructor. Disallowed.
1.290 + * @stable ICU 2.4
1.291 + */
1.292 + UnicodeSetIterator(const UnicodeSetIterator&); // disallow
1.293 +
1.294 + /** Assignment operator. Disallowed.
1.295 + * @stable ICU 2.4
1.296 + */
1.297 + UnicodeSetIterator& operator=(const UnicodeSetIterator&); // disallow
1.298 +
1.299 + /** Load range
1.300 + * @stable ICU 2.4
1.301 + */
1.302 + virtual void loadRange(int32_t range);
1.303 +
1.304 +};
1.305 +
1.306 +inline UBool UnicodeSetIterator::isString() const {
1.307 + return codepoint == (UChar32)IS_STRING;
1.308 +}
1.309 +
1.310 +inline UChar32 UnicodeSetIterator::getCodepoint() const {
1.311 + return codepoint;
1.312 +}
1.313 +
1.314 +inline UChar32 UnicodeSetIterator::getCodepointEnd() const {
1.315 + return codepointEnd;
1.316 +}
1.317 +
1.318 +
1.319 +U_NAMESPACE_END
1.320 +
1.321 +#endif
