Gecko [ dallasfobana2023.variationbd.com ]

Name	Size	Permission
alphaindex.h	26.54 KB	-rw-r--r--
appendable.h	8.54 KB	-rw-r--r--
basictz.h	9.99 KB	-rw-r--r--
brkiter.h	27.86 KB	-rw-r--r--
bytestream.h	10.75 KB	-rw-r--r--
bytestrie.h	20.8 KB	-rw-r--r--
bytestriebuilder.h	7.48 KB	-rw-r--r--
calendar.h	106.52 KB	-rw-r--r--
caniter.h	7.47 KB	-rw-r--r--
casemap.h	25.42 KB	-rw-r--r--
char16ptr.h	7.22 KB	-rw-r--r--
chariter.h	24.06 KB	-rw-r--r--
choicfmt.h	24 KB	-rw-r--r--
coleitr.h	13.78 KB	-rw-r--r--
coll.h	56.3 KB	-rw-r--r--
compactdecimalformat.h	6.88 KB	-rw-r--r--
curramt.h	3.67 KB	-rw-r--r--
currpinf.h	7.3 KB	-rw-r--r--
currunit.h	4.02 KB	-rw-r--r--
datefmt.h	40.72 KB	-rw-r--r--
dbbi.h	1.19 KB	-rw-r--r--
dcfmtsym.h	20.94 KB	-rw-r--r--
decimfmt.h	87.54 KB	-rw-r--r--
displayoptions.h	7.08 KB	-rw-r--r--
docmain.h	7.3 KB	-rw-r--r--
dtfmtsym.h	38.23 KB	-rw-r--r--
dtintrv.h	3.85 KB	-rw-r--r--
dtitvfmt.h	49.26 KB	-rw-r--r--
dtitvinf.h	18.63 KB	-rw-r--r--
dtptngen.h	28.64 KB	-rw-r--r--
dtrule.h	8.69 KB	-rw-r--r--
edits.h	20.73 KB	-rw-r--r--
enumset.h	2.08 KB	-rw-r--r--
errorcode.h	4.84 KB	-rw-r--r--
fieldpos.h	8.7 KB	-rw-r--r--
filteredbrk.h	5.37 KB	-rw-r--r--
fmtable.h	24.45 KB	-rw-r--r--
format.h	12.5 KB	-rw-r--r--
formattednumber.h	6.15 KB	-rw-r--r--
formattedvalue.h	9.75 KB	-rw-r--r--
fpositer.h	3.03 KB	-rw-r--r--
gender.h	3.35 KB	-rw-r--r--
gregocal.h	30.03 KB	-rw-r--r--
icudataver.h	1.02 KB	-rw-r--r--
icuplug.h	12.1 KB	-rw-r--r--
idna.h	12.71 KB	-rw-r--r--
listformatter.h	8.59 KB	-rw-r--r--
localebuilder.h	11.08 KB	-rw-r--r--
localematcher.h	26.83 KB	-rw-r--r--
localpointer.h	19.44 KB	-rw-r--r--
locdspnm.h	7.12 KB	-rw-r--r--
locid.h	48.27 KB	-rw-r--r--
measfmt.h	11.42 KB	-rw-r--r--
measunit.h	107.38 KB	-rw-r--r--
measure.h	4.69 KB	-rw-r--r--
messagepattern.h	33.72 KB	-rw-r--r--
msgfmt.h	44.21 KB	-rw-r--r--
normalizer2.h	34.73 KB	-rw-r--r--
normlzr.h	30.97 KB	-rw-r--r--
nounit.h	2.25 KB	-rw-r--r--
numberformatter.h	90.03 KB	-rw-r--r--
numberrangeformatter.h	25.32 KB	-rw-r--r--
numfmt.h	50.26 KB	-rw-r--r--
numsys.h	7.23 KB	-rw-r--r--
parseerr.h	3.08 KB	-rw-r--r--
parsepos.h	5.57 KB	-rw-r--r--
platform.h	27.8 KB	-rw-r--r--
plurfmt.h	25.25 KB	-rw-r--r--
plurrule.h	20.64 KB	-rw-r--r--
ptypes.h	3.49 KB	-rw-r--r--
putil.h	6.32 KB	-rw-r--r--
rbbi.h	32.07 KB	-rw-r--r--
rbnf.h	49.92 KB	-rw-r--r--
rbtz.h	15.77 KB	-rw-r--r--
regex.h	84.45 KB	-rw-r--r--
region.h	9.2 KB	-rw-r--r--
reldatefmt.h	22.36 KB	-rw-r--r--
rep.h	9.38 KB	-rw-r--r--
resbund.h	18.11 KB	-rw-r--r--
schriter.h	6.1 KB	-rw-r--r--
scientificnumberformatter.h	6.44 KB	-rw-r--r--
search.h	22.24 KB	-rw-r--r--
selfmt.h	14.35 KB	-rw-r--r--
simpleformatter.h	12.6 KB	-rw-r--r--
simplenumberformatter.h	8.88 KB	-rw-r--r--
simpletz.h	45.65 KB	-rw-r--r--
smpdtfmt.h	71.85 KB	-rw-r--r--
sortkey.h	11.19 KB	-rw-r--r--
std_string.h	1.05 KB	-rw-r--r--
strenum.h	9.96 KB	-rw-r--r--
stringoptions.h	5.79 KB	-rw-r--r--
stringpiece.h	10.05 KB	-rw-r--r--
stringtriebuilder.h	15.5 KB	-rw-r--r--
stsearch.h	21.44 KB	-rw-r--r--
symtable.h	4.28 KB	-rw-r--r--
tblcoll.h	36.94 KB	-rw-r--r--
timezone.h	45.67 KB	-rw-r--r--
tmunit.h	3.4 KB	-rw-r--r--
tmutamt.h	4.91 KB	-rw-r--r--
tmutfmt.h	7.42 KB	-rw-r--r--
translit.h	65.83 KB	-rw-r--r--
tzfmt.h	42.96 KB	-rw-r--r--
tznames.h	16.85 KB	-rw-r--r--
tzrule.h	34.86 KB	-rw-r--r--
tztrans.h	6.13 KB	-rw-r--r--
ubidi.h	89.61 KB	-rw-r--r--
ubiditransform.h	12.71 KB	-rw-r--r--
ubrk.h	24.43 KB	-rw-r--r--
ucal.h	64.28 KB	-rw-r--r--
ucasemap.h	15.21 KB	-rw-r--r--
ucat.h	5.35 KB	-rw-r--r--
uchar.h	145.7 KB	-rw-r--r--
ucharstrie.h	22.56 KB	-rw-r--r--
ucharstriebuilder.h	7.48 KB	-rw-r--r--
uchriter.h	13.42 KB	-rw-r--r--
uclean.h	11.21 KB	-rw-r--r--
ucnv.h	83.46 KB	-rw-r--r--
ucnv_cb.h	6.58 KB	-rw-r--r--
ucnv_err.h	20.98 KB	-rw-r--r--
ucnvsel.h	6.24 KB	-rw-r--r--
ucol.h	62.7 KB	-rw-r--r--
ucoleitr.h	9.82 KB	-rw-r--r--
uconfig.h	12.31 KB	-rw-r--r--
ucpmap.h	5.54 KB	-rw-r--r--
ucptrie.h	22.51 KB	-rw-r--r--
ucsdet.h	14.69 KB	-rw-r--r--
ucurr.h	16.72 KB	-rw-r--r--
udat.h	62.36 KB	-rw-r--r--
udata.h	15.63 KB	-rw-r--r--
udateintervalformat.h	11.93 KB	-rw-r--r--
udatpg.h	30.13 KB	-rw-r--r--
udisplaycontext.h	5.94 KB	-rw-r--r--
udisplayoptions.h	8.86 KB	-rw-r--r--
uenum.h	7.79 KB	-rw-r--r--
ufieldpositer.h	4.41 KB	-rw-r--r--
uformattable.h	10.97 KB	-rw-r--r--
uformattednumber.h	8.09 KB	-rw-r--r--
uformattedvalue.h	12.25 KB	-rw-r--r--
ugender.h	2.06 KB	-rw-r--r--
uidna.h	33.43 KB	-rw-r--r--
uiter.h	22.75 KB	-rw-r--r--
uldnames.h	10.48 KB	-rw-r--r--
ulistformatter.h	10.78 KB	-rw-r--r--
uloc.h	54.66 KB	-rw-r--r--
ulocale.h	6.35 KB	-rw-r--r--
ulocbuilder.h	16.72 KB	-rw-r--r--
ulocdata.h	11.3 KB	-rw-r--r--
umachine.h	15 KB	-rw-r--r--
umisc.h	1.34 KB	-rw-r--r--
umsg.h	24.25 KB	-rw-r--r--
umutablecptrie.h	8.3 KB	-rw-r--r--
unifilt.h	4 KB	-rw-r--r--
unifunct.h	4.05 KB	-rw-r--r--
unimatch.h	6.1 KB	-rw-r--r--
unirepl.h	3.38 KB	-rw-r--r--
uniset.h	66.85 KB	-rw-r--r--
unistr.h	171.35 KB	-rw-r--r--
unorm.h	20.55 KB	-rw-r--r--
unorm2.h	25.71 KB	-rw-r--r--
unum.h	55.16 KB	-rw-r--r--
unumberformatter.h	19.68 KB	-rw-r--r--
unumberoptions.h	5.23 KB	-rw-r--r--
unumberrangeformatter.h	15.35 KB	-rw-r--r--
unumsys.h	7.26 KB	-rw-r--r--
uobject.h	10.66 KB	-rw-r--r--
upluralrules.h	8.79 KB	-rw-r--r--
uregex.h	71.99 KB	-rw-r--r--
uregion.h	9.81 KB	-rw-r--r--
ureldatefmt.h	16.98 KB	-rw-r--r--
urename.h	140.82 KB	-rw-r--r--
urep.h	5.38 KB	-rw-r--r--
ures.h	36.65 KB	-rw-r--r--
uscript.h	27.8 KB	-rw-r--r--
usearch.h	39.21 KB	-rw-r--r--
uset.h	45.61 KB	-rw-r--r--
usetiter.h	9.63 KB	-rw-r--r--
ushape.h	18 KB	-rw-r--r--
usimplenumberformatter.h	7.46 KB	-rw-r--r--
uspoof.h	80.32 KB	-rw-r--r--
usprep.h	8.19 KB	-rw-r--r--
ustdio.h	38.56 KB	-rw-r--r--
ustream.h	1.89 KB	-rw-r--r--
ustring.h	72.13 KB	-rw-r--r--
ustringtrie.h	3.15 KB	-rw-r--r--
utext.h	58.1 KB	-rw-r--r--
utf.h	7.87 KB	-rw-r--r--
utf16.h	23.35 KB	-rw-r--r--
utf32.h	763 B	-rw-r--r--
utf8.h	30.83 KB	-rw-r--r--
utf_old.h	45.8 KB	-rw-r--r--
utmscale.h	13.78 KB	-rw-r--r--
utrace.h	17.18 KB	-rw-r--r--
utrans.h	25.54 KB	-rw-r--r--
utypes.h	31.06 KB	-rw-r--r--
uvernum.h	6.33 KB	-rw-r--r--
uversion.h	5.99 KB	-rw-r--r--
vtzone.h	20.69 KB	-rw-r--r--

Code Editor : bytestrie.h

// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2010-2012, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  bytestrie.h
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
*   created on: 2010sep25
*   created by: Markus W. Scherer
*/

#ifndef __BYTESTRIE_H__
#define __BYTESTRIE_H__

/**
 * \file
 * \brief C++ API: Trie for mapping byte sequences to integer values.
 */

#include "unicode/utypes.h"

#if U_SHOW_CPLUSPLUS_API

#include "unicode/stringpiece.h"
#include "unicode/uobject.h"
#include "unicode/ustringtrie.h"

class BytesTrieTest;

U_NAMESPACE_BEGIN

class ByteSink;
class BytesTrieBuilder;
class CharString;
class UVector32;

/**
 * Light-weight, non-const reader class for a BytesTrie.
 * Traverses a byte-serialized data structure with minimal state,
 * for mapping byte sequences to non-negative integer values.
 *
 * This class owns the serialized trie data only if it was constructed by
 * the builder's build() method.
 * The public constructor and the copy constructor only alias the data (only copy the pointer).
 * There is no assignment operator.
 *
 * This class is not intended for public subclassing.
 * @stable ICU 4.8
 */
class U_COMMON_API BytesTrie : public UMemory {
public:
    /**
     * Constructs a BytesTrie reader instance.
     *
     * The trieBytes must contain a copy of a byte sequence from the BytesTrieBuilder,
     * starting with the first byte of that sequence.
     * The BytesTrie object will not read more bytes than
     * the BytesTrieBuilder generated in the corresponding build() call.
     *
     * The array is not copied/cloned and must not be modified while
     * the BytesTrie object is in use.
     *
     * @param trieBytes The byte array that contains the serialized trie.
     * @stable ICU 4.8
     */
    BytesTrie(const void *trieBytes)
            : ownedArray_(nullptr), bytes_(static_cast<const uint8_t *>(trieBytes)),
              pos_(bytes_), remainingMatchLength_(-1) {}

/**
     * Destructor.
     * @stable ICU 4.8
     */
    ~BytesTrie();

/**
     * Copy constructor, copies the other trie reader object and its state,
     * but not the byte array which will be shared. (Shallow copy.)
     * @param other Another BytesTrie object.
     * @stable ICU 4.8
     */
    BytesTrie(const BytesTrie &other)
            : ownedArray_(nullptr), bytes_(other.bytes_),
              pos_(other.pos_), remainingMatchLength_(other.remainingMatchLength_) {}

/**
     * Resets this trie to its initial state.
     * @return *this
     * @stable ICU 4.8
     */
    BytesTrie &reset() {
        pos_=bytes_;
        remainingMatchLength_=-1;
        return *this;
    }

/**
     * Returns the state of this trie as a 64-bit integer.
     * The state value is never 0.
     *
     * @return opaque state value
     * @see resetToState64
     * @stable ICU 65
     */
    uint64_t getState64() const {
        return (static_cast<uint64_t>(remainingMatchLength_ + 2) << kState64RemainingShift) |
            (uint64_t)(pos_ - bytes_);
    }

/**
     * Resets this trie to the saved state.
     * Unlike resetToState(State), the 64-bit state value
     * must be from getState64() from the same trie object or
     * from one initialized the exact same way.
     * Because of no validation, this method is faster.
     *
     * @param state The opaque trie state value from getState64().
     * @return *this
     * @see getState64
     * @see resetToState
     * @see reset
     * @stable ICU 65
     */
    BytesTrie &resetToState64(uint64_t state) {
        remainingMatchLength_ = static_cast<int32_t>(state >> kState64RemainingShift) - 2;
        pos_ = bytes_ + (state & kState64PosMask);
        return *this;
    }

/**
     * BytesTrie state object, for saving a trie's current state
     * and resetting the trie back to this state later.
     * @stable ICU 4.8
     */
    class State : public UMemory {
    public:
        /**
         * Constructs an empty State.
         * @stable ICU 4.8
         */
        State() { bytes=nullptr; }
    private:
        friend class BytesTrie;

const uint8_t *bytes;
        const uint8_t *pos;
        int32_t remainingMatchLength;
    };

/**
     * Saves the state of this trie.
     * @param state The State object to hold the trie's state.
     * @return *this
     * @see resetToState
     * @stable ICU 4.8
     */
    const BytesTrie &saveState(State &state) const {
        state.bytes=bytes_;
        state.pos=pos_;
        state.remainingMatchLength=remainingMatchLength_;
        return *this;
    }

/**
     * Resets this trie to the saved state.
     * If the state object contains no state, or the state of a different trie,
     * then this trie remains unchanged.
     * @param state The State object which holds a saved trie state.
     * @return *this
     * @see saveState
     * @see reset
     * @stable ICU 4.8
     */
    BytesTrie &resetToState(const State &state) {
        if(bytes_==state.bytes && bytes_!=nullptr) {
            pos_=state.pos;
            remainingMatchLength_=state.remainingMatchLength;
        }
        return *this;
    }

/**
     * Determines whether the byte sequence so far matches, whether it has a value,
     * and whether another input byte can continue a matching byte sequence.
     * @return The match/value Result.
     * @stable ICU 4.8
     */
    UStringTrieResult current() const;

/**
     * Traverses the trie from the initial state for this input byte.
     * Equivalent to reset().next(inByte).
     * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff.
     *               Values below -0x100 and above 0xff will never match.
     * @return The match/value Result.
     * @stable ICU 4.8
     */
    inline UStringTrieResult first(int32_t inByte) {
        remainingMatchLength_=-1;
        if(inByte<0) {
            inByte+=0x100;
        }
        return nextImpl(bytes_, inByte);
    }

/**
     * Traverses the trie from the current state for this input byte.
     * @param inByte Input byte value. Values -0x100..-1 are treated like 0..0xff.
     *               Values below -0x100 and above 0xff will never match.
     * @return The match/value Result.
     * @stable ICU 4.8
     */
    UStringTrieResult next(int32_t inByte);

/**
     * Traverses the trie from the current state for this byte sequence.
     * Equivalent to
     * \code
     * Result result=current();
     * for(each c in s)
     *   if(!USTRINGTRIE_HAS_NEXT(result)) return USTRINGTRIE_NO_MATCH;
     *   result=next(c);
     * return result;
     * \endcode
     * @param s A string or byte sequence. Can be nullptr if length is 0.
     * @param length The length of the byte sequence. Can be -1 if NUL-terminated.
     * @return The match/value Result.
     * @stable ICU 4.8
     */
    UStringTrieResult next(const char *s, int32_t length);

/**
     * Returns a matching byte sequence's value if called immediately after
     * current()/first()/next() returned USTRINGTRIE_INTERMEDIATE_VALUE or USTRINGTRIE_FINAL_VALUE.
     * getValue() can be called multiple times.
     *
     * Do not call getValue() after USTRINGTRIE_NO_MATCH or USTRINGTRIE_NO_VALUE!
     * @return The value for the byte sequence so far.
     * @stable ICU 4.8
     */
    inline int32_t getValue() const {
        const uint8_t *pos=pos_;
        int32_t leadByte=*pos++;
        // U_ASSERT(leadByte>=kMinValueLead);
        return readValue(pos, leadByte>>1);
    }

/**
     * Determines whether all byte sequences reachable from the current state
     * map to the same value.
     * @param uniqueValue Receives the unique value, if this function returns true.
     *                    (output-only)
     * @return true if all byte sequences reachable from the current state
     *         map to the same value.
     * @stable ICU 4.8
     */
    inline UBool hasUniqueValue(int32_t &uniqueValue) const {
        const uint8_t *pos=pos_;
        // Skip the rest of a pending linear-match node.
        return pos!=nullptr && findUniqueValue(pos+remainingMatchLength_+1, false, uniqueValue);
    }

/**
     * Finds each byte which continues the byte sequence from the current state.
     * That is, each byte b for which it would be next(b)!=USTRINGTRIE_NO_MATCH now.
     * @param out Each next byte is appended to this object.
     *            (Only uses the out.Append(s, length) method.)
     * @return the number of bytes which continue the byte sequence from here
     * @stable ICU 4.8
     */
    int32_t getNextBytes(ByteSink &out) const;

/**
     * Iterator for all of the (byte sequence, value) pairs in a BytesTrie.
     * @stable ICU 4.8
     */
    class U_COMMON_API Iterator : public UMemory {
    public:
        /**
         * Iterates from the root of a byte-serialized BytesTrie.
         * @param trieBytes The trie bytes.
         * @param maxStringLength If 0, the iterator returns full strings/byte sequences.
         *                        Otherwise, the iterator returns strings with this maximum length.
         * @param errorCode Standard ICU error code. Its input value must
         *                  pass the U_SUCCESS() test, or else the function returns
         *                  immediately. Check for U_FAILURE() on output or use with
         *                  function chaining. (See User Guide for details.)
         * @stable ICU 4.8
         */
        Iterator(const void *trieBytes, int32_t maxStringLength, UErrorCode &errorCode);

/**
         * Iterates from the current state of the specified BytesTrie.
         * @param trie The trie whose state will be copied for iteration.
         * @param maxStringLength If 0, the iterator returns full strings/byte sequences.
         *                        Otherwise, the iterator returns strings with this maximum length.
         * @param errorCode Standard ICU error code. Its input value must
         *                  pass the U_SUCCESS() test, or else the function returns
         *                  immediately. Check for U_FAILURE() on output or use with
         *                  function chaining. (See User Guide for details.)
         * @stable ICU 4.8
         */
        Iterator(const BytesTrie &trie, int32_t maxStringLength, UErrorCode &errorCode);

/**
         * Destructor.
         * @stable ICU 4.8
         */
        ~Iterator();

/**
         * Resets this iterator to its initial state.
         * @return *this
         * @stable ICU 4.8
         */
        Iterator &reset();

/**
         * @return true if there are more elements.
         * @stable ICU 4.8
         */
        UBool hasNext() const;

/**
         * Finds the next (byte sequence, value) pair if there is one.
         *
         * If the byte sequence is truncated to the maximum length and does not
         * have a real value, then the value is set to -1.
         * In this case, this "not a real value" is indistinguishable from
         * a real value of -1.
         * @param errorCode Standard ICU error code. Its input value must
         *                  pass the U_SUCCESS() test, or else the function returns
         *                  immediately. Check for U_FAILURE() on output or use with
         *                  function chaining. (See User Guide for details.)
         * @return true if there is another element.
         * @stable ICU 4.8
         */
        UBool next(UErrorCode &errorCode);

/**
         * @return The NUL-terminated byte sequence for the last successful next().
         * @stable ICU 4.8
         */
        StringPiece getString() const;
        /**
         * @return The value for the last successful next().
         * @stable ICU 4.8
         */
        int32_t getValue() const { return value_; }

private:
        UBool truncateAndStop();

const uint8_t *branchNext(const uint8_t *pos, int32_t length, UErrorCode &errorCode);

const uint8_t *bytes_;
        const uint8_t *pos_;
        const uint8_t *initialPos_;
        int32_t remainingMatchLength_;
        int32_t initialRemainingMatchLength_;

CharString *str_;
        int32_t maxLength_;
        int32_t value_;

// The stack stores pairs of integers for backtracking to another
        // outbound edge of a branch node.
        // The first integer is an offset from bytes_.
        // The second integer has the str_->length() from before the node in bits 15..0,
        // and the remaining branch length in bits 24..16. (Bits 31..25 are unused.)
        // (We could store the remaining branch length minus 1 in bits 23..16 and not use bits 31..24,
        // but the code looks more confusing that way.)
        UVector32 *stack_;
    };

private:
    friend class BytesTrieBuilder;
    friend class ::BytesTrieTest;

/**
     * Constructs a BytesTrie reader instance.
     * Unlike the public constructor which just aliases an array,
     * this constructor adopts the builder's array.
     * This constructor is only called by the builder.
     */
    BytesTrie(void *adoptBytes, const void *trieBytes)
            : ownedArray_(static_cast<uint8_t *>(adoptBytes)),
              bytes_(static_cast<const uint8_t *>(trieBytes)),
              pos_(bytes_), remainingMatchLength_(-1) {}

// No assignment operator.
    BytesTrie &operator=(const BytesTrie &other) = delete;

inline void stop() {
        pos_=nullptr;
    }

// Reads a compact 32-bit integer.
    // pos is already after the leadByte, and the lead byte is already shifted right by 1.
    static int32_t readValue(const uint8_t *pos, int32_t leadByte);
    static inline const uint8_t *skipValue(const uint8_t *pos, int32_t leadByte) {
        // U_ASSERT(leadByte>=kMinValueLead);
        if(leadByte>=(kMinTwoByteValueLead<<1)) {
            if(leadByte<(kMinThreeByteValueLead<<1)) {
                ++pos;
            } else if(leadByte<(kFourByteValueLead<<1)) {
                pos+=2;
            } else {
                pos+=3+((leadByte>>1)&1);
            }
        }
        return pos;
    }
    static inline const uint8_t *skipValue(const uint8_t *pos) {
        int32_t leadByte=*pos++;
        return skipValue(pos, leadByte);
    }

// Reads a jump delta and jumps.
    static const uint8_t *jumpByDelta(const uint8_t *pos);

static inline const uint8_t *skipDelta(const uint8_t *pos) {
        int32_t delta=*pos++;
        if(delta>=kMinTwoByteDeltaLead) {
            if(delta<kMinThreeByteDeltaLead) {
                ++pos;
            } else if(delta<kFourByteDeltaLead) {
                pos+=2;
            } else {
                pos+=3+(delta&1);
            }
        }
        return pos;
    }

static inline UStringTrieResult valueResult(int32_t node) {
        return (UStringTrieResult)(USTRINGTRIE_INTERMEDIATE_VALUE-(node&kValueIsFinal));
    }

// Handles a branch node for both next(byte) and next(string).
    UStringTrieResult branchNext(const uint8_t *pos, int32_t length, int32_t inByte);

// Requires remainingLength_<0.
    UStringTrieResult nextImpl(const uint8_t *pos, int32_t inByte);

// Helper functions for hasUniqueValue().
    // Recursively finds a unique value (or whether there is not a unique one)
    // from a branch.
    static const uint8_t *findUniqueValueFromBranch(const uint8_t *pos, int32_t length,
                                                    UBool haveUniqueValue, int32_t &uniqueValue);
    // Recursively finds a unique value (or whether there is not a unique one)
    // starting from a position on a node lead byte.
    static UBool findUniqueValue(const uint8_t *pos, UBool haveUniqueValue, int32_t &uniqueValue);

// Helper functions for getNextBytes().
    // getNextBytes() when pos is on a branch node.
    static void getNextBranchBytes(const uint8_t *pos, int32_t length, ByteSink &out);
    static void append(ByteSink &out, int c);

// BytesTrie data structure
    //
    // The trie consists of a series of byte-serialized nodes for incremental
    // string/byte sequence matching. The root node is at the beginning of the trie data.
    //
    // Types of nodes are distinguished by their node lead byte ranges.
    // After each node, except a final-value node, another node follows to
    // encode match values or continue matching further bytes.
    //
    // Node types:
    //  - Value node: Stores a 32-bit integer in a compact, variable-length format.
    //    The value is for the string/byte sequence so far.
    //    One node bit indicates whether the value is final or whether
    //    matching continues with the next node.
    //  - Linear-match node: Matches a number of bytes.
    //  - Branch node: Branches to other nodes according to the current input byte.
    //    The node byte is the length of the branch (number of bytes to select from)
    //    minus 1. It is followed by a sub-node:
    //    - If the length is at most kMaxBranchLinearSubNodeLength, then
    //      there are length-1 (key, value) pairs and then one more comparison byte.
    //      If one of the key bytes matches, then the value is either a final value for
    //      the string/byte sequence so far, or a "jump" delta to the next node.
    //      If the last byte matches, then matching continues with the next node.
    //      (Values have the same encoding as value nodes.)
    //    - If the length is greater than kMaxBranchLinearSubNodeLength, then
    //      there is one byte and one "jump" delta.
    //      If the input byte is less than the sub-node byte, then "jump" by delta to
    //      the next sub-node which will have a length of length/2.
    //      (The delta has its own compact encoding.)
    //      Otherwise, skip the "jump" delta to the next sub-node
    //      which will have a length of length-length/2.

// Node lead byte values.

// 00..0f: Branch node. If node!=0 then the length is node+1, otherwise
    // the length is one more than the next byte.

// For a branch sub-node with at most this many entries, we drop down
    // to a linear search.
    static const int32_t kMaxBranchLinearSubNodeLength=5;

// 10..1f: Linear-match node, match 1..16 bytes and continue reading the next node.
    static const int32_t kMinLinearMatch=0x10;
    static const int32_t kMaxLinearMatchLength=0x10;

// 20..ff: Variable-length value node.
    // If odd, the value is final. (Otherwise, intermediate value or jump delta.)
    // Then shift-right by 1 bit.
    // The remaining lead byte value indicates the number of following bytes (0..4)
    // and contains the value's top bits.
    static const int32_t kMinValueLead=kMinLinearMatch+kMaxLinearMatchLength;  // 0x20
    // It is a final value if bit 0 is set.
    static const int32_t kValueIsFinal=1;

// Compact value: After testing bit 0, shift right by 1 and then use the following thresholds.
    static const int32_t kMinOneByteValueLead=kMinValueLead/2;  // 0x10
    static const int32_t kMaxOneByteValue=0x40;  // At least 6 bits in the first byte.

static const int32_t kMinTwoByteValueLead=kMinOneByteValueLead+kMaxOneByteValue+1;  // 0x51
    static const int32_t kMaxTwoByteValue=0x1aff;

static const int32_t kMinThreeByteValueLead=kMinTwoByteValueLead+(kMaxTwoByteValue>>8)+1;  // 0x6c
    static const int32_t kFourByteValueLead=0x7e;

// A little more than Unicode code points. (0x11ffff)
    static const int32_t kMaxThreeByteValue=((kFourByteValueLead-kMinThreeByteValueLead)<<16)-1;

static const int32_t kFiveByteValueLead=0x7f;

// Compact delta integers.
    static const int32_t kMaxOneByteDelta=0xbf;
    static const int32_t kMinTwoByteDeltaLead=kMaxOneByteDelta+1;  // 0xc0
    static const int32_t kMinThreeByteDeltaLead=0xf0;
    static const int32_t kFourByteDeltaLead=0xfe;
    static const int32_t kFiveByteDeltaLead=0xff;

static const int32_t kMaxTwoByteDelta=((kMinThreeByteDeltaLead-kMinTwoByteDeltaLead)<<8)-1;  // 0x2fff
    static const int32_t kMaxThreeByteDelta=((kFourByteDeltaLead-kMinThreeByteDeltaLead)<<16)-1;  // 0xdffff

// For getState64():
    // The remainingMatchLength_ is -1..14=(kMaxLinearMatchLength=0x10)-2
    // so we need at least 5 bits for that.
    // We add 2 to store it as a positive value 1..16=kMaxLinearMatchLength.
    static constexpr int32_t kState64RemainingShift = 59;
    static constexpr uint64_t kState64PosMask = (UINT64_C(1) << kState64RemainingShift) - 1;

uint8_t *ownedArray_;

// Fixed value referencing the BytesTrie bytes.
    const uint8_t *bytes_;

// Iterator variables.

// Pointer to next trie byte to read. nullptr if no more matches.
    const uint8_t *pos_;
    // Remaining length of a linear-match node, minus 1. Negative if not in such a node.
    int32_t remainingMatchLength_;
};

U_NAMESPACE_END

#endif /* U_SHOW_CPLUSPLUS_API */

#endif  // __BYTESTRIE_H__

${this.title}

Code Editor : bytestrie.h