/*
*******************************************************************************
* Copyright (C) 2013-2014, International Business Machines
* Corporation and others.  All Rights Reserved.
*******************************************************************************
* CollationFastLatin.java, ported from collationfastlatin.h/.cpp
*
* C++ version created on: 2013aug09
* created by: Markus W. Scherer
*/

package com.ibm.icu.impl.coll;

import com.ibm.icu.lang.UScript;
import com.ibm.icu.text.Collator;

public final class CollationFastLatin /* all static */ {
    /**
     * Fast Latin format version (one byte 1..FF).
     * Must be incremented for any runtime-incompatible changes,
     * in particular, for changes to any of the following constants.
     *
     * When the major version number of the main data format changes,
     * we can reset this fast Latin version to 1.
     */
    public static final int VERSION = 1;

    public static final int LATIN_MAX = 0x17f;
    public static final int LATIN_LIMIT = LATIN_MAX + 1;

    static final int LATIN_MAX_UTF8_LEAD = 0xc5;  // UTF-8 lead byte of LATIN_MAX

    static final int PUNCT_START = 0x2000;
    static final int PUNCT_LIMIT = 0x2040;

    // excludes U+FFFE & U+FFFF
    static final int NUM_FAST_CHARS = LATIN_LIMIT + (PUNCT_LIMIT - PUNCT_START);

    // Note on the supported weight ranges:
    // Analysis of UCA 6.3 and CLDR 23 non-search tailorings shows that
    // the CEs for characters in the above ranges, excluding expansions with length >2,
    // excluding contractions of >2 characters, and other restrictions
    // (see the builder's getCEsFromCE32()),
    // use at most about 150 primary weights,
    // where about 94 primary weights are possibly-variable (space/punct/symbol/currency),
    // at most 4 secondary before-common weights,
    // at most 4 secondary after-common weights,
    // at most 16 secondary high weights (in secondary CEs), and
    // at most 4 tertiary after-common weights.
    // The following ranges are designed to support slightly more weights than that.
    // (en_US_POSIX is unusual: It creates about 64 variable + 116 Latin primaries.)

    // Digits may use long primaries (preserving more short ones)
    // or short primaries (faster) without changing this data structure.
    // (If we supported numeric collation, then digits would have to have long primaries
    // so that special handling does not affect the fast path.)

    static final int SHORT_PRIMARY_MASK = 0xfc00;  // bits 15..10
    static final int INDEX_MASK = 0x3ff;  // bits 9..0 for expansions & contractions
    static final int SECONDARY_MASK = 0x3e0;  // bits 9..5
    static final int CASE_MASK = 0x18;  // bits 4..3
    static final int LONG_PRIMARY_MASK = 0xfff8;  // bits 15..3
    static final int TERTIARY_MASK = 7;  // bits 2..0
    static final int CASE_AND_TERTIARY_MASK = CASE_MASK | TERTIARY_MASK;

    static final int TWO_SHORT_PRIMARIES_MASK =
            (SHORT_PRIMARY_MASK << 16) | SHORT_PRIMARY_MASK;  // 0xfc00fc00
    static final int TWO_LONG_PRIMARIES_MASK =
            (LONG_PRIMARY_MASK << 16) | LONG_PRIMARY_MASK;  // 0xfff8fff8
    static final int TWO_SECONDARIES_MASK =
            (SECONDARY_MASK << 16) | SECONDARY_MASK;  // 0x3e003e0
    static final int TWO_CASES_MASK =
            (CASE_MASK << 16) | CASE_MASK;  // 0x180018
    static final int TWO_TERTIARIES_MASK =
            (TERTIARY_MASK << 16) | TERTIARY_MASK;  // 0x70007

    /**
     * Contraction with one fast Latin character.
     * Use INDEX_MASK to find the start of the contraction list after the fixed table.
     * The first entry contains the default mapping.
     * Otherwise use CONTR_CHAR_MASK for the contraction character index
     * (in ascending order).
     * Use CONTR_LENGTH_SHIFT for the length of the entry
     * (1=BAIL_OUT, 2=one CE, 3=two CEs).
     *
     * Also, U+0000 maps to a contraction entry, so that the fast path need not
     * check for NUL termination.
     * It usually maps to a contraction list with only the completely ignorable default value.
     */
    static final int CONTRACTION = 0x400;
    /**
     * An expansion encodes two CEs.
     * Use INDEX_MASK to find the pair of CEs after the fixed table.
     *
     * The higher a mini CE value, the easier it is to process.
     * For expansions and higher, no context needs to be considered.
     */
    static final int EXPANSION = 0x800;
    /**
     * Encodes one CE with a long/low mini primary (there are 128).
     * All potentially-variable primaries must be in this range,
     * to make the short-primary path as fast as possible.
     */
    static final int MIN_LONG = 0xc00;
    static final int LONG_INC = 8;
    static final int MAX_LONG = 0xff8;
    /**
     * Encodes one CE with a short/high primary (there are 60),
     * plus a secondary CE if the secondary weight is high.
     * Fast handling: At least all letter primaries should be in this range.
     */
    static final int MIN_SHORT = 0x1000;
    static final int SHORT_INC = 0x400;
    /** The highest primary weight is reserved for U+FFFF. */
    static final int MAX_SHORT = SHORT_PRIMARY_MASK;

    static final int MIN_SEC_BEFORE = 0;  // must add SEC_OFFSET
    static final int SEC_INC = 0x20;
    static final int MAX_SEC_BEFORE = MIN_SEC_BEFORE + 4 * SEC_INC;  // 5 before common
    static final int COMMON_SEC = MAX_SEC_BEFORE + SEC_INC;
    static final int MIN_SEC_AFTER = COMMON_SEC + SEC_INC;
    static final int MAX_SEC_AFTER = MIN_SEC_AFTER + 5 * SEC_INC;  // 6 after common
    static final int MIN_SEC_HIGH = MAX_SEC_AFTER + SEC_INC;  // 20 high secondaries
    static final int MAX_SEC_HIGH = SECONDARY_MASK;

    /**
     * Lookup: Add this offset to secondary weights, except for completely ignorable CEs.
     * Must be greater than any special value, e.g., MERGE_WEIGHT.
     * The exact value is not relevant for the format version.
     */
    static final int SEC_OFFSET = SEC_INC;
    static final int COMMON_SEC_PLUS_OFFSET = COMMON_SEC + SEC_OFFSET;

    static final int TWO_SEC_OFFSETS =
            (SEC_OFFSET << 16) | SEC_OFFSET;  // 0x200020
    static final int TWO_COMMON_SEC_PLUS_OFFSET =
            (COMMON_SEC_PLUS_OFFSET << 16) | COMMON_SEC_PLUS_OFFSET;

    static final int LOWER_CASE = 8;  // case bits include this offset
    static final int TWO_LOWER_CASES = (LOWER_CASE << 16) | LOWER_CASE;  // 0x80008

    static final int COMMON_TER = 0;  // must add TER_OFFSET
    static final int MAX_TER_AFTER = 7;  // 7 after common

    /**
     * Lookup: Add this offset to tertiary weights, except for completely ignorable CEs.
     * Must be greater than any special value, e.g., MERGE_WEIGHT.
     * Must be greater than case bits as well, so that with combined case+tertiary weights
     * plus the offset the tertiary bits does not spill over into the case bits.
     * The exact value is not relevant for the format version.
     */
    static final int TER_OFFSET = SEC_OFFSET;
    static final int COMMON_TER_PLUS_OFFSET = COMMON_TER + TER_OFFSET;

    static final int TWO_TER_OFFSETS = (TER_OFFSET << 16) | TER_OFFSET;
    static final int TWO_COMMON_TER_PLUS_OFFSET =
            (COMMON_TER_PLUS_OFFSET << 16) | COMMON_TER_PLUS_OFFSET;

    static final int MERGE_WEIGHT = 3;
    static final int EOS = 2;  // end of string
    static final int BAIL_OUT = 1;

    /**
     * Contraction result first word bits 8..0 contain the
     * second contraction character, as a char index 0..NUM_FAST_CHARS-1.
     * Each contraction list is terminated with a word containing CONTR_CHAR_MASK.
     */
    static final int CONTR_CHAR_MASK = 0x1ff;
    /**
     * Contraction result first word bits 10..9 contain the result length:
     * 1=bail out, 2=one mini CE, 3=two mini CEs
     */
    static final int CONTR_LENGTH_SHIFT = 9;

    /**
     * Comparison return value when the regular comparison must be used.
     * The exact value is not relevant for the format version.
     */
    public static final int BAIL_OUT_RESULT = -2;

    static int getCharIndex(char c) {
        if(c <= LATIN_MAX) {
            return c;
        } else if(PUNCT_START <= c && c < PUNCT_LIMIT) {
            return c - (PUNCT_START - LATIN_LIMIT);
        } else {
            // Not a fast Latin character.
            // Note: U+FFFE & U+FFFF are forbidden in tailorings
            // and thus do not occur in any contractions.
            return -1;
        }
    }

    /**
     * Computes the options value for the compare functions
     * and writes the precomputed primary weights.
     * Returns -1 if the Latin fastpath is not supported for the data and settings.
     * The capacity must be LATIN_LIMIT.
     */
    public static int getOptions(CollationData data, CollationSettings settings,
            char[] primaries) {
        char[] header = data.fastLatinTableHeader;
        if(header == null) { return -1; }
        assert((header[0] >> 8) == VERSION);
        assert(primaries.length == LATIN_LIMIT);
        if(primaries.length != LATIN_LIMIT) { return -1; }

        int miniVarTop;
        if((settings.options & CollationSettings.ALTERNATE_MASK) == 0) {
            // No mini primaries are variable, set a variableTop just below the
            // lowest long mini primary.
            miniVarTop = MIN_LONG - 1;
        } else {
            int v1 = (int)(settings.variableTop >> 24);
            int headerLength = header[0] & 0xff;
            int i = headerLength - 1;
            if(i <= 0 || v1 > (header[i] & 0x7f)) {
                return -1;  // variableTop >= digits, should not occur
            }
            while(i > 1 && v1 <= (header[i - 1] & 0x7f)) { --i; }
            // In the table header, the miniVarTop is in bits 15..7, with 4 zero bits 19..16 implied.
            // Shift right to make it comparable with long mini primaries in bits 15..3.
            miniVarTop = (header[i] & 0xff80) >> 4;
        }

        byte[] reorderTable = settings.reorderTable;
        if(reorderTable != null) {
            char[] scripts = data.scripts;
            int length = data.scripts.length;
            int prevLastByte = 0;
            for(int i = 0; i < length;) {
                // reordered last byte of the group
                int lastByte = reorderTable[scripts[i] & 0xff] & 0xff;
                if(lastByte < prevLastByte) {
                    // The permutation affects the groups up to Latin.
                    return -1;
                }
                if(scripts[i + 2] == UScript.LATIN) { break; }
                i = i + 2 + scripts[i + 1];
                prevLastByte = lastByte;
            }
        }

        char[] table = data.fastLatinTable;  // skip the header
        for(int c = 0; c < LATIN_LIMIT; ++c) {
            int p = table[c];
            if(p >= MIN_SHORT) {
                p &= SHORT_PRIMARY_MASK;
            } else if(p > miniVarTop) {
                p &= LONG_PRIMARY_MASK;
            } else {
                p = 0;
            }
            primaries[c] = (char)p;
        }
        if((settings.options & CollationSettings.NUMERIC) != 0) {
            // Bail out for digits.
            for(int c = 0x30; c <= 0x39; ++c) { primaries[c] = 0; }
        }

        // Shift the miniVarTop above other options.
        return (miniVarTop << 16) | settings.options;
    }

    public static int compareUTF16(char[] table, char[] primaries, int options,
            CharSequence left, CharSequence right, int startIndex) {
        // This is a modified copy of CollationCompare.compareUpToQuaternary(),
        // optimized for common Latin text.
        // Keep them in sync!

        int variableTop = options >> 16;  // see getOptions()
        options &= 0xffff;  // needed for CollationSettings.getStrength() to work

        // Check for supported characters, fetch mini CEs, and compare primaries.
        int leftIndex = startIndex, rightIndex = startIndex;
        /**
         * Single mini CE or a pair.
         * The current mini CE is in the lower 16 bits, the next one is in the upper 16 bits.
         * If there is only one, then it is in the lower bits, and the upper bits are 0.
         */
        int leftPair = 0, rightPair = 0;
        for(;;) {
            // We fetch CEs until we get a non-ignorable primary or reach the end.
            while(leftPair == 0) {
                if(leftIndex == left.length()) {
                    leftPair = EOS;
                    break;
                }
                int c = left.charAt(leftIndex++);
                if(c <= LATIN_MAX) {
                    leftPair = primaries[c];
                    if(leftPair != 0) { break; }
                    if(c <= 0x39 && c >= 0x30 && (options & CollationSettings.NUMERIC) != 0) {
                        return BAIL_OUT_RESULT;
                    }
                    leftPair = table[c];
                } else if(PUNCT_START <= c && c < PUNCT_LIMIT) {
                    leftPair = table[c - PUNCT_START + LATIN_LIMIT];
                } else {
                    leftPair = lookup(table, c);
                }
                if(leftPair >= MIN_SHORT) {
                    leftPair &= SHORT_PRIMARY_MASK;
                    break;
                } else if(leftPair > variableTop) {
                    leftPair &= LONG_PRIMARY_MASK;
                    break;
                } else {
                    long pairAndInc = nextPair(table, c, leftPair, left, leftIndex);
                    if(pairAndInc < 0) {
                        ++leftIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    leftPair = (int)pairAndInc;
                    if(leftPair == BAIL_OUT) { return BAIL_OUT_RESULT; }
                    leftPair = getPrimaries(variableTop, leftPair);
                }
            }

            while(rightPair == 0) {
                if(rightIndex == right.length()) {
                    rightPair = EOS;
                    break;
                }
                int c = right.charAt(rightIndex++);
                if(c <= LATIN_MAX) {
                    rightPair = primaries[c];
                    if(rightPair != 0) { break; }
                    if(c <= 0x39 && c >= 0x30 && (options & CollationSettings.NUMERIC) != 0) {
                        return BAIL_OUT_RESULT;
                    }
                    rightPair = table[c];
                } else if(PUNCT_START <= c && c < PUNCT_LIMIT) {
                    rightPair = table[c - PUNCT_START + LATIN_LIMIT];
                } else {
                    rightPair = lookup(table, c);
                }
                if(rightPair >= MIN_SHORT) {
                    rightPair &= SHORT_PRIMARY_MASK;
                    break;
                } else if(rightPair > variableTop) {
                    rightPair &= LONG_PRIMARY_MASK;
                    break;
                } else {
                    long pairAndInc = nextPair(table, c, rightPair, right, rightIndex);
                    if(pairAndInc < 0) {
                        ++rightIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    rightPair = (int)pairAndInc;
                    if(rightPair == BAIL_OUT) { return BAIL_OUT_RESULT; }
                    rightPair = getPrimaries(variableTop, rightPair);
                }
            }

            if(leftPair == rightPair) {
                if(leftPair == EOS) { break; }
                leftPair = rightPair = 0;
                continue;
            }
            int leftPrimary = leftPair & 0xffff;
            int rightPrimary = rightPair & 0xffff;
            if(leftPrimary != rightPrimary) {
                // Return the primary difference.
                return (leftPrimary < rightPrimary) ? Collation.LESS : Collation.GREATER;
            }
            if(leftPair == EOS) { break; }
            leftPair >>>= 16;
            rightPair >>>= 16;
        }
        // In the following, we need to re-fetch each character because we did not buffer the CEs,
        // but we know that the string is well-formed and
        // only contains supported characters and mappings.

        // We might skip the secondary level but continue with the case level
        // which is turned on separately.
        if(CollationSettings.getStrength(options) >= Collator.SECONDARY) {
            leftIndex = rightIndex = startIndex;
            leftPair = rightPair = 0;
            for(;;) {
                while(leftPair == 0) {
                    if(leftIndex == left.length()) {
                        leftPair = EOS;
                        break;
                    }
                    int c = left.charAt(leftIndex++);
                    if(c <= LATIN_MAX) {
                        leftPair = table[c];
                    } else if(PUNCT_START <= c && c < PUNCT_LIMIT) {
                        leftPair = table[c - PUNCT_START + LATIN_LIMIT];
                    } else {
                        leftPair = lookup(table, c);
                    }
                    if(leftPair >= MIN_SHORT) {
                        leftPair = getSecondariesFromOneShortCE(leftPair);
                        break;
                    } else if(leftPair > variableTop) {
                        leftPair = COMMON_SEC_PLUS_OFFSET;
                        break;
                    } else {
                        long pairAndInc = nextPair(table, c, leftPair, left, leftIndex);
                        if(pairAndInc < 0) {
                            ++leftIndex;
                            pairAndInc = ~pairAndInc;
                        }
                        leftPair = getSecondaries(variableTop, (int)pairAndInc);
                    }
                }

                while(rightPair == 0) {
                    if(rightIndex == right.length()) {
                        rightPair = EOS;
                        break;
                    }
                    int c = right.charAt(rightIndex++);
                    if(c <= LATIN_MAX) {
                        rightPair = table[c];
                    } else if(PUNCT_START <= c && c < PUNCT_LIMIT) {
                        rightPair = table[c - PUNCT_START + LATIN_LIMIT];
                    } else {
                        rightPair = lookup(table, c);
                    }
                    if(rightPair >= MIN_SHORT) {
                        rightPair = getSecondariesFromOneShortCE(rightPair);
                        break;
                    } else if(rightPair > variableTop) {
                        rightPair = COMMON_SEC_PLUS_OFFSET;
                        break;
                    } else {
                        long pairAndInc = nextPair(table, c, rightPair, right, rightIndex);
                        if(pairAndInc < 0) {
                            ++rightIndex;
                            pairAndInc = ~pairAndInc;
                        }
                        rightPair = getSecondaries(variableTop, (int)pairAndInc);
                    }
                }

                if(leftPair == rightPair) {
                    if(leftPair == EOS) { break; }
                    leftPair = rightPair = 0;
                    continue;
                }
                int leftSecondary = leftPair & 0xffff;
                int rightSecondary = rightPair & 0xffff;
                if(leftSecondary != rightSecondary) {
                    if((options & CollationSettings.BACKWARD_SECONDARY) != 0) {
                        // Full support for backwards secondary requires backwards contraction matching
                        // and moving backwards between merge separators.
                        return BAIL_OUT_RESULT;
                    }
                    return (leftSecondary < rightSecondary) ? Collation.LESS : Collation.GREATER;
                }
                if(leftPair == EOS) { break; }
                leftPair >>>= 16;
                rightPair >>>= 16;
            }
        }

        if((options & CollationSettings.CASE_LEVEL) != 0) {
            boolean strengthIsPrimary = CollationSettings.getStrength(options) == Collator.PRIMARY;
            leftIndex = rightIndex = startIndex;
            leftPair = rightPair = 0;
            for(;;) {
                while(leftPair == 0) {
                    if(leftIndex == left.length()) {
                        leftPair = EOS;
                        break;
                    }
                    int c = left.charAt(leftIndex++);
                    leftPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                    if(leftPair < MIN_LONG) {
                        long pairAndInc = nextPair(table, c, leftPair, left, leftIndex);
                        if(pairAndInc < 0) {
                            ++leftIndex;
                            pairAndInc = ~pairAndInc;
                        }
                        leftPair = (int)pairAndInc;
                    }
                    leftPair = getCases(variableTop, strengthIsPrimary, leftPair);
                }

                while(rightPair == 0) {
                    if(rightIndex == right.length()) {
                        rightPair = EOS;
                        break;
                    }
                    int c = right.charAt(rightIndex++);
                    rightPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                    if(rightPair < MIN_LONG) {
                        long pairAndInc = nextPair(table, c, rightPair, right, rightIndex);
                        if(pairAndInc < 0) {
                            ++rightIndex;
                            pairAndInc = ~pairAndInc;
                        }
                        rightPair = (int)pairAndInc;
                    }
                    rightPair = getCases(variableTop, strengthIsPrimary, rightPair);
                }

                if(leftPair == rightPair) {
                    if(leftPair == EOS) { break; }
                    leftPair = rightPair = 0;
                    continue;
                }
                int leftCase = leftPair & 0xffff;
                int rightCase = rightPair & 0xffff;
                if(leftCase != rightCase) {
                    if((options & CollationSettings.UPPER_FIRST) == 0) {
                        return (leftCase < rightCase) ? Collation.LESS : Collation.GREATER;
                    } else {
                        return (leftCase < rightCase) ? Collation.GREATER : Collation.LESS;
                    }
                }
                if(leftPair == EOS) { break; }
                leftPair >>>= 16;
                rightPair >>>= 16;
            }
        }
        if(CollationSettings.getStrength(options) <= Collator.SECONDARY) { return Collation.EQUAL; }

        // Remove the case bits from the tertiary weight when caseLevel is on or caseFirst is off.
        boolean withCaseBits = CollationSettings.isTertiaryWithCaseBits(options);

        leftIndex = rightIndex = startIndex;
        leftPair = rightPair = 0;
        for(;;) {
            while(leftPair == 0) {
                if(leftIndex == left.length()) {
                    leftPair = EOS;
                    break;
                }
                int c = left.charAt(leftIndex++);
                leftPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                if(leftPair < MIN_LONG) {
                    long pairAndInc = nextPair(table, c, leftPair, left, leftIndex);
                    if(pairAndInc < 0) {
                        ++leftIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    leftPair = (int)pairAndInc;
                }
                leftPair = getTertiaries(variableTop, withCaseBits, leftPair);
            }

            while(rightPair == 0) {
                if(rightIndex == right.length()) {
                    rightPair = EOS;
                    break;
                }
                int c = right.charAt(rightIndex++);
                rightPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                if(rightPair < MIN_LONG) {
                    long pairAndInc = nextPair(table, c, rightPair, right, rightIndex);
                    if(pairAndInc < 0) {
                        ++rightIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    rightPair = (int)pairAndInc;
                }
                rightPair = getTertiaries(variableTop, withCaseBits, rightPair);
            }

            if(leftPair == rightPair) {
                if(leftPair == EOS) { break; }
                leftPair = rightPair = 0;
                continue;
            }
            int leftTertiary = leftPair & 0xffff;
            int rightTertiary = rightPair & 0xffff;
            if(leftTertiary != rightTertiary) {
                if(CollationSettings.sortsTertiaryUpperCaseFirst(options)) {
                    // Pass through EOS and MERGE_WEIGHT
                    // and keep real tertiary weights larger than the MERGE_WEIGHT.
                    // Tertiary CEs (secondary ignorables) are not supported in fast Latin.
                    if(leftTertiary > MERGE_WEIGHT) {
                        leftTertiary ^= CASE_MASK;
                    }
                    if(rightTertiary > MERGE_WEIGHT) {
                        rightTertiary ^= CASE_MASK;
                    }
                }
                return (leftTertiary < rightTertiary) ? Collation.LESS : Collation.GREATER;
            }
            if(leftPair == EOS) { break; }
            leftPair >>>= 16;
            rightPair >>>= 16;
        }
        if(CollationSettings.getStrength(options) <= Collator.TERTIARY) { return Collation.EQUAL; }

        leftIndex = rightIndex = startIndex;
        leftPair = rightPair = 0;
        for(;;) {
            while(leftPair == 0) {
                if(leftIndex == left.length()) {
                    leftPair = EOS;
                    break;
                }
                int c = left.charAt(leftIndex++);
                leftPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                if(leftPair < MIN_LONG) {
                    long pairAndInc = nextPair(table, c, leftPair, left, leftIndex);
                    if(pairAndInc < 0) {
                        ++leftIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    leftPair = (int)pairAndInc;
                }
                leftPair = getQuaternaries(variableTop, leftPair);
            }

            while(rightPair == 0) {
                if(rightIndex == right.length()) {
                    rightPair = EOS;
                    break;
                }
                int c = right.charAt(rightIndex++);
                rightPair = (c <= LATIN_MAX) ? table[c] : lookup(table, c);
                if(rightPair < MIN_LONG) {
                    long pairAndInc = nextPair(table, c, rightPair, right, rightIndex);
                    if(pairAndInc < 0) {
                        ++rightIndex;
                        pairAndInc = ~pairAndInc;
                    }
                    rightPair = (int)pairAndInc;
                }
                rightPair = getQuaternaries(variableTop, rightPair);
            }

            if(leftPair == rightPair) {
                if(leftPair == EOS) { break; }
                leftPair = rightPair = 0;
                continue;
            }
            int leftQuaternary = leftPair & 0xffff;
            int rightQuaternary = rightPair & 0xffff;
            if(leftQuaternary != rightQuaternary) {
                return (leftQuaternary < rightQuaternary) ? Collation.LESS : Collation.GREATER;
            }
            if(leftPair == EOS) { break; }
            leftPair >>>= 16;
            rightPair >>>= 16;
        }
        return Collation.EQUAL;
    }

    private static int lookup(char[] table, int c) {
        assert(c > LATIN_MAX);
        if(PUNCT_START <= c && c < PUNCT_LIMIT) {
            return table[c - PUNCT_START + LATIN_LIMIT];
        } else if(c == 0xfffe) {
            return MERGE_WEIGHT;
        } else if(c == 0xffff) {
            return MAX_SHORT | COMMON_SEC | LOWER_CASE | COMMON_TER;
        } else {
            return BAIL_OUT;
        }
    }

    /**
     * Java returns a negative result (use the '~' operator) if sIndex is to be incremented.
     * C++ modifies sIndex.
     */
    private static long nextPair(char[] table, int c, int ce, CharSequence s16, int sIndex) {
        if(ce >= MIN_LONG || ce < CONTRACTION) {
            return ce;  // simple or special mini CE
        } else if(ce >= EXPANSION) {
            int index = NUM_FAST_CHARS + (ce & INDEX_MASK);
            return ((long)table[index + 1] << 16) | table[index];
        } else /* ce >= CONTRACTION */ {
            // Contraction list: Default mapping followed by
            // 0 or more single-character contraction suffix mappings.
            int index = NUM_FAST_CHARS + (ce & INDEX_MASK);
            boolean inc = false;  // true if the next char is consumed.
            if(sIndex != s16.length()) {
                // Read the next character.
                int c2;
                int nextIndex = sIndex;
                c2 = s16.charAt(nextIndex++);
                if(c2 > LATIN_MAX) {
                    if(PUNCT_START <= c2 && c2 < PUNCT_LIMIT) {
                        c2 = c2 - PUNCT_START + LATIN_LIMIT;  // 2000..203F -> 0180..01BF
                    } else if(c2 == 0xfffe || c2 == 0xffff) {
                        c2 = -1;  // U+FFFE & U+FFFF cannot occur in contractions.
                    } else {
                        return BAIL_OUT;
                    }
                }
                // Look for the next character in the contraction suffix list,
                // which is in ascending order of single suffix characters.
                int i = index;
                int head = table[i];  // first skip the default mapping
                int x;
                do {
                    i += head >> CONTR_LENGTH_SHIFT;
                    head = table[i];
                    x = head & CONTR_CHAR_MASK;
                } while(x < c2);
                if(x == c2) {
                    index = i;
                    inc = true;
                }
            }
            // Return the CE or CEs for the default or contraction mapping.
            int length = table[index] >> CONTR_LENGTH_SHIFT;
            if(length == 1) {
                return BAIL_OUT;
            }
            ce = table[index + 1];
            long result;
            if(length == 2) {
                result = ce;
            } else {
                result = ((long)table[index + 2] << 16) | ce;
            }
            return inc ? ~result : result;
        }
    }

    private static int getPrimaries(int variableTop, int pair) {
        int ce = pair & 0xffff;
        if(ce >= MIN_SHORT) { return pair & TWO_SHORT_PRIMARIES_MASK; }
        if(ce > variableTop) { return pair & TWO_LONG_PRIMARIES_MASK; }
        if(ce >= MIN_LONG) { return 0; }  // variable
        return pair;  // special mini CE
    }

    private static int getSecondariesFromOneShortCE(int ce) {
        ce &= SECONDARY_MASK;
        if(ce < MIN_SEC_HIGH) {
            return ce + SEC_OFFSET;
        } else {
            return ((ce + SEC_OFFSET) << 16) | COMMON_SEC_PLUS_OFFSET;
        }
    }

    private static int getSecondaries(int variableTop, int pair) {
        if(pair <= 0xffff) {
            // one mini CE
            if(pair >= MIN_SHORT) {
                pair = getSecondariesFromOneShortCE(pair);
            } else if(pair > variableTop) {
                pair = COMMON_SEC_PLUS_OFFSET;
            } else if(pair >= MIN_LONG) {
                pair = 0;  // variable
            }
            // else special mini CE
        } else {
            int ce = pair & 0xffff;
            if(ce >= MIN_SHORT) {
                pair = (pair & TWO_SECONDARIES_MASK) + TWO_SEC_OFFSETS;
            } else if(ce > variableTop) {
                pair = TWO_COMMON_SEC_PLUS_OFFSET;
            } else {
                assert(ce >= MIN_LONG);
                pair = 0;  // variable
            }
        }
        return pair;
    }

    private static int getCases(int variableTop, boolean strengthIsPrimary, int pair) {
        // Primary+caseLevel: Ignore case level weights of primary ignorables.
        // Otherwise: Ignore case level weights of secondary ignorables.
        // For details see the comments in the CollationCompare class.
        // Tertiary CEs (secondary ignorables) are not supported in fast Latin.
        if(pair <= 0xffff) {
            // one mini CE
            if(pair >= MIN_SHORT) {
                // A high secondary weight means we really have two CEs,
                // a primary CE and a secondary CE.
                int ce = pair;
                pair &= CASE_MASK;  // explicit weight of primary CE
                if(!strengthIsPrimary && (ce & SECONDARY_MASK) >= MIN_SEC_HIGH) {
                    pair |= LOWER_CASE << 16;  // implied weight of secondary CE
                }
            } else if(pair > variableTop) {
                pair = LOWER_CASE;
            } else if(pair >= MIN_LONG) {
                pair = 0;  // variable
            }
            // else special mini CE
        } else {
            // two mini CEs, same primary groups, neither expands like above
            int ce = pair & 0xffff;
            if(ce >= MIN_SHORT) {
                if(strengthIsPrimary && (pair & (SHORT_PRIMARY_MASK << 16)) == 0) {
                    pair &= CASE_MASK;
                } else {
                    pair &= TWO_CASES_MASK;
                }
            } else if(ce > variableTop) {
                pair = TWO_LOWER_CASES;
            } else {
                assert(ce >= MIN_LONG);
                pair = 0;  // variable
            }
        }
        return pair;
    }

    private static int getTertiaries(int variableTop, boolean withCaseBits, int pair) {
        if(pair <= 0xffff) {
            // one mini CE
            if(pair >= MIN_SHORT) {
                // A high secondary weight means we really have two CEs,
                // a primary CE and a secondary CE.
                int ce = pair;
                if(withCaseBits) {
                    pair = (pair & CASE_AND_TERTIARY_MASK) + TER_OFFSET;
                    if((ce & SECONDARY_MASK) >= MIN_SEC_HIGH) {
                        pair |= (LOWER_CASE | COMMON_TER_PLUS_OFFSET) << 16;
                    }
                } else {
                    pair = (pair & TERTIARY_MASK) + TER_OFFSET;
                    if((ce & SECONDARY_MASK) >= MIN_SEC_HIGH) {
                        pair |= COMMON_TER_PLUS_OFFSET << 16;
                    }
                }
            } else if(pair > variableTop) {
                pair = (pair & TERTIARY_MASK) + TER_OFFSET;
                if(withCaseBits) {
                    pair |= LOWER_CASE;
                }
            } else if(pair >= MIN_LONG) {
                pair = 0;  // variable
            }
            // else special mini CE
        } else {
            // two mini CEs, same primary groups, neither expands like above
            int ce = pair & 0xffff;
            if(ce >= MIN_SHORT) {
                if(withCaseBits) {
                    pair &= TWO_CASES_MASK | TWO_TERTIARIES_MASK;
                } else {
                    pair &= TWO_TERTIARIES_MASK;
                }
                pair += TWO_TER_OFFSETS;
            } else if(ce > variableTop) {
                pair = (pair & TWO_TERTIARIES_MASK) + TWO_TER_OFFSETS;
                if(withCaseBits) {
                    pair |= TWO_LOWER_CASES;
                }
            } else {
                assert(ce >= MIN_LONG);
                pair = 0;  // variable
            }
        }
        return pair;
    }

    private static int getQuaternaries(int variableTop, int pair) {
        // Return the primary weight of a variable CE,
        // or the maximum primary weight for a non-variable, not-completely-ignorable CE.
        if(pair <= 0xffff) {
            // one mini CE
            if(pair >= MIN_SHORT) {
                // A high secondary weight means we really have two CEs,
                // a primary CE and a secondary CE.
                if((pair & SECONDARY_MASK) >= MIN_SEC_HIGH) {
                    pair = TWO_SHORT_PRIMARIES_MASK;
                } else {
                    pair = SHORT_PRIMARY_MASK;
                }
            } else if(pair > variableTop) {
                pair = SHORT_PRIMARY_MASK;
            } else if(pair >= MIN_LONG) {
                pair &= LONG_PRIMARY_MASK;  // variable
            }
            // else special mini CE
        } else {
            // two mini CEs, same primary groups, neither expands like above
            int ce = pair & 0xffff;
            if(ce > variableTop) {
                pair = TWO_SHORT_PRIMARIES_MASK;
            } else {
                assert(ce >= MIN_LONG);
                pair &= TWO_LONG_PRIMARIES_MASK;  // variable
            }
        }
        return pair;
    }

    private CollationFastLatin() {}  // no constructor
}