michael@0: /* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 4 -*-
michael@0:  * vim: set ts=8 sts=4 et sw=4 tw=99:
michael@0:  *
michael@0:  * Copyright (C) 2012 Apple Inc. All rights reserved.
michael@0:  *
michael@0:  * Redistribution and use in source and binary forms, with or without
michael@0:  * modification, are permitted provided that the following conditions
michael@0:  * are met:
michael@0:  * 1. Redistributions of source code must retain the above copyright
michael@0:  *    notice, this list of conditions and the following disclaimer.
michael@0:  * 2. Redistributions in binary form must reproduce the above copyright
michael@0:  *    notice, this list of conditions and the following disclaimer in the
michael@0:  *    documentation and/or other materials provided with the distribution.
michael@0:  *
michael@0:  * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY
michael@0:  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
michael@0:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
michael@0:  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL APPLE INC. OR
michael@0:  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
michael@0:  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
michael@0:  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
michael@0:  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
michael@0:  * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
michael@0:  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
michael@0:  * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
michael@0:  */
michael@0: 
michael@0: #ifndef yarr_YarrCanonicalizeUCS2_h
michael@0: #define yarr_YarrCanonicalizeUCS2_h
michael@0: 
michael@0: #include <stdint.h>
michael@0: 
michael@0: #include "yarr/wtfbridge.h"
michael@0: 
michael@0: namespace JSC { namespace Yarr {
michael@0: 
michael@0: // This set of data (autogenerated using YarrCanonicalizeUCS2.js into YarrCanonicalizeUCS2.cpp)
michael@0: // provides information for each UCS2 code point as to the set of code points that it should
michael@0: // match under the ES5.1 case insensitive RegExp matching rules, specified in 15.10.2.8.
michael@0: enum UCS2CanonicalizationType {
michael@0:     CanonicalizeUnique,               // No canonically equal values, e.g. 0x0.
michael@0:     CanonicalizeSet,                  // Value indicates a set in characterSetInfo.
michael@0:     CanonicalizeRangeLo,              // Value is positive delta to pair, E.g. 0x41 has value 0x20, -> 0x61.
michael@0:     CanonicalizeRangeHi,              // Value is positive delta to pair, E.g. 0x61 has value 0x20, -> 0x41.
michael@0:     CanonicalizeAlternatingAligned,   // Aligned consequtive pair, e.g. 0x1f4,0x1f5.
michael@0:     CanonicalizeAlternatingUnaligned  // Unaligned consequtive pair, e.g. 0x241,0x242.
michael@0: };
michael@0: struct UCS2CanonicalizationRange { uint16_t begin, end, value, type; };
michael@0: extern const size_t UCS2_CANONICALIZATION_RANGES;
michael@0: extern const uint16_t* const characterSetInfo[];
michael@0: extern const UCS2CanonicalizationRange rangeInfo[];
michael@0: 
michael@0: // This table is similar to the full rangeInfo table, however this maps from UCS2 codepoints to
michael@0: // the set of Latin1 codepoints that could match.
michael@0: enum LatinCanonicalizationType {
michael@0:     CanonicalizeLatinSelf,     // This character is in the Latin1 range, but has no canonical equivalent in the range.
michael@0:     CanonicalizeLatinMask0x20, // One of a pair of characters, under the mask 0x20.
michael@0:     CanonicalizeLatinOther,    // This character is not in the Latin1 range, but canonicalizes to another that is.
michael@0:     CanonicalizeLatinInvalid   // Cannot match against Latin1 input.
michael@0: };
michael@0: struct LatinCanonicalizationRange { uint16_t begin, end, value, type; };
michael@0: extern const size_t LATIN_CANONICALIZATION_RANGES;
michael@0: extern const LatinCanonicalizationRange latinRangeInfo[];
michael@0: 
michael@0: // This searches in log2 time over ~364 entries, so should typically result in 8 compares.
michael@0: inline const UCS2CanonicalizationRange* rangeInfoFor(UChar ch)
michael@0: {
michael@0:     const UCS2CanonicalizationRange* info = rangeInfo;
michael@0:     size_t entries = UCS2_CANONICALIZATION_RANGES;
michael@0: 
michael@0:     while (true) {
michael@0:         size_t candidate = entries >> 1;
michael@0:         const UCS2CanonicalizationRange* candidateInfo = info + candidate;
michael@0:         if (ch < candidateInfo->begin)
michael@0:             entries = candidate;
michael@0:         else if (ch <= candidateInfo->end)
michael@0:             return candidateInfo;
michael@0:         else {
michael@0:             info = candidateInfo + 1;
michael@0:             entries -= (candidate + 1);
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: // Should only be called for characters that have one canonically matching value.
michael@0: inline UChar getCanonicalPair(const UCS2CanonicalizationRange* info, UChar ch)
michael@0: {
michael@0:     ASSERT(ch >= info->begin && ch <= info->end);
michael@0:     switch (info->type) {
michael@0:     case CanonicalizeRangeLo:
michael@0:         return ch + info->value;
michael@0:     case CanonicalizeRangeHi:
michael@0:         return ch - info->value;
michael@0:     case CanonicalizeAlternatingAligned:
michael@0:         return ch ^ 1;
michael@0:     case CanonicalizeAlternatingUnaligned:
michael@0:         return ((ch - 1) ^ 1) + 1;
michael@0:     default:
michael@0:         ASSERT_NOT_REACHED();
michael@0:     }
michael@0:     ASSERT_NOT_REACHED();
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: // Returns true if no other UCS2 codepoint can match this value.
michael@0: inline bool isCanonicallyUnique(UChar ch)
michael@0: {
michael@0:     return rangeInfoFor(ch)->type == CanonicalizeUnique;
michael@0: }
michael@0: 
michael@0: // Returns true if values are equal, under the canonicalization rules.
michael@0: inline bool areCanonicallyEquivalent(UChar a, UChar b)
michael@0: {
michael@0:     const UCS2CanonicalizationRange* info = rangeInfoFor(a);
michael@0:     switch (info->type) {
michael@0:     case CanonicalizeUnique:
michael@0:         return a == b;
michael@0:     case CanonicalizeSet: {
michael@0:         for (const uint16_t* set = characterSetInfo[info->value]; (a = *set); ++set) {
michael@0:             if (a == b)
michael@0:                 return true;
michael@0:         }
michael@0:         return false;
michael@0:     }
michael@0:     case CanonicalizeRangeLo:
michael@0:         return (a == b) || (a + info->value == b);
michael@0:     case CanonicalizeRangeHi:
michael@0:         return (a == b) || (a - info->value == b);
michael@0:     case CanonicalizeAlternatingAligned:
michael@0:         return (a | 1) == (b | 1);
michael@0:     case CanonicalizeAlternatingUnaligned:
michael@0:         return ((a - 1) | 1) == ((b - 1) | 1);
michael@0:     }
michael@0: 
michael@0:     ASSERT_NOT_REACHED();
michael@0:     return false;
michael@0: }
michael@0: 
michael@0: } } // JSC::Yarr
michael@0: 
michael@0: #endif /* yarr_YarrCanonicalizeUCS2_h */