Changeset 5727 for icGREP/icgrepdevel
 Timestamp:
 Nov 7, 2017, 6:12:56 PM (21 months ago)
 Location:
 icGREP/icgrepdevel/icgrep
 Files:

 4 edited
Legend:
 Unmodified
 Added
 Removed

icGREP/icgrepdevel/icgrep/UCD/unicode_set.cpp
r5706 r5727 32 32 // 33 33 // Select the correct builtin scan function, dependent on whatever 34 // bitquad_t resolves to, when scan_forw rad_zeroes<bitquad_t> is called.34 // bitquad_t resolves to, when scan_forward_zeroes<bitquad_t> is called. 35 35 template <typename T> int scan_forward_zeroes(T x); 36 36 template <> inline int scan_forward_zeroes<unsigned int>(unsigned int x){return __builtin_ctz(x);} … … 126 126 bool UnicodeSet::empty() const { 127 127 return (mRuns.size() == 1) && typeOf(mRuns.front()) == Empty; 128 } 129 130 /**  * 131 * @brief full 132 **  */ 133 bool UnicodeSet::full() const { 134 return (mRuns.size() == 1) && typeOf(mRuns.front()) == Full; 128 135 } 129 136 … … 424 431 void UnicodeSet::insert(const codepoint_t cp) { 425 432 426 if (LLVM_UNLIKELY(cp > =UNICODE_MAX)) {433 if (LLVM_UNLIKELY(cp > UNICODE_MAX)) { 427 434 throw std::runtime_error(std::to_string(cp) + " exceeds maximum code point."); 428 435 } … … 696 703 assert (n == 1); 697 704 698 if (LLVM_UNLIKELY(mMinCodePoint > = 0x110000)) {705 if (LLVM_UNLIKELY(mMinCodePoint > UNICODE_MAX)) { 699 706 throw std::runtime_error("UnicodeSet iterator exceeded maximum code point."); 700 707 } … … 702 709 bool found = false; 703 710 // Find the start of our interval 704 while ( mBaseCodePoint < 0x110000) {711 while ( mBaseCodePoint <= UNICODE_MAX ) { 705 712 // Find the first nonempty block 706 713 if (typeOf(*mRunIterator) != Mixed) { … … 738 745 739 746 if (!found) { 740 assert (mBaseCodePoint == 0x110000);741 mMinCodePoint = 0x110000;747 assert (mBaseCodePoint == (UNICODE_MAX+1)); 748 mMinCodePoint = (UNICODE_MAX+1); 742 749 return; 743 750 } … … 747 754 found = false; 748 755 // Find the end of our interval 749 while ( mBaseCodePoint < 0x110000) {756 while ( mBaseCodePoint <= UNICODE_MAX ) { 750 757 751 758 // Find the first nonFull block … … 783 790 } 784 791 } 785 // if the very last block is a mixed block and we go past it, the last code point of the range is 0x10FFFF792 // if the very last block is a mixed block and we go past it, the last code point of the range is UNICODE_MAX 786 793 if (!found) { 787 assert (mBaseCodePoint == 0x110000);788 mMaxCodePoint = 0x10FFFF;794 assert (mBaseCodePoint == (UNICODE_MAX+1)); 795 mMaxCodePoint = UNICODE_MAX; 789 796 } 790 797 791 798 assert (mMinCodePoint <= mMaxCodePoint); 792 799 } 793 794 /**  * 795 * @brief Empty Set Constructor796 **  */ 797 UnicodeSet::UnicodeSet( )800 801 /**  * 802 * @brief Empty/Full Set Constructor 803 **  */ 804 UnicodeSet::UnicodeSet(run_type_t emptyOrFull) 798 805 : mRuns(mAllocator) 799 806 , mQuads(mAllocator) 800 807 { 801 append_run(Empty, UNICODE_QUAD_COUNT, mRuns); 808 assert((emptyOrFull == Empty)  (emptyOrFull == Full)); 809 append_run(emptyOrFull, UNICODE_QUAD_COUNT, mRuns); 802 810 assert (verify(mRuns, mQuads)); 803 811 } 804 812 805 813 /**  * 806 814 * @brief Singleton Set Constructor … … 850 858 assert (std::is_sorted(begin, end, [](const interval_t l, const interval_t r) { 851 859 assert (l.first <= l.second); 852 assert (l.second < UNICODE_MAX);860 assert (l.second <= UNICODE_MAX); 853 861 assert (r.first <= r.second); 854 assert (r.second < UNICODE_MAX);862 assert (r.second <= UNICODE_MAX); 855 863 return l.second < r.first; 856 864 })); 
icGREP/icgrepdevel/icgrep/UCD/unicode_set.h
r5632 r5727 98 98 99 99 inline iterator end() const { 100 return iterator(mRuns.cend(), mQuads.cend(), 0x110000); 101 } 102 103 bool empty() const; 104 100 return iterator(mRuns.cend(), mQuads.cend(), UNICODE_MAX+1); 101 } 102 103 bool empty() const; // The set has no members 104 105 bool full() const; // The set has the full set of possible Unicode codepoints. 106 105 107 bool contains(const codepoint_t codepoint) const; 106 108 … … 134 136 bool operator<(const UnicodeSet & other) const; 135 137 136 UnicodeSet( );138 UnicodeSet(run_type_t emptyOrFull = Empty); 137 139 UnicodeSet(const codepoint_t codepoint); 138 140 UnicodeSet(const codepoint_t lo, const codepoint_t hi); … … 141 143 UnicodeSet(std::initializer_list<interval_t>::iterator begin, std::initializer_list<interval_t>::iterator end); 142 144 UnicodeSet(const std::vector<interval_t>::iterator begin, const std::vector<interval_t>::iterator end); 143 145 144 146 inline void swap(UnicodeSet & other); 145 147 inline void swap(UnicodeSet && other); 
icGREP/icgrepdevel/icgrep/cc/multiplex_CCs.cpp
r5630 r5727 16 16 // The breakpoints may be determined by iterating through the interval 17 17 // representation of each CC. For each interval (lo, hi), lo and hi+1 18 // are breakpoints. 18 // are breakpoints. 19 19 // 20 20 // For each breakpoint, a bitset is computed identifying the source CCs whose … … 79 79 // Start a new range. 80 80 range_lo = bkpt_entry.first; 81 if (range_lo > UCD::UNICODE_MAX) continue; 81 82 current_set ^= bkpt_entry.second; 82 83 auto idx_iter = CC_set_to_exclusive_set_map.find(current_set); 
icGREP/icgrepdevel/icgrep/re/re_analysis.cpp
r5723 r5727 89 89 return subtractCC(matchableCodepoints(diff>getLH()), matchableCodepoints(diff>getRH())); 90 90 } else if (const Intersect * e = dyn_cast<Intersect>(re)) { 91 return intersectCC(matchableCodepoints( diff>getLH()), matchableCodepoints(diff>getRH()));91 return intersectCC(matchableCodepoints(e>getLH()), matchableCodepoints(e>getRH())); 92 92 } else if (isa<Any>(re)) { 93 93 return makeCC(0, 0x10FFFF);
Note: See TracChangeset
for help on using the changeset viewer.