1//===--- UnicodeCharRanges.h - Types and functions for character ranges ---===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9#ifndef LLVM_SUPPORT_UNICODECHARRANGES_H
10#define LLVM_SUPPORT_UNICODECHARRANGES_H
11
12#include "llvm/ADT/ArrayRef.h"
13#include "llvm/ADT/SmallPtrSet.h"
14#include "llvm/Support/Compiler.h"
15#include "llvm/Support/Debug.h"
16#include "llvm/Support/Mutex.h"
17#include "llvm/Support/MutexGuard.h"
18#include "llvm/Support/raw_ostream.h"
19
20#include <algorithm>
21
22namespace llvm {
23namespace sys {
24
25/// \brief Represents a closed range of Unicode code points [Lower, Upper].
26struct UnicodeCharRange {
27  uint32_t Lower;
28  uint32_t Upper;
29};
30
31inline bool operator<(uint32_t Value, UnicodeCharRange Range) {
32  return Value < Range.Lower;
33}
34inline bool operator<(UnicodeCharRange Range, uint32_t Value) {
35  return Range.Upper < Value;
36}
37
38/// \brief Holds a reference to an ordered array of UnicodeCharRange and allows
39/// to quickly check if a code point is contained in the set represented by this
40/// array.
41class UnicodeCharSet {
42public:
43  typedef llvm::ArrayRef<UnicodeCharRange> CharRanges;
44
45  /// \brief Constructs a UnicodeCharSet instance from an array of
46  /// UnicodeCharRanges.
47  ///
48  /// Array pointed by \p Ranges should have the lifetime at least as long as
49  /// the UnicodeCharSet instance, and should not change. Array is validated by
50  /// the constructor, so it makes sense to create as few UnicodeCharSet
51  /// instances per each array of ranges, as possible.
52  UnicodeCharSet(CharRanges Ranges) : Ranges(Ranges) {
53    assert(rangesAreValid());
54  }
55
56  /// \brief Returns true if the character set contains the Unicode code point
57  /// \p C.
58  bool contains(uint32_t C) const {
59    return std::binary_search(Ranges.begin(), Ranges.end(), C);
60  }
61
62private:
63  /// \brief Returns true if each of the ranges is a proper closed range
64  /// [min, max], and if the ranges themselves are ordered and non-overlapping.
65  bool rangesAreValid() const {
66    uint32_t Prev = 0;
67    for (CharRanges::const_iterator I = Ranges.begin(), E = Ranges.end();
68         I != E; ++I) {
69      if (I != Ranges.begin() && Prev >= I->Lower) {
70        DEBUG(llvm::dbgs() << "Upper bound 0x");
71        DEBUG(llvm::dbgs().write_hex(Prev));
72        DEBUG(llvm::dbgs() << " should be less than succeeding lower bound 0x");
73        DEBUG(llvm::dbgs().write_hex(I->Lower) << "\n");
74        return false;
75      }
76      if (I->Upper < I->Lower) {
77        DEBUG(llvm::dbgs() << "Upper bound 0x");
78        DEBUG(llvm::dbgs().write_hex(I->Lower));
79        DEBUG(llvm::dbgs() << " should not be less than lower bound 0x");
80        DEBUG(llvm::dbgs().write_hex(I->Upper) << "\n");
81        return false;
82      }
83      Prev = I->Upper;
84    }
85
86    return true;
87  }
88
89  const CharRanges Ranges;
90};
91
92} // namespace sys
93} // namespace llvm
94
95
96#endif // LLVM_SUPPORT_UNICODECHARRANGES_H
97