UnicodeCharRanges.h revision dce4a407a24b04eebc6a376f8e62b41aaa7b071f 
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#include <algorithm>
20
21namespace llvm {
22namespace sys {
23
24#define DEBUG_TYPE "unicode"
25
26/// \brief Represents a closed range of Unicode code points [Lower, Upper].
27struct UnicodeCharRange {
28  uint32_t Lower;
29  uint32_t Upper;
30};
31
32inline bool operator<(uint32_t Value, UnicodeCharRange Range) {
33  return Value < Range.Lower;
34}
35inline bool operator<(UnicodeCharRange Range, uint32_t Value) {
36  return Range.Upper < Value;
37}
38
39/// \brief Holds a reference to an ordered array of UnicodeCharRange and allows
40/// to quickly check if a code point is contained in the set represented by this
41/// array.
42class UnicodeCharSet {
43public:
44  typedef ArrayRef<UnicodeCharRange> CharRanges;
45
46  /// \brief Constructs a UnicodeCharSet instance from an array of
47  /// UnicodeCharRanges.
48  ///
49  /// Array pointed by \p Ranges should have the lifetime at least as long as
50  /// the UnicodeCharSet instance, and should not change. Array is validated by
51  /// the constructor, so it makes sense to create as few UnicodeCharSet
52  /// instances per each array of ranges, as possible.
53  UnicodeCharSet(CharRanges Ranges) : Ranges(Ranges) {
54    assert(rangesAreValid());
55  }
56
57  /// \brief Returns true if the character set contains the Unicode code point
58  /// \p C.
59  bool contains(uint32_t C) const {
60    return std::binary_search(Ranges.begin(), Ranges.end(), C);
61  }
62
63private:
64  /// \brief Returns true if each of the ranges is a proper closed range
65  /// [min, max], and if the ranges themselves are ordered and non-overlapping.
66  bool rangesAreValid() const {
67    uint32_t Prev = 0;
68    for (CharRanges::const_iterator I = Ranges.begin(), E = Ranges.end();
69         I != E; ++I) {
70      if (I != Ranges.begin() && Prev >= I->Lower) {
71        DEBUG(dbgs() << "Upper bound 0x");
72        DEBUG(dbgs().write_hex(Prev));
73        DEBUG(dbgs() << " should be less than succeeding lower bound 0x");
74        DEBUG(dbgs().write_hex(I->Lower) << "\n");
75        return false;
76      }
77      if (I->Upper < I->Lower) {
78        DEBUG(dbgs() << "Upper bound 0x");
79        DEBUG(dbgs().write_hex(I->Lower));
80        DEBUG(dbgs() << " should not be less than lower bound 0x");
81        DEBUG(dbgs().write_hex(I->Upper) << "\n");
82        return false;
83      }
84      Prev = I->Upper;
85    }
86
87    return true;
88  }
89
90  const CharRanges Ranges;
91};
92
93#undef DEBUG_TYPE // "unicode"
94
95} // namespace sys
96} // namespace llvm
97
98
99#endif // LLVM_SUPPORT_UNICODECHARRANGES_H
100