RapidFuzz
Loading...
Searching...
No Matches
SplittedSentenceView.hpp
1#pragma once
2#include <algorithm>
3#include <rapidfuzz/details/Range.hpp>
4#include <rapidfuzz/details/type_traits.hpp>
5
6namespace rapidfuzz {
7namespace detail {
8
9template <typename InputIt>
10class SplittedSentenceView {
11public:
12 using CharT = iter_value_t<InputIt>;
13
14 SplittedSentenceView(RangeVec<InputIt> sentence) noexcept(
15 std::is_nothrow_move_constructible<RangeVec<InputIt>>::value)
16 : m_sentence(std::move(sentence))
17 {}
18
19 size_t dedupe();
20 size_t size() const;
21
22 size_t length() const
23 {
24 return size();
25 }
26
27 bool empty() const
28 {
29 return m_sentence.empty();
30 }
31
32 size_t word_count() const
33 {
34 return m_sentence.size();
35 }
36
37 std::vector<CharT> join() const;
38
39 const RangeVec<InputIt>& words() const
40 {
41 return m_sentence;
42 }
43
44private:
45 RangeVec<InputIt> m_sentence;
46};
47
48template <typename InputIt>
49size_t SplittedSentenceView<InputIt>::dedupe()
50{
51 size_t old_word_count = word_count();
52 m_sentence.erase(std::unique(m_sentence.begin(), m_sentence.end()), m_sentence.end());
53 return old_word_count - word_count();
54}
55
56template <typename InputIt>
57size_t SplittedSentenceView<InputIt>::size() const
58{
59 if (m_sentence.empty()) return 0;
60
61 // there is a whitespace between each word
62 size_t result = m_sentence.size() - 1;
63 for (const auto& word : m_sentence) {
64 result += static_cast<size_t>(std::distance(word.begin(), word.end()));
65 }
66
67 return result;
68}
69
70template <typename InputIt>
71auto SplittedSentenceView<InputIt>::join() const -> std::vector<CharT>
72{
73 if (m_sentence.empty()) {
74 return std::vector<CharT>();
75 }
76
77 auto sentence_iter = m_sentence.begin();
78 std::vector<CharT> joined(sentence_iter->begin(), sentence_iter->end());
79 ++sentence_iter;
80 for (; sentence_iter != m_sentence.end(); ++sentence_iter) {
81 joined.push_back(0x20);
82 joined.insert(joined.end(), sentence_iter->begin(), sentence_iter->end());
83 }
84 return joined;
85}
86
87} // namespace detail
88} // namespace rapidfuzz