| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | #include "makoveeva_s_number_of_sentence/mpi/include/ops_mpi.hpp" | ||
| 2 | |||
| 3 | #include <mpi.h> | ||
| 4 | |||
| 5 | #include <cstddef> | ||
| 6 | #include <string> | ||
| 7 | #include <vector> | ||
| 8 | |||
| 9 | #include "makoveeva_s_number_of_sentence/common/include/common.hpp" | ||
| 10 | |||
| 11 | namespace makoveeva_s_number_of_sentence { | ||
| 12 | |||
| 13 |
1/2✓ Branch 1 taken 54 times.
✗ Branch 2 not taken.
|
54 | SentencesCounterMPI::SentencesCounterMPI(const InType &in) { |
| 14 | SetTypeOfTask(GetStaticTypeOfTask()); | ||
| 15 | GetInput() = in; | ||
| 16 | 54 | GetOutput() = 0; | |
| 17 | 54 | } | |
| 18 | |||
| 19 | 54 | bool SentencesCounterMPI::ValidationImpl() { | |
| 20 | 54 | return true; | |
| 21 | } | ||
| 22 | |||
| 23 | 54 | bool SentencesCounterMPI::PreProcessingImpl() { | |
| 24 | 54 | return true; | |
| 25 | } | ||
| 26 | |||
| 27 | ✗ | bool SentencesCounterMPI::IsSentenceEnding(char character) { | |
| 28 | ✗ | return character == '.' || character == '!' || character == '?'; | |
| 29 | } | ||
| 30 | |||
| 31 | ✗ | int SentencesCounterMPI::SkipRepeatedPunctuation(const std::string &text, int current_position) { | |
| 32 | 79 | auto position = static_cast<std::size_t>(current_position); | |
| 33 |
8/12✓ Branch 0 taken 8 times.
✓ Branch 1 taken 2 times.
✓ Branch 2 taken 5 times.
✓ Branch 3 taken 3 times.
✓ Branch 4 taken 69 times.
✓ Branch 5 taken 29 times.
✓ Branch 6 taken 19 times.
✓ Branch 7 taken 50 times.
✗ Branch 8 not taken.
✗ Branch 9 not taken.
✗ Branch 10 not taken.
✗ Branch 11 not taken.
|
108 | while (position < text.length() && IsSentenceEnding(text[position])) { |
| 34 | 24 | position++; | |
| 35 | } | ||
| 36 | 84 | return static_cast<int>(position); | |
| 37 | } | ||
| 38 | |||
| 39 |
2/2✓ Branch 0 taken 51 times.
✓ Branch 1 taken 1 times.
|
52 | int SentencesCounterMPI::ProcessTextSegment(const std::string &text_segment, char previous_char) { |
| 40 | int sentence_count = 0; | ||
| 41 | std::size_t index = 0; | ||
| 42 | const auto segment_length = text_segment.length(); | ||
| 43 | |||
| 44 |
2/2✓ Branch 0 taken 51 times.
✓ Branch 1 taken 1 times.
|
52 | if (segment_length > 0) { |
| 45 |
2/2✓ Branch 0 taken 9 times.
✓ Branch 1 taken 42 times.
|
51 | char first_char = text_segment[0]; |
| 46 | |||
| 47 | if (IsSentenceEnding(previous_char) && IsSentenceEnding(first_char)) { | ||
| 48 | 5 | index = static_cast<std::size_t>(SkipRepeatedPunctuation(text_segment, 0)); | |
| 49 | } | ||
| 50 | } | ||
| 51 | |||
| 52 |
2/2✓ Branch 0 taken 435 times.
✓ Branch 1 taken 52 times.
|
487 | while (index < segment_length) { |
| 53 |
2/2✓ Branch 0 taken 79 times.
✓ Branch 1 taken 356 times.
|
435 | char current_char = text_segment[index]; |
| 54 | |||
| 55 | if (IsSentenceEnding(current_char)) { | ||
| 56 | 79 | sentence_count++; | |
| 57 | 158 | index = static_cast<std::size_t>(SkipRepeatedPunctuation(text_segment, static_cast<int>(index + 1))); | |
| 58 | } else { | ||
| 59 | 356 | index++; | |
| 60 | } | ||
| 61 | } | ||
| 62 | |||
| 63 | 52 | return sentence_count; | |
| 64 | } | ||
| 65 | |||
| 66 | 54 | bool SentencesCounterMPI::RunImpl() { | |
| 67 | 54 | int rank = 0; | |
| 68 | 54 | int size = 0; | |
| 69 | 54 | MPI_Comm_rank(MPI_COMM_WORLD, &rank); | |
| 70 | 54 | MPI_Comm_size(MPI_COMM_WORLD, &size); | |
| 71 | |||
| 72 | std::string full_text; | ||
| 73 | 54 | int text_length = 0; | |
| 74 | |||
| 75 |
2/2✓ Branch 0 taken 27 times.
✓ Branch 1 taken 27 times.
|
54 | if (rank == 0) { |
| 76 | full_text = GetInput(); | ||
| 77 | 27 | text_length = static_cast<int>(full_text.length()); | |
| 78 | } | ||
| 79 | |||
| 80 |
1/2✓ Branch 1 taken 54 times.
✗ Branch 2 not taken.
|
54 | MPI_Bcast(&text_length, 1, MPI_INT, 0, MPI_COMM_WORLD); |
| 81 | |||
| 82 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 52 times.
|
54 | if (text_length == 0) { |
| 83 | 2 | GetOutput() = 0; | |
| 84 | 2 | return true; | |
| 85 | } | ||
| 86 | |||
| 87 | 52 | int base_chunk_size = text_length / size; | |
| 88 | 52 | int remaining_chars = text_length % size; | |
| 89 | |||
| 90 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | std::vector<int> chunk_sizes(static_cast<std::size_t>(size)); |
| 91 |
1/4✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
✗ Branch 3 not taken.
✗ Branch 4 not taken.
|
52 | std::vector<int> displacements(static_cast<std::size_t>(size)); |
| 92 | |||
| 93 | int current_displacement = 0; | ||
| 94 |
2/2✓ Branch 0 taken 104 times.
✓ Branch 1 taken 52 times.
|
156 | for (int i = 0; i < size; ++i) { |
| 95 |
2/2✓ Branch 0 taken 82 times.
✓ Branch 1 taken 22 times.
|
186 | chunk_sizes[static_cast<std::size_t>(i)] = base_chunk_size + (i < remaining_chars ? 1 : 0); |
| 96 | 104 | displacements[static_cast<std::size_t>(i)] = current_displacement; | |
| 97 | 104 | current_displacement += chunk_sizes[static_cast<std::size_t>(i)]; | |
| 98 | } | ||
| 99 | |||
| 100 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | int local_chunk_size = chunk_sizes[static_cast<std::size_t>(rank)]; |
| 101 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | std::string local_chunk(static_cast<std::size_t>(local_chunk_size), '\0'); |
| 102 | |||
| 103 |
3/4✓ Branch 0 taken 26 times.
✓ Branch 1 taken 26 times.
✓ Branch 3 taken 52 times.
✗ Branch 4 not taken.
|
78 | MPI_Scatterv(rank == 0 ? full_text.data() : nullptr, chunk_sizes.data(), displacements.data(), MPI_CHAR, |
| 104 | local_chunk.data(), local_chunk_size, MPI_CHAR, 0, MPI_COMM_WORLD); | ||
| 105 | |||
| 106 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | std::vector<char> boundary_chars(static_cast<std::size_t>(size)); |
| 107 |
2/2✓ Branch 0 taken 26 times.
✓ Branch 1 taken 26 times.
|
52 | if (rank == 0) { |
| 108 |
2/2✓ Branch 0 taken 52 times.
✓ Branch 1 taken 26 times.
|
78 | for (int i = 0; i < size; ++i) { |
| 109 |
2/2✓ Branch 0 taken 26 times.
✓ Branch 1 taken 26 times.
|
52 | int chunk_start = displacements[static_cast<std::size_t>(i)]; |
| 110 | 52 | boundary_chars[static_cast<std::size_t>(i)] = | |
| 111 |
2/2✓ Branch 0 taken 26 times.
✓ Branch 1 taken 26 times.
|
52 | (chunk_start > 0) ? full_text[static_cast<std::size_t>(chunk_start - 1)] : '\0'; |
| 112 | } | ||
| 113 | } | ||
| 114 | |||
| 115 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | char previous_char = '\0'; |
| 116 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | MPI_Scatter(boundary_chars.data(), 1, MPI_CHAR, &previous_char, 1, MPI_CHAR, 0, MPI_COMM_WORLD); |
| 117 | |||
| 118 | 52 | int local_sentence_count = ProcessTextSegment(local_chunk, previous_char); | |
| 119 | |||
| 120 | 52 | int total_sentences = 0; | |
| 121 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | MPI_Reduce(&local_sentence_count, &total_sentences, 1, MPI_INT, MPI_SUM, 0, MPI_COMM_WORLD); |
| 122 | |||
| 123 |
1/2✓ Branch 1 taken 52 times.
✗ Branch 2 not taken.
|
52 | MPI_Bcast(&total_sentences, 1, MPI_INT, 0, MPI_COMM_WORLD); |
| 124 | |||
| 125 |
1/2✓ Branch 0 taken 52 times.
✗ Branch 1 not taken.
|
52 | GetOutput() = total_sentences; |
| 126 | return true; | ||
| 127 | } | ||
| 128 | |||
| 129 | 54 | bool SentencesCounterMPI::PostProcessingImpl() { | |
| 130 | 54 | return true; | |
| 131 | } | ||
| 132 | |||
| 133 | } // namespace makoveeva_s_number_of_sentence | ||
| 134 |