KMP-MPI/main.c

#include <stdio.h>
#include <string.h>
#include "kmp.h"
#include "util.h"
#include <mpi.h>
#include <malloc.h>

#define MASTER 0
#define DEFAULT_TAG 17

void find_end(int, char *, char *, int **, int *);

void initialize();

void distribute_text();

void distribute_pattern();

void search_for_splitted_pattern(int *residue, int *match_number, int **matches);

void calculate_absolute_indices(int match_number, int *matches);

void collect_results(int *match_number, int *matches);

int rank, size;
int text_len;       // length of all text
int private_text_len;   // length of private text
int pattern_len;    //length of the pattern
char *text;
char *pattern;
int remain = 0;
int *text_piece;
int *displacements;
char *private_text;
int *match_numbers;
int *total_matches;
int total_match_number;

int main() {
    MPI_Init(NULL, NULL);

    MPI_Comm_size(MPI_COMM_WORLD, &size);
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);

    if (rank == MASTER)
        initialize();

    distribute_text();

    printf("%d -> input: %s\n", rank, private_text);
    printf("%d -> text len: %d\n", rank, private_text_len);

    distribute_pattern();

    // create and build lps array
    int lps[pattern_len];
    create_lps(pattern, pattern_len, lps);

    // search for matches
    int match_number = 0;
    int residue;
    int *matches = search_pattern(private_text, pattern, lps, &match_number, &residue);

    search_for_splitted_pattern(&residue, &match_number, &matches);

    calculate_absolute_indices(match_number, matches);

    printf("%d -> result number: %d\n", rank, match_number);

    collect_results(&match_number, matches);

    MPI_Finalize();

    if (rank == MASTER) {
        printf("total matches: %d\n", total_match_number);
        printf("matches index: ");
        print_array(total_matches, total_match_number);
    }
}

void find_end(int residue, char *pattern, char *text, int **matches, int *match_number) {
    int pattern_index = residue;
    int text_index = 0;
    int text_len = strlen(text);
    int pattern_len = strlen(pattern);

    // TODO: match splitted on 3+ node
    if (pattern_len - residue > text_len) return;

    while (pattern_index < pattern_len && pattern[pattern_index] == text[text_index]) {
        pattern_index++;
        text_index++;
    }

    if (pattern_index != pattern_len) return;

    *match_number = *match_number + 1;
    *matches = (int *) realloc(*matches, *match_number * sizeof(int));
    (*matches)[*match_number - 1] = -residue;
}

void initialize() {
    text = read_file("data/text.txt", &text_len);
    pattern = read_file("data/pattern.txt", &pattern_len);

    printf("text: %s\n", text);
    printf("pattern: %s\n", pattern);

    private_text_len = text_len / size;
    remain = text_len % size;

    if (pattern_len > private_text_len)
        MPI_Abort(MPI_COMM_WORLD, 1);

    text_piece = (int *) malloc(sizeof(int) * size);
    displacements = (int *) malloc(sizeof(int) * size);

    displacements[0] = 0;
    for (int i = 0; i < size - 1; ++i) {
        text_piece[i] = private_text_len;
        displacements[i + 1] = displacements[i] + private_text_len;
    }

    text_piece[size - 1] = private_text_len + remain;
}

void distribute_text() {
    // distribution of the length of the text portion to each process
    MPI_Scatter(text_piece, 1, MPI_INT, &private_text_len, 1, MPI_INT, MASTER, MPI_COMM_WORLD);

    // allocation of space for text
    private_text = (char *) malloc(sizeof(char) * (private_text_len + 1));

    // distribution of the text portion to each process
    // scatterv is necessary because the last process receives a larger portion of text in case it is not perfectly divisible.
    MPI_Scatterv(
            text,
            text_piece,
            displacements,
            MPI_CHAR,
            private_text,
            private_text_len,
            MPI_CHAR, MASTER,
            MPI_COMM_WORLD
    );
    private_text[private_text_len] = '\0';
}

void distribute_pattern() {
    // distribution of the length of the pattern to each process
    MPI_Bcast(&pattern_len, 1, MPI_INT, MASTER, MPI_COMM_WORLD);

    // allocation of space for the pattern
    // the master has already done so previously
    if (rank != MASTER) {
        pattern = (char *) malloc(sizeof(char) * (pattern_len + 1));
    }

    // distribution of the pattern to each process
    MPI_Bcast(pattern, pattern_len + 1, MPI_CHAR, MASTER, MPI_COMM_WORLD);
}

void search_for_splitted_pattern(int *residue, int *match_number, int **matches) {
    // send the residue to the next process
    if (rank + 1 != size) {
        MPI_Send(residue, 1, MPI_INT, rank + 1, DEFAULT_TAG, MPI_COMM_WORLD);
    }

    // receiving the residue from the previous process
    if (rank != 0) {
        MPI_Recv(residue, 1, MPI_INT, rank - 1, DEFAULT_TAG, MPI_COMM_WORLD, MPI_STATUS_IGNORE);
        if (*residue != 0)
            // search for a split match
            find_end(*residue, pattern, private_text, matches, match_number);
    }

}

void calculate_absolute_indices(int match_number, int *matches) {
    // sending the remain to the last process
    // this is necessary for the calculation of shift amount
    if (rank == MASTER)
        MPI_Send(&remain, 1, MPI_INT, size - 1, DEFAULT_TAG, MPI_COMM_WORLD);
    else if (rank == size - 1)
        MPI_Recv(&remain, 1, MPI_INT, MASTER, DEFAULT_TAG, MPI_COMM_WORLD, MPI_STATUS_IGNORE);

    // transformation of match indices from relative to absolute position.
    int shift = rank * (private_text_len - remain); // the rest is zero for all but the last process
    apply_shift(shift, matches, match_number);
}

void collect_results(int *match_number, int *matches) {
    if (rank == MASTER) {
        // allocation for the array containing the number of matches of each process
        match_numbers = (int *) malloc(sizeof(int) * size);
    }

    // collection of the number of matches from each process
    MPI_Gather(match_number, 1, MPI_INT, match_numbers, 1, MPI_INT, MASTER, MPI_COMM_WORLD);

    if (rank == MASTER) {
        // preparation of the data structures needed to receive match indices from all processes
        total_match_number = sum_array(match_numbers, size);
        total_matches = (int *) malloc(sizeof(int) * total_match_number);

        displacements[0] = 0;
        for (int i = 0; i < size - 1; ++i) {
            displacements[i + 1] = displacements[i] + match_numbers[i];
        }
    }

    // collection of match indices from each process
    // gatherv is necessary because each process will have a different number of matches
    MPI_Gatherv(
            matches,
            *match_number,
            MPI_INT,
            total_matches,
            match_numbers,
            displacements,
            MPI_INT, MASTER,
            MPI_COMM_WORLD
    );
}