From b0da2dcda6f975e77d92340bf3b406d8e47a2e25 Mon Sep 17 00:00:00 2001
From: "dario.genga" <dario.genga@etu.hesge.ch>
Date: Tue, 7 Jun 2022 15:11:19 +0200
Subject: [PATCH] Add export of data from universe
Added a 'dimensions' property for the universe in order to know the
size of a data. This is required for saving each dimension value of a
point.
Added the signature of the 'remove_string_trailing_newline' method in
the 'files_utils.h' file.
Updated the 'open_file' method by adding a mode parameter.
Renamed some variables for more clarity.
---
files_utils.c | 19 ++++++++-----------
files_utils.h | 7 ++++++-
kmeans.c | 43 ++++++++++++++++++++++++++++++++++++++-----
kmeans.h | 4 ++++
main.c | 18 ++++++++++++++++++
output_data.txt | 10 ++++++++++
6 files changed, 84 insertions(+), 17 deletions(-)
create mode 100644 output_data.txt
diff --git a/files_utils.c b/files_utils.c
index 9a8738f..567e56b 100644
--- a/files_utils.c
+++ b/files_utils.c
@@ -4,38 +4,35 @@
#include "files_utils.h"
-FILE* open_file(char *file_path) {
+FILE* open_file(char *file_path, char *mode) {
// Open the file
- FILE *fp = fopen(file_path, "r");
- if (fp == NULL)
+ FILE *file = fopen(file_path, mode);
+ if (file == NULL)
{
perror("Error while opening the file.\n");
exit(EXIT_FAILURE);
}
- return fp;
+ return file;
}
int count_file_lines(char *file_path) {
- FILE *fp = open_file(file_path);
+ FILE *file = open_file(file_path, "r");
int current_line = 0;
char *line = NULL;
- size_t len = 0;
+ size_t buffer_size = 0;
- while(getline(&line, &len, fp) != -1) {
+ while(getline(&line, &buffer_size, file) != -1) {
// Count the lines that are not empty
if (strcmp(line, "\n") != 0) {
current_line++;
}
}
- fclose(fp);
+ fclose(file);
free(line);
return current_line;
}
-/// Remove the newline character in the string. The char must be at the end of the line.
-/// \param str The string to modify.
-/// \see https://siongui.github.io/2013/01/09/c-remove-string-trailing-newline-carriage-return/
void remove_string_trailing_newline(char *str) {
if (str == NULL)
return;
diff --git a/files_utils.h b/files_utils.h
index 3d78c57..45ffb24 100644
--- a/files_utils.h
+++ b/files_utils.h
@@ -12,11 +12,16 @@
/// Open a file.
/// \param file_path The path to the file.
/// \return The pointer of the FILE object.
-FILE *open_file(char *file_path);
+FILE *open_file(char *file_path, char *mode);
/// Count the number of lines in the file.
/// \param file_path The path to the file.
/// \return The number of line in the file.
int count_file_lines(char *file_path);
+/// Remove the newline character in the string. The char must be at the end of the line.
+/// \param str The string to modify.
+/// \see https://siongui.github.io/2013/01/09/c-remove-string-trailing-newline-carriage-return/
+void remove_string_trailing_newline(char *str);
+
#endif
diff --git a/kmeans.c b/kmeans.c
index fcf9aa7..3d2547f 100644
--- a/kmeans.c
+++ b/kmeans.c
@@ -8,6 +8,7 @@ kmeans* kmeans_create_empty() {
universe->points_array = NULL;
universe->nb_points = 0;
universe->k = 0;
+ universe->dimensions = 0;
universe->clusters_array = NULL;
return universe;
@@ -101,26 +102,27 @@ void init_from_cmd_arguments(kmeans *universe) {
void read_data_source(kmeans* universe, char* source_file) {
char *line = NULL;
- size_t len = 0;
+ size_t buffer_size = 0;
ssize_t line_length;
int current_line = 0;
int dimensions = 0;
int point_index = 0;
- // Count the number of points
+ // Count the number of data (points)
int nb_data = count_file_lines(source_file) - 2; // The two first line are not for the data
universe->nb_points = nb_data;
initialize_points_array(universe);
// Open the file and read it, line by line
- FILE *fp = open_file(source_file);
- while((line_length = getline(&line, &len, fp)) != -1) {
+ FILE *file = open_file(source_file, "r");
+ while((line_length = getline(&line, &buffer_size, file)) != -1) {
// Remove newline char at end of line
remove_string_trailing_newline(line);
// Get the number of dimensions for each data
if (current_line == LINE_INDEX_DIMENSIONS) {
dimensions = (int)strtol(line, NULL, 10);
+ universe->dimensions = dimensions;
}
// Get the number of cluster
if (current_line == LINE_INDEX_CLUSTER) {
@@ -140,12 +142,43 @@ void read_data_source(kmeans* universe, char* source_file) {
current_line++;
}
- fclose(fp);
+ // Close the file and free the memory
+ fclose(file);
free(line);
}
void write_data_output(kmeans *universe, char* output_file) {
+ // Open the file
+ FILE *file = open_file(output_file, "w");
+ // Write the number of dimensions
+ fprintf(file, "%d\n", universe->dimensions);
+ // Write the number of clusters
+ fprintf(file, "%d\n", universe->k);
+
+ // Start writing the data of each cluster
+ for (int i = 0; i < universe->k; i++) {
+ // Prints the cluster symbol
+ fprintf(file, CLUSTER_SYMBOL);
+ fprintf(file, "\n");
+
+ for (int j = 0; j < universe->nb_points; j++) {
+ // Prints the point of the cluster
+ if (universe->points_array[j]->cluster == universe->clusters_array[i]) {
+ for (int d = 0; d < universe->dimensions; d++) {
+ fprintf(file, "%.2f", universe->points_array[j]->value[d]);
+ if (d < universe->dimensions - 1) {
+ fprintf(file, SEPARATOR);
+ } else {
+ fprintf(file, "\n");
+ }
+ }
+ }
+ }
+ }
+
+ // Close and save the file
+ fclose(file);
}
void init_clusters(kmeans *universe) {
diff --git a/kmeans.h b/kmeans.h
index f415495..69b86f9 100644
--- a/kmeans.h
+++ b/kmeans.h
@@ -14,6 +14,8 @@
#define LINE_INDEX_DIMENSIONS 0
#define LINE_INDEX_CLUSTER 1
#define LINE_INDEX_CONTENT 2
+#define CLUSTER_SYMBOL "*"
+#define SEPARATOR ";"
/// A group who contains points.
typedef struct _cluster {
@@ -39,6 +41,8 @@ typedef struct _point {
typedef struct _kmeans {
/// The number of clusters in the universe.
int k;
+ /// The number of dimensions for each data in the universe.
+ int dimensions;
/// A 'k' size array of clusters, who contains the data.
struct _cluster** clusters_array;
/// An array of points, representing the data in the universe.
diff --git a/main.c b/main.c
index e66bf46..eacdf12 100644
--- a/main.c
+++ b/main.c
@@ -7,9 +7,27 @@
int main() {
char* path = "./source_data.txt";
+ char* output = "./output_data.txt";
kmeans *universe = kmeans_create_empty();
read_data_source(universe, path);
+ // Custom clustering for testing
+ for (int i = 0; i < universe->nb_points; i++) {
+ int c = 0;
+ if (i % 2 == 0) {
+ c = 1;
+ }
+ else if (i % 3 == 0) {
+ c = 2;
+ }
+ else {
+ c = 0;
+ }
+ universe->points_array[i]->cluster = universe->clusters_array[c];
+ }
+
+ write_data_output(universe, output);
+
destroy_universe(universe);
return EXIT_SUCCESS;
}
diff --git a/output_data.txt b/output_data.txt
new file mode 100644
index 0000000..1c8a7f5
--- /dev/null
+++ b/output_data.txt
@@ -0,0 +1,10 @@
+2
+3
+*
+2.30;33.65
+*
+1.00;24.00
+3.00;4.00
+-1.00;5.00
+*
+5.00;34.00
--
GitLab