From c4ba647889fdb4ea218115fc5d6a3b5903b51018 Mon Sep 17 00:00:00 2001
From: "dario.genga" <dario.genga@etu.hesge.ch>
Date: Tue, 31 May 2022 16:31:42 +0200
Subject: [PATCH] Add import of data from file
The universe can now be created from data in a text file.
Added the files_utils who contains methods to open a file, count the
number of lines and remove trailing newline.
Added methods to create an empty kmeans universe and to initialize
empty clusters array and points array.
Added a method who create a point from a line in the text file.
Fixed a memory leak caused by non freed point value.
---
files_utils.c | 46 +++++++++++++++++
files_utils.h | 22 ++++++++
kmeans.c | 133 ++++++++++++++++++++++++++++++++++++++++++++----
kmeans.h | 10 ++++
main.c | 14 ++---
makefile | 4 +-
source_data.txt | 7 +++
7 files changed, 214 insertions(+), 22 deletions(-)
create mode 100644 files_utils.c
create mode 100644 files_utils.h
create mode 100644 source_data.txt
diff --git a/files_utils.c b/files_utils.c
new file mode 100644
index 0000000..9a8738f
--- /dev/null
+++ b/files_utils.c
@@ -0,0 +1,46 @@
+// Project : K-means
+// Author : Dario GENGA
+
+#include "files_utils.h"
+
+
+FILE* open_file(char *file_path) {
+ // Open the file
+ FILE *fp = fopen(file_path, "r");
+ if (fp == NULL)
+ {
+ perror("Error while opening the file.\n");
+ exit(EXIT_FAILURE);
+ }
+ return fp;
+}
+
+int count_file_lines(char *file_path) {
+ FILE *fp = open_file(file_path);
+ int current_line = 0;
+ char *line = NULL;
+ size_t len = 0;
+
+ while(getline(&line, &len, fp) != -1) {
+ // Count the lines that are not empty
+ if (strcmp(line, "\n") != 0) {
+ current_line++;
+ }
+ }
+
+ fclose(fp);
+ 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;
+
+ int length = strlen(str);
+ if (str[length-1] == '\n')
+ str[length-1] = '\0';
+}
diff --git a/files_utils.h b/files_utils.h
new file mode 100644
index 0000000..3d78c57
--- /dev/null
+++ b/files_utils.h
@@ -0,0 +1,22 @@
+// Project : K-means
+// Author : Dario GENGA
+
+
+#ifndef _FILES_UTILS_H
+#define _FILES_UTILS_H
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+/// Open a file.
+/// \param file_path The path to the file.
+/// \return The pointer of the FILE object.
+FILE *open_file(char *file_path);
+
+/// 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);
+
+#endif
diff --git a/kmeans.c b/kmeans.c
index 80f2259..fcf9aa7 100644
--- a/kmeans.c
+++ b/kmeans.c
@@ -3,16 +3,80 @@
#include "kmeans.h"
+kmeans* kmeans_create_empty() {
+ kmeans* universe = malloc(sizeof(kmeans));
+ universe->points_array = NULL;
+ universe->nb_points = 0;
+ universe->k = 0;
+ universe->clusters_array = NULL;
+
+ return universe;
+}
+
+/// Create empty clusters for the universe
+/// \param universe The universe who contains the cluster to create.
+void initialize_clusters_array(kmeans* universe) {
+ universe->clusters_array = malloc(sizeof(cluster) * universe->k);
+
+ for (int i = 0; i < universe->k; i++) {
+ universe->clusters_array[i] = cluster_create(NULL);
+ }
+}
+
+/// Initialize the array of points in the universe.
+/// \param universe The universe who contains the points.
+void initialize_points_array(kmeans* universe) {
+ universe->points_array = malloc(sizeof(point) * universe->nb_points);
+ for (int i = 0; i < universe->nb_points; i++) {
+ universe->points_array[i] = NULL;
+ }
+}
+
+/// Create a point from the values in a string.
+/// \param line The string who contains the data of the point.
+/// \param dimensions The number of dimension in the point.
+/// \return The point created from the string.
+point* create_point_from_string(char *line, int dimensions) {
+ // Skip the parsing if the line doesn't contain data
+ if (strcmp(line, "\n") == 0) {
+ return NULL;
+ }
+
+ int i = 0;
+ point* p;
+ char *token;
+ const char separator[2] = ",";
+ double* data = malloc(sizeof(double) * dimensions);
+
+ // Parse the line
+ token = strtok(line, separator);
+ while(token != NULL) {
+ // Convert the string value to double, then save it
+ double value = atof(token);
+ data[i] = value;
+
+ // Get the next value
+ token = strtok(NULL, separator);
+ i++;
+ }
+ // Verify the dimensions of the point
+ if (i != dimensions) {
+ printf("Bad dimensions for the point.\n");
+ exit(EXIT_FAILURE);
+ }
+
+ // Create the point and return it
+ p = point_create(data);
+ return p;
+}
+
kmeans* kmeans_create(int k, point** data, int nb_points) {
kmeans* universe = malloc(sizeof(kmeans));
universe->points_array = data;
universe->nb_points = nb_points;
universe->k = k;
- universe->clusters_array = malloc(sizeof(cluster) * k);
-
- for (int i = 0; i < k; i++) {
- universe->clusters_array[i] = cluster_create(NULL);
- }
+ universe->clusters_array = NULL;
+ initialize_clusters_array(universe);
return universe;
}
@@ -35,8 +99,49 @@ void init_from_cmd_arguments(kmeans *universe) {
}
-void read_data_source(kmeans *universe, char* source_file) {
+void read_data_source(kmeans* universe, char* source_file) {
+ char *line = NULL;
+ size_t len = 0;
+ ssize_t line_length;
+ int current_line = 0;
+ int dimensions = 0;
+ int point_index = 0;
+
+ // Count the number of 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) {
+ // 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);
+ }
+ // Get the number of cluster
+ if (current_line == LINE_INDEX_CLUSTER) {
+ universe->k = (int)strtol(line, NULL, 10);
+ initialize_clusters_array(universe);
+ }
+ // Retrieve the data
+ if (current_line >= LINE_INDEX_CONTENT) {
+ point* p = create_point_from_string(line, dimensions);
+
+ if (p != NULL) {
+ universe->points_array[point_index] = p;
+ point_index++;
+ }
+ }
+ current_line++;
+ }
+
+ fclose(fp);
+ free(line);
}
void write_data_output(kmeans *universe, char* output_file) {
@@ -68,6 +173,10 @@ void destroy_point(point* p) {
if (p->label != NULL)
free(p->label);
+ if (p->value != NULL) {
+ free(p->value);
+ }
+
free(p);
}
}
@@ -81,11 +190,15 @@ void destroy_cluster(cluster* clstr) {
void destroy_universe(kmeans* kmeans) {
if (kmeans != NULL) {
- for (int i = 0; i < kmeans->nb_points; i++) {
- destroy_point(kmeans->points_array[i]);
+ if (kmeans->points_array != NULL) {
+ for (int i = 0; i < kmeans->nb_points; i++) {
+ destroy_point(kmeans->points_array[i]);
+ }
}
- for (int i = 0; i < kmeans->k; i++) {
- destroy_cluster(kmeans->clusters_array[i]);
+ if (kmeans->clusters_array != NULL) {
+ for (int i = 0; i < kmeans->k; i++) {
+ destroy_cluster(kmeans->clusters_array[i]);
+ }
}
free(kmeans->points_array);
diff --git a/kmeans.h b/kmeans.h
index 498de3d..f415495 100644
--- a/kmeans.h
+++ b/kmeans.h
@@ -7,7 +7,13 @@
#include <stdio.h>
#include <stdlib.h>
#include <stdbool.h>
+#include <string.h>
#include <math.h>
+#include "files_utils.h"
+
+#define LINE_INDEX_DIMENSIONS 0
+#define LINE_INDEX_CLUSTER 1
+#define LINE_INDEX_CONTENT 2
/// A group who contains points.
typedef struct _cluster {
@@ -41,6 +47,10 @@ typedef struct _kmeans {
int nb_points;
} kmeans;
+/// Create an empty kmeans universe.
+/// \return The kmeans object of the universe.
+kmeans* kmeans_create_empty();
+
/// Create the kmeans universe.
/// \param k The number of clusters in the universe.
/// \param data The array of points
diff --git a/main.c b/main.c
index c586a14..e66bf46 100644
--- a/main.c
+++ b/main.c
@@ -6,18 +6,10 @@
#include "kmeans.h"
int main() {
- // The code below verify the create and destroy methods.
- double d = 5.5;
- double b = 3.3;
- point* gravity = point_create(&b);
- point* p = point_create(&d);
- cluster* c = cluster_create(gravity);
+ char* path = "./source_data.txt";
+ kmeans *universe = kmeans_create_empty();
+ read_data_source(universe, path);
- point** points = malloc(sizeof(points) * 1);
- points[0] = p;
- kmeans* universe = kmeans_create(3, points, 1);
-
- destroy_cluster(c);
destroy_universe(universe);
return EXIT_SUCCESS;
}
diff --git a/makefile b/makefile
index afe312c..47c8687 100644
--- a/makefile
+++ b/makefile
@@ -1,9 +1,11 @@
LIB=-lm
CC=gcc -Wall -Wextra -g
-main: kmeans.o main.o
+main: files_utils.o kmeans.o main.o
$(CC) $^ -fsanitize=address -fsanitize=leak -o $@ $(LIB)
+files_utils.o: files_utils.c files_utils.h
+ $(CC) -c $< $(LIB)
kmeans.o: kmeans.c kmeans.h
$(CC) -c $< $(LIB)
main.o: main.c
diff --git a/source_data.txt b/source_data.txt
new file mode 100644
index 0000000..782668a
--- /dev/null
+++ b/source_data.txt
@@ -0,0 +1,7 @@
+2
+3
+1,24
+2.3,33.65
+3,4
+5,34
+-1,5
--
GitLab