diff --git a/doc/Makefile b/doc/Makefile
index 05189ad57601a6ccd7781a3f0ccd282b1d6e7499..acba9dfc7b74658f961f868ffe1f1d4932b91db4 100644
--- a/doc/Makefile
+++ b/doc/Makefile
@@ -7,7 +7,7 @@ read: ${PDF}
firefox $^
%.pdf: %.md Makefile
- pandoc --pdf-engine=xelatex -o $@ $<
+ pandoc --pdf-engine=lualatex -t beamer -o $@ $<
clean:
rm -rf ${PDF}
diff --git a/doc/kmeans.md b/doc/kmeans.md
index cdd2c5c593ea1ee974ce43d80f4d50fed66588c9..dd06909e5805cd64fcf021f41779b180b98b7edd 100644
--- a/doc/kmeans.md
+++ b/doc/kmeans.md
@@ -2,6 +2,7 @@
title: K-Means - Une Implémentation
author: Boris Stefanovic
date: 2022-05-24
+theme: "Frankfurt"
geometry: "margin=40mm"
mainfont: DejaVu Sans
header-includes:
diff --git a/src/common.h b/src/common.h
index e2fddef821d2ad35b1dcecfe986f04553fb47ec4..5d079229665fe858fc92ea0eebe8d2809a043f7e 100644
--- a/src/common.h
+++ b/src/common.h
@@ -9,7 +9,7 @@
typedef int64_t int_t;
-typedef double fp_t;
+typedef double fpt_t;
#endif //PROG_KMEANS_COMMON_H
diff --git a/src/distance.c b/src/distance.c
index 922e24035a305bc6bb7dcbb10aba233f85a16cc0..a854e6cb876ebd132d4302f36441f229b1707fb2 100644
--- a/src/distance.c
+++ b/src/distance.c
@@ -9,7 +9,17 @@
#define ERROR -1.0
-double distance_euclid_int(const vector_int_t* p1, const vector_int_t* p2) {
+int_t abs_diff_int(const int_t a1, const int_t a2) {
+ int_t diff = a2 - a1;
+ return diff >= 0 ? diff : -diff;
+}
+
+fpt_t abs_diff_fpt(const fpt_t a1, const fpt_t a2) {
+ fpt_t diff = a2 - a1;
+ return diff >= 0.0 ? diff : -diff;
+}
+
+fpt_t distance_euclid_int(const vector_int_t* p1, const vector_int_t* p2) {
if (p1->dim != p2->dim)return ERROR;
int_t acc = 0;
for (size_t i = 0; i < p1->dim; ++i) {
@@ -20,12 +30,40 @@ double distance_euclid_int(const vector_int_t* p1, const vector_int_t* p2) {
return sqrt((double) acc);
}
-int_t abs_diff(const int_t a1, const int_t a2) {
- int_t diff = a2 - a1;
- return diff >= 0 ? diff : -diff;
+fpt_t distance_manhattan_int(const vector_int_t* p1, const vector_int_t* p2) {
+ if (p1->dim != p2->dim)return ERROR;
+ int_t acc = 0;
+ for (size_t i = 0; i < p1->dim; ++i) {
+ int_t diff = p2->data[i] - p1->data[i];
+ int_t item = diff >= 0 ? diff : -diff;
+ acc += item;
+ }
+ return (double) acc;
+}
+
+fpt_t distance_chebyshev_int(const vector_int_t* p1, const vector_int_t* p2) {
+ if (p1->dim != p2->dim)return ERROR;
+ int_t max = ERROR;
+ int_t item;
+ for (size_t i = 0; i < p1->dim; ++i) {
+ item = abs_diff_int(p1->data[i], p2->data[i]);
+ if (item > max) max = item;
+ }
+ return (double) max;
+}
+
+fpt_t distance_euclid_fpt(const vector_int_t* p1, const vector_int_t* p2) {
+ if (p1->dim != p2->dim)return ERROR;
+ int_t acc = 0;
+ for (size_t i = 0; i < p1->dim; ++i) {
+ int_t diff = p2->data[i] - p1->data[i];
+ int_t item = diff * diff;
+ acc += item;
+ }
+ return sqrt((double) acc);
}
-double distance_manhattan_int(const vector_int_t* p1, const vector_int_t* p2) {
+fpt_t distance_manhattan_fpt(const vector_int_t* p1, const vector_int_t* p2) {
if (p1->dim != p2->dim)return ERROR;
int_t acc = 0;
for (size_t i = 0; i < p1->dim; ++i) {
@@ -36,12 +74,12 @@ double distance_manhattan_int(const vector_int_t* p1, const vector_int_t* p2) {
return (double) acc;
}
-double distance_chebyshev_int(const vector_int_t* p1, const vector_int_t* p2) {
+fpt_t distance_chebyshev_fpt(const vector_int_t* p1, const vector_int_t* p2) {
if (p1->dim != p2->dim)return ERROR;
int_t max = ERROR;
int_t item;
for (size_t i = 0; i < p1->dim; ++i) {
- item = abs_diff(p1->data[i], p2->data[i]);
+ item = abs_diff_int(p1->data[i], p2->data[i]);
if (item > max) max = item;
}
return (double) max;
diff --git a/src/distance.h b/src/distance.h
index 47f86eac375cc2d45a7a9b354a2a04b0953b6d3d..a7a2f0ba5aa678d79404ec99730b265d1a57ebfc 100644
--- a/src/distance.h
+++ b/src/distance.h
@@ -15,6 +15,10 @@
*/
+int_t abs_diff_int(int_t a1, int_t a2);
+
+fpt_t abs_diff_fpt(fpt_t a1, fpt_t a2);
+
double distance_euclid_int(const vector_int_t* p1, const vector_int_t* p2);
double distance_manhattan_int(const vector_int_t* p1, const vector_int_t* p2);
diff --git a/src/main.c b/src/main.c
index d675d0b0fcedd2d78b43a1030ec15f7d2ce7fb40..b1cb8d16f4eb2e2b57b686380c788051271ffaa6 100644
--- a/src/main.c
+++ b/src/main.c
@@ -1,13 +1,88 @@
+#define _GNU_SOURCE
+
+#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include "common.h"
+#include "vector.h"
+
+void help(const char* callname) {
+ fprintf(stderr, "\nUSAGE: %s <INPUT_FILE> <OUTPUT_FILE>\n", callname);
+}
+
+int_t read_int(FILE* file) {
+ char* line;
+ size_t len;
+ getline(&line, &len, file);
+ return strtol(line, NULL, 10);
+}
+
+bool read_vector_int(FILE* file, vector_int_t* vector) {
+ // procure line
+ char* line = NULL;
+ size_t len = 0;
+ getline(&line, &len, file);
+ if (len == 0) return false;
+ // tokenise
+ char* toktgt = line;
+ char* token = NULL;
+ for (size_t i = 0; i < vector->dim; ++i, toktgt = NULL) {
+ token = strtok(toktgt, ",");
+ // strtol returns 0 if number not read; desired behaviour:
+ vector->data[i] = token != NULL ? strtol(token, NULL, 10) : 0;
+ }
+ free(line);
+ return true;
+}
+
+vector_int_t* line_to_vector_int(char* line, const size_t dim) {
+ vector_int_t* vector = vector_int_create_zero(dim);
+ char* tgt = line;
+ char* token = NULL;
+ for (size_t i = 0; i < vector->dim; ++i, tgt = NULL) {
+ token = strtok(tgt, ",");
+ // strtol returns 0 if number not read; desired behaviour:
+ vector->data[i] = token != NULL ? strtol(token, NULL, 10) : 0;
+ }
+ return vector;
+}
int main(int argc, char** argv) {
- fprintf(stderr, "USAGE: %s <INPUT_FILE> <OUTPUT_FILE>\n", argv[0]);
- char* ipath = "/dev/stdin";
- char* opath = "/dev/stdout";
- if (argc > 1) ipath = argv[1];
+ if (argc <= 1) help(argv[0]);
+ char* ipath = NULL;
+ char* opath = NULL;
+ if (argc > 1) {
+ ipath = argv[1];
+ if (access(ipath, F_OK) == -1) {
+ fprintf(stderr, "IFILE: [ %s ] file does not exist !", ipath);
+ return EXIT_FAILURE;
+ }
+ }
if (argc > 2) opath = argv[2];
+ // READ
+ FILE* ifile = ipath != NULL ? fopen(ipath, "r") : stdin;
+ const size_t dim = read_int(ifile);
+ const int_t nclusters = read_int(ifile);
+ if (0 <= dim) {
+ printf("DIMENSION MUST BE STRICTLY POSITIVE !\n");
+ return EXIT_FAILURE;
+ }
+ if (0 == nclusters) {
+ printf("NUMBER OF CLUSTERS MUST BE STRICTLY POSITIVE !\n");
+ return EXIT_FAILURE;
+ }
+ char* line = NULL;
+ size_t len = 0;
+ while (getline(&line, &len, ifile) != -1) {
+ vector_int_t* vector = line_to_vector_int(line, dim);
+ //TODO
+ free(line);
+ }
+ // WRITE
+ FILE* ofile = opath != NULL ? fopen(opath, "w") : stdout;
// TODO
return EXIT_SUCCESS;
}
diff --git a/src/vector.c b/src/vector.c
index d30b3ef765c94d6df15aafe2101ce57d1eea55ac..494530f301baa6be8c0a23f642281f6d304c0ac7 100644
--- a/src/vector.c
+++ b/src/vector.c
@@ -8,8 +8,32 @@
vector_int_t* vector_int_create(const size_t dim, const int_t* data) {
vector_int_t* v;
- if ((v = malloc(dim * sizeof(int))) == NULL) return NULL;
+ if ((v = calloc(dim, sizeof(int_t))) == NULL) return NULL;
v->dim = dim;
for (size_t i = 0; i < dim; ++i) v->data[i] = data[i];
return v;
}
+
+vector_fpt_t* vector_fpt_create(const size_t dim, const fpt_t* data) {
+ vector_fpt_t* v;
+ if ((v = calloc(dim, sizeof(fpt_t))) == NULL) return NULL;
+ v->dim = dim;
+ for (size_t i = 0; i < dim; ++i) v->data[i] = data[i];
+ return v;
+}
+
+vector_int_t* vector_int_create_zero(const size_t dim) {
+ vector_int_t* v;
+ if ((v = calloc(dim, sizeof(int_t))) == NULL) return NULL;
+ v->dim = dim;
+ for (size_t i = 0; i < dim; ++i) v->data[i] = 0;
+ return v;
+}
+
+vector_fpt_t* vector_fpt_create_zero(const size_t dim) {
+ vector_fpt_t* v;
+ if ((v = calloc(dim, sizeof(fpt_t))) == NULL) return NULL;
+ v->dim = dim;
+ for (size_t i = 0; i < dim; ++i) v->data[i] = 0.0;
+ return v;
+}
diff --git a/src/vector.h b/src/vector.h
index 2cdd4ad0dc2900b8039bceca7cdc6ed11e77b315..5ca1d546972900a088e3c651442f7b3cc4bf306d 100644
--- a/src/vector.h
+++ b/src/vector.h
@@ -19,7 +19,18 @@ typedef struct vector_int_t_ {
int_t* data;
} vector_int_t;
-vector_int_t* vector_int_create(const size_t dim, const int_t* data);
+typedef struct vector_fpt_t_ {
+ size_t dim;
+ fpt_t* data;
+} vector_fpt_t;
+
+vector_int_t* vector_int_create(size_t dim, const int_t* data);
+
+vector_fpt_t* vector_fpt_create(size_t dim, const fpt_t* data);
+
+vector_int_t* vector_int_create_zero(size_t dim);
+
+vector_fpt_t* vector_fpt_create_zero(size_t dim);
#endif //PROG_KMEANS_VECTOR_H