feat: add read function

CMakeLists.txt

@@ -15,4 +15,4 @@ SHARED
 )
 
 add_executable(hnsw_test src/test.c)
-target_link_libraries(hnsw_test hnswc)
+target_link_libraries(hnsw_test hnswc m)

inc/utils.h

@@ -2,5 +2,4 @@
 typedef char* VecData;
 
 float vec_dist(VecData x, VecData y);
-
-
+void fvecs_read(const char* filename, int* bound, float** vectors, int* num_vectors, int* vector_dimension);

src/test.c

@@ -1,8 +1,27 @@
 #include <stdio.h>
+#include <stdlib.h>
 #include "hnsw.h"
 #include "utils.h"
 
+
 int main() {
-    printf("Hello, world!\n");
+    const char* filename = "../dataset/siftsmall_base.fvecs";
+    int bound[2] = {2, 5};
+    float* vectors;
+    int num_vectors, vector_dimension;
+    fvecs_read(filename, bound, &vectors, &num_vectors, &vector_dimension);
+
+    printf("Number of vectors: %d\n", num_vectors);
+    printf("Vector dimension: %d\n", vector_dimension);
+    int i, j;
+    for (i = 0; i < num_vectors; i++) {
+        printf("Vector %d: ", i);
+        for (j = 0; j < vector_dimension; j++) {
+            printf("%f ", vectors[i * vector_dimension + j]);
+        }
+        printf("\n");
+    }
+
+    free(vectors);
     return 0;
 }

src/utils.c

@@ -0,0 +1,92 @@
+#include "utils.h"
+#include <math.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+
+float vec_dist(VecData x, VecData y) {
+    float dist = 0.0f;
+    while (*x != '\0' && *y != '\0') {
+        int xi = *x - '0';
+        int yi = *y - '0';
+        dist += pow(xi - yi, 2);
+        x++;
+        y++;
+    }
+    return sqrt(dist);
+}
+
+void fvecs_read(const char* filename, int* bound, float** vectors, int* num_vectors, int* vector_dimension) {
+    FILE* fid = fopen(filename, "rb");
+    if (fid == NULL) {
+        fprintf(stderr, "I/O error: Unable to open the file %s\n", filename);
+        exit(EXIT_FAILURE);
+    }
+
+    // Read the vector size
+    int d;
+    fread(&d, sizeof(int), 1, fid);
+
+    // Get the file size
+    fseek(fid, 0, SEEK_END);
+    long file_size = ftell(fid);
+    fseek(fid, 0, SEEK_SET);
+
+    // Get the number of vectors
+    long vec_size = (long) d * sizeof(float);
+    long vec_count = (file_size - sizeof(int)) / vec_size;
+
+    // Apply the bounds if specified
+    int a = 1;
+    int b = vec_count;
+    if (bound != NULL) {
+        if (bound[1] == 1) {
+            b = bound[0];
+        } else if (bound[1] == 2) {
+            a = bound[0];
+            b = bound[1];
+        }
+    }
+
+    // Check if the bounds are valid
+    if (a < 1 || b < a || b > vec_count) {
+        *vectors = NULL;
+        fclose(fid);
+        return;
+    }
+
+    // Compute the number of vectors to read
+    int n = b - a + 1;
+
+    // Read the vectors
+    *vectors = malloc(n * d * sizeof(float));
+    float* ptr = *vectors;
+    for (int i = 0; i < vec_count; i++) {
+        // Read the vector size
+        int vec_d;
+        fread(&vec_d, sizeof(int), 1, fid);
+
+        // Check if the vector size is correct
+        if (vec_d != d) {
+            fprintf(stderr, "Error: Vector %d has incorrect dimension %d (expected %d)\n", i + 1, vec_d, d);
+            fclose(fid);
+            free(*vectors);
+            *vectors = NULL;
+            return;
+        }
+
+        // Read the vector data
+        fread(ptr, sizeof(float), d, fid);
+        ptr += d;
+
+        // Stop reading if we have read enough vectors
+        if (i >= b - 1) {
+            break;
+        }
+    }
+
+    *vector_dimension = d;
+    *num_vectors = n;
+    fclose(fid);
+}