Support variable compression levels

Author: BenBrock

examples/mtx2bsp.c

@@ -5,7 +5,7 @@ int main(int argc, char** argv) {
 
   if (argc < 3) {
     printf("usage: ./mtx2bsp [input.mtx] [output.bsp.h5]:[optional: group] "
-           "[optional: format]\n");
+           "[optional: format] [optional: compression level 0-9]\n");
     printf("\n");
     printf("Description: Convert a Matrix Market file to a Binsparse HDF5 "
            "file.\n");
@@ -29,6 +29,13 @@ int main(int argc, char** argv) {
         "example: ./mtx2bsp chesapeake.mtx chesapeake.bsp.h5:chesapeake CSR\n");
     printf("         - Same as previous example, but matrix will use CSR "
            "format.\n");
+    printf("example: ./mtx2bsp chesapeake.mtx chesapeake.bsp.h5:chesapeake CSR "
+           "5\n");
+    printf("         - Same as previous example, but will use GZip compression "
+           "level 5.\n");
+    printf("           0 is no compression, 1-9 correspond to GZip compression "
+           "levels.\n");
+    printf("           Default is 9.\n");
     return 1;
   }
 
@@ -49,6 +56,12 @@ int main(int argc, char** argv) {
     format_name = argv[3];
   }
 
+  int compression_level = 9;
+
+  if (argc >= 5) {
+    compression_level = atoi(argv[4]);
+  }
+
   char* input_file_extension = bsp_get_file_extension(input_fname);
   char* output_file_extension = bsp_get_file_extension(output_fname);
 
@@ -90,6 +103,8 @@ int main(int argc, char** argv) {
     printf("File has very long comments, not printing.\n");
   }
 
+  printf("Printing with compression level %d.\n", compression_level);
+
   cJSON* user_json = cJSON_CreateObject();
 
   assert(user_json != NULL);
@@ -115,7 +130,8 @@ int main(int argc, char** argv) {
   bsp_print_matrix_info(matrix);
 
   printf(" === Writing to %s... ===\n", output_fname);
-  bsp_write_matrix(output_fname, matrix, group_name, user_json);
+  bsp_write_matrix(output_fname, matrix, group_name, user_json,
+                   compression_level);
   printf(" === Done writing. ===\n");
 
   bsp_destroy_matrix_t(matrix);

examples/simple_matrix_write.c

@@ -15,7 +15,7 @@ int main(int argc, char** argv) {
     printf("%d, %d: %f\n", rowind[i], colind[i], values[i]);
   }
 
-  bsp_write_matrix("test.hdf5", mat, NULL, NULL);
+  bsp_write_matrix("test.hdf5", mat, NULL, NULL, 9);
 
   return 0;
 }

examples/simple_write.c

@@ -13,7 +13,7 @@ int main(int argc, char** argv) {
     bsp_array_write(array, i, i);
   }
 
-  bsp_write_array(f, "test", array);
+  bsp_write_array(f, "test", array, 0);
 
   H5Fclose(f);
   bsp_destroy_array_t(array);

include/binsparse/hdf5_wrapper.h

@@ -7,7 +7,8 @@
 
 // Write an array to a dataset / file
 // Returns 0 on success, nonzero on error.
-int bsp_write_array(hid_t f, char* label, bsp_array_t array) {
+int bsp_write_array(hid_t f, char* label, bsp_array_t array,
+                    int compression_level) {
   if (array.type == BSP_COMPLEX_FLOAT32 || array.type == BSP_COMPLEX_FLOAT64) {
     array = bsp_complex_array_to_fp(array);
   }
@@ -28,7 +29,9 @@ int bsp_write_array(hid_t f, char* label, bsp_array_t array) {
 
   H5Pset_chunk(dcpl, 1, (hsize_t[]){chunk_size});
 
-  H5Pset_deflate(dcpl, 9);
+  if (compression_level > 0) {
+    H5Pset_deflate(dcpl, compression_level);
+  }
 
   hid_t dset =
       H5Dcreate2(f, label, hdf5_standard_type, fspace, lcpl, dcpl, H5P_DEFAULT);

include/binsparse/write_matrix.h

@@ -81,28 +81,32 @@ char* bsp_generate_json(bsp_matrix_t matrix, cJSON* user_json) {
   return string;
 }
 
-int bsp_write_matrix_to_group(hid_t f, bsp_matrix_t matrix, cJSON* user_json) {
-  int result = bsp_write_array(f, "values", matrix.values);
+int bsp_write_matrix_to_group(hid_t f, bsp_matrix_t matrix, cJSON* user_json,
+                              int compression_level) {
+  int result = bsp_write_array(f, "values", matrix.values, compression_level);
 
   if (result != 0)
     return result;
 
   if (matrix.indices_0.size > 0) {
-    result = bsp_write_array(f, "indices_0", matrix.indices_0);
+    result =
+        bsp_write_array(f, "indices_0", matrix.indices_0, compression_level);
     if (result != 0) {
       return result;
     }
   }
 
   if (matrix.indices_1.size > 0) {
-    result = bsp_write_array(f, "indices_1", matrix.indices_1);
+    result =
+        bsp_write_array(f, "indices_1", matrix.indices_1, compression_level);
     if (result != 0) {
       return result;
     }
   }
 
   if (matrix.pointers_to_1.size > 0) {
-    result = bsp_write_array(f, "pointers_to_1", matrix.pointers_to_1);
+    result = bsp_write_array(f, "pointers_to_1", matrix.pointers_to_1,
+                             compression_level);
     if (result != 0) {
       return result;
     }
@@ -117,10 +121,10 @@ int bsp_write_matrix_to_group(hid_t f, bsp_matrix_t matrix, cJSON* user_json) {
 }
 
 int bsp_write_matrix(char* fname, bsp_matrix_t matrix, char* group,
-                     cJSON* user_json) {
+                     cJSON* user_json, int compression_level) {
   if (group == NULL) {
     hid_t f = H5Fcreate(fname, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
-    bsp_write_matrix_to_group(f, matrix, user_json);
+    bsp_write_matrix_to_group(f, matrix, user_json, compression_level);
     H5Fclose(f);
   } else {
     hid_t f;
@@ -130,7 +134,7 @@ int bsp_write_matrix(char* fname, bsp_matrix_t matrix, char* group,
       f = H5Fcreate(fname, H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT);
     }
     hid_t g = H5Gcreate1(f, group, H5P_DEFAULT);
-    bsp_write_matrix_to_group(g, matrix, user_json);
+    bsp_write_matrix_to_group(g, matrix, user_json, compression_level);
     H5Gclose(g);
     H5Fclose(f);
   }