Update mtx2bsp to support multiple formats

Author: BenBrock

examples/check_equivalence.c

@@ -80,39 +80,6 @@ int check_array_equivalence(bsp_array_t array1, bsp_array_t array2) {
   return 0;
 }
 
-typedef struct {
-  char* fname;
-  char* dataset;
-} bsp_fdataset_info_t;
-
-bsp_fdataset_info_t bsp_parse_fdataset_string(char* str) {
-  size_t len = strlen(str);
-
-  int split = -1;
-  for (int i = len - 1; i >= 0; i--) {
-    if (str[i] == ':') {
-      split = i;
-      break;
-    }
-  }
-
-  if (split == -1) {
-    bsp_fdataset_info_t info;
-    info.fname = (char*) malloc(sizeof(char) * (len + 1));
-    strcpy(info.fname, str);
-    info.dataset = NULL;
-    return info;
-  } else {
-    bsp_fdataset_info_t info;
-    info.fname = (char*) malloc(sizeof(char) * (split + 1));
-    strncpy(info.fname, str, split);
-    info.fname[split] = '\0';
-    info.dataset = (char*) malloc(sizeof(char) * (len - split));
-    strcpy(info.dataset, &str[split + 1]);
-    return info;
-  }
-}
-
 int main(int argc, char** argv) {
   if (argc < 3) {
     printf(

examples/mtx2bsp.c

@@ -4,8 +4,31 @@
 int main(int argc, char** argv) {
 
   if (argc < 3) {
-    printf("usage: ./mtx2bsp [inputfile_name.mtx] [outputfile_name.bsp.hdf5] "
-           "[optional: dataset]\n");
+    printf("usage: ./mtx2bsp [input.mtx] [output.bsp.h5]:[optional: group] "
+           "[optional: format]\n");
+    printf("\n");
+    printf("Description: Convert a Matrix Market file to a Binsparse HDF5 "
+           "file.\n");
+    printf("             Users can optionally provide an HDF5 group to store "
+           "the\n");
+    printf("             file in as well as a specific format. The default "
+           "format\n");
+    printf("             is row-sorted COO (COOR).\n");
+    printf("\n");
+    printf("example: ./mtx2bsp chesapeake.mtx chesapeake.bsp.h5\n");
+    printf("        - Convert Matrix Market file `chesapeake.mtx` to Binsparse "
+           "HDF5 file `chesapeake.bsp.h5`.\n");
+    printf("        - Matrix will be stored in root group.\n");
+    printf("        - Matrix will be stored in COOR format.\n");
+    printf("\n");
+    printf("example: ./mtx2bsp chesapeake.mtx chesapeake.bsp.h5:chesapeake\n");
+    printf("         - Same as previous example, but matrix will be stored in "
+           "HDF5 group `chesapeake`.\n");
+    printf("\n");
+    printf(
+        "example: ./mtx2bsp chesapeake.mtx chesapeake.bsp.h5:chesapeake CSR\n");
+    printf("         - Same as previous example, but matrix will use CSR "
+           "format.\n");
     return 1;
   }
 
@@ -15,16 +38,47 @@ int main(int argc, char** argv) {
   bool perform_suitesparse_declamping = true;
 
   char* input_fname = argv[1];
-  char* output_fname = argv[2];
 
-  char* group_name = NULL;
+  bsp_fdataset_info_t info2 = bsp_parse_fdataset_string(argv[2]);
+  char* output_fname = info2.fname;
+  char* group_name = info2.dataset;
+
+  char* format_name = NULL;
 
   if (argc >= 4) {
-    group_name = argv[3];
+    format_name = argv[3];
+  }
+
+  char* input_file_extension = bsp_get_file_extension(input_fname);
+  char* output_file_extension = bsp_get_file_extension(output_fname);
+
+  if (input_file_extension == NULL ||
+      strcmp(input_file_extension, ".mtx") != 0) {
+    fprintf(stderr,
+            "error: input file \"%s\" is not a Matrix Market file. "
+            "(Its extension is not '.mtx'.)\n",
+            input_fname);
+    return 1;
+  }
+
+  if (output_file_extension == NULL ||
+      (strcmp(output_file_extension, ".h5") != 0 &&
+       strcmp(output_file_extension, ".hdf5") != 0)) {
+    fprintf(stderr,
+            "error: output file \"%s\" is not an HDF5 file. "
+            "(Its extension is not '.h5' or '.hdf5'.)\n",
+            output_fname);
+    return 1;
   }
 
   bsp_mm_metadata m = bsp_mmread_metadata(input_fname);
 
+  bsp_matrix_format_t format = BSP_COOR;
+  if (format_name != NULL) {
+    format = bsp_get_matrix_format(format_name);
+    assert(format != 0);
+  }
+
   printf("%lu x %lu matrix with %lu nonzeros.\n", m.nrows, m.ncols, m.nnz);
   printf(
       "Matrix Market format is \"%s\" with type \"%s\" and structure \"%s\"\n",
@@ -52,6 +106,12 @@ int main(int argc, char** argv) {
 
   matrix = bsp_matrix_minimize_values(matrix);
 
+  if (format != BSP_COOR) {
+    bsp_matrix_t converted_matrix = bsp_convert_matrix(matrix, format);
+    bsp_destroy_matrix_t(matrix);
+    matrix = converted_matrix;
+  }
+
   bsp_print_matrix_info(matrix);
 
   printf(" === Writing to %s... ===\n", output_fname);

include/binsparse/convert_matrix.h

@@ -5,9 +5,9 @@
 
 bsp_matrix_t bsp_convert_matrix(bsp_matrix_t matrix,
                                 bsp_matrix_format_t format) {
-  // If already in desired format, do nothing.
+  // Throw an error if matrix already in desired format.
   if (matrix.format == format) {
-    return matrix;
+    assert(false);
   }
 
   if (format == BSP_COOR) {

include/binsparse/detail/detail.h

@@ -1,3 +1,4 @@
 #pragma once
 
 #include <binsparse/detail/declamp_values.h>
+#include <binsparse/detail/parse_dataset.h>

include/binsparse/detail/parse_dataset.h

@@ -0,0 +1,46 @@
+#pragma once
+
+#include <string.h>
+
+typedef struct {
+  char* fname;
+  char* dataset;
+} bsp_fdataset_info_t;
+
+bsp_fdataset_info_t bsp_parse_fdataset_string(char* str) {
+  size_t len = strlen(str);
+
+  int split = -1;
+  for (int i = len - 1; i >= 0; i--) {
+    if (str[i] == ':') {
+      split = i;
+      break;
+    }
+  }
+
+  if (split == -1) {
+    bsp_fdataset_info_t info;
+    info.fname = (char*) malloc(sizeof(char) * (len + 1));
+    strcpy(info.fname, str);
+    info.dataset = NULL;
+    return info;
+  } else {
+    bsp_fdataset_info_t info;
+    info.fname = (char*) malloc(sizeof(char) * (split + 1));
+    strncpy(info.fname, str, split);
+    info.fname[split] = '\0';
+    info.dataset = (char*) malloc(sizeof(char) * (len - split));
+    strcpy(info.dataset, &str[split + 1]);
+    return info;
+  }
+}
+
+char* bsp_get_file_extension(char* file_name) {
+  int64_t len = strlen(file_name);
+  for (int64_t i = len - 1; i >= 0; i--) {
+    if (file_name[i] == '.') {
+      return &file_name[i];
+    }
+  }
+  return NULL;
+}