static int dump_file(const char *input_filename) {
int input_format = format(input_filename);
readstat_parser_t *parser = readstat_parser_init();
readstat_error_t error = READSTAT_OK;
printf("Format: %s\n", format_name(input_format));
readstat_set_error_handler(parser, &handle_error);
readstat_set_info_handler(parser, &dump_info);
readstat_set_metadata_handler(parser, &dump_metadata);
error = parse_file(parser, input_filename, input_format, NULL);
readstat_parser_free(parser);
if (error != READSTAT_OK) {
fprintf(stderr, "Error processing %s: %s\n", input_filename, readstat_error_message(error));
return 1;
}
return 0;
}
int main(int argc, char** argv) {
struct timeval start_time, end_time;
readstat_error_t error = READSTAT_OK;
char *input_filename = NULL;
char *catalog_filename = NULL;
char *output_filename = NULL;
if (argc == 2 && (strcmp(argv[1], "-v") == 0 || strcmp(argv[1], "--version") == 0)) {
print_version();
return 0;
} else if (argc == 2 && (strcmp(argv[1], "-h") == 0 || strcmp(argv[1], "--help") == 0)) {
print_usage(argv[0]);
return 0;
} if (argc == 3) {
if (!can_read(argv[1]) || !can_write(argv[2])) {
print_usage(argv[0]);
return 1;
}
input_filename = argv[1];
output_filename = argv[2];
} else if (argc == 4) {
if (!can_read(argv[1]) || !is_catalog(argv[2]) || !can_write(argv[3])) {
print_usage(argv[0]);
return 1;
}
input_filename = argv[1];
catalog_filename = argv[2];
output_filename = argv[3];
} else {
print_usage(argv[0]);
return 1;
}
int input_format = format(input_filename);
int output_format = format(output_filename);
gettimeofday(&start_time, NULL);
int fd = open(output_filename, O_CREAT | O_WRONLY | O_EXCL, 0644);
if (fd == -1) {
dprintf(STDERR_FILENO, "Error opening %s for writing: %s\n", output_filename, strerror(errno));
return 1;
}
readstat_parser_t *pass1_parser = readstat_parser_init();
readstat_parser_t *pass2_parser = readstat_parser_init();
readstat_writer_t *writer = readstat_writer_init();
readstat_writer_set_file_label(writer, "Created by ReadStat <https://github.com/WizardMac/ReadStat>");
rs_ctx_t *rs_ctx = ctx_init();
rs_ctx->writer = writer;
rs_ctx->out_fd = fd;
rs_ctx->out_format = output_format;
readstat_set_data_writer(writer, &write_data);
// Pass 1 - Collect fweight and value labels
readstat_set_error_handler(pass1_parser, &handle_error);
readstat_set_info_handler(pass1_parser, &handle_info);
readstat_set_value_label_handler(pass1_parser, &handle_value_label);
readstat_set_fweight_handler(pass1_parser, &handle_fweight);
if (catalog_filename) {
error = parse_file(pass1_parser, catalog_filename, RS_FORMAT_SAS_CATALOG, rs_ctx);
} else {
error = parse_file(pass1_parser, input_filename, input_format, rs_ctx);
}
if (error != READSTAT_OK)
goto cleanup;
// Pass 2 - Parse full file
readstat_set_error_handler(pass2_parser, &handle_error);
readstat_set_info_handler(pass2_parser, &handle_info);
readstat_set_variable_handler(pass2_parser, &handle_variable);
readstat_set_value_handler(pass2_parser, &handle_value);
error = parse_file(pass2_parser, input_filename, input_format, rs_ctx);
if (error != READSTAT_OK)
goto cleanup;
gettimeofday(&end_time, NULL);
dprintf(STDERR_FILENO, "Converted %ld variables and %ld rows in %.2lf seconds\n",
rs_ctx->var_count, rs_ctx->row_count,
(end_time.tv_sec + 1e-6 * end_time.tv_usec) -
(start_time.tv_sec + 1e-6 * start_time.tv_usec));
cleanup:
readstat_parser_free(pass1_parser);
readstat_parser_free(pass2_parser);
readstat_writer_free(writer);
ctx_free(rs_ctx);
close(fd);
if (error != READSTAT_OK) {
dprintf(STDERR_FILENO, "%s\n", readstat_error_message(error));
unlink(output_filename);
return 1;
}
return 0;
}
static int convert_file(const char *input_filename, const char *catalog_filename, const char *output_filename,
rs_module_t *modules, int modules_count) {
readstat_error_t error = READSTAT_OK;
const char *error_filename = NULL;
struct timeval start_time, end_time;
int input_format = format(input_filename);
rs_module_t *module = rs_module_for_filename(modules, modules_count, output_filename);
gettimeofday(&start_time, NULL);
readstat_parser_t *pass1_parser = readstat_parser_init();
readstat_parser_t *pass2_parser = readstat_parser_init();
rs_ctx_t *rs_ctx = calloc(1, sizeof(rs_ctx_t));
void *module_ctx = module->init(output_filename);
if (module_ctx == NULL)
goto cleanup;
rs_ctx->module = module;
rs_ctx->module_ctx = module_ctx;
// Pass 1 - Collect fweight and value labels
readstat_set_error_handler(pass1_parser, &handle_error);
readstat_set_info_handler(pass1_parser, &handle_info);
readstat_set_value_label_handler(pass1_parser, &handle_value_label);
readstat_set_fweight_handler(pass1_parser, &handle_fweight);
if (catalog_filename) {
error = parse_file(pass1_parser, catalog_filename, RS_FORMAT_SAS_CATALOG, rs_ctx);
error_filename = catalog_filename;
} else {
error = parse_file(pass1_parser, input_filename, input_format, rs_ctx);
error_filename = input_filename;
}
if (error != READSTAT_OK)
goto cleanup;
// Pass 2 - Parse full file
readstat_set_error_handler(pass2_parser, &handle_error);
readstat_set_info_handler(pass2_parser, &handle_info);
readstat_set_metadata_handler(pass2_parser, &handle_metadata);
readstat_set_note_handler(pass2_parser, &handle_note);
readstat_set_variable_handler(pass2_parser, &handle_variable);
readstat_set_value_handler(pass2_parser, &handle_value);
error = parse_file(pass2_parser, input_filename, input_format, rs_ctx);
error_filename = input_filename;
if (error != READSTAT_OK)
goto cleanup;
gettimeofday(&end_time, NULL);
fprintf(stderr, "Converted %ld variables and %ld rows in %.2lf seconds\n",
rs_ctx->var_count, rs_ctx->row_count,
(end_time.tv_sec + 1e-6 * end_time.tv_usec) -
(start_time.tv_sec + 1e-6 * start_time.tv_usec));
cleanup:
readstat_parser_free(pass1_parser);
readstat_parser_free(pass2_parser);
if (module->finish) {
module->finish(rs_ctx->module_ctx);
}
free(rs_ctx);
if (error != READSTAT_OK) {
fprintf(stderr, "Error processing %s: %s\n", error_filename, readstat_error_message(error));
unlink(output_filename);
return 1;
}
return 0;
}