rs_ctx_t *ctx_init() {
rs_ctx_t *ctx = malloc(sizeof(rs_ctx_t));
ctx->label_set_dict = ck_hash_table_init(1024);
return ctx;
}
static readstat_error_t read_por_file_data(readstat_por_ctx_t *ctx) {
int i;
char string[256];
char error_buf[1024];
int retval = 0;
readstat_error_t rs_retval = READSTAT_OK;
while (1) {
for (i=0; i<ctx->var_count; i++) {
spss_varinfo_t *info = &ctx->varinfo[i];
readstat_value_t value = { .type = info->type };
if (info->type == READSTAT_TYPE_STRING) {
retval = read_string(ctx, string, sizeof(string));
if (i == 0 && retval == 1) {
return 0;
} else if (retval == -1) {
if (ctx->error_handler) {
snprintf(error_buf, sizeof(error_buf), "Error in %s\n", info->name);
ctx->error_handler(error_buf, ctx->user_ctx);
}
rs_retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
value.v.string_value = string;
// printf("String value: %s\n", string);
} else if (info->type == READSTAT_TYPE_DOUBLE) {
retval = read_double(ctx, &value.v.double_value);
if (i == 0 && retval == 1) {
return READSTAT_OK;
} else if (retval != 0) {
if (ctx->error_handler) {
snprintf(error_buf, sizeof(error_buf), "Error in %s\n", info->name);
ctx->error_handler(error_buf, ctx->user_ctx);
}
rs_retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
spss_tag_missing_double(&value, &info->missingness);
}
ctx->value_handler(ctx->obs_count, i, value, ctx->user_ctx);
}
ctx->obs_count++;
rs_retval = por_update_progress(ctx);
if (rs_retval != READSTAT_OK)
break;
}
cleanup:
return rs_retval;
}
readstat_error_t readstat_parse_por(readstat_parser_t *parser, const char *filename, void *user_ctx) {
readstat_error_t retval = READSTAT_OK;
unsigned char reverse_lookup[256];
char vanity[200];
readstat_por_ctx_t *ctx = calloc(1, sizeof(readstat_por_ctx_t));
ctx->space = ' ';
ctx->var_dict = ck_hash_table_init(1024);
ctx->info_handler = parser->info_handler;
ctx->variable_handler = parser->variable_handler;
ctx->value_handler = parser->value_handler;
ctx->value_label_handler = parser->value_label_handler;
ctx->error_handler = parser->error_handler;
ctx->progress_handler = parser->progress_handler;
ctx->user_ctx = user_ctx;
if ((ctx->fd = readstat_open(filename)) == -1) {
free(ctx);
return READSTAT_ERROR_OPEN;
}
if ((ctx->file_size = lseek(ctx->fd, 0, SEEK_END)) == -1) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (lseek(ctx->fd, 0, SEEK_SET) == -1) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (read_bytes(ctx, vanity, sizeof(vanity)) != sizeof(vanity)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (read_bytes(ctx, reverse_lookup, sizeof(reverse_lookup)) != sizeof(reverse_lookup)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
ctx->space = reverse_lookup[126];
int i;
for (i=0; i<256; i++)
ctx->lookup[reverse_lookup[i]] = unicode_lookup[i];
unsigned char check[9];
char tr_check[9];
if (read_bytes(ctx, check, sizeof(check)-1) != sizeof(check)-1) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
check[8] = '\0';
utf8_encode(check, sizeof(check), tr_check, sizeof(tr_check), ctx->lookup);
if (strcmp("SPSSPORT", tr_check) != 0) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
ctx->var_offset = -1;
unsigned char version;
char string[256];
if (read_bytes(ctx, &version, sizeof(version)) != sizeof(version)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (read_string(ctx, string, sizeof(string)) == -1) { /* creation date */
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (read_string(ctx, string, sizeof(string)) == -1) { /* creation time */
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
while (1) {
uint16_t tr_tag = read_tag(ctx);
switch (tr_tag) {
case '1': /* product ID */
case '2': /* author ID */
case '3': /* sub-product ID */
if (read_string(ctx, string, sizeof(string)) == -1) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
break;
case '4': /* variable count */
retval = read_variable_count_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case '6': /* case weight */
retval = read_case_weight_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case '7': /* variable */
retval = read_variable_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case '8': /* missing value */
retval = read_missing_value_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'B': /* missing value range */
retval = read_missing_value_range_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case '9': /* LO THRU x */
retval = read_missing_value_lo_range_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'A': /* x THRU HI */
retval = read_missing_value_hi_range_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'C': /* variable label */
retval = read_variable_label_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'D': /* value label */
retval = read_value_label_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'E': /* document record */
retval = read_document_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
break;
case 'F': /* file data */
if (ctx->var_offset != ctx->var_count - 1) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
for (i=0; i<ctx->var_count; i++) {
char label_name_buf[256];
spss_varinfo_t *info = &ctx->varinfo[i];
info->missingness = spss_missingness_for_info(info);
readstat_variable_t *variable = spss_init_variable_for_info(info);
snprintf(label_name_buf, sizeof(label_name_buf), POR_LABEL_NAME_PREFIX "%d", info->labels_index);
int cb_retval = ctx->variable_handler(i, variable,
info->labels_index == -1 ? NULL : label_name_buf,
user_ctx);
spss_free_variable(variable);
if (cb_retval) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
}
if (parser->fweight_handler && ctx->fweight_name[0]) {
for (i=0; i<ctx->var_count; i++) {
spss_varinfo_t *info = &ctx->varinfo[i];
if (strcmp(info->name, ctx->fweight_name) == 0) {
if (parser->fweight_handler(i, user_ctx)) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
break;
}
}
}
retval = read_por_file_data(ctx);
if (retval != READSTAT_OK)
goto cleanup;
if (parser->info_handler) {
if (parser->info_handler(ctx->obs_count, ctx->var_count, ctx->user_ctx)) {
retval = READSTAT_ERROR_USER_ABORT;
}
}
goto cleanup;
break;
default:
retval = READSTAT_ERROR_PARSE;
goto cleanup;
break;
}
}
cleanup:
readstat_close(ctx->fd);
por_ctx_free(ctx);
return retval;
}