static readstat_error_t dta_read_map(int fd, dta_ctx_t *ctx) {
if (!ctx->file_is_xmlish)
return 0;
readstat_error_t retval = READSTAT_OK;
if ((retval = dta_read_tag(fd, ctx, "<map>")) != READSTAT_OK) {
goto cleanup;
}
uint64_t map_buffer[14];
if (read(fd, map_buffer, sizeof(map_buffer)) != sizeof(map_buffer)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
ctx->data_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[9]) : map_buffer[9];
ctx->strls_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[10]) : map_buffer[10];
ctx->value_labels_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[11]) : map_buffer[11];
if ((retval = dta_read_tag(fd, ctx, "</map>")) != READSTAT_OK) {
goto cleanup;
}
cleanup:
return retval;
}
static readstat_error_t sav_parse_machine_floating_point_record(const void *data, sav_ctx_t *ctx) {
sav_machine_floating_point_info_record_t fp_info;
memcpy(&fp_info, data, sizeof(sav_machine_floating_point_info_record_t));
ctx->missing_double = ctx->bswap ? byteswap8(fp_info.sysmis) : fp_info.sysmis;
ctx->highest_double = ctx->bswap ? byteswap8(fp_info.highest) : fp_info.highest;
ctx->lowest_double = ctx->bswap ? byteswap8(fp_info.lowest) : fp_info.lowest;
return READSTAT_OK;
}
static readstat_error_t dta_read_map(dta_ctx_t *ctx) {
if (!ctx->file_is_xmlish)
return READSTAT_OK;
readstat_error_t retval = READSTAT_OK;
uint64_t map_buffer[14];
if ((retval = dta_read_chunk(ctx, "<map>", map_buffer,
sizeof(map_buffer), "</map>")) != READSTAT_OK) {
goto cleanup;
}
ctx->data_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[9]) : map_buffer[9];
ctx->strls_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[10]) : map_buffer[10];
ctx->value_labels_offset = ctx->machine_needs_byte_swap ? byteswap8(map_buffer[11]) : map_buffer[11];
cleanup:
return retval;
}
static sas7bcat_block_t *sas7bcat_block_for_label_set(readstat_label_set_t *r_label_set) {
size_t len = 0;
size_t name_len = strlen(r_label_set->name);
int j;
char name[32];
len += 106;
if (name_len > 8) {
len += 32; // long name
if (name_len > 32) {
name_len = 32;
}
}
memcpy(&name[0], r_label_set->name, name_len);
for (j=0; j<r_label_set->value_labels_count; j++) {
readstat_value_label_t *value_label = readstat_get_value_label(r_label_set, j);
len += 30; // Value: 14-byte header + 16-byte padded value
len += 8 + 2 + value_label->label_len + 1;
}
sas7bcat_block_t *block = calloc(1, sizeof(sas7bcat_block_t) + len);
block->len = len;
off_t begin = 106;
int32_t count = r_label_set->value_labels_count;
memcpy(&block->data[38], &count, sizeof(int32_t));
memcpy(&block->data[42], &count, sizeof(int32_t));
if (name_len > 8) {
block->data[2] = (char)0x80;
memcpy(&block->data[8], name, 8);
memset(&block->data[106], ' ', 32);
memcpy(&block->data[106], name, name_len);
begin += 32;
} else {
memset(&block->data[8], ' ', 8);
memcpy(&block->data[8], name, name_len);
}
char *lbp1 = &block->data[begin];
char *lbp2 = &block->data[begin+r_label_set->value_labels_count*30];
for (j=0; j<r_label_set->value_labels_count; j++) {
readstat_value_label_t *value_label = readstat_get_value_label(r_label_set, j);
lbp1[2] = 24; // size - 6
int32_t index = j;
memcpy(&lbp1[10], &index, sizeof(int32_t));
if (r_label_set->type == READSTAT_TYPE_STRING) {
size_t string_len = value_label->string_key_len;
if (string_len > 16)
string_len = 16;
memset(&lbp1[14], ' ', 16);
memcpy(&lbp1[14], value_label->string_key, string_len);
} else {
uint64_t big_endian_value;
double double_value = -1.0 * value_label->double_key;
memcpy(&big_endian_value, &double_value, sizeof(double));
if (machine_is_little_endian()) {
big_endian_value = byteswap8(big_endian_value);
}
memcpy(&lbp1[22], &big_endian_value, sizeof(uint64_t));
}
int16_t label_len = value_label->label_len;
memcpy(&lbp2[8], &label_len, sizeof(int16_t));
memcpy(&lbp2[10], value_label->label, label_len);
lbp1 += 30;
lbp2 += 8 + 2 + value_label->label_len + 1;
}
return block;
}
readstat_error_t sas_read_header(int fd, sas_header_info_t *ctx,
readstat_error_handler error_handler, void *user_ctx) {
sas_header_start_t header_start;
sas_header_end_t header_end;
int retval = READSTAT_OK;
char error_buf[1024];
if (read(fd, &header_start, sizeof(sas_header_start_t)) < sizeof(sas_header_start_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (memcmp(header_start.magic, sas7bdat_magic_number, sizeof(sas7bdat_magic_number)) != 0 &&
memcmp(header_start.magic, sas7bcat_magic_number, sizeof(sas7bcat_magic_number)) != 0) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (header_start.a1 == SAS_ALIGNMENT_OFFSET_4) {
ctx->pad1 = 4;
}
if (header_start.a2 == SAS_ALIGNMENT_OFFSET_4) {
ctx->u64 = 1;
}
int bswap = 0;
if (header_start.endian == SAS_ENDIAN_BIG) {
bswap = machine_is_little_endian();
ctx->little_endian = 0;
} else if (header_start.endian == SAS_ENDIAN_LITTLE) {
bswap = !machine_is_little_endian();
ctx->little_endian = 1;
} else {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
int i;
for (i=0; i<sizeof(_charset_table)/sizeof(_charset_table[0]); i++) {
if (header_start.encoding == _charset_table[i].code) {
ctx->encoding = _charset_table[i].name;
break;
}
}
if (ctx->encoding == NULL) {
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "Unsupported character set code: %d\n", header_start.encoding);
error_handler(error_buf, user_ctx);
}
retval = READSTAT_ERROR_UNSUPPORTED_CHARSET;
goto cleanup;
}
if (readstat_lseek(fd, 196 + ctx->pad1, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "ReadStat: Failed to seek to position %d\n", 196 + ctx->pad1);
error_handler(error_buf, user_ctx);
}
goto cleanup;
}
uint32_t header_size, page_size;
if (read(fd, &header_size, sizeof(uint32_t)) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (read(fd, &page_size, sizeof(uint32_t)) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
ctx->header_size = bswap ? byteswap4(header_size) : header_size;
ctx->page_size = bswap ? byteswap4(page_size) : page_size;
if (ctx->header_size < 1024) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (ctx->u64) {
uint64_t page_count;
if (read(fd, &page_count, sizeof(uint64_t)) < sizeof(uint64_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
ctx->page_count = bswap ? byteswap8(page_count) : page_count;
} else {
uint32_t page_count;
if (read(fd, &page_count, sizeof(uint32_t)) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
ctx->page_count = bswap ? byteswap4(page_count) : page_count;
}
if (readstat_lseek(fd, 8, SEEK_CUR) == -1) {
retval = READSTAT_ERROR_SEEK;
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "ReadStat: Failed to seek forward by %d\n", 8);
error_handler(error_buf, user_ctx);
}
goto cleanup;
}
if (read(fd, &header_end, sizeof(sas_header_end_t)) < sizeof(sas_header_end_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (strncmp(header_end.release, "9.0000M0", sizeof(header_end.release)) == 0) {
/* A bit of a hack, but most SAS installations are running a minor update */
ctx->vendor = READSTAT_VENDOR_STAT_TRANSFER;
} else {
ctx->vendor = READSTAT_VENDOR_SAS;
}
if (readstat_lseek(fd, ctx->header_size, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "ReadStat: Failed to seek to position %lld\n", ctx->header_size);
error_handler(error_buf, user_ctx);
}
goto cleanup;
}
cleanup:
return retval;
}
uint64_t sas_read8(const char *data, int bswap) {
uint64_t tmp;
memcpy(&tmp, data, 8);
return bswap ? byteswap8(tmp) : tmp;
}
readstat_error_t sas_read_header(readstat_io_t *io, sas_header_info_t *hinfo,
readstat_error_handler error_handler, void *user_ctx) {
sas_header_start_t header_start;
sas_header_end_t header_end;
int retval = READSTAT_OK;
char error_buf[1024];
struct tm epoch_tm = { .tm_year = 60, .tm_mday = 1 };
time_t epoch = mktime(&epoch_tm);
if (io->read(&header_start, sizeof(sas_header_start_t), io->io_ctx) < sizeof(sas_header_start_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (memcmp(header_start.magic, sas7bdat_magic_number, sizeof(sas7bdat_magic_number)) != 0 &&
memcmp(header_start.magic, sas7bcat_magic_number, sizeof(sas7bcat_magic_number)) != 0) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (header_start.a1 == SAS_ALIGNMENT_OFFSET_4) {
hinfo->pad1 = 4;
}
if (header_start.a2 == SAS_ALIGNMENT_OFFSET_4) {
hinfo->u64 = 1;
}
int bswap = 0;
if (header_start.endian == SAS_ENDIAN_BIG) {
bswap = machine_is_little_endian();
hinfo->little_endian = 0;
} else if (header_start.endian == SAS_ENDIAN_LITTLE) {
bswap = !machine_is_little_endian();
hinfo->little_endian = 1;
} else {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
int i;
for (i=0; i<sizeof(_charset_table)/sizeof(_charset_table[0]); i++) {
if (header_start.encoding == _charset_table[i].code) {
hinfo->encoding = _charset_table[i].name;
break;
}
}
if (hinfo->encoding == NULL) {
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "Unsupported character set code: %d\n", header_start.encoding);
error_handler(error_buf, user_ctx);
}
retval = READSTAT_ERROR_UNSUPPORTED_CHARSET;
goto cleanup;
}
memcpy(hinfo->file_label, header_start.file_label, sizeof(header_start.file_label));
if (io->seek(hinfo->pad1, READSTAT_SEEK_CUR, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
double creation_time, modification_time;
if (io->read(&creation_time, sizeof(double), io->io_ctx) < sizeof(double)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (io->read(&modification_time, sizeof(double), io->io_ctx) < sizeof(double)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
hinfo->creation_time = bswap ? byteswap_double(creation_time) + epoch : creation_time + epoch;
hinfo->modification_time = bswap ? byteswap_double(creation_time) + epoch : creation_time + epoch;
if (io->seek(16, READSTAT_SEEK_CUR, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
uint32_t header_size, page_size;
if (io->read(&header_size, sizeof(uint32_t), io->io_ctx) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (io->read(&page_size, sizeof(uint32_t), io->io_ctx) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
hinfo->header_size = bswap ? byteswap4(header_size) : header_size;
hinfo->page_size = bswap ? byteswap4(page_size) : page_size;
if (hinfo->header_size < 1024) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if (hinfo->u64) {
uint64_t page_count;
if (io->read(&page_count, sizeof(uint64_t), io->io_ctx) < sizeof(uint64_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
hinfo->page_count = bswap ? byteswap8(page_count) : page_count;
} else {
uint32_t page_count;
if (io->read(&page_count, sizeof(uint32_t), io->io_ctx) < sizeof(uint32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
hinfo->page_count = bswap ? byteswap4(page_count) : page_count;
}
if (io->seek(8, READSTAT_SEEK_CUR, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
if (error_handler) {
snprintf(error_buf, sizeof(error_buf), "ReadStat: Failed to seek forward by %d\n", 8);
error_handler(error_buf, user_ctx);
}
goto cleanup;
}
if (io->read(&header_end, sizeof(sas_header_end_t), io->io_ctx) < sizeof(sas_header_end_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
int major, minor, revision;
if (sscanf(header_end.release, "%1d.%04dM%1d", &major, &minor, &revision) == 3) {
hinfo->major_version = major;
hinfo->minor_version = minor;
hinfo->revision = revision;
}
if (major == 9 && minor == 0 && revision == 0) {
/* A bit of a hack, but most SAS installations are running a minor update */
hinfo->vendor = READSTAT_VENDOR_STAT_TRANSFER;
} else {
hinfo->vendor = READSTAT_VENDOR_SAS;
}
if (io->seek(hinfo->header_size, READSTAT_SEEK_SET, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
if (error_handler) {
snprintf(error_buf, sizeof(error_buf),
"ReadStat: Failed to seek to position %" PRId64 "\n", hinfo->header_size);
error_handler(error_buf, user_ctx);
}
goto cleanup;
}
cleanup:
return retval;
}
