int arsParamSave( char *filename, ARSParam *sparam )
{
FILE *fp;
fp = fopen( filename, "wb" );
if( fp == NULL ) return -1;
#ifdef AR_LITTLE_ENDIAN
byteswap2( sparam );
#endif
if( fwrite( sparam, sizeof(ARSParam), 1, fp ) != 1 ) {
fclose(fp);
#ifdef AR_LITTLE_ENDIAN
byteswap2( sparam );
#endif
return -1;
}
#ifdef AR_LITTLE_ENDIAN
byteswap2( sparam );
#endif
fclose(fp);
return 0;
}
static readstat_error_t xport_read_labels_v8(xport_ctx_t *ctx, int label_count) {
readstat_error_t retval = READSTAT_OK;
uint16_t labeldef[3];
int i;
for (i=0; i<label_count; i++) {
int index, name_len, label_len;
if (read_bytes(ctx, labeldef, sizeof(labeldef)) != sizeof(labeldef)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (machine_is_little_endian()) {
index = byteswap2(labeldef[0]);
name_len = byteswap2(labeldef[1]);
label_len = byteswap2(labeldef[2]);
} else {
index = labeldef[0];
name_len = labeldef[1];
label_len = labeldef[2];
}
if (index >= ctx->var_count) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
char name[name_len];
char label[label_len];
readstat_variable_t *variable = ctx->variables[index];
if (read_bytes(ctx, name, name_len) != name_len ||
read_bytes(ctx, label, label_len) != label_len) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
retval = readstat_convert(variable->name, sizeof(variable->name),
name, name_len, NULL);
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_convert(variable->label, sizeof(variable->label),
label, label_len, NULL);
if (retval != READSTAT_OK)
goto cleanup;
}
retval = xport_skip_rest_of_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
retval = xport_read_obs_header_record(ctx);
if (retval != READSTAT_OK)
goto cleanup;
cleanup:
return retval;
}
// Save this (preloaded) marker to disk.
int JpegMarker::Save(FILE *pFile) {
const int location = ftell(pFile);
if (pFile == NULL)
throw("Null File in JpegMarker::Save.");
unsigned short markerswapped = byteswap2(marker_);
int rv = fwrite(&markerswapped, sizeof(unsigned short), 1, pFile);
if (rv != 1) return 0;
unsigned short slice_or_length = length_;
if (marker_ == Jpeg::jpeg_sos)
slice_or_length = slice_;
ByteSwapInPlace(&slice_or_length, 1);
rv = fwrite(&slice_or_length, sizeof(unsigned short), 1, pFile);
if (rv != 1) return 0;
if (marker_ == Jpeg::jpeg_sos) {
WriteWithStuffBytes(pFile);
// Add the EOI
unsigned char eoi0 = 0xff;
unsigned char eoi1 = 0xd9;
fwrite(&eoi0, sizeof(eoi0), 1, pFile);
fwrite(&eoi1, sizeof(eoi1), 1, pFile);
} else {
rv = fwrite(&data_[0], sizeof(char), data_.size(), pFile);
if (rv != data_.size()) return 0;
}
if (debug > 0)
printf("Saved marker %04x length %u at %d\n", marker_, length_, location);
return 1;
}
void xport_namestr_bswap(xport_namestr_t *namestr) {
if (!machine_is_little_endian())
return;
namestr->ntype = byteswap2(namestr->ntype);
namestr->nhfun = byteswap2(namestr->nhfun);
namestr->nlng = byteswap2(namestr->nlng);
namestr->nvar0 = byteswap2(namestr->nlng);
namestr->nfl = byteswap2(namestr->nfl);
namestr->nfd = byteswap2(namestr->nfd);
namestr->nfj = byteswap2(namestr->nfj);
namestr->nifl = byteswap2(namestr->nifl);
namestr->nifd = byteswap2(namestr->nifd);
namestr->npos = byteswap4(namestr->npos);
namestr->labeln = byteswap2(namestr->labeln);
}
static readstat_error_t dta_read_descriptors(int fd, dta_ctx_t *ctx) {
if (dta_read_tag(fd, ctx, "<variable_types>") != READSTAT_OK)
return -1;
size_t buffer_len = ctx->nvar * ctx->typlist_entry_len;
unsigned char *buffer = malloc(buffer_len);
if (read(fd, buffer, buffer_len) != buffer_len) {
free(buffer);
return -1;
}
int i;
if (ctx->typlist_entry_len == 1) {
for (i=0; i<ctx->nvar; i++) {
ctx->typlist[i] = buffer[i];
}
} else if (ctx->typlist_entry_len == 2) {
memcpy(ctx->typlist, buffer, buffer_len);
if (ctx->machine_needs_byte_swap) {
for (i=0; i<ctx->nvar; i++) {
ctx->typlist[i] = byteswap2(ctx->typlist[i]);
}
}
}
free(buffer);
if (dta_read_tag(fd, ctx, "</variable_types>") != READSTAT_OK)
return -1;
if (dta_read_tag(fd, ctx, "<varnames>") != READSTAT_OK ||
read(fd, ctx->varlist, ctx->varlist_len) != ctx->varlist_len ||
dta_read_tag(fd, ctx, "</varnames>") != READSTAT_OK)
return -1;
if (dta_read_tag(fd, ctx, "<sortlist>") != READSTAT_OK ||
read(fd, ctx->srtlist, ctx->srtlist_len) != ctx->srtlist_len ||
dta_read_tag(fd, ctx, "</sortlist>") != READSTAT_OK)
return -1;
if (dta_read_tag(fd, ctx, "<formats>") != READSTAT_OK ||
read(fd, ctx->fmtlist, ctx->fmtlist_len) != ctx->fmtlist_len ||
dta_read_tag(fd, ctx, "</formats>") != READSTAT_OK)
return -1;
if (dta_read_tag(fd, ctx, "<value_label_names>") != READSTAT_OK ||
read(fd, ctx->lbllist, ctx->lbllist_len) != ctx->lbllist_len ||
dta_read_tag(fd, ctx, "</value_label_names>") != READSTAT_OK)
return -1;
if (dta_read_tag(fd, ctx, "<variable_labels>") != READSTAT_OK ||
read(fd, ctx->variable_labels, ctx->variable_labels_len) != ctx->variable_labels_len ||
dta_read_tag(fd, ctx, "</variable_labels>") != READSTAT_OK)
return -1;
return 0;
}
static readstat_error_t dta_skip_expansion_fields(int fd, dta_ctx_t *ctx) {
if (ctx->file_is_xmlish) {
if (readstat_lseek(fd, ctx->data_offset, SEEK_SET) == -1)
return READSTAT_ERROR_SEEK;
return READSTAT_OK;
}
if (ctx->expansion_len_len == 0)
return READSTAT_OK;
while (1) {
size_t len;
char data_type;
if (ctx->expansion_len_len == 2) {
dta_short_expansion_field_t expansion_field;
if (read(fd, &expansion_field, sizeof(expansion_field)) != sizeof(expansion_field))
return READSTAT_ERROR_READ;
if (ctx->machine_needs_byte_swap) {
len = byteswap2(expansion_field.len);
} else {
len = expansion_field.len;
}
data_type = expansion_field.data_type;
} else {
dta_expansion_field_t expansion_field;
if (read(fd, &expansion_field, sizeof(expansion_field)) != sizeof(expansion_field))
return READSTAT_ERROR_READ;
if (ctx->machine_needs_byte_swap) {
len = byteswap4(expansion_field.len);
} else {
len = expansion_field.len;
}
data_type = expansion_field.data_type;
}
if (data_type == 0 && len == 0)
return READSTAT_OK;
if (data_type != 1)
return READSTAT_ERROR_PARSE;
if (readstat_lseek(fd, len, SEEK_CUR) == -1)
return READSTAT_ERROR_SEEK;
}
return READSTAT_ERROR_PARSE;
}
static readstat_error_t dta_read_descriptors(dta_ctx_t *ctx) {
readstat_error_t retval = READSTAT_OK;
size_t buffer_len = ctx->nvar * ctx->typlist_entry_len;
unsigned char *buffer = NULL;
int i;
buffer = malloc(buffer_len);
if ((retval = dta_read_chunk(ctx, "<variable_types>", buffer,
buffer_len, "</variable_types>")) != READSTAT_OK)
goto cleanup;
if (ctx->typlist_entry_len == 1) {
for (i=0; i<ctx->nvar; i++) {
ctx->typlist[i] = buffer[i];
}
} else if (ctx->typlist_entry_len == 2) {
memcpy(ctx->typlist, buffer, buffer_len);
if (ctx->machine_needs_byte_swap) {
for (i=0; i<ctx->nvar; i++) {
ctx->typlist[i] = byteswap2(ctx->typlist[i]);
}
}
}
if ((retval = dta_read_chunk(ctx, "<varnames>", ctx->varlist,
ctx->varlist_len, "</varnames>")) != READSTAT_OK)
goto cleanup;
if ((retval = dta_read_chunk(ctx, "<sortlist>", ctx->srtlist,
ctx->srtlist_len, "</sortlist>")) != READSTAT_OK)
goto cleanup;
if ((retval = dta_read_chunk(ctx, "<formats>", ctx->fmtlist,
ctx->fmtlist_len, "</formats>")) != READSTAT_OK)
goto cleanup;
if ((retval = dta_read_chunk(ctx, "<value_label_names>", ctx->lbllist,
ctx->lbllist_len, "</value_label_names>")) != READSTAT_OK)
goto cleanup;
if ((retval = dta_read_chunk(ctx, "<variable_labels>", ctx->variable_labels,
ctx->variable_labels_len, "</variable_labels>")) != READSTAT_OK)
goto cleanup;
cleanup:
if (buffer)
free(buffer);
return retval;
}
/* {{{ void byteswapbuffer(void *pData, unsigned int uElements, unsigned int uElementSize) */
void byteswapbuffer(void *pData, unsigned int uElements, unsigned int uElementSize) {
unsigned char *pbyData = (unsigned char *)pData;
unsigned int cuElements = uElements;
while(cuElements > 0) {
if(uElementSize == 2)
byteswap2(pbyData);
else if(uElementSize == 4)
byteswap4(pbyData);
else
return;
pbyData += uElementSize;
cuElements--;
}
}
readstat_error_t readstat_parse_dta(readstat_parser_t *parser, const char *filename, void *user_ctx) {
readstat_error_t retval = READSTAT_OK;
int i;
size_t record_len = 0;
int fd = -1;
char *buf = NULL;
dta_header_t header;
dta_ctx_t *ctx = NULL;
char str_buf[2048];
char *long_string = NULL;
size_t file_size = 0;
if ((fd = readstat_open(filename)) == -1) {
retval = READSTAT_ERROR_OPEN;
goto cleanup;
}
char magic[4];
if (read(fd, magic, 4) != 4) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
file_size = readstat_lseek(fd, 0, SEEK_END);
if (file_size == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (readstat_lseek(fd, 0, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (strncmp(magic, "<sta", 4) == 0) {
retval = dta_read_xmlish_preamble(fd, ctx, &header);
} else {
if (read(fd, &header, sizeof(header)) != sizeof(header)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
}
if ((ctx = dta_ctx_init(header.nvar, header.nobs, header.byteorder, header.ds_format)) == NULL) {
retval = READSTAT_ERROR_MALLOC;
goto cleanup;
}
ctx->user_ctx = user_ctx;
ctx->file_size = file_size;
ctx->progress_handler = parser->progress_handler;
retval = dta_update_progress(fd, ctx);
if (retval != READSTAT_OK)
goto cleanup;
if (parser->info_handler) {
if (parser->info_handler(ctx->nobs, ctx->nvar, user_ctx)) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
}
if (ctx->file_is_xmlish) {
uint16_t label_len = 0;
unsigned char timestamp_len;
if ((retval = dta_read_tag(fd, ctx, "<label>")) != READSTAT_OK) {
goto cleanup;
}
if (ctx->data_label_len_len == 2) {
if (read(fd, &label_len, sizeof(uint16_t)) != sizeof(uint16_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
label_len = ctx->machine_needs_byte_swap ? byteswap2(label_len) : label_len;
} else if (ctx->data_label_len_len == 1) {
unsigned char label_len_char;
if (read(fd, &label_len_char, sizeof(unsigned char)) != sizeof(unsigned char)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
label_len = label_len_char;
}
if (readstat_lseek(fd, label_len, SEEK_CUR) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if ((retval = dta_read_tag(fd, ctx, "</label>")) != READSTAT_OK) {
goto cleanup;
}
if ((retval = dta_read_tag(fd, ctx, "<timestamp>")) != READSTAT_OK) {
goto cleanup;
}
if (read(fd, ×tamp_len, 1) != 1) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (readstat_lseek(fd, timestamp_len, SEEK_CUR) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if ((retval = dta_read_tag(fd, ctx, "</timestamp>")) != READSTAT_OK) {
goto cleanup;
}
} else {
if (readstat_lseek(fd, ctx->data_label_len, SEEK_CUR) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
if (ctx->time_stamp_len) {
if (readstat_lseek(fd, ctx->time_stamp_len, SEEK_CUR) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
}
}
if ((retval = dta_read_tag(fd, ctx, "</header>")) != READSTAT_OK) {
goto cleanup;
}
if (dta_read_map(fd, ctx) != READSTAT_OK) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (dta_read_descriptors(fd, ctx) != READSTAT_OK) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
for (i=0; i<ctx->nvar; i++) {
size_t max_len;
readstat_types_t type = dta_type_info(ctx->typlist[i], &max_len, ctx);
record_len += max_len;
if (type == READSTAT_TYPE_STRING)
max_len++; /* might append NULL */
if (parser->variable_handler) {
readstat_variable_t *variable = dta_init_variable(ctx, i, type);
const char *value_labels = NULL;
if (ctx->lbllist[ctx->lbllist_entry_len*i])
value_labels = &ctx->lbllist[ctx->lbllist_entry_len*i];
int cb_retval = parser->variable_handler(i, variable, value_labels, user_ctx);
free(variable);
if (cb_retval) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
}
}
if ((retval = dta_skip_expansion_fields(fd, ctx)) != READSTAT_OK) {
goto cleanup;
}
if (record_len == 0) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
if ((retval = dta_read_tag(fd, ctx, "<data>")) != READSTAT_OK) {
goto cleanup;
}
if ((retval = dta_update_progress(fd, ctx)) != READSTAT_OK) {
goto cleanup;
}
if ((buf = malloc(record_len)) == NULL) {
retval = READSTAT_ERROR_MALLOC;
goto cleanup;
}
for (i=0; i<ctx->nobs; i++) {
if (read(fd, buf, record_len) != record_len) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
int j;
off_t offset = 0;
for (j=0; j<ctx->nvar; j++) {
size_t max_len;
readstat_value_t value;
memset(&value, 0, sizeof(readstat_value_t));
value.type = dta_type_info(ctx->typlist[j], &max_len, ctx);
if (value.type == READSTAT_TYPE_STRING) {
readstat_convert(str_buf, sizeof(str_buf), &buf[offset], max_len, ctx->converter);
value.v.string_value = str_buf;
} else if (value.type == READSTAT_TYPE_LONG_STRING) {
uint32_t v, o;
v = *((uint32_t *)&buf[offset]);
o = *((uint32_t *)&buf[offset+4]);
if (ctx->machine_needs_byte_swap) {
v = byteswap4(v);
o = byteswap4(o);
}
if (v > 0 && o > 0) {
off_t cur_pos = readstat_lseek(fd, 0, SEEK_CUR);
if (cur_pos == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
retval = dta_read_long_string(fd, ctx, v, o, &long_string);
if (retval != READSTAT_OK) {
goto cleanup;
}
value.v.string_value = long_string;
if (readstat_lseek(fd, cur_pos, SEEK_SET) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
}
} else if (value.type == READSTAT_TYPE_CHAR) {
char byte = buf[offset];
if (ctx->machine_is_twos_complement) {
byte = ones_to_twos_complement1(byte);
}
if (byte > DTA_MAX_CHAR) {
value.is_system_missing = 1;
if (byte > DTA_MISSING_CHAR) {
value.tag = 'a' + (byte - DTA_MISSING_CHAR_A);
}
}
value.v.char_value = byte;
} else if (value.type == READSTAT_TYPE_INT16) {
int16_t num = *((int16_t *)&buf[offset]);
if (ctx->machine_needs_byte_swap) {
num = byteswap2(num);
}
if (ctx->machine_is_twos_complement) {
num = ones_to_twos_complement2(num);
}
if (num > DTA_MAX_INT16) {
value.is_system_missing = 1;
if (num > DTA_MISSING_INT16) {
value.tag = 'a' + (num - DTA_MISSING_INT16_A);
}
}
value.v.i16_value = num;
} else if (value.type == READSTAT_TYPE_INT32) {
int32_t num = *((int32_t *)&buf[offset]);
if (ctx->machine_needs_byte_swap) {
num = byteswap4(num);
}
if (ctx->machine_is_twos_complement) {
num = ones_to_twos_complement4(num);
}
if (num > DTA_MAX_INT32) {
value.is_system_missing = 1;
if (num > DTA_MISSING_INT32) {
value.tag = 'a' + (num - DTA_MISSING_INT32_A);
}
}
value.v.i32_value = num;
} else if (value.type == READSTAT_TYPE_FLOAT) {
uint32_t num = *((uint32_t *)&buf[offset]);
float f_num = NAN;
if (ctx->machine_needs_byte_swap) {
num = byteswap4(num);
}
if (num > DTA_MAX_FLOAT) {
value.is_system_missing = 1;
if (num > DTA_MISSING_FLOAT) {
value.tag = 'a' + ((num - DTA_MISSING_FLOAT_A) >> 11);
}
} else {
uint16_t sas_read2(const char *data, int bswap) {
uint16_t tmp;
memcpy(&tmp, data, 2);
return bswap ? byteswap2(tmp) : tmp;
}
static readstat_error_t xport_write_variables(readstat_writer_t *writer) {
readstat_error_t retval = READSTAT_OK;
int i;
long offset = 0;
int num_long_labels = 0;
int any_has_long_format = 0;
for (i=0; i<writer->variables_count; i++) {
int needs_long_record = 0;
readstat_variable_t *variable = readstat_get_variable(writer, i);
size_t width = xport_variable_width(variable->type, variable->user_width);
xport_namestr_t namestr = {
.nvar0 = i,
.nlng = width,
.npos = offset
};
if (readstat_variable_get_type_class(variable) == READSTAT_TYPE_CLASS_STRING) {
namestr.ntype = SAS_COLUMN_TYPE_CHR;
} else {
namestr.ntype = SAS_COLUMN_TYPE_NUM;
}
copypad(namestr.nname, sizeof(namestr.nname), variable->name);
copypad(namestr.nlabel, sizeof(namestr.nlabel), variable->label);
if (variable->format[0]) {
int decimals = 0;
int width = 0;
char name[24];
sscanf(variable->format, "%s%d.%d", name, &width, &decimals);
copypad(namestr.nform, sizeof(namestr.nform), name);
namestr.nfl = width;
namestr.nfd = decimals;
copypad(namestr.niform, sizeof(namestr.niform), name);
namestr.nifl = width;
namestr.nifd = decimals;
if (strlen(name) > 8) {
any_has_long_format = 1;
needs_long_record = 1;
}
}
namestr.nfj = (variable->alignment == READSTAT_ALIGNMENT_RIGHT);
if (writer->version == 8) {
copypad(namestr.longname, sizeof(namestr.longname), variable->name);
size_t label_len = strlen(variable->label);
if (label_len > 40) {
needs_long_record = 1;
}
namestr.labeln = label_len;
}
if (needs_long_record) {
num_long_labels++;
}
offset += width;
xport_namestr_bswap(&namestr);
retval = xport_write_bytes(writer, &namestr, sizeof(xport_namestr_t));
if (retval != READSTAT_OK)
goto cleanup;
}
retval = xport_finish_record(writer);
if (retval != READSTAT_OK)
goto cleanup;
if (writer->version == 8 && num_long_labels) {
xport_header_record_t header = {
.name = "LABELV8",
.num1 = num_long_labels };
if (any_has_long_format) {
strcpy(header.name, "LABELV9");
}
retval = xport_write_header_record_v8(writer, &header);
if (retval != READSTAT_OK)
goto cleanup;
for (i=0; i<writer->variables_count; i++) {
readstat_variable_t *variable = readstat_get_variable(writer, i);
size_t label_len = strlen(variable->label);
size_t name_len = strlen(variable->name);
int has_long_label = 0;
int has_long_format = 0;
int format_len = 0;
char format_name[24];
memset(format_name, 0, sizeof(format_name));
has_long_label = (label_len > 40);
if (variable->format[0]) {
int decimals = 2;
int width = 8;
int matches = sscanf(variable->format, "%s%d.%d", format_name, &width, &decimals);
if (matches < 1) {
retval = READSTAT_ERROR_BAD_FORMAT_STRING;
goto cleanup;
}
format_len = strlen(format_name);
if (format_len > 8) {
has_long_format = 1;
}
}
if (has_long_format) {
uint16_t labeldef[5] = { i, name_len, format_len, format_len, label_len };
if (machine_is_little_endian()) {
labeldef[0] = byteswap2(labeldef[0]);
labeldef[1] = byteswap2(labeldef[1]);
labeldef[2] = byteswap2(labeldef[2]);
labeldef[3] = byteswap2(labeldef[3]);
labeldef[4] = byteswap2(labeldef[4]);
}
retval = readstat_write_bytes(writer, labeldef, sizeof(labeldef));
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, variable->name);
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, format_name);
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, format_name);
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, variable->label);
if (retval != READSTAT_OK)
goto cleanup;
} else if (has_long_label) {
uint16_t labeldef[3] = { i, name_len, label_len };
if (machine_is_little_endian()) {
labeldef[0] = byteswap2(labeldef[0]);
labeldef[1] = byteswap2(labeldef[1]);
labeldef[2] = byteswap2(labeldef[2]);
}
retval = readstat_write_bytes(writer, labeldef, sizeof(labeldef));
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, variable->name);
if (retval != READSTAT_OK)
goto cleanup;
retval = readstat_write_string(writer, variable->label);
if (retval != READSTAT_OK)
goto cleanup;
}
}
retval = xport_finish_record(writer);
if (retval != READSTAT_OK)
goto cleanup;
}
cleanup:
return retval;
}
static readstat_error_t dta_handle_rows(dta_ctx_t *ctx) {
readstat_io_t *io = ctx->io;
char *buf = NULL;
char str_buf[2048];
int i;
readstat_error_t retval = READSTAT_OK;
if ((buf = malloc(ctx->record_len)) == NULL) {
retval = READSTAT_ERROR_MALLOC;
goto cleanup;
}
for (i=0; i<ctx->row_limit; i++) {
if (io->read(buf, ctx->record_len, io->io_ctx) != ctx->record_len) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
int j;
off_t offset = 0;
for (j=0; j<ctx->nvar; j++) {
size_t max_len;
readstat_value_t value;
memset(&value, 0, sizeof(readstat_value_t));
value.type = dta_type_info(ctx->typlist[j], &max_len, ctx);
if (value.type == READSTAT_TYPE_STRING) {
readstat_convert(str_buf, sizeof(str_buf), &buf[offset], max_len, ctx->converter);
value.v.string_value = str_buf;
} else if (value.type == READSTAT_TYPE_STRING_REF) {
dta_strl_t key;
dta_interpret_strl_vo_bytes(ctx, (unsigned char *)&buf[offset], &key);
dta_strl_t **found = bsearch(&key, ctx->strls, ctx->strls_count, sizeof(dta_strl_t *), &dta_compare_strls);
if (found) {
value.v.string_value = (*found)->data;
}
value.type = READSTAT_TYPE_STRING;
} else if (value.type == READSTAT_TYPE_INT8) {
int8_t byte = buf[offset];
if (ctx->machine_is_twos_complement) {
byte = ones_to_twos_complement1(byte);
}
if (byte > ctx->max_int8) {
if (ctx->supports_tagged_missing && byte > DTA_113_MISSING_INT8) {
value.tag = 'a' + (byte - DTA_113_MISSING_INT8_A);
value.is_tagged_missing = 1;
} else {
value.is_system_missing = 1;
}
}
value.v.i8_value = byte;
} else if (value.type == READSTAT_TYPE_INT16) {
int16_t num = *((int16_t *)&buf[offset]);
if (ctx->machine_needs_byte_swap) {
num = byteswap2(num);
}
if (ctx->machine_is_twos_complement) {
num = ones_to_twos_complement2(num);
}
if (num > ctx->max_int16) {
if (ctx->supports_tagged_missing && num > DTA_113_MISSING_INT16) {
value.tag = 'a' + (num - DTA_113_MISSING_INT16_A);
value.is_tagged_missing = 1;
} else {
value.is_system_missing = 1;
}
}
value.v.i16_value = num;
} else if (value.type == READSTAT_TYPE_INT32) {
int32_t num = *((int32_t *)&buf[offset]);
if (ctx->machine_needs_byte_swap) {
num = byteswap4(num);
}
if (ctx->machine_is_twos_complement) {
num = ones_to_twos_complement4(num);
}
if (num > ctx->max_int32) {
if (ctx->supports_tagged_missing && num > DTA_113_MISSING_INT32) {
value.tag = 'a' + (num - DTA_113_MISSING_INT32_A);
value.is_tagged_missing = 1;
} else {
value.is_system_missing = 1;
}
}
value.v.i32_value = num;
} else if (value.type == READSTAT_TYPE_FLOAT) {
int32_t num = *((int32_t *)&buf[offset]);
float f_num = NAN;
if (ctx->machine_needs_byte_swap) {
num = byteswap4(num);
}
if (num > ctx->max_float) {
if (ctx->supports_tagged_missing && num > DTA_113_MISSING_FLOAT) {
value.tag = 'a' + ((num - DTA_113_MISSING_FLOAT_A) >> 11);
value.is_tagged_missing = 1;
} else {
value.is_system_missing = 1;
}
} else {
static readstat_error_t dta_read_expansion_fields(dta_ctx_t *ctx) {
readstat_error_t retval = READSTAT_OK;
readstat_io_t *io = ctx->io;
char *buffer = NULL;
if (ctx->expansion_len_len == 0)
return READSTAT_OK;
if (ctx->file_is_xmlish && !ctx->note_handler) {
if (io->seek(ctx->data_offset, READSTAT_SEEK_SET, io->io_ctx) == -1) {
return READSTAT_ERROR_SEEK;
}
return READSTAT_OK;
}
retval = dta_read_tag(ctx, "<characteristics>");
if (retval != READSTAT_OK)
goto cleanup;
while (1) {
size_t len;
char data_type;
if (ctx->file_is_xmlish) {
char start[4];
if (io->read(start, sizeof(start), io->io_ctx) != sizeof(start)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
if (memcmp(start, "</ch", sizeof(start)) == 0) {
retval = dta_read_tag(ctx, "aracteristics>");
if (retval != READSTAT_OK)
goto cleanup;
break;
} else if (memcmp(start, "<ch>", sizeof(start)) != 0) {
retval = READSTAT_ERROR_PARSE;
goto cleanup;
}
data_type = 1;
} else {
if (io->read(&data_type, 1, io->io_ctx) != 1) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
}
if (ctx->expansion_len_len == 2) {
int16_t len16;
if (io->read(&len16, sizeof(int16_t), io->io_ctx) != sizeof(int16_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
len = ctx->machine_needs_byte_swap ? byteswap2(len16) : len16;
} else {
int32_t len32;
if (io->read(&len32, sizeof(int32_t), io->io_ctx) != sizeof(int32_t)) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
len = ctx->machine_needs_byte_swap ? byteswap2(len32) : len32;
}
if (data_type == 0 && len == 0)
break;
if (data_type != 1 || len > (1<<20)) {
retval = READSTAT_ERROR_NOTE_IS_TOO_LONG;
goto cleanup;
}
if (ctx->note_handler && len >= 2 * ctx->ch_metadata_len) {
buffer = realloc(buffer, len + 1);
buffer[len] = '\0';
if (io->read(buffer, len, io->io_ctx) != len) {
retval = READSTAT_ERROR_READ;
goto cleanup;
}
int index = 0;
if (strncmp(&buffer[0], "_dta", 4) == 0 &&
sscanf(&buffer[ctx->ch_metadata_len], "note%d", &index) == 1) {
if (ctx->note_handler(index, &buffer[2*ctx->ch_metadata_len], ctx->user_ctx)) {
retval = READSTAT_ERROR_USER_ABORT;
goto cleanup;
}
}
} else {
if (io->seek(len, READSTAT_SEEK_CUR, io->io_ctx) == -1) {
retval = READSTAT_ERROR_SEEK;
goto cleanup;
}
}
retval = dta_read_tag(ctx, "</ch>");
if (retval != READSTAT_OK)
goto cleanup;
}
cleanup:
if (buffer)
free(buffer);
return retval;
}
