int FileReader::extract_record(RecordBlock &block, const RecordDelima &rec_delima, const RecordDelima &col_delima, int64_t max_rec_extracted)
{
int64_t pos = 0;
int64_t rec_nr = 0;
int64_t last_pos = 0;
int ret = 0;
bool has_partial_rec = false;
std::string &buffer = block.buffer();
while (pos < buffer_size_) {
char *p = reader_buffer_ + pos;
last_pos = pos;
if (rec_nr >= max_rec_extracted) {
TBSYS_LOG(DEBUG, "extract rec count[%ld] reaches max_rec_extracted[%ld]", rec_nr, max_rec_extracted);
break;
}
if ((has_partial_rec = is_partial_record(pos, rec_delima, col_delima)) == true) {
TBSYS_LOG(DEBUG, "partial record meet, will be proceed next time");
break;
}
rec_nr++;
col_delima.append_delima(reader_buffer_, pos, buffer_size_); /* replace rec_delima with col_delima */
rec_delima.skip_delima(pos); /* skip rec_delima */
/* include rec_delima in buffer */
int32_t psize = static_cast<int32_t>(pos - last_pos);
append_int32(buffer, psize);
last_pos = pos;
buffer.append(p, psize); /* append length of record here */
}
TBSYS_LOG(DEBUG, "record block actual used size[%ld], rec_nr[%ld]", pos, rec_nr);
if ((has_partial_rec && last_pos < buffer_size_) /* partial record reside in buffer */
|| (rec_nr >= max_rec_extracted && last_pos < buffer_size_)) {
shrink(last_pos);
} else {
buffer_size_ = 0;
}
if (rec_nr == 0 && buffer_size_ == kReadBufferSize) {
TBSYS_LOG(ERROR, "record size too big, max size=%d", kReadBufferSize);
ret = -1;
}
if (ret == 0) {
block.set_rec_num(rec_nr);
}
return ret;
}
char *rec_state_bits_to_str(DEV_RECORD *rec)
{
static char buf[200];
buf[0] = 0;
if (rec->state_bits & REC_NO_HEADER) {
strcat(buf, _("Nohdr,"));
}
if (is_partial_record(rec)) {
strcat(buf, _("partial,"));
}
if (rec->state_bits & REC_BLOCK_EMPTY) {
strcat(buf, _("empty,"));
}
if (rec->state_bits & REC_NO_MATCH) {
strcat(buf, _("Nomatch,"));
}
if (rec->state_bits & REC_CONTINUATION) {
strcat(buf, _("cont,"));
}
if (buf[0]) {
buf[strlen(buf)-1] = 0;
}
return buf;
}