private HBaseSubScanSpec regionInfoToSubScanSpec(HRegionInfo ri) {
HBaseScanSpec spec = hbaseScanSpec;
return new HBaseSubScanSpec()
.setTableName(spec.getTableName())
.setRegionServer(regionsToScan.get(ri).getHostname())
.setStartRow(
(!isNullOrEmpty(spec.getStartRow()) && ri.containsRow(spec.getStartRow()))
? spec.getStartRow()
: ri.getStartKey())
.setStopRow(
(!isNullOrEmpty(spec.getStopRow()) && ri.containsRow(spec.getStopRow()))
? spec.getStopRow()
: ri.getEndKey())
.setSerializedFilter(spec.getSerializedFilter());
}
@Override
public ScanStats getScanStats() {
int rowCount =
(int)
((scanSizeInBytes / statsCalculator.getAvgRowSizeInBytes())
* (hbaseScanSpec.getFilter() != null ? 0.5 : 1));
// the following calculation is not precise since 'columns' could specify CFs while
// getColsPerRow() returns the number of qualifier.
float diskCost =
scanSizeInBytes
* ((columns == null || columns.isEmpty())
? 1
: columns.size() / statsCalculator.getColsPerRow());
return new ScanStats(GroupScanProperty.NO_EXACT_ROW_COUNT, rowCount, 1, diskCost);
}
private void init() {
logger.debug("Getting region locations");
try {
HTable table = new HTable(storagePluginConfig.getHBaseConf(), hbaseScanSpec.getTableName());
this.hTableDesc = table.getTableDescriptor();
NavigableMap<HRegionInfo, ServerName> regionsMap = table.getRegionLocations();
statsCalculator =
new TableStatsCalculator(table, hbaseScanSpec, storagePlugin.getContext().getConfig());
boolean foundStartRegion = false;
regionsToScan = new TreeMap<HRegionInfo, ServerName>();
for (Entry<HRegionInfo, ServerName> mapEntry : regionsMap.entrySet()) {
HRegionInfo regionInfo = mapEntry.getKey();
if (!foundStartRegion
&& hbaseScanSpec.getStartRow() != null
&& hbaseScanSpec.getStartRow().length != 0
&& !regionInfo.containsRow(hbaseScanSpec.getStartRow())) {
continue;
}
foundStartRegion = true;
regionsToScan.put(regionInfo, mapEntry.getValue());
scanSizeInBytes += statsCalculator.getRegionSizeInBytes(regionInfo.getRegionName());
if (hbaseScanSpec.getStopRow() != null
&& hbaseScanSpec.getStopRow().length != 0
&& regionInfo.containsRow(hbaseScanSpec.getStopRow())) {
break;
}
}
table.close();
} catch (IOException e) {
throw new DrillRuntimeException(
"Error getting region info for table: " + hbaseScanSpec.getTableName(), e);
}
verifyColumns();
}