Examples with ExtractPrepareException org.apache.gobblin.source.extractor.exception.ExtractPrepareException used on opensource projects

Example 1 with ExtractPrepareException

use of org.apache.gobblin.source.extractor.exception.ExtractPrepareException in project incubator-gobblin by apache.

the class QueryBasedExtractor method build.

/**
 * build schema, record count and high water mark
 */
public Extractor<S, D> build() throws ExtractPrepareException {
    String watermarkColumn = this.workUnitState.getProp(ConfigurationKeys.EXTRACT_DELTA_FIELDS_KEY);
    long lwm = partition.getLowWatermark();
    long hwm = partition.getHighWatermark();
    log.info("Low water mark: " + lwm + "; and High water mark: " + hwm);
    WatermarkType watermarkType;
    if (StringUtils.isBlank(this.workUnitState.getProp(ConfigurationKeys.SOURCE_QUERYBASED_WATERMARK_TYPE))) {
        watermarkType = null;
    } else {
        watermarkType = WatermarkType.valueOf(this.workUnitState.getProp(ConfigurationKeys.SOURCE_QUERYBASED_WATERMARK_TYPE).toUpperCase());
    }
    log.info("Source Entity is " + this.entity);
    try {
        this.setTimeOut(this.workUnitState.getPropAsInt(ConfigurationKeys.SOURCE_CONN_TIMEOUT, ConfigurationKeys.DEFAULT_CONN_TIMEOUT));
        this.extractMetadata(this.schema, this.entity, this.workUnit);
        if (StringUtils.isNotBlank(watermarkColumn)) {
            if (partition.isLastPartition()) {
                // Get a more accurate high watermark from the source
                long adjustedHighWatermark = this.getLatestWatermark(watermarkColumn, watermarkType, lwm, hwm);
                log.info("High water mark from source: " + adjustedHighWatermark);
                // Else, consider the low watermark as high water mark(with no delta).i.e, don't move the pointer
                if (adjustedHighWatermark == ConfigurationKeys.DEFAULT_WATERMARK_VALUE) {
                    adjustedHighWatermark = getLowWatermarkWithNoDelta(lwm);
                }
                this.highWatermark = adjustedHighWatermark;
            } else {
                this.highWatermark = hwm;
            }
            log.info("High water mark for the current run: " + highWatermark);
            this.setRangePredicates(watermarkColumn, watermarkType, lwm, highWatermark);
        }
        // if it is set to true, skip count calculation and set source count to -1
        if (!Boolean.valueOf(this.workUnitState.getProp(ConfigurationKeys.SOURCE_QUERYBASED_SKIP_COUNT_CALC))) {
            this.sourceRecordCount = this.getSourceCount(this.schema, this.entity, this.workUnit, this.predicateList);
        } else {
            log.info("Skip count calculation");
            this.sourceRecordCount = -1;
        }
        if (this.sourceRecordCount == 0) {
            log.info("Record count is 0; Setting fetch status to false to skip readRecord()");
            this.setFetchStatus(false);
        }
    } catch (SchemaException e) {
        throw new ExtractPrepareException("Failed to get schema for this object; error - " + e.getMessage(), e);
    } catch (HighWatermarkException e) {
        throw new ExtractPrepareException("Failed to get high watermark; error - " + e.getMessage(), e);
    } catch (RecordCountException e) {
        throw new ExtractPrepareException("Failed to get record count; error - " + e.getMessage(), e);
    } catch (Exception e) {
        throw new ExtractPrepareException("Failed to prepare the extract build; error - " + e.getMessage(), e);
    }
    return this;
}