Examples with GetTableDataSupport com.thinkbiganalytics.ingest.GetTableDataSupport used on opensource projects

Example 1 with GetTableDataSupport

use of com.thinkbiganalytics.ingest.GetTableDataSupport in project kylo by Teradata.

the class GetTableData method onTrigger.

@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) throws ProcessException {
    FlowFile flowFile = null;
    if (context.hasIncomingConnection()) {
        flowFile = session.get();
        // we know that we should run only if we have a FlowFile.
        if (flowFile == null && context.hasNonLoopConnection()) {
            return;
        }
    }
    final FlowFile incoming = flowFile;
    final ComponentLog logger = getLog();
    final DBCPService dbcpService = context.getProperty(JDBC_SERVICE).asControllerService(DBCPService.class);
    final MetadataProviderService metadataService = context.getProperty(METADATA_SERVICE).asControllerService(MetadataProviderService.class);
    final String loadStrategy = context.getProperty(LOAD_STRATEGY).getValue();
    final String categoryName = context.getProperty(FEED_CATEGORY).evaluateAttributeExpressions(incoming).getValue();
    final String feedName = context.getProperty(FEED_NAME).evaluateAttributeExpressions(incoming).getValue();
    final String tableName = context.getProperty(TABLE_NAME).evaluateAttributeExpressions(incoming).getValue();
    final String fieldSpecs = context.getProperty(TABLE_SPECS).evaluateAttributeExpressions(incoming).getValue();
    final String dateField = context.getProperty(DATE_FIELD).evaluateAttributeExpressions(incoming).getValue();
    final Integer queryTimeout = context.getProperty(QUERY_TIMEOUT).asTimePeriod(TimeUnit.SECONDS).intValue();
    final Integer overlapTime = context.getProperty(OVERLAP_TIME).evaluateAttributeExpressions(incoming).asTimePeriod(TimeUnit.SECONDS).intValue();
    final Integer backoffTime = context.getProperty(BACKOFF_PERIOD).asTimePeriod(TimeUnit.SECONDS).intValue();
    final String unitSize = context.getProperty(UNIT_SIZE).getValue();
    final String outputType = context.getProperty(OUTPUT_TYPE).getValue();
    String outputDelimiter = context.getProperty(OUTPUT_DELIMITER).evaluateAttributeExpressions(incoming).getValue();
    final String delimiter = StringUtils.isBlank(outputDelimiter) ? "," : outputDelimiter;
    final PropertyValue waterMarkPropName = context.getProperty(HIGH_WATER_MARK_PROP).evaluateAttributeExpressions(incoming);
    final String[] selectFields = parseFields(fieldSpecs);
    final LoadStrategy strategy = LoadStrategy.valueOf(loadStrategy);
    final StopWatch stopWatch = new StopWatch(true);
    try (final Connection conn = dbcpService.getConnection()) {
        FlowFile outgoing = (incoming == null ? session.create() : incoming);
        final AtomicLong nrOfRows = new AtomicLong(0L);
        final LastFieldVisitor visitor = new LastFieldVisitor(dateField, null);
        final FlowFile current = outgoing;
        outgoing = session.write(outgoing, new OutputStreamCallback() {

            @Override
            public void process(final OutputStream out) throws IOException {
                ResultSet rs = null;
                try {
                    GetTableDataSupport support = new GetTableDataSupport(conn, queryTimeout);
                    if (strategy == LoadStrategy.FULL_LOAD) {
                        rs = support.selectFullLoad(tableName, selectFields);
                    } else if (strategy == LoadStrategy.INCREMENTAL) {
                        String waterMarkValue = getIncrementalWaterMarkValue(current, waterMarkPropName);
                        LocalDateTime waterMarkTime = LocalDateTime.parse(waterMarkValue, DATE_TIME_FORMAT);
                        Date lastLoadDate = toDate(waterMarkTime);
                        visitor.setLastModifyDate(lastLoadDate);
                        rs = support.selectIncremental(tableName, selectFields, dateField, overlapTime, lastLoadDate, backoffTime, GetTableDataSupport.UnitSizes.valueOf(unitSize));
                    } else {
                        throw new RuntimeException("Unsupported loadStrategy [" + loadStrategy + "]");
                    }
                    if (GetTableDataSupport.OutputType.DELIMITED.equals(GetTableDataSupport.OutputType.valueOf(outputType))) {
                        nrOfRows.set(JdbcCommon.convertToDelimitedStream(rs, out, (strategy == LoadStrategy.INCREMENTAL ? visitor : null), delimiter));
                    } else if (GetTableDataSupport.OutputType.AVRO.equals(GetTableDataSupport.OutputType.valueOf(outputType))) {
                        avroSchema = JdbcCommon.createSchema(rs);
                        nrOfRows.set(JdbcCommon.convertToAvroStream(rs, out, (strategy == LoadStrategy.INCREMENTAL ? visitor : null), avroSchema));
                    } else {
                        throw new RuntimeException("Unsupported output format type [" + outputType + "]");
                    }
                } catch (final SQLException e) {
                    throw new IOException("SQL execution failure", e);
                } finally {
                    if (rs != null) {
                        try {
                            if (rs.getStatement() != null) {
                                rs.getStatement().close();
                            }
                            rs.close();
                        } catch (SQLException e) {
                            getLog().error("Error closing sql statement and resultset");
                        }
                    }
                }
            }
        });
        // set attribute how many rows were selected
        outgoing = session.putAttribute(outgoing, RESULT_ROW_COUNT, Long.toString(nrOfRows.get()));
        // set output format type and avro schema for feed setup, if available
        outgoing = session.putAttribute(outgoing, "db.table.output.format", outputType);
        String avroSchemaForFeedSetup = (avroSchema != null) ? JdbcCommon.getAvroSchemaForFeedSetup(avroSchema) : EMPTY_STRING;
        outgoing = session.putAttribute(outgoing, "db.table.avro.schema", avroSchemaForFeedSetup);
        session.getProvenanceReporter().modifyContent(outgoing, "Retrieved " + nrOfRows.get() + " rows", stopWatch.getElapsed(TimeUnit.MILLISECONDS));
        // Terminate flow file if no work
        Long rowcount = nrOfRows.get();
        outgoing = session.putAttribute(outgoing, ComponentAttributes.NUM_SOURCE_RECORDS.key(), String.valueOf(rowcount));
        if (nrOfRows.get() == 0L) {
            logger.info("{} contains no data; transferring to 'nodata'", new Object[] { outgoing });
            session.transfer(outgoing, REL_NO_DATA);
        } else {
            logger.info("{} contains {} records; transferring to 'success'", new Object[] { outgoing, nrOfRows.get() });
            if (strategy == LoadStrategy.INCREMENTAL) {
                String newWaterMarkStr = format(visitor.getLastModifyDate());
                outgoing = setIncrementalWaterMarkValue(session, outgoing, waterMarkPropName, newWaterMarkStr);
                logger.info("Recorded load status feed {} date {}", new Object[] { feedName, newWaterMarkStr });
            }
            session.transfer(outgoing, REL_SUCCESS);
        }
    } catch (final Exception e) {
        if (incoming == null) {
            logger.error("Unable to execute SQL select from table due to {}. No incoming flow file to route to failure", new Object[] { e });
        } else {
            logger.error("Unable to execute SQL select from table due to {}; routing to failure", new Object[] { incoming, e });
            session.transfer(incoming, REL_FAILURE);
        }
    }
}