SDC-12486. Avoid BigQuery.getTable(TableId) throw a NullPointerExcept…

…ion when TableId contains an empty name BigQuery.getTable(tableId) throws a NullPointerException when called with a tableId with an empty name (empty, not null). This is a bug in the Google client implementation. This workaround intends to prevent our software to crash in this case and to show a warning to the user instead while sending their record to error. Change-Id: Ifbdc01242d54ed4ccb077431b8cb6d993b88a90d Reviewed-on: https://review.streamsets.net/c/25791 Tested-by: StreamSets CI <streamsets-ci-spam@streamsets.com> Reviewed-by: Xavier Baqués <xavi@streamsets.com>
edp963 · Sep 17, 2019 · ee35154 · ee35154
1 parent d2737a5
commit ee35154
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 4 deletions.
diff --git a/...-lib/src/main/java/com/streamsets/pipeline/stage/bigquery/destination/BigQueryTarget.java b/...-lib/src/main/java/com/streamsets/pipeline/stage/bigquery/destination/BigQueryTarget.java
@@ -145,17 +145,25 @@ public void write(Batch batch) throws StageException {
           String datasetName = dataSetEval.eval(elVars, conf.datasetEL, String.class);
           String tableName = tableNameELEval.eval(elVars, conf.tableNameEL, String.class);
           TableId tableId = TableId.of(datasetName, tableName);
-          if (tableIdExistsCache.get(tableId)) {
+
+          // BigQuery.getTable(tableId) throws a NullPointerException if the table name is empty
+          // This has been reported to Google and they marked it as a bug in their issue tracker
+          // Send the record to error and show a more friendly message instead
+          if(tableName.isEmpty()) {
+            getContext().toError(record, Errors.BIGQUERY_18);
+          }
+          else if(!tableIdExistsCache.get(tableId)) {
+            getContext().toError(record, Errors.BIGQUERY_17, datasetName, tableName, conf.credentials.projectId);
+          }
+          else {
             List<Record> tableIdRecords = tableIdToRecords.computeIfAbsent(tableId, t -> new ArrayList<>());
             tableIdRecords.add(record);
-          } else {
-            getContext().toError(record, Errors.BIGQUERY_17, datasetName, tableName, conf.credentials.projectId);
           }
         } catch (ELEvalException e) {
           LOG.error("Error evaluating DataSet/TableName EL", e);
           getContext().toError(record, Errors.BIGQUERY_10, e);
         } catch (ExecutionException e){
-          LOG.error("Error when checking exists for tableId, Reason : {}", e);
+          LOG.error("Error when checking exists for tableId, Reason : {}", e.getMessage(), e);
           Throwable rootCause = Throwables.getRootCause(e);
           getContext().toError(record, Errors.BIGQUERY_13, rootCause);
         }

diff --git a/google-cloud-lib/src/main/java/com/streamsets/pipeline/stage/bigquery/lib/Errors.java b/google-cloud-lib/src/main/java/com/streamsets/pipeline/stage/bigquery/lib/Errors.java
@@ -32,6 +32,7 @@ public enum Errors implements ErrorCode {
   BIGQUERY_15("Error evaluated Row Id, value evaluates to empty"),
   BIGQUERY_16("Root field of record should be a Map or a List Map"),
   BIGQUERY_17("Data set '{}' or Table '{}' does not exist in Big Query under project '{}'"),
+  BIGQUERY_18("Table name is empty or table expression refers to an empty or non-existing field"),
   ;
 
   private final String msg;