MCOL-5746 Cpimport: convert blob data from ascii hex when reads from STDIN

writeengine/bulk/we_columninfo.cpp

@@ -1661,7 +1661,8 @@ int ColumnInfo::closeDctnryStore(bool bAbort)
 // Update dictionary store file with string column parquet data, and return the assigned
 // tokens (tokenbuf) to be stored in the corresponding column token file.
 //--------------------------------------------------------------------------------------
-int ColumnInfo::updateDctnryStoreParquet(std::shared_ptr<arrow::Array> columnData, int tokenPos, const int totalRow, char* tokenBuf)
+int ColumnInfo::updateDctnryStoreParquet(std::shared_ptr<arrow::Array> columnData, int tokenPos,
+                                         const int totalRow, char* tokenBuf)
 {
   long long truncCount = 0;
 
@@ -1688,7 +1689,6 @@ int ColumnInfo::updateDctnryStoreParquet(std::shared_ptr<arrow::Array> columnDat
   }
 
   incSaturatedCnt(truncCount);
-  
   return NO_ERROR;
 }
 
@@ -1707,7 +1707,7 @@ int ColumnInfo::updateDctnryStore(char* buf, ColPosPair** pos, const int totalRo
   // column.
   // This only applies to default text mode.  This step is bypassed for
   // binary imports, because in that case, the data is already true binary.
-  if (((curCol.colType == WR_VARBINARY) || (curCol.colType == WR_BLOB)) &&
+  if (((curCol.colType == WR_VARBINARY) || (curCol.colType == WR_BLOB && fpTableInfo->readFromSTDIN())) &&
       (fpTableInfo->getImportDataMode() == IMPORT_DATA_TEXT))
   {
 #ifdef PROFILE

writeengine/bulk/we_tableinfo.cpp

@@ -2463,5 +2463,10 @@ int TableInfo::allocateBRMColumnExtent(OID columnOID, uint16_t dbRoot, uint32_t&
   return rc;
 }
 
+bool TableInfo::readFromSTDIN()
+{
+  return fReadFromStdin;
+}
+
 }  // namespace WriteEngine
 // end of namespace

writeengine/bulk/we_tableinfo.h

@@ -80,7 +80,7 @@ class TableInfo : public WeUIDGID
   FILE* fHandle;           // Handle to the input load file
   int fCurrentReadBuffer;  // Id of current buffer being popu-
   //   lated by the read thread
-  RID fTotalReadRows;               // Total number of rows read
+  RID fTotalReadRows;      // Total number of rows read
   unsigned fTotalErrRows;  // Total error rows among all input
   //   for this table.  Is volatile to
   //   insure parser & reader threads
@@ -174,24 +174,25 @@ class TableInfo : public WeUIDGID
   boost::uuids::uuid fJobUUID;              // Job UUID
   std::vector<BRM::LBID_t> fDictFlushBlks;  // dict blks to be flushed from cache
 
-  std::shared_ptr<arrow::RecordBatchReader> fParquetReader;   // Batch reader to read batches of data
+  std::shared_ptr<arrow::RecordBatchReader> fParquetReader;  // Batch reader to read batches of data
-  std::unique_ptr<parquet::arrow::FileReader> fReader;        // Reader to read parquet file
+  std::unique_ptr<parquet::arrow::FileReader> fReader;       // Reader to read parquet file
   //--------------------------------------------------------------------------
   // Private Functions
   //--------------------------------------------------------------------------
 
-  int changeTableLockState();                       // Change state of table lock to cleanup
+  int changeTableLockState();           // Change state of table lock to cleanup
-  void closeTableFile();                            // Close current tbl file; free buffer
+  void closeTableFile();                // Close current tbl file; free buffer
-  void closeOpenDbFiles();                          // Close DB files left open at job's end
+  void closeOpenDbFiles();              // Close DB files left open at job's end
-  int confirmDBFileChanges();                       // Confirm DB file changes (on HDFS)
+  int confirmDBFileChanges();           // Confirm DB file changes (on HDFS)
-  void deleteTempDBFileChanges();                   // Delete DB temp swap files (on HDFS)
+  void deleteTempDBFileChanges();       // Delete DB temp swap files (on HDFS)
-  int finishBRM();                                  // Finish reporting updates for BRM
+  int finishBRM();                      // Finish reporting updates for BRM
-  void freeProcessingBuffers();                     // Free up Processing Buffers
+  void freeProcessingBuffers();         // Free up Processing Buffers
-  bool isBufferAvailable(bool report);              // Is tbl buffer available for reading
+  bool isBufferAvailable(bool report);  // Is tbl buffer available for reading
-  int openTableFileParquet(int64_t &totalRowsParquet);  // Open parquet data file and set batch reader for each buffer
+  int openTableFileParquet(
-  int openTableFile();                              // Open data file and set the buffer
+      int64_t& totalRowsParquet);        // Open parquet data file and set batch reader for each buffer
-  void reportTotals(double elapsedSec);             // Report summary totals
+  int openTableFile();                   // Open data file and set the buffer
-  void sleepMS(long int ms);                        // Sleep method
+  void reportTotals(double elapsedSec);  // Report summary totals
+  void sleepMS(long int ms);             // Sleep method
   // Compare column HWM with the examplar HWM.
   int compareHWMs(const int smallestColumnId, const int widerColumnId, const uint32_t smallerColumnWidth,
                   const uint32_t widerColumnWidth, const std::vector<DBRootExtentInfo>& segFileInfo,
@@ -465,6 +466,8 @@ class TableInfo : public WeUIDGID
 
   void setJobUUID(const boost::uuids::uuid& jobUUID);
 
+  bool readFromSTDIN();
+
  public:
   friend class BulkLoad;
   friend class ColumnInfo;