Github user sohami commented on a diff in the pull request:

    https://github.com/apache/drill/pull/1111#discussion_r166213941
  
    --- Diff: 
contrib/storage-hive/core/src/main/java/org/apache/drill/exec/store/hive/HiveUtilities.java
 ---
    @@ -507,5 +509,51 @@ public static boolean 
hasHeaderOrFooter(HiveTableWithColumnCache table) {
         int skipFooter = retrieveIntProperty(tableProperties, 
serdeConstants.FOOTER_COUNT, -1);
         return skipHeader > 0 || skipFooter > 0;
       }
    +
    +  /**
    +   * This method checks whether the schema evolution properties are set in 
job conf for the input format. If they
    +   * aren't set, method sets the column names and types from 
table/partition properties or storage descriptor.
    +   * @param job the job to update
    +   * @param properties table or partition properties
    +   * @param isAcidTable true if the table is transactional, false otherwise
    +   * @param sd storage descriptor
    +   */
    +  public static void setColumnTypes(JobConf job, Properties properties, 
boolean isAcidTable, StorageDescriptor sd) {
    +
    +    // No work is needed, if schema evolution is used
    +    if (Utilities.isSchemaEvolutionEnabled(job, isAcidTable) && 
job.get(IOConstants.SCHEMA_EVOLUTION_COLUMNS) != null &&
    +        job.get(IOConstants.SCHEMA_EVOLUTION_COLUMNS_TYPES) != null) {
    +      return;
    +    }
    +
    +    String colNames;
    +    String colTypes;
    +
    +    // Try to get get column names and types from table or partition 
properties. If they are absent there, get columns
    +    // data from storage descriptor of the table
    +    if (properties.containsKey(serdeConstants.LIST_COLUMNS) && 
properties.containsKey(serdeConstants.LIST_COLUMN_TYPES)) {
    +      colNames = job.get(serdeConstants.LIST_COLUMNS);
    +      colTypes = job.get(serdeConstants.LIST_COLUMN_TYPES);
    +    } else {
    +      StringBuilder colNamesBuilder = new StringBuilder();
    +      StringBuilder colTypesBuilder = new StringBuilder();
    +      boolean isFirst = true;
    +      for(FieldSchema col: sd.getCols()) {
    +        if (isFirst) {
    +          isFirst = false;
    +        } else {
    +          colNamesBuilder.append(',');
    +          colTypesBuilder.append(',');
    +        }
    +        colNamesBuilder.append(col.getName());
    +        colTypesBuilder.append(col.getType());
    +      }
    +      colNames = colNamesBuilder.toString();
    +      colTypes = colTypesBuilder.toString();
    --- End diff --
    
    how about changing the loop as below:
    
    ```
    final StringBuilder colNamesBuilder = new StringBuilder();
    final StringBuilder colTypesBuilder = new StringBuilder();
    
         for(FieldSchema col: sd.getCols()) {
               colNamesBuilder.append(col.getName());
               colTypesBuilder.append(col.getType());
               colNamesBuilder.append(',');
               colTypesBuilder.append(',');
          }
          colNames = colNamesBuilder.substring(0, colNamesBuilder.length() - 1);
          colTypes = colTypesBuilder.substring(0, colTypesBuilder.length() - 1);
    ```



---

Reply via email to