Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Avro >> mail # user >> Error parsing an 'array<union{record1, record2}>' in an avdl file


Copy link to this message
-
Error parsing an 'array<union{record1, record2}>' in an avdl file
Hi Avro-ers,
I have an avdl file that I am getting the following parse error in:"
org.apache.avro.compiler.idl.ParseException: Encountered " "record"
"record "" at line <Pointer to the line of the first record after
'AndFilter' record def>." The relevant section of my avdl file looks like:

  record ColumnRangeFilter {
    /** The minimum qualifier bound. null means none.*/
    union{null, string} min_qualifier;

    /** Whether to include the lower bound. */
    boolean min_included;

    /** The maximum qualifier bound. null means none.*/
    union{null, string} max_qualifier;

    /** Whether to include the upper bound. */
    boolean max_included;
  }

  /**
   * Specification of a RegexQualifierFilter.
   */
   record RegexQualifierFilter {
     /** The regular expression for qualifiers that should be
      *  accepted by this filter. The expression is matched against the
full qualifier (as
      *  if it implicitly starts with '^' and ends with '$'.
      */
      string regex;
   }

  /**
   * Specification of an 'and' filter.
   */
  record AndFilter {
    /** A List of filters to apply the logical operator to.
    array<union{RegexQualifierFilter, ColumnRangeFilter}> filter_list;
  }

When I remove the "AndFilter" record declaration from my file, the file is
parsed correctly and all of the classes are generated. Is there a
different/better way to declare an array of union type in an avdl file? My
(hopefully temporary) solution is to make another record with one field
that is of type 'union{RegexQualifierFilter, ColumnRangeFilter}' and use
that to parameterize the array.
Thanks,
Juliet