Status OrcSchemaResolver::ValidatePrimitiveType()

in be/src/exec/orc/orc-metadata-utils.cc [463:533]

• 64 lines of code
• 40 McCabe index (conditional complexity)

Status OrcSchemaResolver::ValidatePrimitiveType(const ColumnType& type,
    const orc::Type& orc_type) const {
  switch (orc_type.getKind()) {
    case orc::TypeKind::BOOLEAN:
      if (type.type == TYPE_BOOLEAN) return Status::OK();
      break;
    case orc::TypeKind::BYTE:
      if (type.type == TYPE_TINYINT || type.type == TYPE_SMALLINT
          || type.type == TYPE_INT || type.type == TYPE_BIGINT) {
        return Status::OK();
      }
      break;
    case orc::TypeKind::SHORT:
      if (type.type == TYPE_SMALLINT || type.type == TYPE_INT
          || type.type == TYPE_BIGINT) {
        return Status::OK();
      }
      break;
    case orc::TypeKind::INT:
      if (type.type == TYPE_INT || type.type == TYPE_BIGINT) return Status::OK();
      break;
    case orc::TypeKind::LONG:
      if (type.type == TYPE_BIGINT) return Status::OK();
      break;
    case orc::TypeKind::FLOAT:
    case orc::TypeKind::DOUBLE:
      if (type.type == TYPE_FLOAT || type.type == TYPE_DOUBLE) return Status::OK();
      break;
    case orc::TypeKind::STRING:
    case orc::TypeKind::VARCHAR:
    case orc::TypeKind::CHAR:
    case orc::TypeKind::BINARY:
      // orc::TypeKind::BINARY is handled as TYPE_STRING, TYPE_BINARY is not used.
      if (type.type == TYPE_STRING || type.type == TYPE_VARCHAR
          || type.type == TYPE_CHAR) {
        return Status::OK();
      }
      break;
    case orc::TypeKind::TIMESTAMP:
    case orc::TypeKind::TIMESTAMP_INSTANT:
      if (type.type == TYPE_TIMESTAMP) return Status::OK();
      break;
    case orc::TypeKind::DECIMAL: {
      if (type.type != TYPE_DECIMAL || type.scale != orc_type.getScale()) break;
      bool overflow = false;
      int orc_precision = orc_type.getPrecision();
      if (orc_precision == 0 || orc_precision > ColumnType::MAX_DECIMAL8_PRECISION) {
        // For ORC decimals whose precision is larger than 18, its value can't fit into
        // an int64 (10^19 > 2^63). So we should use int128 (16 bytes) for this case.
        // The possible byte sizes for Impala decimals are 4, 8, 16.
        // We mark it as overflow if the target byte size is not 16.
        overflow = (type.GetByteSize() != 16);
      } else if (orc_type.getPrecision() > ColumnType::MAX_DECIMAL4_PRECISION) {
        // For ORC decimals whose precision <= 18 and > 9, int64 and int128 can fit them.
        // We only mark it as overflow if the target byte size is 4.
        overflow = (type.GetByteSize() == 4);
      }
      if (!overflow) return Status::OK();
      return Status(Substitute(
          "Column $0 in ORC file '$1' can't be truncated to table column $2",
          orc_type.toString(), filename_, type.DebugString()));
    }
    case orc::TypeKind::DATE:
      if (type.type == TYPE_DATE) return Status::OK();
      break;
    default: break;
  }
  return Status(Substitute(
      "Type mismatch: table column $0 is map to column $1 in ORC file '$2'",
      type.DebugString(), orc_type.toString(), filename_));
}