Method addToOutput

bolt/exec/NestedLoopJoinProbe.cpp:338–508

Main join loop.

// Main join loop.
bool NestedLoopJoinProbe::addToOutput() {
  BOLT_CHECK_NOT_NULL(input_);

  // First, create a new output vector. By default, allocate space for
  // outputBatchSize_ rows. The output always generates dictionaries wrapped
  // around the probe vector being processed.
  //
  // Since cross join batches can be returned without filter evaluation, no need
  // to prepare output here.
  if (!isCrossJoin()) {
    prepareOutput();
  }

  const auto startProbeRow = probeRow_;
  // probeRow_ is temporarily modified during the loop (for addOutputRow /
  // addProbeMismatchRow), but must be restored to startProbeRow on every exit
  // so that advanceProbe() can correctly advance by probeRowCount_.
  auto probeRowGuard = folly::makeGuard([&] { probeRow_ = startProbeRow; });

  while (!hasProbedAllBuildData()) {
    const auto& currentBuild = buildVectors_.value()[buildIndex_];

    // Empty build vector; move to the next.
    if (currentBuild->size() == 0) {
      ++buildIndex_;
      filterResultRow_ = 0;
      continue;
    }

    // If this is a cross join, there is no filter to evaluate. We can just
    // return the output vector directly. Also don't need to bother about adding
    // mismatched rows.
    if (isCrossJoin()) {
      output_ = getNextCrossProductBatch(
          currentBuild, outputType_, identityProjections_, buildProjections_);
      numOutputRows_ = output_->size();
      probeRowHasMatch_ = true;
      ++buildIndex_;
      filterResultRow_ = 0;
      return false;
    }

    // Handle LeftSemiProjectJoin with no join condition before evaluating the
    // filter
    if (isLeftSemiProjectNoCondition()) {
      handleLeftSemiProjectNoCondition();
      return true;
    }

    // Only re-calculate the filter if we have a new build vector.
    if (filterResultRow_ == 0) {
      evaluateJoinFilter(currentBuild);
    }

    const auto buildRowCount = currentBuild->size();

    // Iterate over the filter results. For each match, add an output record.
    // Use probeOffset and buildIdx