merge-sort-sql-results

<?php
#
# Input multiple SELECT statements from different SQL queries
# get back a single stream of rows with records merged if
# they have the same key
#

function genOneTwoThree() {
  foreach ([1=>'one',2=>'two',3=>'three'] as $key => $value) {
#    echo "...ready to yield: $key => $number\n";
    yield $key => $value;
    
  }
}

function genTwoThreeFour() {
  foreach ([1=>'uno',2=>'does',4=>'quatro'] as $key => $value) {
#    echo "...ready to yield: $key => $number\n";
    yield $key => $value;
    
  }
}

$emptyIterator = new EmptyIterator();

/**
 * mergeGenerators Combines multiple generators and has unique keys
 * I use it to merge sort queries from multiple databases and
 * combine records that have duplicate keys across them
 *
 * @access public
 * @param array $generators Yields key => some object and keys sorted
 * @param callable $merger Takes arroy of 2 or more objects and returns one
 * @return void
 */
function mergeGenerators(array $generators, callable $merger) {
  while (count($generators) > 0) {
    // Find minimum input key
    $minimumKey = null;
    foreach ($generators as $i => $generator) {
      if (!$generator->valid()) {
        unset($generators[$i]);
        continue;
      }
      if (is_null($minimumKey) || $generator->key() < $minimumKey) {
        $minimumKey = $generator->key();
      }
    }
    if (count($generators) == 0) {
      return;
    }

    // Merge all minimum values
    $outputItems = array();
    foreach ($generators as $generator) {
      while ($generator->key() == $minimumKey) {
        $outputItems[] = $generator->current();
        $generator->next();
      }
    }
    yield $minimumKey => count($outputItems) > 1 ? $merger($outputItems) : $outputItems[0];
  }
}

function combiner($items) {
  return max($items);
}

$a = mergeGenerators([genOneTwoThree(), $emptyIterator, genTwoThreeFour()], 'combiner');

echo "\n\nYIELDED:\n";
foreach ($a as $key => $value) {
  var_dump($key, $value);
}

Something like this also works

    // Merge all minimum values
    $outputItems = array();
    foreach ($generators as $generator) {
      while ($generator->key() == $minimumKey) {
        $outputItems[] = $generator->current();
      }
    }
    yield $minimumKey => count($outputItems) > 1 ? $merger($outputItems) : $outputItems[0];

Regarding ^^ https://gist.github.com/fulldecent/a4cac03087cadec4c275a54317a6f623#gistcomment-1949315, the latest version now includes this approach

Cleaner function

  /**
   * Combines multiple generators and has unique keys
   * I use it to merge sort queries from multiple databases and
   * combine records that have duplicate keys across them
   *
   * http://stackoverflow.com/q/37973732/300224
   *
   * @access public
   * @param array $generators Yields key => some object and keys sorted by keyComparitor
   * @param callable $keyComparitor Puts keys in order, returns <0, 0, >0
   * @param callable $merger Takes arroy of 2 or more objects and returns one
   * @return void
   */
  static function mergeGenerators(array $generators, callable $keyComparitor, callable $merger)
  {
    while (count($generators) > 0) {
      $generators = array_filter($generators, function(Iterator $g){return $g->valid();});
      if (empty($generators)) {
        return;
      }
      usort($generators, function(Iterator $a, Iterator $b) use ($keyComparitor) {
        return $keyComparitor($a->key(), $b->key());
      });
      $minimumKey = $generators[0]->key();
      $outputItems = [];
      foreach ($generators as $generator) {
        if ($keyComparitor($generator->key(), $minimumKey) !== 0) break; // Sorted, later ones must also not be same
        while ($keyComparitor($generator->key(), $minimumKey) === 0) {
          $outputItems[] = $generator->current();
          $generator->next();
        }
      }
      yield $minimumKey => count($outputItems) > 1 ? $merger($outputItems) : $outputItems[0];
    }
  }