function compareComplexMulti($a1, $a2)
{
$intersection = array_intersect($a1, $a2);
// similarity of names, no matter of name order
$weightIdentical = count($intersection);
$weightTotal = max(count($a1), count($a2));
$weightSimilar = 0;
$a1Available = array_diff($a1, $intersection);
$weightSimilarTotal = count($a1Available);
$a2Available = array_diff($a2, $intersection);
while (count($a1Available)) {
$v1 = array_shift($a1Available);
// build comparison array
$comparison = array();
$exactMatch = false;
foreach ($a2Available as $v2) {
if ($v1 == $v2) {
$exactMatch = true;
break;
}
}
if ($exactMatch) {
$comparison[] = 1;
} else {
foreach ($a2Available as $v2) {
// calculate similarity based on weights
$comparison[] = compareComplex($v1, $v2);
}
}
// get maximal match
arsort($comparison);
$index = array_shift(array_keys($comparison));
$weightSimilar += $comparison[$index] / 100;
unset($a2Available[$index]);
}
unset($a1Available);
unset($a2Available);
unset($comparison);
unset($intersection);
#echo "$weightIdentical - $weightSimilar - $weightTotal<br>";
$similarity = ($weightIdentical + $weightSimilar) / $weightTotal * 100;
return $similarity;
}
function compareTwoContacts($c1, $c2)
{
global $thresholdUsernameSimilarity, $thresholdNameSimilarity;
logMsg('DEBUG', "Comparing " . $c1['email'] . " with " . $c2['email']);
// merge primary name with secondary names
$c1Names = $c1['secondaryNames'];
$c2Names = $c2['secondaryNames'];
if ($c1['name']) {
$c1Names[] = $c1['name'];
}
if ($c2['name']) {
$c2Names[] = $c2['name'];
}
// if we have names
if (count($c1Names) && count($c2Names)) {
// clean names
array_walk($c1Names, 'cleanName');
array_walk($c2Names, 'cleanName');
// check for exact items
$commonNames = array_intersect($c1Names, $c2Names);
}
$c1Username = $c1['usernames'][0];
cleanName($c1Username);
$c2Username = $c2['usernames'][0];
cleanName($c2Username);
// check similarity
if (!empty($commonNames)) {
// full names match
return 1;
} elseif ($c1Username && $c2Username && ($usernameSimilarity = compareComplex($c1Username, $c2Username)) && $usernameSimilarity > $thresholdUsernameSimilarity) {
// very high username similarity
logMsg('DEBUG', "Username similarity between " . $c1Username . " and " . $c2Username . " is {$usernameSimilarity}");
return 1;
} elseif (count($c1Names) && count($c2Names)) {
// check names similarity
$nameSimilarity = 0;
foreach ($c1Names as $c1Name) {
if (!$c1Name) {
continue;
}
$c1NamesSplit = explode(' ', $c1Name);
$nameSimilarity = 0;
foreach ($c2Names as $c2Name) {
if (!$c2Name) {
continue;
}
$c2NamesSplit = explode(' ', $c2Name);
$similarity = compareComplexMulti($c1NamesSplit, $c2NamesSplit);
if ($similarity > $nameSimilarity) {
logMsg('DEBUG', "Name similarity between " . $c1Name . " and " . $c2Name . " is {$similarity}");
$nameSimilarity = $similarity;
}
}
}
unset($c1Names);
unset($c2Names);
unset($c1NamesSplit);
unset($c2NamesSplit);
return $nameSimilarity > $thresholdUsernameSimilarity ? 1 : 0;
} else {
unset($c1Names);
unset($c2Names);
return 0;
}
}