function cluster($strings)
{
$result = new stdclass();
$n = count($strings);
// clean
for ($i = 0; $i < $n; $i++) {
$strings[$i] = finger_print(trim($strings[$i]));
}
$map = array();
$inv_map = array();
$count = 0;
foreach ($strings as $k => $v) {
$map[$k] = $count;
$inv_map[$count] = $k;
$count++;
}
// Create adjacency matrix and fill with 0's
$X = array();
for ($i = 0; $i < $n; $i++) {
$X[$i] = array();
for ($j = 0; $j < $n; $j++) {
$X[$i][$j] = 0;
}
}
$nodes = '';
$edges = '';
// Compare names using approximate string matching
$i = 0;
foreach ($strings as $k1 => $v1) {
$nodes .= "node {$i} [label=\"" . $v1 . "\"];\n";
if ($i < $n - 1) {
$j = 0;
foreach ($strings as $k2 => $v2) {
if ($j > $i && $j < $n) {
// Find longest common subsequence for this pair of cleaned names
$lcs = new LongestCommonSequence($v1, $v2);
$d = $lcs->score();
// Filter by longest common substring (to ensure we have a "meaningful"
// match), that is, so that we avoid subsequences that have little continuity
$str = '';
$lcstr = LongestCommonSubstring($v1, $v2, $str);
if ($lcstr >= 4) {
// Ignore matches just on date, we want more than that
if (is_numeric(trim($str))) {
} else {
// If longest common subsequence is > 70% of the length of both strings
// we accept it.
if ($d / strlen($v1) >= 0.7 || $d / strlen($v2) >= 0.7) {
$X[$map[$k1]][$map[$k2]] = 1;
$X[$map[$k2]][$map[$k1]] = 1;
$edges .= $i . " -- " . $j . " [label=\"" . $lcstr . "\"];\n";
}
}
} else {
// If just a short match is it the start if the string (e.g., an abbreviation)
$abbreviation = false;
if (strlen($v1) == $d) {
if (strpos($v2, $v1, 0) === false) {
} else {
$abbreviation = true;
}
} else {
if (strpos($v1, $v2, 0) === false) {
} else {
$abbreviation = true;
}
}
// Accept abbreviation
if ($abbreviation) {
$X[$map[$k1]][$map[$k2]] = 1;
$X[$map[$k2]][$map[$k1]] = 1;
$edges .= $i . " -- " . $j . " [label=\"" . $lcstr . "\"];\n";
}
}
}
$j++;
}
$i++;
}
}
$result->graph = "graph {\n" . $nodes . $edges . "}\n";
//echo $graph;
// Get components of adjacency matrix
$components = get_components($X);
$result->clusters = array();
foreach ($components as $component) {
$cluster = array();
foreach ($component as $k => $v) {
$member = new stdclass();
$member->id = $inv_map[$v];
$member->string = $strings[$inv_map[$v]];
$cluster[] = $member;
}
$result->clusters[] = $cluster;
}
//print_r($map);
if (0) {
print_r($c);
}
return $result;
}
echo get_projects($_GET);
} else {
if (strpos($uri, '/companies') !== false) {
echo get_company($_GET);
} else {
if (strpos($uri, '/company_reviews') !== false) {
echo get_company_review($_GET);
} else {
if (strpos($uri, '/add_review_company') !== false) {
echo add_company_review($_POST);
} else {
if (strpos($uri, '/get_account') !== false) {
echo get_account($_GET);
} else {
if (strpos($uri, '/components') !== false) {
echo get_components($_GET);
} else {
if (strpos($uri, '/documents') !== false) {
echo get_documents($_GET);
} else {
if (strpos($uri, '/products') !== false) {
echo get_products($_GET);
} else {
if (strpos($uri, '/add_to_project') !== false) {
echo add_to_project($_POST);
} else {
if (strpos($uri, '/remove_from_project') !== false) {
echo remove_from_project($_POST);
} else {
if (strpos($uri, '/uploadServiceImage') !== false) {
echo upload_service_image($_POST, $_FILES);
$component_b = get_component_b($id);
$family = get_families($id);
// components A
$comp_a_data = '';
if ($component_a) {
foreach ($component_a->result() as $gaji) {
$component = get_components($gaji->gaji_component_id);
$comp_a_data .= '["' . $gaji->gaji_component_id . '", "' . $component->comp_name . '", "' . $component->comp_type . '", "' . $gaji->gaji_daily_value . '", "' . number_format($gaji->gaji_amount_value, 2, ".", ",") . '"],';
}
$comp_a_data = substr($comp_a_data, 0, strlen($comp_a_data) - 1);
}
// components B
$comp_b_data = '';
if ($component_b) {
foreach ($component_b->result() as $gaji) {
$component = get_components($gaji->gaji_component_id);
$comp_b_data .= '["' . $gaji->gaji_component_id . '", "' . $component->comp_name . '", "' . $component->comp_type . '", "' . $gaji->gaji_daily_value . '", "' . number_format($gaji->gaji_amount_value, 2, ".", ",") . '"],';
}
$comp_b_data = substr($comp_b_data, 0, strlen($comp_b_data) - 1);
}
// family
$families = '';
if ($family) {
foreach ($family->result() as $f) {
$families .= '["' . $f->staff_fam_order . '", "' . $f->staff_fam_name . '", "' . $f->staff_fam_birthdate . '", "' . $f->staff_fam_birthplace . '", "' . $f->staff_fam_sex . '", "' . $f->staff_fam_relation . '"],';
}
$families = substr($families, 0, strlen($families) - 1);
}
?>
<script type="text/javascript">
var comp_a_data = [<?php
