Index: www/sites/all/modules/custom/np_scan/np_scan.archive.inc
===================================================================
--- www/sites/all/modules/custom/np_scan/np_scan.archive.inc	(revision 34194)
+++ www/sites/all/modules/custom/np_scan/np_scan.archive.inc	(working copy)
@@ -47,7 +47,7 @@
           foreach (array('general', 'photo', 'video') as $key) {
             if (isset($result['days'][$key])) {
               // get the index of the last updated bucket
-              $last_update = _fetch_scan_bucket_index('scanurl', $result['updated']->sec, 'days');
+              $last_update = scan_api_bucket_index('scanurl', 'days', $result['updated']->sec);
               $sum = 0;
               foreach ($result['days'][$key] as $index => $value) {
                 // ignore velocity values and keys which have not been updated recently enough
@@ -401,7 +401,7 @@
     if ($cursor->hasNext()) {
       $result = $cursor->getNext();
       $last_update_start = $result['updated']->sec;
-      $last_update_index = _fetch_scan_bucket_index('scan', $result['updated']->sec, 'days');
+      $last_update_index = scan_api_bucket_index('scan', 'days', $result['updated']->sec);
       foreach ($result['days'] as $index => $count) {
         if (!is_int($index) && in_array($index, array('velocity', 'prev_velocity', 'velocity7'))) {
           continue;
Index: www/sites/all/modules/custom/np_scan_stats/np_scan_stats.proto.inc
===================================================================
--- www/sites/all/modules/custom/np_scan_stats/np_scan_stats.proto.inc	(revision 34194)
+++ www/sites/all/modules/custom/np_scan_stats/np_scan_stats.proto.inc	(working copy)
@@ -813,7 +813,7 @@
   $updated = $record['updated']->sec;
   foreach (array('minutes', 'hours', 'days') as $type) {
     if (isset($record[$type])) {
-      $index = _np_scan_stats_fetch_scan_bucket_index($collection, $updated, $type);
+      $index = scan_api_bucket_index($collection, $type, $updated);
       if (isset($record[$type][$index])) {
         $record[$type][$index] .= ' (*' . t('updated') . ')';
       }
@@ -821,7 +821,7 @@
 //      drupal_set_message(t('current %type index of %index missing', array('%type' => $type, '%index' => $index)));
       }
       // Validate that the velocities.
-      $size = _np_scan_stats_fetch_collection_time_bucket_size($collection, $type);
+      $size = scan_api_interval_count($collection, $type);
       $velocity = 0;
       foreach ($record[$type] as $key => $value) {
         if (is_numeric($key)) {
@@ -854,68 +854,6 @@
   return $record;
 }
 
-/**
- * Caculate the correct bin index for scan.{minutes|hours|days}
- * based on scan.updated or current time (default)
- *
- * Using scan.updated tells us the last updated bin index for minutes|hours|days
- *
- * Copied from scan_api.module.
- */
-function _np_scan_stats_fetch_scan_bucket_index($collection, $updated = NULL, $type) {
-  $updated = isset($updated) ? $updated : mktime(date('H') , 0, 0 );
-
-  // see #2113, we need to add 1 to the doy because of a bug in the worker.
-  switch($type) {
-    case 'minutes':
-      $index = intval(date('i', $updated) * _np_scan_stats_fetch_collection_time_bucket_size($collection, $type) / 60 ); // (* 12 / 60 )  Every 5 minutes
-      break;
-    case 'hours':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated) + 1);
-      $index = (date('G', $updated) +  24* $days_since_y2k) % _np_scan_stats_fetch_collection_time_bucket_size($collection, $type);
-      break;
-    case 'days':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated) + 1);
-      $index = $days_since_y2k % _np_scan_stats_fetch_collection_time_bucket_size($collection, $type);
-      break;
-  }
-
-  return $index;
-}
-
-/**
- * Returns size of time bucket size (ex. scan.hours == 48, scan.days == 30, etc)
- * Keep this lookup table in sync with https://apps.d2.nowpublic.com/trac/wiki/Scan/MongoSchema
- *
- * Copied from scan_api.module.
- */
-function _np_scan_stats_fetch_collection_time_bucket_size($collection, $time_type) {
-  $lookup = array(
-    'scan' => array(
-      'minutes' => 12, 'hours'   => 48, 'days'    => 30,
-    ),
-    'hashtag' => array(
-      'minutes' => 12, 'hours'   => 24, 'days'    => 30,
-    ),
-    'keyword' => array(
-      'minutes' => 12, 'hours'   => 24, 'days'    => 30,
-    ),
-    'location' => array(
-      'minutes' => 12, 'hours'   => 24, 'days'    => 30,
-    ),
-    'retweet' => array(
-      'minutes' => 12, 'hours'   => 24, 'days'    => 30,
-    ),
-    'scanurl' => array(
-      'minutes' => 12, 'hours'   => 48, 'days'    => 30,
-    ),
-    'url' => array(
-      'minutes' => 0, 'hours'   => 6, 'days'    => 0,
-    ),
-  );
-  return isset($lookup[$collection], $lookup[$collection][$time_type]) ? $lookup[$collection][$time_type] : 0;
-}
-
 function np_scan_stats_proto_options_form($form_state) {
   $form = array();
   $form['limit'] = array(
Index: www/sites/all/modules/custom/scan_api/scan_api.module
===================================================================
--- www/sites/all/modules/custom/scan_api/scan_api.module	(revision 34194)
+++ www/sites/all/modules/custom/scan_api/scan_api.module	(working copy)
@@ -1304,58 +1304,48 @@
   db_set_active();
   print _scan_api_format($return, $format);
 }
+
 /**
- * Caculate the correct bin index for scan.{minutes|hours|days}
- * based on scan.updated or current time (default)
- *
- * Using scan.updated tells us the last updated bin index for minutes|hours|days
+ * Caculate the correct bin index for a collection at a point in time.
  */
-function _fetch_scan_bucket_index($collection, $updated = NULL, $type) {
-  $updated = isset($updated) ? $updated : mktime(date('H') , 0, 0 );
+function scan_api_bucket_index($collection, $time_type, $when = NULL) {
+  $when = isset($when) ? $when : time();
+  $interval_size_ = scan_api_interval_size($collection, $time_type);
+  $interval_count_ = scan_api_interval_count($collection, $time_type);
+  
+  // This is a direct port of the C++ code.
 
-  // see #2113, we need to add 1 to the doy because of a bug in the worker.
-  switch($type) {
-    case 'minutes':
-      $index = intval(date('i', $updated) * _fetch_collection_time_bucket_size($collection, 'minutes') / 60 ); // (* 12 / 60 )  Every 5 minutes
-      break;
-    case 'hours':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated)) + 1;
-      $index = (date('G', $updated) +  24* $days_since_y2k) % _fetch_collection_time_bucket_size($collection, 'hours');
-      break;
-    case 'days':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated)) + 1;
-      $index = $days_since_y2k % _fetch_collection_time_bucket_size($collection, 'days');
-      break;
-  }
-
-  return $index;
+  // int i = when % (interval_size_ * interval_count_);
+  $i = $when % ($interval_size_ * $interval_count_);
+  // i = floor(i / interval_size_);
+  $i = floor($i / $interval_size_);
+  // Do the integer cast explicitly, as PHP does not have static typing.
+  // In the C++ version, the cast is implicit as i is declared as int.
+  $i = intval($i);
+  return $i;
 }
+
 /**
- * Helper function that returns how much time in seconds would it take to update the whole time type bin
- * ex it would take 12 * 5 * 60 = 60 * 60 = 3600 seconds to go through the scan.minutes bin
- * One use case of this API is to find out whether a bin have been stale for more than a full cycle
- * which leads to zeroing out the whole bin. See top of scan_api_reorder_scan_time_buckets()
+ * Get interval length (in seconds).
  */
-function _bucket_size_to_time($collection, $time_type) {
-  $time = 0;
-  switch($time_type) {
-    case 'minutes':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 5 * 60;
-      break;
-    case 'hours':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 60 * 60;
-      break;
-    case 'days':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 60 * 60 * 24;
-      break;
+function scan_api_interval_size($collection, $time_type) {
+  if ($time_type == 'minutes') {
+    return 5; // 60 seconds / 12 divisions.
   }
-  return $time;
+  elseif ($time_type == 'days') {
+    return 86400; // 60 (seconds) * 60 (minutes) * 24 (hours)
+  }
+  elseif ($time_type == 'hours') {
+    return 3600; // 60 (seconds) * 60 (minutes)
+  }
+  assert("Invalid time_type!");
 }
+
 /**
- * Returns size of time bucket size (ex. scan.hours == 48, scan.days == 30, etc)
- * Keep this lookup table in sync with https://apps.d2.nowpublic.com/trac/wiki/Scan/MongoSchema
+ * Get number of intervals in interval array.
+ * This is based on https://apps.d2.nowpublic.com/trac/wiki/Scan/MongoSchema
  */
-function _fetch_collection_time_bucket_size($collection, $time_type) {
+function scan_api_interval_count($collection, $time_type) {
   $lookup = array(
     'scan' => array(
       'minutes' => 12, 'hours'   => 48, 'days'    => 30,
@@ -1376,10 +1366,13 @@
       'minutes' => 12, 'hours'   => 48, 'days'    => 30,
     ),
     'url' => array(
-      'minutes' => 0, 'hours'   => 6, 'days'    => 0,
+      'hours'   => 6,
     ),
   );
-  return isset($lookup[$collection], $lookup[$collection][$time_type]) ? $lookup[$collection][$time_type] : 0;
+  if (isset($lookup[$collection], $lookup[$collection][$time_type])) {
+    return $lookup[$collection][$time_type];
+  }
+  assert("Count entry missing from interval count table!");
 }
 
 /**
@@ -1388,22 +1381,23 @@
  * which makes it not easy to manage. here's API for that.
  */
 function scan_api_reorder_scan_time_buckets($bucket, $collection, $time_type, $last_updated = NULL) {
+  $bucket_cycle_time = scan_api_interval_size($collection, $time_type) * scan_api_interval_count($collection, $time_type);
   // bucket have been outdated for more than one full cycle, flush it completely
-  if ( time() > (_bucket_size_to_time($collection, $time_type) + $last_updated) ) {
-    $size = _fetch_collection_time_bucket_size($collection, $time_type);
+  if ( time() > ($bucket_cycle_time + $last_updated) ) {
+    $size = scan_api_interval_count($collection, $time_type);
     $bucket = array_fill(0, $size, 0 ) ;
   }
   else {
   // the following 6 lines normalize our bucket since sometimes buckets may have gaps making behavior unpredictable
-    $size = _fetch_collection_time_bucket_size($collection, $time_type);
+    $size = scan_api_interval_count($collection, $time_type);
     $new_bucket = array_fill(0, $size, 0 );
     foreach($bucket as $index => $value ) {
       $new_bucket[$index] = $value;
     }
     $bucket = $new_bucket;
     // Flatten and flush stale bins for buckets that have been stale for less than full cycle
-    $current_index = _fetch_scan_bucket_index($collection, null, $time_type);
-    $last_updated_index = _fetch_scan_bucket_index($collection, $last_updated, $time_type);
+    $current_index = scan_api_bucket_index($collection, $time_type);
+    $last_updated_index = scan_api_bucket_index($collection, $time_type, $last_updated);
     // Remove non numerical indexes, leaving only time buckets
     foreach($bucket as $index => $value) {
       if (!is_numeric($index)) {
Index: view/sites/all/modules/scan_api/scan_api.module
===================================================================
--- view/sites/all/modules/scan_api/scan_api.module	(revision 34194)
+++ view/sites/all/modules/scan_api/scan_api.module	(working copy)
@@ -1365,57 +1365,48 @@
   db_set_active();
   print _scan_api_format($return, $format);
 }
+
 /**
- * Caculate the correct bin index for scan.{minutes|hours|days}
- * based on scan.updated or current time (default)
- *
- * Using scan.updated tells us the last updated bin index for minutes|hours|days
+ * Caculate the correct bin index for a collection at a point in time.
  */
-function _fetch_scan_bucket_index($collection, $updated = NULL, $type) {
-  $updated = isset($updated) ? $updated : mktime(date('H') , 0, 0 );
+function scan_api_bucket_index($collection, $time_type, $when = NULL) {
+  $when = isset($when) ? $when : time();
+  $interval_size_ = scan_api_interval_size($collection, $time_type);
+  $interval_count_ = scan_api_interval_count($collection, $time_type);
+  
+  // This is a direct port of the C++ code.
 
-  switch($type) {
-    case 'minutes':
-      $index = intval(date('i', $updated) * _fetch_collection_time_bucket_size($collection, 'minutes') / 60 ); // (* 12 / 60 )  Every 5 minutes
-      break;
-    case 'hours':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated) - 1);
-      $index = (date('G', $updated) +  24* $days_since_y2k) % _fetch_collection_time_bucket_size($collection, 'hours');
-      break;
-    case 'days':
-      $days_since_y2k = intval(((date('Y', $updated) - 2000) * 365.2475) + date('z', $updated) - 1);
-      $index = $days_since_y2k % _fetch_collection_time_bucket_size($collection, 'days');
-      break;
-  }
-
-  return $index;
+  // int i = when % (interval_size_ * interval_count_);
+  $i = $when % ($interval_size_ * $interval_count_);
+  // i = floor(i / interval_size_);
+  $i = floor($i / $interval_size_);
+  // Do the integer cast explicitly, as PHP does not have static typing.
+  // In the C++ version, the cast is implicit as i is declared as int.
+  $i = intval($i);
+  return $i;
 }
+
 /**
- * Helper function that returns how much time in seconds would it take to update the whole time type bin
- * ex it would take 12 * 5 * 60 = 60 * 60 = 3600 seconds to go through the scan.minutes bin
- * One use case of this API is to find out whether a bin have been stale for more than a full cycle
- * which leads to zeroing out the whole bin. See top of scan_api_reorder_scan_time_buckets()
+ * Get interval length (in seconds).
  */
-function _bucket_size_to_time($collection, $time_type) {
-  $time = 0;
-  switch($time_type) {
-    case 'minutes':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 5 * 60;
-      break;
-    case 'hours':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 60 * 60;
-      break;
-    case 'days':
-      $time = _fetch_collection_time_bucket_size($collection, $time_type) * 60 * 60 * 24;
-      break;
+function scan_api_interval_size($collection, $time_type) {
+  if ($time_type == 'minutes') {
+    return 5; // 60 seconds / 12 divisions.
   }
-  return $time;
+  elseif ($time_type == 'days') {
+    return 86400; // 60 (seconds) * 60 (minutes) * 24 (hours)
+  }
+  elseif ($time_type == 'hours') {
+    return 3600; // 60 (seconds) * 60 (minutes)
+  }
+  assert("Invalid time_type!");
 }
+
 /**
- * Returns size of time bucket size (ex. scan.hours == 48, scan.days == 30, etc)
- * Keep this lookup table in sync with https://apps.d2.nowpublic.com/trac/wiki/Scan/MongoSchema
+ * Get number of intervals in interval array.
+ * This is based on https://apps.d2.nowpublic.com/trac/wiki/Scan/MongoSchema
  */
-function _fetch_collection_time_bucket_size($collection, $time_type) {
+function scan_api_interval_count($collection, $time_type) {
   $lookup = array(
     'scan' => array(
       'minutes' => 12, 'hours'   => 48, 'days'    => 30,
@@ -1436,10 +1427,13 @@
       'minutes' => 12, 'hours'   => 48, 'days'    => 30,
     ),
     'url' => array(
-      'minutes' => 0, 'hours'   => 6, 'days'    => 0,
+      'hours'   => 6,
     ),
   );
-  return isset($lookup[$collection], $lookup[$collection][$time_type]) ? $lookup[$collection][$time_type] : 0;
+  if (isset($lookup[$collection], $lookup[$collection][$time_type])) {
+    return $lookup[$collection][$time_type];
+  }
+  assert("Count entry missing from interval count table!");
 }
 
 /**
@@ -1448,22 +1442,23 @@
  * which makes it not easy to manage. here's API for that.
  */
 function scan_api_reorder_scan_time_buckets($bucket, $collection, $time_type, $last_updated = NULL) {
+  $bucket_cycle_time = scan_api_interval_size($collection, $time_type) * scan_api_interval_count($collection, $time_type);
   // bucket have been outdated for more than one full cycle, flush it completely
-  if ( time() > (_bucket_size_to_time($collection, $time_type) + $last_updated) ) {
-    $size = _fetch_collection_time_bucket_size($collection, $time_type);
+  if ( time() > ($bucket_cycle_time + $last_updated) ) {
+    $size = scan_api_interval_count($collection, $time_type);
     $bucket = array_fill(0, $size, 0 ) ;
   }
   else {
   // the following 6 lines normalize our bucket since sometimes buckets may have gaps making behavior unpredictable
-    $size = _fetch_collection_time_bucket_size($collection, $time_type);
+    $size = scan_api_interval_count($collection, $time_type);
     $new_bucket = array_fill(0, $size, 0 );
     foreach($bucket as $index => $value ) {
       $new_bucket[$index] = $value;
     }
     $bucket = $new_bucket;
     // Flatten and flush stale bins for buckets that have been stale for less than full cycle
-    $current_index = _fetch_scan_bucket_index($collection, null, $time_type);
-    $last_updated_index = _fetch_scan_bucket_index($collection, $last_updated, $time_type);
+    $current_index = scan_api_bucket_index($collection, $time_type);
+    $last_updated_index = scan_api_bucket_index($collection, $time_type, $last_updated);
     // Remove non numerical indexes, leaving only time buckets
     foreach($bucket as $index => $value) {
       if (!is_numeric($index)) {