Stats: Add self-sponsored hours and contributor counts.

Fixes #124

Some variables/keys to clarify their names, or makes them less verbose. See #198, #200.

plugins/wporg-5ftf/includes/contributor.php

@@ -367,6 +367,8 @@ function get_contributor_user_ids( $contributor_posts ) {
 		$wpdb->prepare( $query, $usernames )
 	);
 
+	$user_ids = array_map( 'absint', $user_ids );
+
 	return $user_ids;
 }
 

plugins/wporg-5ftf/includes/stats.php

@@ -58,47 +58,64 @@ function schedule_cron_jobs() {
 
 /**
  * Record a snapshot of the current stats, so we can track trends over time.
+ *
+ * "Self-sponsored" contributors are those that volunteer their time rather than being paid by a company.
  */
 function record_snapshot() {
 	$stats = get_snapshot_data();
 
-	bump_stats_extra( 'five-for-the-future', 'Company-sponsored hours', $stats['confirmed_company_hours'] );
+	bump_stats_extra( 'five-for-the-future', 'Self-sponsored hours', $stats['self_sponsored_hours'] );
-	bump_stats_extra( 'five-for-the-future', 'Company-sponsored contributors', $stats['confirmed_company_contributors'] );
+	bump_stats_extra( 'five-for-the-future', 'Self-sponsored contributors', $stats['self_sponsored_contributors'] );
-	bump_stats_extra( 'five-for-the-future', 'Companies', $stats['confirmed_companies'] );
+	bump_stats_extra( 'five-for-the-future', 'Companies', $stats['companies'] );
+	bump_stats_extra( 'five-for-the-future', 'Company-sponsored hours', $stats['company_sponsored_hours'] );
+	bump_stats_extra( 'five-for-the-future', 'Company-sponsored contributors', $stats['company_sponsored_contributors'] );
 
-	foreach ( $stats['confirmed_team_company_contributors'] as $team => $contributors ) {
+	foreach ( array( 'team_company_sponsored_contributors', 'team_self_sponsored_contributors' ) as $key ) {
+		foreach ( $stats[ $key ] as $team => $contributors ) {
 			// The labels are listed alphabetically in MC, so starting them all with "Team" groups them together and
 			// makes the interface easier to use.
-		$grouped_name = sprintf( 'Team %s company-sponsored contributors', str_replace( ' Team', '', $team ) );
+			$grouped_name = sprintf(
+				'Team %s %s-sponsored contributors',
+				str_replace( ' Team', '', $team ),
+				str_contains( $key, 'self' ) ? 'self' : 'company'
+			);
+
 			bump_stats_extra( 'five-for-the-future', $grouped_name, $contributors );
 		}
 	}
+}
 
 /**
  * Calculate the stats for the current snapshot.
  *
+ * This will be processing a large amount of data, so `unset()` is used throughout the function on variables that
+ * are no longer needed. That should help to avoid out-of-memory errors.
+ *
  * @return array
  */
 function get_snapshot_data() {
 	$snapshot_data = array(
-		'confirmed_company_hours'             => 0,
+		'company_sponsored_hours'             => 0,
-		'confirmed_team_company_contributors' => array(),
+		'self_sponsored_hours'                => 0,
+		'team_company_sponsored_contributors' => array(),
+		'team_self_sponsored_contributors'    => array(),
 	);
 
-	$confirmed_companies = new WP_Query( array(
+	$companies = new WP_Query( array(
 		'post_type'   => Pledge\CPT_ID,
 		'post_status' => 'publish',
 		'numberposts' => 1, // We only need `found_posts`, not the posts themselves.
 	) );
 
-	$snapshot_data['confirmed_companies'] = $confirmed_companies->found_posts;
+	$snapshot_data['companies'] = $companies->found_posts;
+	unset( $companies );
 
 	/*
 	 * A potential future optimization would be make WP_Query only return the `post_title`. The `fields` parameter
 	 * doesn't currently support `post_title`, but it may be possible with filters like `posts_fields`
 	 * or `posts_fields_request`. That was premature at the time this code was written, though.
 	 */
-	$confirmed_company_contributors = get_posts( array(
+	$company_sponsored_contributors = get_posts( array(
 		'post_type'   => Contributor\CPT_ID,
 		'post_status' => 'publish',
 		'numberposts' => -1,
@@ -112,37 +129,35 @@ function get_snapshot_data() {
 	 * but `WP_Query` doesn't support `DISTINCT` directly, and it's premature at this point. It may be possible
 	 * with the filters mentioned above.
 	 */
-	$confirmed_user_ids                              = array_unique( Contributor\get_contributor_user_ids( $confirmed_company_contributors ) );
+	$company_contributor_user_ids = array_unique( Contributor\get_contributor_user_ids( $company_sponsored_contributors ) );
-	$snapshot_data['confirmed_company_contributors'] = count( $confirmed_user_ids );
+	unset( $company_sponsored_contributors );
-	$company_contributors_profile_data               = XProfile\get_xprofile_contribution_data( $confirmed_user_ids );
 
-	foreach ( $company_contributors_profile_data as $profile_data ) {
+	$all_contributor_profiles                        = XProfile\get_all_xprofile_contributor_hours_teams();
-		switch ( (int) $profile_data['field_id'] ) {
+	$snapshot_data['company_sponsored_contributors'] = count( $company_contributor_user_ids );
-			case XProfile\FIELD_IDS['hours_per_week']:
+	$snapshot_data['self_sponsored_contributors']    = count( $all_contributor_profiles ) - count( $company_contributor_user_ids );
-				$snapshot_data['confirmed_company_hours'] += absint( $profile_data['value'] );
-				break;
 
-			case XProfile\FIELD_IDS['team_names']:
+	foreach ( $all_contributor_profiles as $profile ) {
-				/*
+		$attribution_prefix = in_array( $profile->user_id, $company_contributor_user_ids, true )
-				 * BuddyPress validates the team name(s) the user provides before saving them in the database, so
+			? 'company_sponsored'
-				 * it should be safe to unserialize, and to assume that they're valid.
+			: 'self_sponsored';
-				 *
-				 * The database stores team _names_ rather than _IDs_, though, so if a team is ever renamed, this
-				 * data will be distorted.
-				 */
-				$associated_teams = (array) maybe_unserialize( $profile_data['value'] );
 
-				foreach ( $associated_teams as $team ) {
+		$team_contributor_key = sprintf( 'team_%s_contributors', $attribution_prefix );
-					if ( isset( $snapshot_data['confirmed_team_company_contributors'][ $team ] ) ) {
+
-						$snapshot_data['confirmed_team_company_contributors'][ $team ]++;
+		$snapshot_data[ $attribution_prefix . '_hours'] += $profile->hours_per_week;
+
+		foreach ( $profile->team_names as $team ) {
+			if ( isset( $snapshot_data[ $team_contributor_key ][ $team ] ) ) {
+				$snapshot_data[ $team_contributor_key ][ $team ] ++;
 			} else {
-						$snapshot_data['confirmed_team_company_contributors'][ $team ] = 1;
+				$snapshot_data[ $team_contributor_key ][ $team ] = 1;
 			}
 		}
+	}
+	unset( $all_contributor_profiles );
 
-				break;
+	// Alphabetize so that they appear in a consistent order in the MC interface.
-		}
+	ksort( $snapshot_data['team_company_sponsored_contributors'] );
-	}
+	ksort( $snapshot_data['team_self_sponsored_contributors'] );
 
 	return $snapshot_data;
 }
@@ -150,6 +165,9 @@ function get_snapshot_data() {
 /**
  * Render the shortcode to display stats.
  *
+ * @deprecated Stats were originally kept in these posts, but are currently stored in MC. This is kept so that we
+ * have a historical record.
+ *
  * @return string
  */
 function render_shortcode() {

plugins/wporg-5ftf/includes/xprofile.php

@@ -2,7 +2,7 @@
 namespace WordPressDotOrg\FiveForTheFuture\XProfile;
 
 use WordPressDotOrg\FiveForTheFuture\Contributor;
-use wpdb;
+use WPDB;
 
 /*
  * The IDs of the xprofile fields we need. Better to use the numerical IDs than the field labels,
@@ -16,13 +16,66 @@ const FIELD_IDS = array(
 
 defined( 'WPINC' ) || die();
 
+/**
+ * Get the xprofile `hours_per_week` and `team_names` for all contributors, regardless of sponsorship status.
+ *
+ * The "Sponsored" field is not retrieved because it's usually not needed, and including it would significantly
+ * hurt performance.
+ */
+function get_all_xprofile_contributor_hours_teams() : array {
+	global $wpdb;
+
+	// This might need a `LIMIT` in the future as more users save values, but it's performant as of August 2022.
+	// `LIMIT`ing it would require batch processing, which would add a significant amount of complexity.
+	// A better alternative might be to add a cron job to delete rows from `bpmain_bp_xprofile_data` where
+	// `hours_per_week` is < 1, or `teams_names` is a (serialized) empty array. BuddyPress saves those as
+	// values rather than deleting them, and that significantly increases the number of rows returned.
+	$users = $wpdb->get_results( $wpdb->prepare( '
+		SELECT user_id, GROUP_CONCAT( field_id ) AS field_ids, GROUP_CONCAT( value ) AS field_values
+		FROM `bpmain_bp_xprofile_data`
+		WHERE field_id IN ( %d, %d )
+		GROUP BY user_id',
+		FIELD_IDS['hours_per_week'],
+		FIELD_IDS['team_names']
+	) );
+
+	$field_names = array_flip( FIELD_IDS );
+
+	foreach ( $users as $user_index => & $user ) {
+		$fields = explode( ',', $user->field_ids );
+		$values = explode( ',', $user->field_values );
+
+		foreach ( $fields as $field_index => $id ) {
+			/*
+			 * BuddyPress validates the team name(s) the user provides before saving them in the database, so
+			 * it should be safe to unserialize, and to assume that they're valid.
+			 *
+			 * The database stores team _names_ rather than _IDs_, though, so if a team is ever renamed, this
+			 * data will be distorted.
+			 */
+			$user->{$field_names[ $id ]} = maybe_unserialize( $values[ $field_index ] );
+		}
+		unset( $user->field_ids, $user->field_values ); // Remove the concatenated data now that it's exploded.
+
+		$user->user_id        = absint( $user->user_id );
+		$user->hours_per_week = absint( $user->hours_per_week ?? 0 );
+		$user->team_names     = (array) $user->team_names ?? array();
+
+		if ( 0 >= $user->hours_per_week || empty( $user->team_names ) ) {
+			unset( $users[ $user_index ] );
+		}
+	}
+
+	return $users;
+}
+
 /**
  * Pull relevant data from profiles.wordpress.org.
  *
  * Note that this does not unserialize anything, it just pulls the raw values from the database table. If you
  * want unserialized data, use `prepare_xprofile_contribution_data()`.
  *
- * @global wpdb $wpdb
+ * @global WPDB $wpdb
  *
  * @param array $user_ids
  *