Fix incorrect analytics count. Closes #712.

cmd/init.go

@@ -283,19 +283,25 @@ func readQueries(sqlFile string, db *sqlx.DB, fs stuffbin.FileSystem) goyesql.Qu
 
 // prepareQueries queries prepares a query map and returns a *Queries
 func prepareQueries(qMap goyesql.Queries, db *sqlx.DB, ko *koanf.Koanf) *models.Queries {
-	// The campaign view/click count queries have a COUNT(%s) placeholder that should either
+	var (
-	// be substituted with * to pull non-unique rows when individual subscriber tracking is off
+		countQuery = "get-campaign-analytics-counts"
-	// as all subscriber_ids will be null, or with DISTINCT subscriber_id when tracking is on
+		linkSel    = "*"
-	// to only pull unique rows per subscriber.
+	)
-	sel := "*"
 	if ko.Bool("privacy.individual_tracking") {
-		sel = "DISTINCT subscriber_id"
+		countQuery = "get-campaign-analytics-unique-counts"
+		linkSel = "DISTINCT subscriber_id"
 	}
 
-	keys := []string{"get-campaign-view-counts", "get-campaign-click-counts", "get-campaign-link-counts"}
+	// These don't exist in the SQL file but are in the queries struct to be prepared.
-	for _, k := range keys {
+	qMap["get-campaign-view-counts"] = &goyesql.Query{
-		qMap[k].Query = fmt.Sprintf(qMap[k].Query, sel)
+		Query: fmt.Sprintf(qMap[countQuery].Query, "campaign_views"),
+		Tags:  map[string]string{"name": "get-campaign-view-counts"},
 	}
+	qMap["get-campaign-click-counts"] = &goyesql.Query{
+		Query: fmt.Sprintf(qMap[countQuery].Query, "link_clicks"),
+		Tags:  map[string]string{"name": "get-campaign-click-counts"},
+	}
+	qMap["get-campaign-link-counts"].Query = fmt.Sprintf(qMap["get-campaign-link-counts"].Query, linkSel)
 
 	// Scan and prepare all queries.
 	var q models.Queries

models/queries.go

@@ -56,10 +56,16 @@ type Queries struct {
 	GetCampaignForPreview *sqlx.Stmt `query:"get-campaign-for-preview"`
 	GetCampaignStats      *sqlx.Stmt `query:"get-campaign-stats"`
 	GetCampaignStatus     *sqlx.Stmt `query:"get-campaign-status"`
+
+	// These two queries are read as strings and based on settings.individual_tracking=on/off,
+	// are interpolated and copied to view and click counts. Same query, different tables.
+	GetCampaignAnalyticsCounts       string     `query:"get-campaign-analytics-counts"`
+	GetCampaignAnalyticsCountsUnique string     `query:"get-campaign-analytics-unique-counts"`
 	GetCampaignViewCounts            *sqlx.Stmt `query:"get-campaign-view-counts"`
 	GetCampaignClickCounts           *sqlx.Stmt `query:"get-campaign-click-counts"`
 	GetCampaignLinkCounts            *sqlx.Stmt `query:"get-campaign-link-counts"`
 	GetCampaignBounceCounts          *sqlx.Stmt `query:"get-campaign-bounce-counts"`
+
 	NextCampaigns            *sqlx.Stmt `query:"next-campaigns"`
 	NextCampaignSubscribers  *sqlx.Stmt `query:"next-campaign-subscribers"`
 	GetOneCampaignSubscriber *sqlx.Stmt `query:"get-one-campaign-subscriber"`

queries.sql

@@ -569,27 +569,28 @@ u AS (
 )
 SELECT * FROM camps;
 
--- name: get-campaign-view-counts
+-- name: get-campaign-analytics-unique-counts
--- raw: true
--- %s = * or DISTINCT subscriber_id (prepared based on based on individual tracking=on/off). Prepared on boot.
 WITH intval AS (
     -- For intervals < a week, aggregate counts hourly, otherwise daily.
     SELECT CASE WHEN (EXTRACT (EPOCH FROM ($3::TIMESTAMP - $2::TIMESTAMP)) / 86400) >= 7 THEN 'day' ELSE 'hour' END
-)
+),
-SELECT campaign_id, COUNT(%s) AS "count", DATE_TRUNC((SELECT * FROM intval), created_at) AS "timestamp"
+uniqIDs AS (
-    FROM campaign_views
+    SELECT DISTINCT ON(subscriber_id) subscriber_id, campaign_id, DATE_TRUNC((SELECT * FROM intval), created_at) AS "timestamp"
+    FROM %s
     WHERE campaign_id=ANY($1) AND created_at >= $2 AND created_at <= $3
-    GROUP BY campaign_id, "timestamp" ORDER BY "timestamp" ASC;
+    ORDER BY subscriber_id, "timestamp"
+)
+SELECT COUNT(*) AS "count", campaign_id, "timestamp"
+    FROM uniqIDs GROUP BY campaign_id, "timestamp";
 
--- name: get-campaign-click-counts
+-- name: get-campaign-analytics-counts
 -- raw: true
--- %s = * or DISTINCT subscriber_id (prepared based on based on individual tracking=on/off). Prepared on boot.
 WITH intval AS (
     -- For intervals < a week, aggregate counts hourly, otherwise daily.
     SELECT CASE WHEN (EXTRACT (EPOCH FROM ($3::TIMESTAMP - $2::TIMESTAMP)) / 86400) >= 7 THEN 'day' ELSE 'hour' END
 )
-SELECT campaign_id, COUNT(%s) AS "count", DATE_TRUNC((SELECT * FROM intval), created_at) AS "timestamp"
+SELECT campaign_id, COUNT(*) AS "count", DATE_TRUNC((SELECT * FROM intval), created_at) AS "timestamp"
-    FROM link_clicks
+    FROM %s
     WHERE campaign_id=ANY($1) AND created_at >= $2 AND created_at <= $3
     GROUP BY campaign_id, "timestamp" ORDER BY "timestamp" ASC;