From a3acdc7041564592cd2891f9992ea2c9db95d6cd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ferdinand=20M=C3=BCtsch?= <ferdinand@muetsch.io>
Date: Fri, 18 Mar 2022 12:29:43 +0100
Subject: [PATCH] fix: duration aggregation for heartbeats with identical
 timestamps (resolve #340)

---
 scripts/aggregate_durations.sql      |  9 +++++++++
 scripts/count_duplicates_by_user.sql | 13 +++++++++++++
 services/duration.go                 |  6 +++++-
 services/duration_test.go            | 11 +++++++++++
 4 files changed, 38 insertions(+), 1 deletion(-)
 create mode 100644 scripts/aggregate_durations.sql
 create mode 100644 scripts/count_duplicates_by_user.sql

diff --git a/scripts/aggregate_durations.sql b/scripts/aggregate_durations.sql
new file mode 100644
index 0000000..e1a1b2a
--- /dev/null
+++ b/scripts/aggregate_durations.sql
@@ -0,0 +1,9 @@
+SELECT project, language, editor, operating_system, machine, branch, SUM(GREATEST(1, diff)) as 'sum'
+FROM (
+         SELECT project, language, editor, operating_system, machine, branch, TIME_TO_SEC(LEAST(TIMEDIFF(time, LAG(time) over w), '00:02:00')) as 'diff'
+         FROM heartbeats
+         WHERE user_id = 'n1try'
+             WINDOW w AS (ORDER BY time)
+     ) s2
+WHERE diff IS NOT NULL
+GROUP BY project, language, editor, operating_system, machine, branch;
\ No newline at end of file
diff --git a/scripts/count_duplicates_by_user.sql b/scripts/count_duplicates_by_user.sql
new file mode 100644
index 0000000..751a9e5
--- /dev/null
+++ b/scripts/count_duplicates_by_user.sql
@@ -0,0 +1,13 @@
+SELECT s2.user_id, sum(c) as count, total, (sum(c) / total) as ratio
+FROM (
+         SELECT time,
+                user_id,
+                entity,
+                COUNT(time) as c
+         FROM heartbeats
+         GROUP BY time, user_id, entity
+         HAVING COUNT(time) > 1
+     ) s2
+         LEFT JOIN (SELECT user_id, count(id) AS total FROM heartbeats GROUP BY user_id) s3 ON s2.user_id = s3.user_id
+GROUP BY user_id
+ORDER BY count DESC;
\ No newline at end of file
diff --git a/services/duration.go b/services/duration.go
index 42ab705..6dedfbe 100644
--- a/services/duration.go
+++ b/services/duration.go
@@ -80,8 +80,12 @@ func (srv *DurationService) Get(from, to time.Time, user *models.User, filters *
 
 	for _, list := range mapping {
 		for _, d := range list {
+			// will only happen if two heartbeats with different hashes (e.g. different project) have the same timestamp
+			// that, in turn, will most likely only happen for mysql, where `time` column's precision was set to second for a while
+			// assume that two non-identical heartbeats with identical time are sub-second apart from each other, so round up to expectancy value
+			// also see https://github.com/muety/wakapi/issues/340
 			if d.Duration == 0 {
-				d.Duration = HeartbeatDiffThreshold
+				d.Duration = 500 * time.Millisecond
 			}
 			durations = append(durations, d)
 		}
diff --git a/services/duration_test.go b/services/duration_test.go
index cb3e6aa..7f793f5 100644
--- a/services/duration_test.go
+++ b/services/duration_test.go
@@ -65,6 +65,17 @@ func (suite *DurationServiceTestSuite) SetupSuite() {
 			Machine:         TestMachine1,
 			Time:            models.CustomTime(suite.TestStartTime.Add(30 * time.Second)), // 0:30
 		},
+		// duplicate of previous one
+		{
+			ID:              rand.Uint64(),
+			UserID:          TestUserId,
+			Project:         TestProject1,
+			Language:        TestLanguageGo,
+			Editor:          TestEditorGoland,
+			OperatingSystem: TestOsLinux,
+			Machine:         TestMachine1,
+			Time:            models.CustomTime(suite.TestStartTime.Add(30 * time.Second)), // 0:30
+		},
 		{
 			ID:              rand.Uint64(),
 			UserID:          TestUserId,