Update query snapshots

PostHog · Apr 26, 2024 · eba7345 · eba7345
1 parent e31fa29
commit eba7345
Show file tree

Hide file tree

Showing 4 changed files with 502 additions and 1,024 deletions.
diff --git a/posthog/queries/funnels/test/__snapshots__/test_funnel_time_to_convert.ambr b/posthog/queries/funnels/test/__snapshots__/test_funnel_time_to_convert.ambr
@@ -20,7 +20,7 @@
                      AND latest_1 <= latest_0 + INTERVAL 7 DAY
                      AND latest_1 <= latest_2
                      AND latest_2 <= latest_0 + INTERVAL 7 DAY, 3, if(latest_0 <= latest_1
-                                                                      AND latest_1 <= latest_0 + INTERVAL 7 DAY, 2, 1)) AS steps,
+                                                                      AND latest_1 <= latest_0 + INTERVAL 7 DAY, 2, 1)) AS steps ,
                   if(isNotNull(latest_1)
                      AND latest_1 <= latest_0 + INTERVAL 7 DAY, dateDiff('second', toDateTime(latest_0), toDateTime(latest_1)), NULL) step_1_conversion_time,
                   if(isNotNull(latest_2)
@@ -86,17 +86,19 @@
               steps
      HAVING steps = max_steps),
        histogram_params AS
-    (/* Binning ensures that each sample belongs to a bin in results */ /* If bin_count is not a custom number, it's calculated in bin_count_expression */ SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
-                                                                                                                                                                  ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
-                                                                                                                                                                  round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
-                                                                                                                                                                  count() AS sample_count,
-                                                                                                                                                                  least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
-                                                                                                                                                                  ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw, /* Use 60 seconds as fallback bin width in case of only one sample */ if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
-     FROM step_runs -- We only need to check step to_step here, because it depends on all the other ones being NOT NULL too
+    (SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
+            ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
+            round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
+            count() AS sample_count,
+            least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
+            ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw,
+            if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
+     FROM step_runs
+     WHERE step_1_average_conversion_time_inner IS NOT NULL ),
 
-     WHERE step_1_average_conversion_time_inner IS NOT NULL ), /* Below CTEs make histogram_params columns available to the query below as straightforward identifiers */
     (SELECT bin_width_seconds
-     FROM histogram_params) AS bin_width_seconds, /* bin_count is only made available as an identifier if it had to be calculated */
+     FROM histogram_params) AS bin_width_seconds,
+
     (SELECT bin_count
      FROM histogram_params) AS bin_count,
 
@@ -112,12 +114,12 @@
          person_count,
          histogram_average_conversion_time AS average_conversion_time
   FROM
-    (/* Calculating bins from step runs */ SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
-                                                  count() AS person_count
+    (SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
+            count() AS person_count
      FROM step_runs
      GROUP BY bin_from_seconds) results
   RIGHT OUTER JOIN
-    (/* Making sure bin_count bins are returned */ /* Those not present in the results query due to lack of data simply get person_count 0 */ SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
+    (SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
      FROM system.numbers
      LIMIT ifNull(bin_count, 0) + 1) fill USING (bin_from_seconds)
   ORDER BY bin_from_seconds SETTINGS max_ast_elements=1000000,
@@ -446,17 +448,19 @@
               steps
      HAVING steps = max_steps),
        histogram_params AS
-    (/* Binning ensures that each sample belongs to a bin in results */ /* If bin_count is not a custom number, it's calculated in bin_count_expression */ SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
-                                                                                                                                                                  ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
-                                                                                                                                                                  round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
-                                                                                                                                                                  count() AS sample_count,
-                                                                                                                                                                  least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
-                                                                                                                                                                  ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw, /* Use 60 seconds as fallback bin width in case of only one sample */ if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
-     FROM step_runs -- We only need to check step to_step here, because it depends on all the other ones being NOT NULL too
+    (SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
+            ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
+            round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
+            count() AS sample_count,
+            least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
+            ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw,
+            if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
+     FROM step_runs
+     WHERE step_1_average_conversion_time_inner IS NOT NULL ),
 
-     WHERE step_1_average_conversion_time_inner IS NOT NULL ), /* Below CTEs make histogram_params columns available to the query below as straightforward identifiers */
     (SELECT bin_width_seconds
-     FROM histogram_params) AS bin_width_seconds, /* bin_count is only made available as an identifier if it had to be calculated */
+     FROM histogram_params) AS bin_width_seconds,
+
     (SELECT bin_count
      FROM histogram_params) AS bin_count,
 
@@ -472,12 +476,12 @@
          person_count,
          histogram_average_conversion_time AS average_conversion_time
   FROM
-    (/* Calculating bins from step runs */ SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
-                                                  count() AS person_count
+    (SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
+            count() AS person_count
      FROM step_runs
      GROUP BY bin_from_seconds) results
   RIGHT OUTER JOIN
-    (/* Making sure bin_count bins are returned */ /* Those not present in the results query due to lack of data simply get person_count 0 */ SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
+    (SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
      FROM system.numbers
      LIMIT ifNull(bin_count, 0) + 1) fill USING (bin_from_seconds)
   ORDER BY bin_from_seconds SETTINGS max_ast_elements=1000000,
@@ -502,7 +506,7 @@
         FROM
           (SELECT *,
                   arraySort([latest_0,latest_1,latest_2]) as event_times,
-                  arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps,
+                  arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps ,
                   arraySort([latest_0,latest_1,latest_2]) as conversion_times,
                   if(isNotNull(conversion_times[2])
                      AND conversion_times[2] <= conversion_times[1] + INTERVAL 7 DAY, dateDiff('second', conversion_times[1], conversion_times[2]), NULL) step_1_conversion_time,
@@ -552,7 +556,7 @@
            WHERE step_0 = 1
            UNION ALL SELECT *,
                             arraySort([latest_0,latest_1,latest_2]) as event_times,
-                            arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps,
+                            arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps ,
                             arraySort([latest_0,latest_1,latest_2]) as conversion_times,
                             if(isNotNull(conversion_times[2])
                                AND conversion_times[2] <= conversion_times[1] + INTERVAL 7 DAY, dateDiff('second', conversion_times[1], conversion_times[2]), NULL) step_1_conversion_time,
@@ -602,7 +606,7 @@
            WHERE step_0 = 1
            UNION ALL SELECT *,
                             arraySort([latest_0,latest_1,latest_2]) as event_times,
-                            arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps,
+                            arraySum([if(latest_0 < latest_1 AND latest_1 <= latest_0 + INTERVAL 7 DAY, 1, 0),if(latest_0 < latest_2 AND latest_2 <= latest_0 + INTERVAL 7 DAY, 1, 0), 1]) AS steps ,
                             arraySort([latest_0,latest_1,latest_2]) as conversion_times,
                             if(isNotNull(conversion_times[2])
                                AND conversion_times[2] <= conversion_times[1] + INTERVAL 7 DAY, dateDiff('second', conversion_times[1], conversion_times[2]), NULL) step_1_conversion_time,
@@ -654,17 +658,19 @@
               steps
      HAVING steps = max_steps),
        histogram_params AS
-    (/* Binning ensures that each sample belongs to a bin in results */ /* If bin_count is not a custom number, it's calculated in bin_count_expression */ SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
-                                                                                                                                                                  ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
-                                                                                                                                                                  round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
-                                                                                                                                                                  count() AS sample_count,
-                                                                                                                                                                  least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
-                                                                                                                                                                  ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw, /* Use 60 seconds as fallback bin width in case of only one sample */ if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
-     FROM step_runs -- We only need to check step to_step here, because it depends on all the other ones being NOT NULL too
+    (SELECT ifNull(floor(min(step_1_average_conversion_time_inner)), 0) AS from_seconds,
+            ifNull(ceil(max(step_1_average_conversion_time_inner)), 1) AS to_seconds,
+            round(avg(step_1_average_conversion_time_inner), 2) AS average_conversion_time,
+            count() AS sample_count,
+            least(60, greatest(1, ceil(cbrt(ifNull(sample_count, 0))))) AS bin_count,
+            ceil((to_seconds - from_seconds) / bin_count) AS bin_width_seconds_raw,
+            if(bin_width_seconds_raw > 0, bin_width_seconds_raw, 60) AS bin_width_seconds
+     FROM step_runs
+     WHERE step_1_average_conversion_time_inner IS NOT NULL ),
 
-     WHERE step_1_average_conversion_time_inner IS NOT NULL ), /* Below CTEs make histogram_params columns available to the query below as straightforward identifiers */
     (SELECT bin_width_seconds
-     FROM histogram_params) AS bin_width_seconds, /* bin_count is only made available as an identifier if it had to be calculated */
+     FROM histogram_params) AS bin_width_seconds,
+
     (SELECT bin_count
      FROM histogram_params) AS bin_count,
 
@@ -680,12 +686,12 @@
          person_count,
          histogram_average_conversion_time AS average_conversion_time
   FROM
-    (/* Calculating bins from step runs */ SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
-                                                  count() AS person_count
+    (SELECT histogram_from_seconds + floor((step_1_average_conversion_time_inner - histogram_from_seconds) / bin_width_seconds) * bin_width_seconds AS bin_from_seconds,
+            count() AS person_count
      FROM step_runs
      GROUP BY bin_from_seconds) results
   RIGHT OUTER JOIN
-    (/* Making sure bin_count bins are returned */ /* Those not present in the results query due to lack of data simply get person_count 0 */ SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
+    (SELECT histogram_from_seconds + number * bin_width_seconds AS bin_from_seconds
      FROM system.numbers
      LIMIT ifNull(bin_count, 0) + 1) fill USING (bin_from_seconds)
   ORDER BY bin_from_seconds SETTINGS max_ast_elements=1000000,

diff --git a/posthog/queries/test/__snapshots__/test_lifecycle.ambr b/posthog/queries/test/__snapshots__/test_lifecycle.ambr
@@ -1,10 +1,8 @@
 # serializer version: 1
 # name: TestLifecycle.test_sampling
   '''
-  WITH 'day' AS selected_period, -- enumerate all requested periods, so we can zero fill as needed.
-   -- NOTE: we use dateSub interval rather than seconds, which means we can handle,
-   -- for instance, month intervals which do not have a fixed number of seconds.
-   periods AS
+  WITH 'day' AS selected_period,
+       periods AS
     (SELECT dateSub(day, number, dateTrunc(selected_period, toDateTime('2020-01-19 23:59:59', 'UTC'))) AS start_of_period
      FROM numbers(dateDiff('day', dateTrunc('day', toDateTime('2020-01-12 00:00:00', 'UTC')), dateTrunc('day', toDateTime('2020-01-19 23:59:59', 'UTC') + INTERVAL 1 day))))
   SELECT groupArray(start_of_period) AS date,
@@ -18,8 +16,7 @@
        (SELECT periods.start_of_period as start_of_period,
                toUInt16(0) AS counts,
                status
-        FROM periods -- Zero fill for each status
-
+        FROM periods
         CROSS JOIN
           (SELECT status
            FROM
@@ -77,10 +74,8 @@
 # ---
 # name: TestLifecycle.test_timezones
   '''
-  WITH 'day' AS selected_period, -- enumerate all requested periods, so we can zero fill as needed.
-   -- NOTE: we use dateSub interval rather than seconds, which means we can handle,
-   -- for instance, month intervals which do not have a fixed number of seconds.
-   periods AS
+  WITH 'day' AS selected_period,
+       periods AS
     (SELECT dateSub(day, number, dateTrunc(selected_period, toDateTime('2020-01-19 23:59:59', 'UTC'))) AS start_of_period
      FROM numbers(dateDiff('day', dateTrunc('day', toDateTime('2020-01-12 00:00:00', 'UTC')), dateTrunc('day', toDateTime('2020-01-19 23:59:59', 'UTC') + INTERVAL 1 day))))
   SELECT groupArray(start_of_period) AS date,
@@ -94,8 +89,7 @@
        (SELECT periods.start_of_period as start_of_period,
                toUInt16(0) AS counts,
                status
-        FROM periods -- Zero fill for each status
-
+        FROM periods
         CROSS JOIN
           (SELECT status
            FROM
@@ -153,10 +147,8 @@
 # ---
 # name: TestLifecycle.test_timezones.1
   '''
-  WITH 'day' AS selected_period, -- enumerate all requested periods, so we can zero fill as needed.
-   -- NOTE: we use dateSub interval rather than seconds, which means we can handle,
-   -- for instance, month intervals which do not have a fixed number of seconds.
-   periods AS
+  WITH 'day' AS selected_period,
+       periods AS
     (SELECT dateSub(day, number, dateTrunc(selected_period, toDateTime('2020-01-19 23:59:59', 'US/Pacific'))) AS start_of_period
      FROM numbers(dateDiff('day', dateTrunc('day', toDateTime('2020-01-12 00:00:00', 'US/Pacific')), dateTrunc('day', toDateTime('2020-01-19 23:59:59', 'US/Pacific') + INTERVAL 1 day))))
   SELECT groupArray(start_of_period) AS date,
@@ -170,8 +162,7 @@
        (SELECT periods.start_of_period as start_of_period,
                toUInt16(0) AS counts,
                status
-        FROM periods -- Zero fill for each status
-
+        FROM periods
         CROSS JOIN
           (SELECT status
            FROM