Update query snapshots

posthog/queries/trends/test/__snapshots__/test_breakdowns_by_current_url.ambr

@@ -19,7 +19,6 @@
 # ---
 # name: TestBreakdownsByCurrentURL.test_breakdown_by_current_url.1
   '''
-
   SELECT groupArray(day_start) as date,
          groupArray(count) AS total,
          breakdown_value
@@ -30,13 +29,36 @@
      FROM
        (SELECT *
         FROM
-          (SELECT toUInt16(0) AS total,
-                  ticks.day_start as day_start,
-                  breakdown_value
+          (-- Create a table with 1 row for each interval for the requested date range
+   -- This acts as a method of zero filling, i.e. when there are no data points
+   -- for a given interval, we'll still have a row for the group by interval with
+   -- a 0 value.
+   --
+   -- It's essentially a cross product of graph "ticks" and breakdown values.
+   --
+   -- TODO: we're relying on num_intervals, seconds_int_interval etc. being passed
+   --       in as a parameter. To reduce the coupling between here and the
+   --       calling code, we could perform calculations for these within the query
+   --       itself based on date_to/date_from. We could also pass in the intervals
+   --       explicitly, although we'll be relying on the date handling between python
+   --       and ClickHouse to be the same.
+   --
+   -- NOTE: there is the ORDER BY ... WITH FILL Expression but I'm not sure how we'd
+   --       handle the edge cases:
+   --
+   --          https://clickhouse.com/docs/en/sql-reference/statements/select/order-by/#orderby-with-fill
+   --
+   SELECT toUInt16(0) AS total,
+          ticks.day_start as day_start,
+          breakdown_value
            FROM
-             (SELECT toStartOfDay(toDateTime('2020-01-12 23:59:59', 'UTC')) - toIntervalDay(number) as day_start
+             (-- Generates all the intervals/ticks in the date range
+   -- NOTE: we build this range by including successive intervals back from the
+   --       upper bound, then including the lower bound in the query also.
+   SELECT toStartOfDay(toDateTime('2020-01-12 23:59:59', 'UTC')) - toIntervalDay(number) as day_start
               FROM numbers(11)
-              UNION ALL SELECT toStartOfDay(toDateTime('2020-01-02 00:00:00', 'UTC')) as day_start) as ticks
+              UNION ALL SELECT toStartOfDay(toDateTime('2020-01-02 00:00:00', 'UTC')) as day_start) as ticks -- Zero fill for all values for the specified breakdown
+
            CROSS JOIN
              (SELECT breakdown_value
               FROM
@@ -84,7 +106,6 @@
 # ---
 # name: TestBreakdownsByCurrentURL.test_breakdown_by_pathname.1
   '''
-
   SELECT groupArray(day_start) as date,
          groupArray(count) AS total,
          breakdown_value
@@ -95,13 +116,36 @@
      FROM
        (SELECT *
         FROM
-          (SELECT toUInt16(0) AS total,
-                  ticks.day_start as day_start,
-                  breakdown_value
+          (-- Create a table with 1 row for each interval for the requested date range
+   -- This acts as a method of zero filling, i.e. when there are no data points
+   -- for a given interval, we'll still have a row for the group by interval with
+   -- a 0 value.
+   --
+   -- It's essentially a cross product of graph "ticks" and breakdown values.
+   --
+   -- TODO: we're relying on num_intervals, seconds_int_interval etc. being passed
+   --       in as a parameter. To reduce the coupling between here and the
+   --       calling code, we could perform calculations for these within the query
+   --       itself based on date_to/date_from. We could also pass in the intervals
+   --       explicitly, although we'll be relying on the date handling between python
+   --       and ClickHouse to be the same.
+   --
+   -- NOTE: there is the ORDER BY ... WITH FILL Expression but I'm not sure how we'd
+   --       handle the edge cases:
+   --
+   --          https://clickhouse.com/docs/en/sql-reference/statements/select/order-by/#orderby-with-fill
+   --
+   SELECT toUInt16(0) AS total,
+          ticks.day_start as day_start,
+          breakdown_value
            FROM
-             (SELECT toStartOfDay(toDateTime('2020-01-12 23:59:59', 'UTC')) - toIntervalDay(number) as day_start
+             (-- Generates all the intervals/ticks in the date range
+   -- NOTE: we build this range by including successive intervals back from the
+   --       upper bound, then including the lower bound in the query also.
+   SELECT toStartOfDay(toDateTime('2020-01-12 23:59:59', 'UTC')) - toIntervalDay(number) as day_start
               FROM numbers(11)
-              UNION ALL SELECT toStartOfDay(toDateTime('2020-01-02 00:00:00', 'UTC')) as day_start) as ticks
+              UNION ALL SELECT toStartOfDay(toDateTime('2020-01-02 00:00:00', 'UTC')) as day_start) as ticks -- Zero fill for all values for the specified breakdown
+
            CROSS JOIN
              (SELECT breakdown_value
               FROM