feat(analytics-platform): Redesign sessions v3 for better partition pruning and indexing

[robbie-c]

Problem

I found that we were not making good use of indexing and partition pruning, this PR fixes that. See some discussion #38809 (comment)

Changes

• Make a separate session_timestamp column which is materialized from the session ID
• Update the where clause extractor to support this
• Add extra logic to the where clause extractor to support point queries

How did you test this code?

• Added some more tests to the HogQL sessions test files
• Updated the where clause extractor v3 tests

👉 Stay up-to-date with PostHog coding conventions for a smoother review.

Changelog: (features only) Is this feature complete?

[greptile-apps]

_{14 files reviewed, 6 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

style: Consider adding type signatures for this function like other similar functions (e.g., fromUnixTimestamp on line 193) to ensure proper type checking and return type specification

Suggested change

	"fromUnixTimestamp64Milli": HogQLFunctionMeta("fromUnixTimestamp64Milli", 1, 1),
	"fromUnixTimestamp64Milli": HogQLFunctionMeta(
	"fromUnixTimestamp64Milli",
	1,
	1,
	signatures=[
	((IntegerType(),), DateTimeType()),
	],
	),

Prompt To Fix With AI

This is a comment left during a code review.
Path: posthog/hogql/functions/clickhouse/datetime.py
Line: 21:21

Comment:
**style:** Consider adding type signatures for this function like other similar functions (e.g., `fromUnixTimestamp` on line 193) to ensure proper type checking and return type specification

```suggestion
    "fromUnixTimestamp64Milli": HogQLFunctionMeta(
        "fromUnixTimestamp64Milli",
        1,
        1,
        signatures=[
            ((IntegerType(),), DateTimeType()),
        ],
    ),
```

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

style: Duplicate test method with identical logic to test_timestamp_unrelated_function above - consider removing or differentiating

Prompt To Fix With AI

This is a comment left during a code review.
Path: posthog/hogql/database/schema/util/test/test_session_v3_where_clause_extractor.py
Line: 140:144

Comment:
**style:** Duplicate test method with identical logic to test_timestamp_unrelated_function above - consider removing or differentiating

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

syntax: Extra comma creates syntax error in SQL

Suggested change

	from sessions
	session_id
	from sessions

Prompt To Fix With AI

This is a comment left during a code review.
Path: posthog/hogql/database/schema/util/test/test_session_v3_where_clause_extractor.py
Line: 612:612

Comment:
**syntax:** Extra comma creates syntax error in SQL

```suggestion
    session_id
    from sessions
```

How can I resolve this? If you propose a fix, please make it concise.

[greptile-apps]

syntax: Missing comma after session.id as session_id, - this will cause a SQL syntax error

Suggested change

	where session_id = {session_id} AND timestamp >= '1970-01-01'
	select
	session.id as session_id
	from events
	where session_id = {session_id} AND timestamp >= '1970-01-01'

Prompt To Fix With AI

This is a comment left during a code review.
Path: posthog/hogql/database/schema/test/test_sessions_v3.py
Line: 641:641

Comment:
**syntax:** Missing comma after `session.id as session_id,` - this will cause a SQL syntax error

```suggestion
                select
                    session.id as session_id
                from events
                where session_id = {session_id} AND timestamp >= '1970-01-01'
```

How can I resolve this? If you propose a fix, please make it concise.

[andyzzhao]

It is possible for session_id_str_to_timestamp_expr to return None? If so, how will be known about it?

[robbie-c]

session_id_str_to_timestamp_expr can be null (it uses accurateCastOrNull(x, 'UUID') behind the scenes)

I don't believe that this being nullable has performance implications, and in testing, if I use a garbage constant string for the UUID (e.g. select * from sessions where session_id = 'garbage'), clickhouse was smart enough to know it doesn't need to load any parts

[lricoy]

Took my time, but really didn't spot anything

[lricoy]

🚢