Fix "Filter internal and test users" when personsOnEventsMode = PersonsOnEventsMode.PERSON_ID_NO_OVERRIDE_PROPERTIES_ON_EVENTS

[danielbachhuber]

When using a data warehouse table with experiments, "Filter internal and test users" fails with this SQL error when personsOnEventsMode = PersonsOnEventsMode.PERSON_ID_NO_OVERRIDE_PROPERTIES_ON_EVENTS:

posthog.errors.CHQueryErrorUnknownIdentifier: Unknown expression or function identifier 'mat_pp_email' in scope

Here's an example of the full failing query from #27067:

SELECT
    ifNull(sum(accurateCastOrNull(e.usage, 'Float64')), 0) AS total,
    toStartOfDay(toDateTime(e.ds, 'UTC')) AS day_start,
    ifNull(nullIf(toString(e__events.`properties___$feature/test-experiment`), ''), '$$_posthog_breakdown_null_$$') AS breakdown_value
FROM
    (SELECT *
     FROM s3('http://host.docker.internal:19000/posthog/test_storage_bucket-posthog.hogql.datawarehouse.trendquery/*.parquet',
             'object_storage_root_user',
             '[HIDDEN]',
             'Parquet',
             '`ds` Date, `id` String, `usage` Int64, `userid` String')) AS e
    ASOF LEFT JOIN
    (SELECT
        replaceRegexpAll(nullIf(nullIf(JSONExtractRaw(events.properties, '$feature/test-experiment'), ''), 'null'), '^"|"$', '') AS `properties___$feature/test-experiment`,
        replaceRegexpAll(nullIf(nullIf(JSONExtractRaw(events.properties, '$host'), ''), 'null'), '^"|"$', '') AS `properties___$host`,
        events.event AS event,
        replaceRegexpAll(nullIf(nullIf(JSONExtractRaw(events.properties, '$user_id'), ''), 'null'), '^"|"$', '') AS `e__events___properties___$user_id`,
        toTimeZone(events.timestamp, 'UTC') AS timestamp,
        events.distinct_id AS distinct_id,
        events.properties AS properties
     FROM events
     WHERE (events.team_id = 2185)
        AND (event = '$feature_flag_called')
        AND (timestamp >= assumeNotNull(parseDateTime64BestEffortOrNull('2023-01-01 00:00:00', 6, 'UTC')))
        AND (timestamp <= assumeNotNull(parseDateTime64BestEffortOrNull('2025-01-06 13:07:18', 6, 'UTC')))) AS e__events
    ON (e__events.event = '$feature_flag_called')
        AND (e.userid = e__events.`e__events___properties___$user_id`)
        AND (e.ds >= e__events.timestamp)
WHERE
    ifNull(toDateTime(e.ds, 'UTC') >= assumeNotNull(parseDateTime64BestEffortOrNull('2023-01-01 00:00:00', 6, 'UTC')), 0)
    AND ifNull(toDateTime(e.ds, 'UTC') <= assumeNotNull(parseDateTime64BestEffortOrNull('2025-01-06 13:07:18', 6, 'UTC')), 0)
    AND ifNull(nullIf(nullIf(mat_pp_email, ''), 'null') NOT ILIKE '%@posthog.com%', 1)
    AND ifNull(NOT match(toString(e__events.`properties___$host`), '^(localhost|127\\.0\\.0\\.1)($|:)'), 1)
    AND (ifNull(e__events.event = '$feature_flag_called', 0)
        AND (ifNull(e__events.`properties___$feature/test-experiment` = 'control', 0)
            OR ifNull(e__events.`properties___$feature/test-experiment` = 'test', 0)))
GROUP BY
    day_start,
    breakdown_value

The underlying problem is that e__events__person.properties___email is resolving to mat_pp_email, but there's no mat_pp_email materialized column on the data warehouse table. The query should resolve to e__events.mat_pp_email.

[danielbachhuber]

Query failure comes from

posthog/posthog/hogql/printer.py

Line 1315 in 89522b1

field = field_type.resolve_database_field(self.context)