feat(web-analytics): HogQL sessions where extractor

posthog/hogql/ast.py

@@ -46,8 +46,17 @@ def resolve_constant_type(self, context: HogQLContext):
     def resolve_database_field(self, context: HogQLContext):
         if isinstance(self.type, FieldType):
             return self.type.resolve_database_field(context)
+        if isinstance(self.type, PropertyType):
+            return self.type.field_type.resolve_database_field(context)
         raise NotImplementedException("FieldAliasType.resolve_database_field not implemented")
 
+    def resolve_table_type(self, context: HogQLContext):
+        if isinstance(self.type, FieldType):
+            return self.type.table_type
+        if isinstance(self.type, PropertyType):
+            return self.type.field_type.table_type
+        raise NotImplementedException("FieldAliasType.resolve_table_type not implemented")
+
 
 @dataclass(kw_only=True)
 class BaseTableType(Type):
@@ -339,6 +348,9 @@ def get_child(self, name: str | int, context: HogQLContext) -> Type:
             f'Can not access property "{name}" on field "{self.name}" of type: {type(database_field).__name__}'
         )
 
+    def resolve_table_type(self, context: HogQLContext):
+        return self.table_type
+
 
 @dataclass(kw_only=True)
 class PropertyType(Type):

posthog/hogql/base.py

@@ -32,7 +32,7 @@ def accept(self, visitor):
             return visit(self)
         if hasattr(visitor, "visit_unknown"):
             return visitor.visit_unknown(self)
-        raise NotImplementedException(f"Visitor has no method {method_name}")
+        raise NotImplementedException(f"{visitor.__class__.__name__} has no method {method_name}")
 
 
 @dataclass(kw_only=True)

posthog/hogql/database/models.py

@@ -3,7 +3,6 @@
 
 from posthog.hogql.base import Expr
 from posthog.hogql.errors import HogQLException, NotImplementedException
-from posthog.schema import HogQLQueryModifiers
 
 if TYPE_CHECKING:
     from posthog.hogql.context import HogQLContext
@@ -126,12 +125,14 @@ def resolve_table(self, context: "HogQLContext") -> Table:
 
 class LazyTable(Table):
     """
-    A table that is replaced with a subquery returned from `lazy_select(requested_fields: Dict[name, chain], modifiers: HogQLQueryModifiers)`
+    A table that is replaced with a subquery returned from `lazy_select(requested_fields: Dict[name, chain], modifiers: HogQLQueryModifiers, node: SelectQuery)`
     """
 
     model_config = ConfigDict(extra="forbid")
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers) -> Any:
+    def lazy_select(
+        self, requested_fields: Dict[str, List[str | int]], context: "HogQLContext", node: "SelectQuery"
+    ) -> Any:
         raise NotImplementedException("LazyTable.lazy_select not overridden")
 
 

posthog/hogql/database/schema/cohort_people.py

@@ -9,7 +9,6 @@
     FieldOrTable,
 )
 from posthog.hogql.database.schema.persons import join_with_persons_table
-from posthog.schema import HogQLQueryModifiers
 
 COHORT_PEOPLE_FIELDS = {
     "person_id": StringDatabaseField(name="person_id"),
@@ -67,7 +66,7 @@ def to_printed_hogql(self):
 class CohortPeople(LazyTable):
     fields: Dict[str, FieldOrTable] = COHORT_PEOPLE_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         return select_from_cohort_people_table(requested_fields)
 
     def to_printed_clickhouse(self, context):

posthog/hogql/database/schema/groups.py

@@ -13,7 +13,6 @@
     FieldOrTable,
 )
 from posthog.hogql.errors import HogQLException
-from posthog.schema import HogQLQueryModifiers
 
 GROUPS_TABLE_FIELDS = {
     "index": IntegerDatabaseField(name="group_type_index"),
@@ -83,7 +82,7 @@ def to_printed_hogql(self):
 class GroupsTable(LazyTable):
     fields: Dict[str, FieldOrTable] = GROUPS_TABLE_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         return select_from_groups_table(requested_fields)
 
     def to_printed_clickhouse(self, context):

posthog/hogql/database/schema/log_entries.py

@@ -9,7 +9,6 @@
     LazyTable,
     FieldOrTable,
 )
-from posthog.schema import HogQLQueryModifiers
 
 LOG_ENTRIES_FIELDS: Dict[str, FieldOrTable] = {
     "team_id": IntegerDatabaseField(name="team_id"),
@@ -35,7 +34,7 @@ def to_printed_hogql(self):
 class ReplayConsoleLogsLogEntriesTable(LazyTable):
     fields: Dict[str, FieldOrTable] = LOG_ENTRIES_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         fields: List[ast.Expr] = [ast.Field(chain=["log_entries"] + chain) for name, chain in requested_fields.items()]
 
         return ast.SelectQuery(
@@ -58,7 +57,7 @@ def to_printed_hogql(self):
 class BatchExportLogEntriesTable(LazyTable):
     fields: Dict[str, FieldOrTable] = LOG_ENTRIES_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         fields: List[ast.Expr] = [ast.Field(chain=["log_entries"] + chain) for name, chain in requested_fields.items()]
 
         return ast.SelectQuery(

posthog/hogql/database/schema/person_distinct_ids.py

@@ -14,7 +14,6 @@
 )
 from posthog.hogql.database.schema.persons import join_with_persons_table
 from posthog.hogql.errors import HogQLException
-from posthog.schema import HogQLQueryModifiers
 
 PERSON_DISTINCT_IDS_FIELDS = {
     "team_id": IntegerDatabaseField(name="team_id"),
@@ -82,7 +81,7 @@ def to_printed_hogql(self):
 class PersonDistinctIdsTable(LazyTable):
     fields: Dict[str, FieldOrTable] = PERSON_DISTINCT_IDS_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         return select_from_person_distinct_ids_table(requested_fields)
 
     def to_printed_clickhouse(self, context):

posthog/hogql/database/schema/persons.py

@@ -123,8 +123,8 @@ def to_printed_hogql(self):
 class PersonsTable(LazyTable):
     fields: Dict[str, FieldOrTable] = PERSONS_FIELDS
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
-        return select_from_persons_table(requested_fields, modifiers)
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
+        return select_from_persons_table(requested_fields, context.modifiers)
 
     def to_printed_clickhouse(self, context):
         return "person"

posthog/hogql/database/schema/persons_pdi.py

@@ -10,7 +10,6 @@
     FieldOrTable,
 )
 from posthog.hogql.errors import HogQLException
-from posthog.schema import HogQLQueryModifiers
 
 
 # :NOTE: We already have person_distinct_ids.py, which most tables link to. This persons_pdi.py is a hack to
@@ -63,7 +62,7 @@ class PersonsPDITable(LazyTable):
         "person_id": StringDatabaseField(name="person_id"),
     }
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         return persons_pdi_select(requested_fields)
 
     def to_printed_clickhouse(self, context):

posthog/hogql/database/schema/session_replay_events.py

@@ -15,7 +15,6 @@
     PersonDistinctIdsTable,
     join_with_person_distinct_ids_table,
 )
-from posthog.schema import HogQLQueryModifiers
 
 RAW_ONLY_FIELDS = ["min_first_timestamp", "max_last_timestamp"]
 
@@ -115,7 +114,7 @@ class SessionReplayEventsTable(LazyTable):
         "first_url": StringDatabaseField(name="first_url"),
     }
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node):
         return select_from_session_replay_events_table(requested_fields)
 
     def to_printed_clickhouse(self, context):

posthog/hogql/database/schema/sessions.py

@@ -1,5 +1,7 @@
 from typing import Dict, List, cast
 
+from posthog.hogql import ast
+from posthog.hogql.context import HogQLContext
 from posthog.hogql.database.models import (
     StringDatabaseField,
     DateTimeDatabaseField,
@@ -11,7 +13,7 @@
     LazyTable,
 )
 from posthog.hogql.database.schema.channel_type import create_channel_type_expr
-from posthog.schema import HogQLQueryModifiers
+from posthog.hogql.database.schema.util.session_where_clause_extractor import SessionMinTimestampWhereClauseExtractor
 
 
 SESSIONS_COMMON_FIELDS: Dict[str, FieldOrTable] = {
@@ -62,7 +64,9 @@ def avoid_asterisk_fields(self) -> List[str]:
         ]
 
 
-def select_from_sessions_table(requested_fields: Dict[str, List[str | int]]):
+def select_from_sessions_table(
+    requested_fields: Dict[str, List[str | int]], node: ast.SelectQuery, context: HogQLContext
+):
     from posthog.hogql import ast
 
     table_name = "raw_sessions"
@@ -134,10 +138,13 @@ def select_from_sessions_table(requested_fields: Dict[str, List[str | int]]):
             )
             group_by_fields.append(ast.Field(chain=cast(list[str | int], [table_name]) + chain))
 
+    where = SessionMinTimestampWhereClauseExtractor(context).get_inner_where(node)
+
     return ast.SelectQuery(
         select=select_fields,
         select_from=ast.JoinExpr(table=ast.Field(chain=[table_name])),
         group_by=group_by_fields,
+        where=where,
     )
 
 
@@ -148,8 +155,8 @@ class SessionsTable(LazyTable):
         "channel_type": StringDatabaseField(name="channel_type"),
     }
 
-    def lazy_select(self, requested_fields: Dict[str, List[str | int]], modifiers: HogQLQueryModifiers):
-        return select_from_sessions_table(requested_fields)
+    def lazy_select(self, requested_fields: Dict[str, List[str | int]], context, node: ast.SelectQuery):
+        return select_from_sessions_table(requested_fields, node, context)
 
     def to_printed_clickhouse(self, context):
         return "sessions"