:hedgehog: PostHog is the leading platform for building self-driving products. Our developer tools – AI observability, analytics, session replay, flags, experiments, error tracking, logs, and more – capture all the context agents need to diagnose problems, uncover opportunities, and ship fixes. Steer it all from Slack, web, desktop, or the MCP.
Python ab-testingai-analyticsanalyticscdpdata-warehouseexperimentsfeature-flagsjavascriptproduct-analyticspythonreactsession-replaysurveystypescriptweb-analytics
"""Fetch survey responses for summarization."""
from datetime import datetime
from typing import cast
from posthog.hogql import ast
from posthog.hogql.parser import parse_expr, parse_select
from posthog.hogql_queries.paginators import HogQLHasMorePaginator
from posthog.models import Team
from products.surveys.backend.responses.fetch_rows import SUBMISSION_GROUPING_KEY
def fetch_responses(
survey_id: str,
question_index: int | None,
question_id: str | None,
start_date: datetime,
end_date: datetime,
team: Team,
limit: int = 100,
exclude_values: list[str] | None = None,
exclude_uuids: set[str] | None = None,
) -> list[str]:
"""
Fetch survey responses for a specific question.
Args:
survey_id: The survey ID
question_index: Index of the question (0-based)
question_id: ID of the question
start_date: Start date for filtering responses
end_date: End date for filtering responses
team: The team to query
limit: Maximum number of responses to fetch
exclude_values: List of values to exclude (e.g., predefined choices for choice questions)
exclude_uuids: Set of response UUIDs to exclude (e.g., archived responses)
Returns:
List of response strings
"""
paginator = HogQLHasMorePaginator(limit=limit, offset=0)
# Merge a submission's events instead of picking a single one: for each
# $survey_submission_id, keep the latest non-null answer to this question. This stops
# multi-event submissions (e.g. AI-feedback manual capture, where each answer can arrive
# on its own event) from dropping answers that only appeared on a non-final event.
# Events without a submission ID are keyed by uuid, so each stays its own response.
base_query = """
SELECT merged_response
FROM (
SELECT
argMaxIf(response, timestamp, isNotNull(response)) AS merged_response,
argMax(uuid, timestamp) AS representative_uuid
FROM (
SELECT
getSurveyResponse({question_index}, {question_id}) AS response,
timestamp,
uuid,
{grouping_key} AS submission_key
FROM events
WHERE event == 'survey sent'
AND timestamp >= {start_date}
AND timestamp <= {end_date}
AND properties.`$survey_id` = {survey_id}
)
GROUP BY submission_key
)
WHERE length(trim(coalesce(merged_response, ''))) > 0
"""
# Add archived response filter if there are UUIDs to exclude. Archiving records the
# submission's representative (latest) uuid, which is what argMax(uuid) surfaces above.
# UUIDs are pre-validated by Django's UUIDField when stored in SurveyResponseArchive
if exclude_uuids:
base_query += " AND representative_uuid NOT IN {exclude_uuids}"
placeholders: dict[str, ast.Expr] = {
"survey_id": ast.Constant(value=survey_id),
"start_date": ast.Constant(value=start_date),
"end_date": ast.Constant(value=end_date),
"question_index": ast.Constant(value=question_index),
"question_id": ast.Constant(value=question_id),
"grouping_key": parse_expr(SUBMISSION_GROUPING_KEY),
}
if exclude_uuids:
placeholders["exclude_uuids"] = ast.Tuple(exprs=[ast.Constant(value=uuid) for uuid in exclude_uuids])
q = parse_select(base_query, placeholders)
query_response = paginator.execute_hogql_query(
team=team,
query_type="survey_response_list_query",
query=cast(ast.SelectQuery, q),
)
responses = [x[0] for x in query_response.results if x[0]]
# Filter out predefined choices for choice questions
if exclude_values:
exclude_set = set(exclude_values)
responses = [r for r in responses if r not in exclude_set]
return responses