mirror of
https://github.com/lingble/chatwoot.git
synced 2025-10-29 10:12:34 +00:00
a509ef826a
This PR collapses multiple queries fetching stats from a single table to a single query ```sql SELECT user_id as user_id, COUNT(CASE WHEN name = 'conversation_resolved' THEN 1 END) as resolved_count, AVG(CASE WHEN name = 'conversation_resolved' THEN value END) as avg_resolution_time, AVG(CASE WHEN name = 'first_response' THEN value END) as avg_first_response_time, AVG(CASE WHEN name = 'reply_time' THEN value END) as avg_reply_time FROM "reporting_events" WHERE "reporting_events"."account_id" = <account_id> AND "reporting_events"."created_at" >= '2025-09-14 18:30:00' AND "reporting_events"."created_at" < '2025-10-14 18:29:59' GROUP BY "reporting_events"."user_id"; ``` ### Why this works? Here's why this optimization is faster based on PostgreSQL internals: - Single Table Scan vs Multiple Scans: Earlier we did 4 sequential scans (or 4 index scans) of the same data, with the same where clause, now in a single scan all 4 `CASE` expressions are evaluated in a single pass. - Shared Buffer Cache Efficiency: PostgreSQL's shared buffer cache stores recently accessed pages, with this, pages are loaded once and re-used for all aggregation, earlier with separate queries we were forced to re-read all from the disk each time - Reduced planning and network overhead (4 vs 1 query) ### How is it tested 1. The specs all pass without making any changes 2. Verified the reports side by side after generating from report seeder #### How to test Generate seed data using the following command ```bash ACCOUNT_ID=1 ENABLE_ACCOUNT_SEEDING=true bundle exec rake db:seed:reports_data ``` Once done download the reports, checkout to this branch and download the reports again and compare them
57 lines
1.7 KiB
Ruby
57 lines
1.7 KiB
Ruby
class V2::Reports::BaseSummaryBuilder
|
|
include DateRangeHelper
|
|
|
|
def build
|
|
load_data
|
|
prepare_report
|
|
end
|
|
|
|
private
|
|
|
|
def load_data
|
|
@conversations_count = fetch_conversations_count
|
|
load_reporting_events_data
|
|
end
|
|
|
|
def load_reporting_events_data
|
|
# Extract the column name for indexing (e.g., 'conversations.team_id' -> 'team_id')
|
|
index_key = group_by_key.to_s.split('.').last
|
|
|
|
results = reporting_events
|
|
.select(
|
|
"#{group_by_key} as #{index_key}",
|
|
"COUNT(CASE WHEN name = 'conversation_resolved' THEN 1 END) as resolved_count",
|
|
"AVG(CASE WHEN name = 'conversation_resolved' THEN #{average_value_key} END) as avg_resolution_time",
|
|
"AVG(CASE WHEN name = 'first_response' THEN #{average_value_key} END) as avg_first_response_time",
|
|
"AVG(CASE WHEN name = 'reply_time' THEN #{average_value_key} END) as avg_reply_time"
|
|
)
|
|
.group(group_by_key)
|
|
.index_by { |record| record.public_send(index_key) }
|
|
|
|
@resolved_count = results.transform_values(&:resolved_count)
|
|
@avg_resolution_time = results.transform_values(&:avg_resolution_time)
|
|
@avg_first_response_time = results.transform_values(&:avg_first_response_time)
|
|
@avg_reply_time = results.transform_values(&:avg_reply_time)
|
|
end
|
|
|
|
def reporting_events
|
|
@reporting_events ||= account.reporting_events.where(created_at: range)
|
|
end
|
|
|
|
def fetch_conversations_count
|
|
# Override this method
|
|
end
|
|
|
|
def group_by_key
|
|
# Override this method
|
|
end
|
|
|
|
def prepare_report
|
|
# Override this method
|
|
end
|
|
|
|
def average_value_key
|
|
ActiveModel::Type::Boolean.new.cast(params[:business_hours]).present? ? :value_in_business_hours : :value
|
|
end
|
|
end
|