data layer

data-layer/api/internal/repo/analytics_repo.go

@@ -0,0 +1,167 @@
+package repo
+
+import (
+	"context"
+	"fmt"
+	"time"
+
+	"github.com/ClickHouse/clickhouse-go/v2"
+	"github.com/ClickHouse/clickhouse-go/v2/lib/driver"
+
+	"github.com/dbiz/cdp/data-layer/api/internal/model"
+	"github.com/dbiz/cdp/data-layer/api/internal/templates"
+)
+
+// AnalyticsRepo runs the higher-level P1 query templates (funnel, retention,
+// session) against ClickHouse. It shares the read connection with EventRepo
+// but lives in its own file because the templates need their own data shapes.
+type AnalyticsRepo struct {
+	ch  driver.Conn
+	tpl *templates.Store
+}
+
+func NewAnalyticsRepo(ch driver.Conn, tpl *templates.Store) *AnalyticsRepo {
+	return &AnalyticsRepo{ch: ch, tpl: tpl}
+}
+
+// ---------------------------------------------------------------------------
+// Funnel
+// ---------------------------------------------------------------------------
+
+type FunnelQuery struct {
+	WorkspaceID   string
+	Steps         []string
+	From          time.Time
+	To            time.Time
+	WindowSeconds uint32
+}
+
+func (r *AnalyticsRepo) Funnel(ctx context.Context, q FunnelQuery) (*model.QueryResult, error) {
+	if len(q.Steps) < 2 {
+		return nil, fmt.Errorf("funnel requires at least 2 steps")
+	}
+
+	type stepTpl struct {
+		Index int
+		Last  bool
+	}
+	stepsTpl := make([]stepTpl, len(q.Steps))
+	for i := range q.Steps {
+		stepsTpl[i] = stepTpl{Index: i, Last: i == len(q.Steps)-1}
+	}
+
+	sql, err := r.tpl.Render("funnel_analysis.sql.tmpl", map[string]any{
+		"Steps":     stepsTpl,
+		"StepCount": len(q.Steps),
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	args := []any{
+		clickhouse.Named("workspace_id", q.WorkspaceID),
+		clickhouse.DateNamed("from", q.From, clickhouse.MilliSeconds),
+		clickhouse.DateNamed("to", q.To, clickhouse.MilliSeconds),
+		clickhouse.Named("window_seconds", q.WindowSeconds),
+	}
+	for i, name := range q.Steps {
+		args = append(args, clickhouse.Named(fmt.Sprintf("step%d", i), name))
+	}
+
+	rows, err := r.ch.Query(ctx, sql, args...)
+	if err != nil {
+		return nil, fmt.Errorf("clickhouse funnel: %w", err)
+	}
+	defer rows.Close()
+	return ScanRows(rows)
+}
+
+// ---------------------------------------------------------------------------
+// Retention
+// ---------------------------------------------------------------------------
+
+type RetentionQuery struct {
+	WorkspaceID  string
+	InitialEvent string
+	ReturnEvent  string
+	From         time.Time
+	To           time.Time
+	Periods      int // e.g. 14 => D0..D13
+}
+
+func (r *AnalyticsRepo) Retention(ctx context.Context, q RetentionQuery) (*model.QueryResult, error) {
+	if q.Periods < 1 {
+		q.Periods = 14
+	}
+	type periodTpl struct {
+		RIndex    int
+		OffsetDay int
+		Last      bool
+	}
+	outer := make([]periodTpl, q.Periods)
+	for i := 0; i < q.Periods; i++ {
+		outer[i] = periodTpl{RIndex: i + 2, OffsetDay: i + 1, Last: i == q.Periods-1}
+	}
+
+	sql, err := r.tpl.Render("retention_cohort.sql.tmpl", map[string]any{
+		"Outer": outer,
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	rows, err := r.ch.Query(ctx, sql,
+		clickhouse.Named("workspace_id", q.WorkspaceID),
+		clickhouse.DateNamed("from", q.From, clickhouse.MilliSeconds),
+		clickhouse.DateNamed("to", q.To, clickhouse.MilliSeconds),
+		clickhouse.Named("initial_event", q.InitialEvent),
+		clickhouse.Named("return_event", q.ReturnEvent),
+	)
+	if err != nil {
+		return nil, fmt.Errorf("clickhouse retention: %w", err)
+	}
+	defer rows.Close()
+	return ScanRows(rows)
+}
+
+// ---------------------------------------------------------------------------
+// Session
+// ---------------------------------------------------------------------------
+
+type SessionQuery struct {
+	WorkspaceID    string
+	UserID         string // optional
+	From           time.Time
+	To             time.Time
+	TimeoutSeconds uint32
+	Limit          int
+	Offset         int
+}
+
+func (r *AnalyticsRepo) Sessions(ctx context.Context, q SessionQuery) (*model.QueryResult, error) {
+	sql, err := r.tpl.Render("session_analysis.sql.tmpl", map[string]any{
+		"HasUserID": q.UserID != "",
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	args := []any{
+		clickhouse.Named("workspace_id", q.WorkspaceID),
+		clickhouse.DateNamed("from", q.From, clickhouse.MilliSeconds),
+		clickhouse.DateNamed("to", q.To, clickhouse.MilliSeconds),
+		clickhouse.Named("timeout_seconds", q.TimeoutSeconds),
+		clickhouse.Named("limit", uint32(q.Limit)),
+		clickhouse.Named("offset", uint32(q.Offset)),
+	}
+	if q.UserID != "" {
+		args = append(args, clickhouse.Named("user_id", q.UserID))
+	}
+
+	rows, err := r.ch.Query(ctx, sql, args...)
+	if err != nil {
+		return nil, fmt.Errorf("clickhouse session: %w", err)
+	}
+	defer rows.Close()
+	return ScanRows(rows)
+}