finance-duck/internal/analytics/query.go

package analytics

import (
	"context"
	"database/sql"
	"fmt"
	"time"
)

// filteredPrefix opens the common table expression every group query reads
// from; the closing parenthesis is supplied with the WHERE clause.
const filteredPrefix = "WITH filtered AS (SELECT t.* FROM transactions t WHERE "

// categoryGroups runs twice, once per compared interval, so a period-over-period
// delta sees exactly the same ancestor rollup on both sides.
const categoryGroups = `SELECT c.id, c.name, t.currency, '', CAST(SUM(t.amount) AS VARCHAR), COUNT(*)
  FROM filtered t JOIN category_ancestors ca ON ca.category_id = t.category_id
  JOIN categories c ON c.id = ca.ancestor_id GROUP BY c.id, c.name, t.currency ORDER BY c.id, t.currency`

func (s *Store) Query(ctx context.Context, filter Filter) (Dashboard, error) {
	empty := Dashboard{
		Totals: []Total{}, Previous: []Total{}, Monthly: []MonthlyPoint{},
		Categories: []Group{}, PreviousCategories: []Group{}, Tags: []Group{},
		Merchants: []Group{}, Accounts: []Group{}, Recurring: []Group{}, Largest: []Group{},
	}
	if err := filter.validate(); err != nil {
		return empty, err
	}
	tx, err := s.db.BeginTx(ctx, nil)
	if err != nil {
		return empty, err
	}
	defer tx.Rollback()
	result := empty
	if result.Totals, err = queryTotals(ctx, tx, filter); err != nil {
		return empty, err
	}
	previous, ok, err := previousFilter(ctx, tx, filter)
	if err != nil {
		return empty, err
	}
	if ok {
		if result.Previous, err = queryTotals(ctx, tx, previous); err != nil {
			return empty, err
		}
		where, args := previous.where()
		if result.PreviousCategories, err = queryGroups(ctx, tx, filteredPrefix+where+") "+categoryGroups, args); err != nil {
			return empty, fmt.Errorf("query previous categories: %w", err)
		}
	}
	where, args := filter.where()
	prefix := filteredPrefix + where + ") "
	if result.Monthly, err = queryMonthly(ctx, tx, prefix, args); err != nil {
		return empty, err
	}
	queries := []struct {
		output *[]Group
		query  string
	}{
		{&result.Categories, categoryGroups},
		{&result.Tags, `SELECT tag.id, tag.name, t.currency, '', CAST(SUM(t.amount) AS VARCHAR), COUNT(*)
   FROM filtered t JOIN transaction_tags tt ON tt.transaction_id = t.id
   JOIN tags tag ON tag.id = tt.tag_id GROUP BY tag.id, tag.name, t.currency ORDER BY tag.id, t.currency`},
		{&result.Merchants, `SELECT t.merchant_id, COALESCE(m.name, 'No merchant'), t.currency, '', CAST(SUM(t.amount) AS VARCHAR), COUNT(*)
   FROM filtered t LEFT JOIN merchants m ON m.id = t.merchant_id
   GROUP BY t.merchant_id, m.name, t.currency ORDER BY t.merchant_id, t.currency`},
		{&result.Accounts, `SELECT a.id, a.display_name, t.currency, '', CAST(SUM(t.amount) AS VARCHAR), COUNT(*)
   FROM filtered t JOIN accounts a ON a.id = t.account_id
   GROUP BY a.id, a.display_name, t.currency ORDER BY a.id, t.currency`},
		{&result.Recurring, `, spaced AS (
   SELECT *, date_diff('day', LAG(booking_date) OVER (
    PARTITION BY merchant_id, account_id, currency, amount ORDER BY booking_date, id), booking_date) AS gap
   FROM filtered WHERE amount < 0 AND merchant_id <> ''
  ), candidates AS (
   SELECT merchant_id, account_id, currency, amount, SUM(amount) AS total, COUNT(*) AS occurrences,
    CASE WHEN MIN(gap) >= 5 AND MAX(gap) <= 9 THEN 'weekly'
     WHEN MIN(gap) >= 26 AND MAX(gap) <= 35 THEN 'monthly'
     WHEN MIN(gap) >= 350 AND MAX(gap) <= 380 THEN 'yearly' ELSE '' END AS cadence
   FROM spaced GROUP BY merchant_id, account_id, currency, amount
   HAVING COUNT(*) >= 3
  )
   SELECT c.merchant_id || ':' || c.account_id || ':' || CAST(c.amount AS VARCHAR),
    m.name, c.currency, c.cadence, CAST(c.total AS VARCHAR), c.occurrences
   FROM candidates c JOIN merchants m ON m.id = c.merchant_id
   WHERE c.cadence <> '' ORDER BY 1, 3`},
		// One row per payee: a rent paid on time every month is six identical
		// rows that explain nothing, so only a merchant's single biggest payment
		// competes. Ranked per currency rather than by a plain LIMIT, so one
		// busy currency cannot crowd another out of its own list. Both windows
		// order by the decimal column, never by its VARCHAR rendering.
		{&result.Largest, `, payments AS (
   SELECT t.id, CASE WHEN COALESCE(m.name, '') <> '' THEN m.name ELSE t.raw_description END AS label,
    t.currency, CAST(t.booking_date AS VARCHAR) AS day, t.amount AS value,
    ROW_NUMBER() OVER (PARTITION BY t.currency,
     CASE WHEN t.merchant_id <> '' THEN 'm:' || t.merchant_id ELSE 'x:' || t.id END
     ORDER BY t.amount, t.id) AS repeats
   FROM filtered t LEFT JOIN merchants m ON m.id = t.merchant_id WHERE t.amount < 0
  ), ranked AS (
   SELECT id, label, currency, day, value,
    ROW_NUMBER() OVER (PARTITION BY currency ORDER BY value, id) AS position
   FROM payments WHERE repeats = 1
  )
   SELECT id, label, currency, day, CAST(value AS VARCHAR), CAST(1 AS BIGINT) FROM ranked
   WHERE position <= 8 ORDER BY currency, position`},
	}
	for _, item := range queries {
		groups, err := queryGroups(ctx, tx, prefix+item.query, args)
		if err != nil {
			return empty, fmt.Errorf("query analytics groups: %w", err)
		}
		*item.output = groups
	}
	if err := tx.Commit(); err != nil {
		return empty, err
	}
	return result, nil
}

// queryMonthly returns one row per month and currency. Months with no activity
// are absent: the caller knows the requested window and fills the gaps.
func queryMonthly(ctx context.Context, tx *sql.Tx, prefix string, args []any) ([]MonthlyPoint, error) {
	rows, err := tx.QueryContext(ctx, prefix+`SELECT strftime(booking_date, '%Y-%m'), currency,
  CAST(SUM(CASE WHEN amount > 0 THEN amount ELSE CAST(0 AS DECIMAL(24,4)) END) AS VARCHAR),
  CAST(SUM(CASE WHEN amount < 0 THEN -amount ELSE CAST(0 AS DECIMAL(24,4)) END) AS VARCHAR),
  CAST(SUM(amount) AS VARCHAR), COUNT(*)
  FROM filtered GROUP BY currency, strftime(booking_date, '%Y-%m') ORDER BY currency, 1`, args...)
	if err != nil {
		return nil, fmt.Errorf("query analytics months: %w", err)
	}
	defer rows.Close()
	result := []MonthlyPoint{}
	for rows.Next() {
		var point MonthlyPoint
		if err := rows.Scan(&point.Period, &point.Currency, &point.Income, &point.Expenses, &point.Net, &point.Count); err != nil {
			return nil, err
		}
		result = append(result, point)
	}
	return result, rows.Err()
}

func queryTotals(ctx context.Context, tx *sql.Tx, filter Filter) ([]Total, error) {
	where, args := filter.where()
	rows, err := tx.QueryContext(ctx, `SELECT t.currency,
  CAST(SUM(CASE WHEN t.amount < 0 THEN -t.amount ELSE CAST(0 AS DECIMAL(24,4)) END) AS VARCHAR),
  CAST(SUM(CASE WHEN t.amount > 0 THEN t.amount ELSE CAST(0 AS DECIMAL(24,4)) END) AS VARCHAR),
  CAST(SUM(t.amount) AS VARCHAR)
  FROM transactions t WHERE `+where+` GROUP BY t.currency ORDER BY t.currency`, args...)
	if err != nil {
		return nil, fmt.Errorf("query analytics totals: %w", err)
	}
	defer rows.Close()
	result := []Total{}
	for rows.Next() {
		var total Total
		if err := rows.Scan(&total.Currency, &total.Expenses, &total.Income, &total.Net); err != nil {
			return nil, err
		}
		result = append(result, total)
	}
	return result, rows.Err()
}

func queryGroups(ctx context.Context, tx *sql.Tx, query string, args []any) ([]Group, error) {
	rows, err := tx.QueryContext(ctx, query, args...)
	if err != nil {
		return nil, err
	}
	defer rows.Close()
	result := []Group{}
	for rows.Next() {
		var group Group
		if err := rows.Scan(&group.ID, &group.Name, &group.Currency, &group.Period, &group.Amount, &group.Count); err != nil {
			return nil, err
		}
		result = append(result, group)
	}
	return result, rows.Err()
}

// Previous is the immediately preceding inclusive interval of equal length.
// For one-sided filters, the missing boundary comes from the matching dataset,
// not the wall clock. All-time queries intentionally have no previous period.
func previousFilter(ctx context.Context, tx *sql.Tx, filter Filter) (Filter, bool, error) {
	if filter.From == "" && filter.To == "" {
		return filter, false, nil
	}
	if filter.From == "" || filter.To == "" {
		bounds := filter
		bounds.From, bounds.To = "", ""
		where, args := bounds.where()
		var first, last sql.NullString
		if err := tx.QueryRowContext(ctx, "SELECT CAST(MIN(t.booking_date) AS VARCHAR), CAST(MAX(t.booking_date) AS VARCHAR) FROM transactions t WHERE "+where, args...).Scan(&first, &last); err != nil {
			return filter, false, err
		}
		if !first.Valid || !last.Valid {
			return filter, false, nil
		}
		if filter.From == "" {
			filter.From = first.String
		}
		if filter.To == "" {
			filter.To = last.String
		}
		if filter.From > filter.To {
			return filter, false, nil
		}
	}
	from, err := time.Parse(time.DateOnly, filter.From)
	if err != nil {
		return filter, false, err
	}
	to, err := time.Parse(time.DateOnly, filter.To)
	if err != nil {
		return filter, false, err
	}
	days := int((to.Unix()-from.Unix())/86400) + 1
	previousFrom, previousTo := from.AddDate(0, 0, -days), from.AddDate(0, 0, -1)
	// ISO date filters cannot describe dates before year zero.
	if previousFrom.Year() < 0 {
		return filter, false, fmt.Errorf("previous period falls outside supported date range")
	}
	filter.From, filter.To = previousFrom.Format(time.DateOnly), previousTo.Format(time.DateOnly)
	return filter, true, nil
}