Files
money/internal/store/store.go
T
nikolaandClaude Opus 5 ebf7770569 Pair transfers from rules.toml
The boolean transfer flag went two commits ago because a one-sided verdict let
half a movement vanish and left the report unbalanced. This is what replaces
it: a [[transfer]] block names both legs, and only a matched pair is dropped
from the report -- both legs together, never one.

Legs pair within five days, nearest date first, and a transaction belongs to at
most one transfer, so the first definition to claim a leg keeps it, exactly as
the first matching rule keeps a tag. The pairing is derived state like the tags:
Engine.Link rewrites the whole transfers table from rules.toml, which is why
retag re-derives both halves of what that file decides, and why it runs over
the whole index rather than a filtered view -- pairing inside one would let a
movement count as a transfer in one report and not in another. An unmatched leg
is not a transfer and keeps counting, surfaced as a warning instead.

Within one currency the amount is the evidence and must be the exact opposite.
Across currencies it is not checked at all: there are no rates here, so the two
numbers are unrelated and the dates carry the pairing alone.

tolerance_pct is the one exception, per definition, for a route where the bank
takes a fee and the two statements genuinely disagree. It defaults to zero and
belongs on the one definition that charges; a global or default tolerance would
loosen every route that does not. The difference it admits is not forgiven --
the pair leaves the report entirely, so a fee hidden inside one would be
spending that appears nowhere. Pair.Fee is what left less what arrived, and
report.Excluded carries it out per currency alongside the legs. It counts only
pairs whose legs are both in view, for the same reason it counts legs and not
transfers: half a pair cannot say what the other half received.

The screens:

- 6 builds a definition against the index as you type, showing the pairs it
  would form and the legs it would catch but leave unpaired. Six fields need
  more room than the rule builder's four, so the form sheds its spacing, then
  its hints, then the borders on unfocused fields.
- 7 lists every definition with what it pairs. Two counts, because they mean
  different things: an unpaired leg is a definition doing something and not
  finishing it, no pairs at all is dead weight. Tol names the tolerance, blank
  where amounts must agree.
- 3 grows a (transfers) row under TOTAL, and a fees row beneath it, or the
  report silently disagrees with the account balances.

Two things that are not part of transfers but are the same day's work:

- ls --uniq lists each account and description once, normalised the way a glob
  sees them, which is the shape of "what still needs a rule?" -- fifty visits
  to one shop are one pattern to write, not fifty rows to read.
- The rule builder's preview now filters to what the glob matches instead of
  marking matches in a full list. The count carries the context the rows no
  longer can: 2 of 7, measured against everything still in view.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
2026-08-15 19:19:12 +02:00

380 lines
11 KiB
Go

// Package store is the SQLite index over the statements. It is entirely
// rebuildable: delete index.db and re-import to get it back. Nothing lives
// only here.
package store
import (
"database/sql"
"fmt"
"os"
"path/filepath"
"strings"
_ "modernc.org/sqlite"
"git.petrovv.com/nikola/money/internal/model"
)
// DB wraps the SQLite handle.
type DB struct {
sql *sql.DB
}
const schema = `
PRAGMA foreign_keys = ON;
CREATE TABLE IF NOT EXISTS accounts (
id INTEGER PRIMARY KEY AUTOINCREMENT,
slug TEXT NOT NULL UNIQUE,
name TEXT NOT NULL,
currency TEXT NOT NULL,
minor_digits INTEGER NOT NULL DEFAULT 2
);
CREATE TABLE IF NOT EXISTS source_files (
id INTEGER PRIMARY KEY AUTOINCREMENT,
account_id INTEGER NOT NULL REFERENCES accounts(id),
path TEXT NOT NULL,
sha256 TEXT NOT NULL,
imported_at TEXT NOT NULL,
UNIQUE(account_id, path)
);
CREATE TABLE IF NOT EXISTS transactions (
id INTEGER PRIMARY KEY AUTOINCREMENT,
account_id INTEGER NOT NULL REFERENCES accounts(id),
source_file_id INTEGER NOT NULL REFERENCES source_files(id),
fingerprint TEXT NOT NULL,
date TEXT NOT NULL,
description TEXT NOT NULL,
amount_minor INTEGER NOT NULL,
type TEXT NOT NULL DEFAULT '',
balance_minor INTEGER,
rule_tag TEXT,
UNIQUE(account_id, fingerprint)
);
CREATE INDEX IF NOT EXISTS idx_txn_date ON transactions(date);
CREATE INDEX IF NOT EXISTS idx_txn_account ON transactions(account_id);
-- One row per matched movement between the user's own accounts, derived from
-- the [[transfer]] blocks in rules.toml. A leg belongs to at most one transfer,
-- which UNIQUE enforces rather than trusting the pairing to be well behaved.
CREATE TABLE IF NOT EXISTS transfers (
id INTEGER PRIMARY KEY AUTOINCREMENT,
def_index INTEGER NOT NULL,
out_txn_id INTEGER NOT NULL UNIQUE REFERENCES transactions(id),
in_txn_id INTEGER NOT NULL UNIQUE REFERENCES transactions(id)
);
`
// Open opens (creating if needed) the index at path.
//
// There is no schema migration: the index caches what the statements and
// rules.toml already say, so a build that changes the schema is answered by
// deleting index.db and importing again, not by patching the old one in place.
// Until it is deleted, queries against it fail on the columns it lacks.
func Open(path string) (*DB, error) {
if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
return nil, fmt.Errorf("create index dir: %w", err)
}
sqlDB, err := sql.Open("sqlite", path)
if err != nil {
return nil, fmt.Errorf("open index %s: %w", path, err)
}
if _, err := sqlDB.Exec(schema); err != nil {
sqlDB.Close()
return nil, fmt.Errorf("apply schema: %w", err)
}
return &DB{sql: sqlDB}, nil
}
// Close releases the underlying handle.
func (d *DB) Close() error { return d.sql.Close() }
// UpsertAccount inserts or updates an account by slug and returns its id.
func (d *DB) UpsertAccount(a model.Account) (int64, error) {
_, err := d.sql.Exec(`
INSERT INTO accounts (slug, name, currency, minor_digits)
VALUES (?, ?, ?, ?)
ON CONFLICT(slug) DO UPDATE SET
name = excluded.name,
currency = excluded.currency,
minor_digits = excluded.minor_digits`,
a.Slug, a.Name, a.Currency, a.MinorDigits)
if err != nil {
return 0, fmt.Errorf("upsert account %s: %w", a.Slug, err)
}
var id int64
if err := d.sql.QueryRow(`SELECT id FROM accounts WHERE slug = ?`, a.Slug).Scan(&id); err != nil {
return 0, fmt.Errorf("read account id %s: %w", a.Slug, err)
}
return id, nil
}
// Accounts lists every known account, ordered by slug.
func (d *DB) Accounts() ([]model.Account, error) {
rows, err := d.sql.Query(`SELECT id, slug, name, currency, minor_digits FROM accounts ORDER BY slug`)
if err != nil {
return nil, fmt.Errorf("list accounts: %w", err)
}
defer rows.Close()
var out []model.Account
for rows.Next() {
var a model.Account
if err := rows.Scan(&a.ID, &a.Slug, &a.Name, &a.Currency, &a.MinorDigits); err != nil {
return nil, err
}
out = append(out, a)
}
return out, rows.Err()
}
// SourceFile records that a statement file was imported, returning its id.
func (d *DB) SourceFile(accountID int64, path, sha, importedAt string) (int64, error) {
_, err := d.sql.Exec(`
INSERT INTO source_files (account_id, path, sha256, imported_at)
VALUES (?, ?, ?, ?)
ON CONFLICT(account_id, path) DO UPDATE SET
sha256 = excluded.sha256,
imported_at = excluded.imported_at`,
accountID, path, sha, importedAt)
if err != nil {
return 0, fmt.Errorf("record source file %s: %w", path, err)
}
var id int64
if err := d.sql.QueryRow(
`SELECT id FROM source_files WHERE account_id = ? AND path = ?`, accountID, path).Scan(&id); err != nil {
return 0, fmt.Errorf("read source file id %s: %w", path, err)
}
return id, nil
}
// SourceFileSHA returns the recorded checksum for a statement file, and whether
// it has been imported before.
func (d *DB) SourceFileSHA(accountID int64, path string) (string, bool, error) {
var sha string
err := d.sql.QueryRow(
`SELECT sha256 FROM source_files WHERE account_id = ? AND path = ?`, accountID, path).Scan(&sha)
if err == sql.ErrNoRows {
return "", false, nil
}
if err != nil {
return "", false, err
}
return sha, true, nil
}
// InsertTransaction adds a transaction unless its fingerprint already exists
// for that account. It reports whether a new row was created.
//
// Existing rows are deliberately left untouched, so re-importing a statement
// that overlaps one already imported adds nothing rather than duplicating it.
func (d *DB) InsertTransaction(t model.Transaction) (bool, error) {
var balance any
if t.BalanceMinor != nil {
balance = *t.BalanceMinor
}
res, err := d.sql.Exec(`
INSERT INTO transactions
(account_id, source_file_id, fingerprint, date, description, amount_minor,
type, balance_minor, rule_tag)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, NULLIF(?, ''))
ON CONFLICT(account_id, fingerprint) DO NOTHING`,
t.AccountID, t.SourceFileID, t.Fingerprint, t.Date, t.Description,
t.AmountMinor, t.Type, balance, t.RuleTag)
if err != nil {
return false, fmt.Errorf("insert transaction: %w", err)
}
n, err := res.RowsAffected()
if err != nil {
return false, err
}
return n > 0, nil
}
// Filter narrows a transaction query.
type Filter struct {
AccountSlug string
// Untagged selects the rows still waiting for a verdict: no tag, and not a
// leg of a matched transfer. A paired leg has been accounted for by the
// transfer that claimed it, so listing it as untagged would ask the user to
// write a rule for something that is already spoken for and that the report
// leaves out anyway.
Untagged bool
Month string // YYYY-MM
Search string // case-insensitive substring of the description
Limit int
}
// Transactions returns rows matching f, newest first.
func (d *DB) Transactions(f Filter) ([]model.Transaction, error) {
q := `
SELECT t.id, t.account_id, a.slug, a.currency, a.minor_digits,
t.fingerprint, t.date, t.description, t.amount_minor,
COALESCE(s.path, ''), t.type, t.balance_minor,
COALESCE(t.rule_tag, ''), x.id
FROM transactions t
JOIN accounts a ON a.id = t.account_id
LEFT JOIN source_files s ON s.id = t.source_file_id
LEFT JOIN transfers x ON x.out_txn_id = t.id OR x.in_txn_id = t.id
WHERE 1 = 1`
var args []any
if f.AccountSlug != "" {
q += ` AND a.slug = ?`
args = append(args, f.AccountSlug)
}
if f.Untagged {
q += ` AND NULLIF(t.rule_tag, '') IS NULL AND x.id IS NULL`
}
if f.Month != "" {
q += ` AND substr(t.date, 1, 7) = ?`
args = append(args, f.Month)
}
q += ` ORDER BY t.date DESC, t.id DESC`
// Search and Limit are applied in Go: SQLite's upper()/LIKE fold ASCII
// only, which would silently fail on Cyrillic statement descriptions.
rows, err := d.sql.Query(q, args...)
if err != nil {
return nil, fmt.Errorf("query transactions: %w", err)
}
defer rows.Close()
needle := model.NormalizeDescription(f.Search)
var out []model.Transaction
for rows.Next() {
var (
t model.Transaction
balance sql.NullInt64
transfer sql.NullInt64
)
if err := rows.Scan(&t.ID, &t.AccountID, &t.AccountSlug, &t.Currency, &t.MinorDigits,
&t.Fingerprint, &t.Date, &t.Description, &t.AmountMinor, &t.SourcePath,
&t.Type, &balance, &t.RuleTag, &transfer); err != nil {
return nil, err
}
if balance.Valid {
v := balance.Int64
t.BalanceMinor = &v
}
if transfer.Valid {
v := transfer.Int64
t.TransferID = &v
}
if needle != "" && !strings.Contains(model.NormalizeDescription(t.Description), needle) {
continue
}
out = append(out, t)
if f.Limit > 0 && len(out) >= f.Limit {
break
}
}
return out, rows.Err()
}
// RuleAssignment is one row's recomputed rule verdict.
type RuleAssignment struct {
ID int64
Tag string
}
// ApplyRuleResults rewrites rule_tag for every listed row in a single
// transaction. The manual column is never touched.
func (d *DB) ApplyRuleResults(rs []RuleAssignment) error {
tx, err := d.sql.Begin()
if err != nil {
return err
}
defer tx.Rollback()
stmt, err := tx.Prepare(
`UPDATE transactions SET rule_tag = NULLIF(?, '') WHERE id = ?`)
if err != nil {
return err
}
defer stmt.Close()
for _, r := range rs {
if _, err := stmt.Exec(r.Tag, r.ID); err != nil {
return fmt.Errorf("apply rules to txn %d: %w", r.ID, err)
}
}
return tx.Commit()
}
// TransferLink is one matched pair, as decided by the transfer definitions.
type TransferLink struct {
DefIndex int
OutID int64
InID int64
}
// ReplaceTransfers rewrites the whole pairing in one transaction. Like the
// tags, it is derived from a file the user edits, so it is replaced wholesale
// rather than patched: a definition removed from rules.toml must take its pairs
// with it.
func (d *DB) ReplaceTransfers(links []TransferLink) error {
tx, err := d.sql.Begin()
if err != nil {
return err
}
defer tx.Rollback()
if _, err := tx.Exec(`DELETE FROM transfers`); err != nil {
return fmt.Errorf("clear transfers: %w", err)
}
stmt, err := tx.Prepare(
`INSERT INTO transfers (def_index, out_txn_id, in_txn_id) VALUES (?, ?, ?)`)
if err != nil {
return err
}
defer stmt.Close()
for _, l := range links {
if _, err := stmt.Exec(l.DefIndex, l.OutID, l.InID); err != nil {
return fmt.Errorf("link transfer %d→%d: %w", l.OutID, l.InID, err)
}
}
return tx.Commit()
}
// Balance sums every transaction in an account.
func (d *DB) Balance(accountID int64) (int64, error) {
var v sql.NullInt64
err := d.sql.QueryRow(
`SELECT SUM(amount_minor) FROM transactions WHERE account_id = ?`, accountID).Scan(&v)
if err != nil {
return 0, err
}
return v.Int64, nil
}
// Count returns the number of transactions in an account.
func (d *DB) Count(accountID int64) (int, error) {
var n int
err := d.sql.QueryRow(
`SELECT COUNT(*) FROM transactions WHERE account_id = ?`, accountID).Scan(&n)
return n, err
}
// Tags lists every tag in use, for completion in the TUI.
func (d *DB) Tags() ([]string, error) {
rows, err := d.sql.Query(`
SELECT DISTINCT rule_tag FROM transactions
WHERE rule_tag IS NOT NULL AND rule_tag != '' ORDER BY rule_tag`)
if err != nil {
return nil, err
}
defer rows.Close()
var out []string
for rows.Next() {
var s string
if err := rows.Scan(&s); err != nil {
return nil, err
}
out = append(out, s)
}
return out, rows.Err()
}