merge: categorizer refactor — :MCI: merchant recovery + real-merchant rules

This commit is contained in:
Munkherdene 2026-08-23 00:49:07 +08:00
commit 9e5d094ca1
5 changed files with 526 additions and 67 deletions

View file

@ -4,11 +4,13 @@ import { useEffect, useMemo, useState } from "react";
import { useRouter } from "next/navigation"; import { useRouter } from "next/navigation";
import { useCategories, useTransactions, todayLocalDate } from "@/api/hooks/reads"; import { useCategories, useTransactions, todayLocalDate } from "@/api/hooks/reads";
import { useCategorize } from "@/api/hooks/mutations"; import { useCategorize } from "@/api/hooks/mutations";
import type { Txn } from "@/api/schemas"; import type { Category, Txn } from "@/api/schemas";
import { Card, EmptyState, IconChip, MercuryButton } from "@/ds"; import { Card, EmptyState, IconChip, MercuryButton } from "@/ds";
import { categoryStyle } from "@/ds/categoryStyle"; import { categoryStyle } from "@/ds/categoryStyle";
import { Icon } from "@/ds/icons"; import { Icon } from "@/ds/icons";
import { MASKED, tugrikRaw } from "@/ds/money"; import { MASKED, tugrikRaw } from "@/ds/money";
import { CategorizeSheet } from "./CategorizeSheet";
import { suggestCategory, extractMerchant, looksLikeCardNumber } from "./suggestCategory";
import { accountingStrings as s } from "./strings"; import { accountingStrings as s } from "./strings";
import { useHiddenAmounts } from "./useHiddenAmounts"; import { useHiddenAmounts } from "./useHiddenAmounts";
@ -26,6 +28,17 @@ function threeMonthsAgo(base: Date = new Date()): Date {
return new Date(base.getFullYear(), base.getMonth() - 3, base.getDate()); return new Date(base.getFullYear(), base.getMonth() - 3, base.getDate());
} }
/** Spending transactions worth reviewing: excludes auto-detected salary
* deposits, any income, and inter-account transfers those aren't expenses
* to categorize. Shared by the uncategorized queue and the coverage
* denominator so both count the same population. */
function isReviewableSpend(t: Txn): boolean {
if (t.salary === true) return false;
if (t.direction === "income") return false;
if (t.transfer === true) return false;
return Boolean(t.matchKey || t.title);
}
/** Groups uncategorized transactions by matchKey, biggest total spend first /** Groups uncategorized transactions by matchKey, biggest total spend first
* a handful of taps then covers most of the uncategorized money instead of * a handful of taps then covers most of the uncategorized money instead of
* burning through many trivial merchants. Ports the grouping in * burning through many trivial merchants. Ports the grouping in
@ -34,6 +47,7 @@ function buildQueue(txns: Txn[]): ReviewItem[] {
const groups = new Map<string, ReviewItem>(); const groups = new Map<string, ReviewItem>();
for (const t of txns) { for (const t of txns) {
if (t.category) continue; // defensive — the fetch already scopes to Uncategorized if (t.category) continue; // defensive — the fetch already scopes to Uncategorized
if (!isReviewableSpend(t)) continue;
const key = t.matchKey || t.title; const key = t.matchKey || t.title;
if (!key) continue; if (!key) continue;
const amount = parseFloat(t.amount) || 0; const amount = parseFloat(t.amount) || 0;
@ -54,11 +68,166 @@ function buildQueue(txns: Txn[]): ReviewItem[] {
return Array.from(groups.values()).sort((a, b) => b.total - a.total); return Array.from(groups.values()).sort((a, b) => b.total - a.total);
} }
// Masked card strings like "554835******6886:13-08-2026 09:41:22" carry a
// trailing timestamp and no real merchant name — collapse them to a short
// "card ending in ####" label instead of dumping the raw string on the row.
const CARD_LIKE = /^(\d{3,})\*{2,}(\d{2,})/;
function prettyMerchant(raw: string): string {
const name = extractMerchant(raw); // recovers "ANTHROPIC"/"TAOBAO" from card strings
if (name && !looksLikeCardNumber(name)) return name;
const head = (raw.split(":")[0] ?? raw).trim();
const m = head.match(CARD_LIKE);
return m ? `Карт •••• ${m[2]}` : head || raw;
}
interface Row extends ReviewItem {
displayName: string;
suggestion: string | null;
}
function toRows(queue: ReviewItem[], categories: Category[]): Row[] {
return queue.map((item) => ({
...item,
displayName: prettyMerchant(item.merchant),
suggestion: suggestCategory(item.merchant, categories),
}));
}
const checkSvg = (
<svg viewBox="0 0 24 24" width={18} height={18} fill="none" stroke="currentColor" strokeWidth={2.5} strokeLinecap="round" strokeLinejoin="round">
<path d="M5 12.5l4.5 4.5L19 7" />
</svg>
);
function ProgressBar({ percent }: { percent: number }) {
return (
<div
style={{
height: 6,
borderRadius: 999,
background: "var(--seed-color-bg-neutral-subtle, #eef0f2)",
overflow: "hidden",
}}
>
<div
style={{
height: "100%",
width: `${Math.max(0, Math.min(100, percent))}%`,
borderRadius: 999,
background: "var(--mercury-success, #1e9e6b)",
transition: "width 300ms ease",
}}
/>
</div>
);
}
function ReviewRow({
row,
hidden,
busy,
removing,
onConfirm,
onPick,
}: {
row: Row;
hidden: boolean;
busy: boolean;
removing: boolean;
onConfirm: (row: Row) => void;
onPick: (row: Row) => void;
}) {
const chip = row.suggestion ? categoryStyle(row.suggestion, false) : null;
return (
<div
style={{
display: "flex",
alignItems: "center",
justifyContent: "space-between",
gap: 12,
padding: "10px 0",
opacity: removing ? 0 : 1,
transform: removing ? "translateX(6px)" : "translateX(0)",
transition: "opacity 220ms ease, transform 220ms ease",
}}
>
<div style={{ display: "flex", alignItems: "center", gap: 12, minWidth: 0 }}>
<IconChip
icon={chip?.icon ?? "receipt"}
tint={chip?.tint ?? "var(--seed-color-bg-neutral-subtle, #eef0f2)"}
fg={chip?.fg ?? "var(--seed-color-fg-neutral-muted, #8b8b8b)"}
/>
<div style={{ display: "flex", flexDirection: "column", gap: 2, minWidth: 0 }}>
<span style={{ fontSize: 15, fontWeight: 700, overflow: "hidden", textOverflow: "ellipsis", whiteSpace: "nowrap" }}>
{row.displayName}
</span>
<span style={{ fontSize: 12, color: "var(--seed-color-fg-neutral-muted, #8b8b8b)" }}>
×{row.count} · {hidden ? MASKED : tugrikRaw(row.total)}
</span>
</div>
</div>
<div style={{ display: "flex", alignItems: "center", gap: 8, flexShrink: 0 }}>
<button
type="button"
onClick={() => onPick(row)}
disabled={busy}
style={{
all: "unset",
cursor: busy ? "default" : "pointer",
display: "flex",
alignItems: "center",
gap: 6,
padding: "8px 12px",
borderRadius: 999,
fontSize: 13,
fontWeight: 600,
opacity: busy ? 0.6 : 1,
background: chip?.tint ?? "var(--seed-color-bg-neutral-subtle, #eef0f2)",
color: chip?.fg ?? "var(--seed-color-fg-neutral-muted, #8b8b8b)",
}}
>
<Icon name={chip?.icon ?? "list"} size={14} />
{chip ? chip.name : s.review.pickCategory}
</button>
{row.suggestion ? (
<button
type="button"
aria-label={s.review.confirm}
onClick={() => onConfirm(row)}
disabled={busy}
style={{
all: "unset",
cursor: busy ? "default" : "pointer",
display: "grid",
placeItems: "center",
width: 32,
height: 32,
borderRadius: "50%",
opacity: busy ? 0.6 : 1,
background: "var(--mercury-success, #1e9e6b)",
color: "#fff",
}}
>
{checkSvg}
</button>
) : null}
</div>
</div>
);
}
/** /**
* Full-screen categorize-review flow at `/accounting/review` (ports * Batch categorize-review screen at `/accounting/review`: every uncategorized
* `CategorizeReviewView.swift`): groups the last three months' uncategorized * merchant from the last three months, biggest spend first, in one scannable
* transactions by merchant, biggest spend first, and asks the user to assign * list with an auto-suggested category per row. Confirming a suggestion (or
* or skip a category one merchant at a time. * picking a different category) posts a retroactive rule via `useCategorize`
* that categorizes all matching past + future transactions server-side the
* row then animates out. "Бүгдийг санал болгосноор ангилах" applies every
* confident suggestion in one pass. Replaces the old one-merchant-at-a-time
* `CategorizeReviewView.swift` port.
*/ */
export function CategorizeReview() { export function CategorizeReview() {
const router = useRouter(); const router = useRouter();
@ -67,40 +236,84 @@ export function CategorizeReview() {
return { from: todayLocalDate(threeMonthsAgo(now)), to: todayLocalDate(now) }; return { from: todayLocalDate(threeMonthsAgo(now)), to: todayLocalDate(now) };
}, []); }, []);
const { data, isLoading } = useTransactions({ from, to, category: "Uncategorized", limit: 500 }); const { data, isLoading } = useTransactions({ from, to, category: "Uncategorized", limit: 500 });
const { data: categories = [] } = useCategories(); // Same 3-month window, unfiltered — only used as the coverage bar's fixed
// denominator (total reviewable spend transactions), not re-rendered as a list.
const { data: allData } = useTransactions({ from, to, limit: 500 });
const { data: categoriesData } = useCategories();
const categories = categoriesData ?? [];
const categorize = useCategorize(); const categorize = useCategorize();
const hidden = useHiddenAmounts(); const hidden = useHiddenAmounts();
// The queue is seeded once from the fetch, then mutated locally (skip // The queue is seeded once from the fetch, then mutated locally (assign
// removes, assign removes on success) — re-deriving it from `data` on every // removes on success) — re-deriving it from `data` on every background
// background refetch (categorize invalidates the transactions cache) would // refetch (categorize invalidates the transactions cache) would otherwise
// otherwise re-insert items the user already handled in this session. // re-insert items the user already handled in this session.
const [queue, setQueue] = useState<ReviewItem[] | null>(null); const [queue, setQueue] = useState<ReviewItem[] | null>(null);
useEffect(() => { useEffect(() => {
if (data && queue === null) setQueue(buildQueue(data)); if (data && queue === null) setQueue(buildQueue(data));
}, [data, queue]); }, [data, queue]);
const [removingKeys, setRemovingKeys] = useState<Set<string>>(new Set());
const [pickerFor, setPickerFor] = useState<ReviewItem | null>(null);
const [bulkRunning, setBulkRunning] = useState(false);
const [bulkProgress, setBulkProgress] = useState({ done: 0, total: 0 });
const rows = useMemo(() => (queue ? toRows(queue, categories) : []), [queue, categories]);
const mainCategories = useMemo(() => categories.filter((c) => c.depth === 1), [categories]); const mainCategories = useMemo(() => categories.filter((c) => c.depth === 1), [categories]);
const totalSpend = useMemo(
() => (allData ? allData.filter(isReviewableSpend).length : null),
[allData],
);
const remainingCount = useMemo(() => rows.reduce((sum, r) => sum + r.count, 0), [rows]);
const percent =
totalSpend && totalSpend > 0 ? Math.round(((totalSpend - remainingCount) / totalSpend) * 100) : null;
function close() { function close() {
router.push("/accounting"); router.push("/accounting");
} }
function skip(item: ReviewItem) { function removeItem(matchKey: string) {
setQueue((q) => (q ? q.filter((i) => i.matchKey !== item.matchKey) : q)); setRemovingKeys((r) => new Set(r).add(matchKey));
setTimeout(() => {
setQueue((q) => (q ? q.filter((i) => i.matchKey !== matchKey) : q));
setRemovingKeys((r) => {
const next = new Set(r);
next.delete(matchKey);
return next;
});
}, 220);
} }
function assign(item: ReviewItem, category: string) { function assign(item: ReviewItem, category: string) {
categorize.mutate( categorize.mutate(
{ matchKey: item.matchKey, category, kind: item.direction }, { matchKey: item.matchKey, category, kind: item.direction },
{ onSuccess: () => setQueue((q) => (q ? q.filter((i) => i.matchKey !== item.matchKey) : q)) }, { onSuccess: () => removeItem(item.matchKey) },
); );
} }
const current = queue?.[0]; async function runBulk() {
const targets = rows.filter((r) => r.suggestion);
if (targets.length === 0) return;
setBulkRunning(true);
setBulkProgress({ done: 0, total: targets.length });
for (const row of targets) {
try {
await categorize.mutateAsync({ matchKey: row.matchKey, category: row.suggestion!, kind: row.direction });
removeItem(row.matchKey);
} catch {
// Leave it in the queue on failure so the user can retry it manually.
}
setBulkProgress((p) => ({ ...p, done: p.done + 1 }));
}
setBulkRunning(false);
}
const confidentCount = rows.filter((r) => r.suggestion).length;
const busy = categorize.isPending || bulkRunning;
return ( return (
<div style={{ display: "flex", flexDirection: "column", gap: 20 }}> <div style={{ display: "flex", flexDirection: "column", gap: 16 }}>
<div style={{ display: "flex", alignItems: "center", justifyContent: "space-between" }}> <div style={{ display: "flex", alignItems: "center", justifyContent: "space-between" }}>
<h1 style={{ fontSize: 18, fontWeight: 700, margin: 0 }}>{s.review.title}</h1> <h1 style={{ fontSize: 18, fontWeight: 700, margin: 0 }}>{s.review.title}</h1>
<button <button
@ -113,64 +326,60 @@ export function CategorizeReview() {
</div> </div>
{isLoading && queue === null ? ( {isLoading && queue === null ? (
<Card style={{ minHeight: 320 }} /> <Card style={{ minHeight: 200 }} />
) : current ? ( ) : rows.length > 0 ? (
<> <>
<p style={{ margin: 0, fontSize: 13, color: "var(--seed-color-fg-neutral-muted, #8b8b8b)" }}> <Card style={{ display: "flex", flexDirection: "column", gap: 8 }}>
{s.review.remaining(queue!.length)} <div style={{ display: "flex", alignItems: "baseline", justifyContent: "space-between" }}>
</p> <span style={{ fontSize: 14, fontWeight: 700 }}>{s.review.remaining(remainingCount)}</span>
{percent !== null ? (
<Card style={{ display: "flex", flexDirection: "column", alignItems: "center", gap: 8, padding: "24px 16px" }}> <span style={{ fontSize: 13, color: "var(--seed-color-fg-neutral-muted, #8b8b8b)" }}>
<IconChip icon="cart" tint="var(--seed-color-bg-neutral-subtle, #eef0f2)" fg="var(--seed-color-fg-neutral)" size={56} /> {s.review.coverage(percent)}
<span style={{ fontSize: 16, fontWeight: 700, textAlign: "center" }}>{current.merchant}</span>
<span style={{ fontSize: 12, color: "var(--seed-color-fg-neutral-muted, #8b8b8b)" }}>
{s.review.transactionCount(current.count)} · {hidden ? MASKED : tugrikRaw(current.total)}
</span> </span>
) : null}
</div>
{percent !== null ? <ProgressBar percent={percent} /> : null}
</Card> </Card>
<p style={{ margin: 0, fontSize: 14, fontWeight: 700 }}>{s.review.question}</p> <MercuryButton variant="primary" onClick={runBulk} disabled={busy || confidentCount === 0}>
{bulkRunning ? s.review.bulkProgress(bulkProgress.done, bulkProgress.total) : s.review.bulkApply}
<div style={{ display: "flex", flexWrap: "wrap", gap: 10 }}>
{mainCategories.map((cat) => {
const style = categoryStyle(cat.name, cat.kind === "income");
return (
<button
key={cat.name}
type="button"
onClick={() => assign(current, cat.name)}
disabled={categorize.isPending}
style={{
all: "unset",
cursor: categorize.isPending ? "default" : "pointer",
display: "flex",
alignItems: "center",
gap: 6,
padding: "10px 14px",
borderRadius: 999,
fontSize: 14,
opacity: categorize.isPending ? 0.6 : 1,
background: "var(--seed-color-bg-neutral-subtle, #eef0f2)",
}}
>
<Icon name={style.icon} size={16} />
{style.name}
</button>
);
})}
</div>
<MercuryButton variant="secondary" onClick={() => skip(current)}>
{s.review.skip}
</MercuryButton> </MercuryButton>
<Card style={{ display: "flex", flexDirection: "column" }}>
{rows.map((row) => (
<ReviewRow
key={row.matchKey}
row={row}
hidden={hidden}
busy={busy}
removing={removingKeys.has(row.matchKey)}
onConfirm={(r) => assign(r, r.suggestion!)}
onPick={(r) => setPickerFor(r)}
/>
))}
</Card>
</> </>
) : ( ) : (
<> <>
<EmptyState icon="calendar-check" title={s.review.done} /> <EmptyState icon="calendar-check" title={s.review.done} hint={s.review.doneHint} />
<MercuryButton variant="primary" onClick={close}> <MercuryButton variant="primary" onClick={close}>
{s.review.close} {s.review.close}
</MercuryButton> </MercuryButton>
</> </>
)} )}
<CategorizeSheet
open={pickerFor !== null}
onOpenChange={(open) => {
if (!open) setPickerFor(null);
}}
categories={mainCategories}
selected={pickerFor ? (suggestCategory(pickerFor.merchant, categories) ?? undefined) : undefined}
onSelect={(cat) => {
if (pickerFor) assign(pickerFor, cat.name);
setPickerFor(null);
}}
/>
</div> </div>
); );
} }

View file

@ -163,7 +163,15 @@ export function TransactionList() {
const categoryOptions = useMemo(() => categoriesIn(all), [all]); const categoryOptions = useMemo(() => categoriesIn(all), [all]);
const hasUncategorized = useMemo( const hasUncategorized = useMemo(
() => all.some((txn) => !txn.category && (txn.matchKey || txn.title)), () =>
all.some(
(txn) =>
!txn.category &&
txn.salary !== true &&
txn.transfer !== true &&
txn.direction !== "income" &&
(txn.matchKey || txn.title),
),
[all], [all],
); );

View file

@ -20,12 +20,16 @@ export const accountingStrings = {
review: { review: {
title: "Ангилалжуулах", title: "Ангилалжуулах",
later: "Дараа нь", later: "Дараа нь",
remaining: (n: number) => `${n} ангилалгүй худалдагч үлдлээ`, remaining: (n: number) => `${n} ангилаагүй үлдлээ`,
question: "Аль ангилалд хамаарах вэ?", coverage: (pct: number) => `${pct}% ангилагдсан`,
skip: "Алгасах",
done: "Бүгд ангилагдлаа",
close: "Хаах",
transactionCount: (n: number) => `${n} гүйлгээ`, transactionCount: (n: number) => `${n} гүйлгээ`,
bulkApply: "Бүгдийг санал болгосноор ангилах",
bulkProgress: (done: number, total: number) => `${done}/${total} ангилж байна…`,
confirm: "Баталгаажуулах",
pickCategory: "Ангилал сонгох",
done: "Бүгд ангилагдлаа",
doneHint: "Шинэ гүйлгээ ирвэл дүрмийн дагуу автоматаар ангилагдана.",
close: "Хаах",
}, },
detail: { detail: {
total: "Нийт", total: "Нийт",

View file

@ -0,0 +1,159 @@
import { describe, it, expect } from "vitest";
import { suggestCategory, extractMerchant, looksLikeCardNumber } from "./suggestCategory";
import type { Category } from "@/api/schemas";
function cat(name: string, kind: "income" | "expense" = "expense", depth = 1): Category {
return { name, kind, depth };
}
const CATEGORIES: Category[] = [
cat("Food & Drink"),
cat("Coffee"),
cat("Transport"),
cat("Shopping"),
cat("Bills & Services"),
cat("Entertainment"),
cat("Groceries"),
cat("Insurance"),
cat("Electronics"),
];
describe("extractMerchant", () => {
it("pulls the real merchant out of a card string's :MCI: segment", () => {
expect(extractMerchant("554835******6886:30-07-2026 11:02:10:MCI:ANTHROPIC 1")).toBe("ANTHROPIC");
expect(extractMerchant("554835******6886:13-08-2026 12:13:52:MCI:WWW HOSTI 5")).toBe("WWW HOSTI");
expect(extractMerchant("554835******6886:11-08-2026 10:07:35:MCI:TAOBAO CO 4")).toBe("TAOBAO CO");
});
it("falls back to the head before the first colon when there's no :MCI: segment", () => {
expect(extractMerchant("ЛИЗИНГ ХХК")).toBe("ЛИЗИНГ ХХК");
expect(extractMerchant("1234******5678")).toBe("1234******5678");
});
});
describe("suggestCategory — real card charges (:MCI: merchant extraction)", () => {
it("suggests Bills & Services for ANTHROPIC (no Software/Subscriptions category present)", () => {
expect(suggestCategory("554835******6886:30-07-2026 11:02:10:MCI:ANTHROPIC 1", CATEGORIES)).toBe(
"Bills & Services",
);
});
it("prefers Software over Bills & Services when the category exists", () => {
const withSoftware = [...CATEGORIES, cat("Software")];
expect(suggestCategory("554835******6886:30-07-2026 11:02:10:MCI:ANTHROPIC 1", withSoftware)).toBe("Software");
});
it("suggests Bills & Services for a hosting merchant (WWW HOSTI)", () => {
expect(suggestCategory("554835******6886:13-08-2026 12:13:52:MCI:WWW HOSTI 5", CATEGORIES)).toBe(
"Bills & Services",
);
});
it("suggests Shopping for TAOBAO", () => {
expect(suggestCategory("554835******6886:11-08-2026 10:07:35:MCI:TAOBAO CO 4", CATEGORIES)).toBe("Shopping");
});
});
describe("suggestCategory — local merchants", () => {
it("matches ТОКИ (non-bank lender) to Loan", () => {
const withLoan = [...CATEGORIES, cat("Loan")];
expect(suggestCategory("ТОКИ ББСБ ХХК", withLoan)).toBe("Loan");
});
it("falls back to Bills & Services when Loan doesn't exist", () => {
expect(suggestCategory("ЛИЗИНГ ХХК", CATEGORIES)).toBe("Bills & Services");
});
it("matches the Cyrillic ЗЭЭЛ keyword (word-boundary-free, Cyrillic-safe)", () => {
const withLoan = [...CATEGORIES, cat("Loan")];
expect(suggestCategory("ХААН ЗЭЭЛ ТӨЛБӨР", withLoan)).toBe("Loan");
});
it("matches TUSHIG to Groceries", () => {
expect(suggestCategory("TUSHIG SUPERMARKET", CATEGORIES)).toBe("Groceries");
});
it("matches other grocery keywords", () => {
expect(suggestCategory("NOMIN SUPERMARKET", CATEGORIES)).toBe("Groceries");
expect(suggestCategory("CU-24 CONVENIENCE", CATEGORIES)).toBe("Groceries");
expect(suggestCategory("ХҮНСНИЙ ДЭЛГҮҮР", CATEGORIES)).toBe("Groceries");
});
it("matches insurance keywords", () => {
expect(suggestCategory("MONGOL ДААТГАЛ LLC", CATEGORIES)).toBe("Insurance");
expect(suggestCategory("SOME INSURANCE CO", CATEGORIES)).toBe("Insurance");
});
it("matches electronics retailers", () => {
expect(suggestCategory("MAGIC TECH STORE", CATEGORIES)).toBe("Electronics");
expect(suggestCategory("ITOPIA MALL", CATEGORIES)).toBe("Electronics");
});
it("matches mobile carriers to Bills & Services", () => {
expect(suggestCategory("MOBICOM PAYMENT", CATEGORIES)).toBe("Bills & Services");
expect(suggestCategory("UNITEL TOP UP", CATEGORIES)).toBe("Bills & Services");
});
it("matches apparel/shopping brands", () => {
expect(suggestCategory("CONVERSE STORE", CATEGORIES)).toBe("Shopping");
expect(suggestCategory("SANT ASAR TRADE", CATEGORIES)).toBe("Shopping");
});
it("matches gaming merchants to Entertainment", () => {
expect(suggestCategory("STEAM GAMES", CATEGORIES)).toBe("Entertainment");
expect(suggestCategory("PGAMING WALLET", CATEGORIES)).toBe("Entertainment");
});
it("matches CAFE/КАФЕ to Coffee, not Food & Drink", () => {
expect(suggestCategory("КАФЕ МОДЕРН", CATEGORIES)).toBe("Coffee");
expect(suggestCategory("TOM CAFE LLC", CATEGORIES)).toBe("Coffee");
expect(suggestCategory("КОФЕ ЦЭГ", CATEGORIES)).toBe("Coffee");
expect(suggestCategory("STARBUCKS COFFEE", CATEGORIES)).toBe("Coffee");
});
it("matches restaurant/food keywords (no CAFE overlap) to Food & Drink", () => {
expect(suggestCategory("KFC ULAANBAATAR", CATEGORIES)).toBe("Food & Drink");
expect(suggestCategory("ХООЛНЫ ГАЗАР", CATEGORIES)).toBe("Food & Drink");
expect(suggestCategory("BURGER KING", CATEGORIES)).toBe("Food & Drink");
});
it("matches transport/fuel keywords", () => {
expect(suggestCategory("UBCAB TRIP", CATEGORIES)).toBe("Transport");
expect(suggestCategory("ШАТАХУУНЫ СТАНЦ", CATEGORIES)).toBe("Transport");
});
});
describe("suggestCategory — no signal", () => {
it("returns null for a masked card number with no :MCI: merchant segment", () => {
expect(suggestCategory("554835******6886:13-08-2026 09:41:22", CATEGORIES)).toBeNull();
expect(suggestCategory("1234******5678", CATEGORIES)).toBeNull();
});
it("returns null when nothing matches", () => {
expect(suggestCategory("TLJ CENTR", CATEGORIES)).toBeNull();
expect(suggestCategory("TSENGELDE", CATEGORIES)).toBeNull();
});
it("returns null when every matching rule's candidates are all absent", () => {
// The phone-carrier rule has a single candidate ("Bills & Services") with
// no fallback, unlike Loan/Insurance/Groceries/Electronics which fall
// back to a broader category — removing it leaves nothing to suggest.
const noBills = CATEGORIES.filter((c) => c.name !== "Bills & Services");
expect(suggestCategory("MOBICOM PAYMENT", noBills)).toBeNull();
});
it("returns null for empty input", () => {
expect(suggestCategory("", CATEGORIES)).toBeNull();
});
});
describe("looksLikeCardNumber", () => {
it("recognizes a masked card head as a card number", () => {
expect(looksLikeCardNumber("554835******6886")).toBe(true);
});
it("does not flag a normal merchant name", () => {
expect(looksLikeCardNumber("ANTHROPIC")).toBe(false);
expect(looksLikeCardNumber("ТОКИ ББСБ ХХК")).toBe(false);
});
});

View file

@ -0,0 +1,79 @@
import type { Category } from "@/api/schemas";
/** Card/online charges arrive as a masked-card string with the real merchant
* buried after the last `:MCI:` segment, e.g.
* `"554835******6886:30-07-2026 11:02:10:MCI:ANTHROPIC 1"` `"ANTHROPIC"`.
* Returns the cleaned merchant name (trailing sequence number stripped), or the
* head before the first colon for normal names. */
export function extractMerchant(raw: string): string {
const s = (raw ?? "").trim();
const mci = s.split(/:MCI:/i);
if (mci.length > 1) {
const m = mci[mci.length - 1].replace(/\s+\d+$/, "").trim();
if (m) return m;
}
return (s.split(":")[0] ?? s).trim();
}
/** True when the string is (still) mostly a masked card number with no merchant
* signal e.g. a card charge with no `:MCI:` merchant segment. */
export function looksLikeCardNumber(s: string): boolean {
const head = (s.split(":")[0] ?? s).trim();
if (/^\d{4,}\*{2,}/.test(head)) return true;
const digits = (s.match(/\d/g) ?? []).length;
const stars = (s.match(/\*/g) ?? []).length;
return stars >= 2 && digits / Math.max(s.length, 1) > 0.4;
}
interface Rule {
pattern: RegExp;
candidates: string[];
}
// Ordered — earlier rules win when a merchant matches more than one. Keyed to
// the REAL uncategorized merchants in the data (incl. the merchant extracted
// from card strings). Each rule's candidates are tried in order; the first that
// exists in the user's category list is used, so a specific child (Groceries,
// Loan, Electronics) falls back to its depth-1 parent when absent. Add new
// merchants here rather than growing suggestCategory()'s logic.
const RULES: Rule[] = [
{ pattern: /ANTHROPIC|OPENAI|CLAUDE|GITHUB|VERCEL|\bAWS\b|HOSTI|\bWWW\b|GOOGLE|NETLIFY/i, candidates: ["Software", "Subscriptions", "Bills & Services"] },
{ pattern: /TAOBAO|ALIEXPRESS|ALIPAY|AMAZON|WISH/i, candidates: ["Shopping"] },
// Note: no `\b` word-boundary around the Cyrillic keywords — JS's `\b` is
// defined in terms of `\w` ([A-Za-z0-9_]), which doesn't include Cyrillic
// letters, so `\bЗЭЭЛ\b` would never match any Cyrillic text at all.
{ pattern: /ЛИЗИНГ|ТОКИ|АВДАР|ББСБ|ЗЭЭЛ|LOAN|LEASING/i, candidates: ["Loan", "Bills & Services"] },
{ pattern: /ДААТГАЛ|INSURANCE/i, candidates: ["Insurance", "Bills & Services"] },
{ pattern: /MOBICOM|UNITEL|SKYTEL|GMOBILE|ONDO/i, candidates: ["Bills & Services"] },
{ pattern: /TUSHIG|CARREFOUR|NOMIN|MART|МАРКЕТ|CU-|GS25|MINII|ХҮНС|GROCER/i, candidates: ["Groceries", "Food & Drink"] },
{ pattern: /MAGIC TEC|ELECTRONI|ITOPIA|TOPAZ|ELECTRO/i, candidates: ["Electronics", "Shopping"] },
{ pattern: /PGAMING|GAMING|STEAM|PLAYSTATION|XBOX/i, candidates: ["Entertainment"] },
// CAFE/КАФЕ live here (not in the Food & Drink rule below) per the latest
// keyword grounding — a bare "cafe" name reads as a coffee spot first.
{ pattern: /КОФЕ|COFFEE|TOM N TOMS|CAFE|КАФЕ/i, candidates: ["Coffee", "Food & Drink"] },
{ pattern: /ХООЛ|RESTAURANT|KFC|PIZZA|BURGER/i, candidates: ["Food & Drink"] },
{ pattern: /TAXI|UBCAB|ТЭЭВЭР|PETROL|ШАТАХУУН|BENZIN/i, candidates: ["Transport"] },
{ pattern: /CONVERSE|WARRIOR|PASTEL|OLYMPIC|НЭКСУС|SANT ASAR|STORE|SHOP|ДЭЛГҮҮР/i, candidates: ["Shopping"] },
];
/**
* Suggests one of the user's existing categories for an uncategorized merchant,
* from a keyword heuristic keyed to the real merchants seen in the data. The
* merchant name is first extracted from any card string (`:MCI:ANTHROPIC`).
* Returns `null` (user picks) when nothing matches, when the matching rule's
* candidates are all absent, or when the string is a card number with no
* recoverable merchant.
*/
export function suggestCategory(merchant: string, categories: Category[]): string | null {
const text = extractMerchant(merchant);
if (!text || looksLikeCardNumber(text)) return null;
const names = new Set(categories.map((c) => c.name));
for (const rule of RULES) {
if (!rule.pattern.test(text)) continue;
for (const candidate of rule.candidates) {
if (names.has(candidate)) return candidate;
}
}
return null;
}