@chainlesschain/personal-data-hub 0.4.28 → 0.4.30

package/__tests__/query-parser.test.js

@@ -1,302 +0,0 @@
-"use strict";
-
-import { describe, it, expect } from "vitest";
-
-const {
-  parseQuery,
-  parseTimeWindow,
-  parseFilters,
-  parseIntent,
-  parseEntityFocus,
-  extractEntityTerm,
-  extractPersonNameCandidate,
-} = require("../lib/query-parser");
-
-// Pin "now" to 2026-05-19 12:00:00 UTC for deterministic windows
-const NOW = new Date("2026-05-19T12:00:00Z").getTime();
-
-describe("parseTimeWindow", () => {
-  it("今天 → start of today through end of today", () => {
-    const w = parseTimeWindow("我今天花了多少？", NOW);
-    expect(w).not.toBeNull();
-    const startOfDay = new Date(NOW);
-    startOfDay.setHours(0, 0, 0, 0);
-    expect(w.since).toBe(startOfDay.getTime());
-    expect(w.until).toBe(startOfDay.getTime() + 86_400_000 - 1);
-  });
-
-  it("昨天 → start of yesterday through end of yesterday", () => {
-    const w = parseTimeWindow("昨天的订单", NOW);
-    expect(w).not.toBeNull();
-    expect(w.until - w.since).toBe(86_400_000 - 1);
-  });
-
-  it("上个月 → full previous calendar month", () => {
-    const w = parseTimeWindow("上个月支出", NOW);
-    expect(w).not.toBeNull();
-    // NOW is 2026-05-19; previous month = 2026-04
-    const apr1 = new Date(2026, 3, 1).getTime();
-    const may1 = new Date(2026, 4, 1).getTime();
-    expect(w.since).toBe(apr1);
-    expect(w.until).toBe(may1 - 1);
-  });
-
-  it("本月 → current calendar month", () => {
-    const w = parseTimeWindow("本月开销", NOW);
-    expect(w).not.toBeNull();
-    expect(w.since).toBe(new Date(2026, 4, 1).getTime());
-    expect(w.until).toBe(new Date(2026, 5, 1).getTime() - 1);
-  });
-
-  it("去年 → full previous calendar year", () => {
-    const w = parseTimeWindow("去年我去过哪些地方", NOW);
-    expect(w.since).toBe(new Date(2025, 0, 1).getTime());
-    expect(w.until).toBe(new Date(2025, 11, 1).getTime() + 31 * 86_400_000 - 1);
-  });
-
-  it("最近 30 天 → past 30-day window ending now", () => {
-    const w = parseTimeWindow("最近 30 天聊过什么", NOW);
-    expect(w.until).toBe(NOW);
-    expect(NOW - w.since).toBe(30 * 86_400_000);
-  });
-
-  it("最近 N 周 / 最近 N 个月 patterns work", () => {
-    const week = parseTimeWindow("最近 2 周", NOW);
-    expect(NOW - week.since).toBe(14 * 86_400_000);
-
-    const months = parseTimeWindow("最近 3 个月", NOW);
-    expect(months.until).toBe(NOW);
-    expect(months.since).toBeLessThan(NOW);
-  });
-
-  it("YYYY 年 M 月 → that calendar month", () => {
-    const w = parseTimeWindow("2024 年 7 月在淘宝下过几单", NOW);
-    expect(w.since).toBe(new Date(2024, 6, 1).getTime());
-    expect(w.until).toBe(new Date(2024, 7, 1).getTime() - 1);
-  });
-
-  it("returns null for question without time clue", () => {
-    expect(parseTimeWindow("妈妈手机号是多少", NOW)).toBeNull();
-  });
-
-  it("returns null for non-string input", () => {
-    expect(parseTimeWindow(null)).toBeNull();
-    expect(parseTimeWindow(undefined)).toBeNull();
-  });
-});
-
-describe("parseFilters", () => {
-  it("identifies subtype via keywords (Chinese + English)", () => {
-    expect(parseFilters("今年在淘宝下了多少单").subtype).toBe("order");
-    expect(parseFilters("上个月总共花了多少").subtype).toBe("payment");
-    expect(parseFilters("转给妈妈多少钱").subtype).toBe("transfer");
-    expect(parseFilters("我今年的收入").subtype).toBe("income");
-    expect(parseFilters("我跟妈妈聊了什么").subtype).toBe("message");
-    expect(parseFilters("我朋友圈发了啥").subtype).toBe("post");
-  });
-
-  it("identifies adapter via keywords (Chinese + English)", () => {
-    expect(parseFilters("淘宝今年下了多少单").adapter).toBe("taobao");
-    expect(parseFilters("支付宝账单").adapter).toBe("alipay-bill");
-    expect(parseFilters("微信里我跟谁聊最多").adapter).toBe("wechat");
-    expect(parseFilters("高德历史足迹").adapter).toBe("amap");
-    expect(parseFilters("DeepSeek 我之前问过啥").adapter).toBe("ai-chat-history");
-  });
-
-  it("returns empty object when no clue", () => {
-    expect(parseFilters("hello world")).toEqual({});
-  });
-});
-
-describe("parseIntent", () => {
-  it("sum-amount when 'total ... money' phrasing", () => {
-    expect(parseIntent("上个月总共花了多少")).toBe("sum-amount");
-    expect(parseIntent("我今年开销加起来")).toBe("sum-amount");
-  });
-
-  it("count when 'how many' phrasing", () => {
-    expect(parseIntent("最近多少次跟妈妈聊过")).toBe("count");
-    expect(parseIntent("我下了几单")).toBe("count");
-  });
-
-  it("latest when 'recent / latest'", () => {
-    expect(parseIntent("最近一次转账")).toBe("latest");
-  });
-
-  it("list as default", () => {
-    expect(parseIntent("妈妈的手机号")).toBe("list");
-  });
-});
-
-describe("parseEntityFocus", () => {
-  it("returns 'persons' for 联系人 / 通讯录 phrasing", () => {
-    expect(parseEntityFocus("我有哪些联系人")).toBe("persons");
-    expect(parseEntityFocus("通讯录里有多少人")).toBe("persons");
-    expect(parseEntityFocus("好友列表谁是张三")).toBe("persons");
-  });
-
-  it("returns 'persons' for phone-number phrasing", () => {
-    expect(parseEntityFocus("妈手机号是多少")).toBe("persons");
-    expect(parseEntityFocus("王医生的电话号码")).toBe("persons");
-    expect(parseEntityFocus("show me my contacts")).toBe("persons");
-  });
-
-  it("returns 'items' for installed-app phrasing", () => {
-    expect(parseEntityFocus("我装了哪些 app")).toBe("items");
-    expect(parseEntityFocus("有哪些游戏")).toBe("items");
-    expect(parseEntityFocus("installed apps")).toBe("items");
-  });
-
-  it("returns null when no focus signal", () => {
-    expect(parseEntityFocus("上个月在淘宝花了多少")).toBeNull();
-    expect(parseEntityFocus("最近的订单")).toBeNull();
-    expect(parseEntityFocus("hello")).toBeNull();
-  });
-
-  it("returns null for non-string / empty input", () => {
-    expect(parseEntityFocus("")).toBeNull();
-    expect(parseEntityFocus(null)).toBeNull();
-    expect(parseEntityFocus(undefined)).toBeNull();
-  });
-});
-
-describe("parseQuery (integration)", () => {
-  it("full parse for spending question", () => {
-    const r = parseQuery("上个月在淘宝总共花了多少钱？", { now: NOW });
-    expect(r.timeWindow.since).toBe(new Date(2026, 3, 1).getTime());
-    expect(r.filters.subtype).toBe("payment");
-    expect(r.filters.adapter).toBe("taobao");
-    expect(r.intent).toBe("sum-amount");
-    expect(r.entityFocus).toBeNull();
-  });
-
-  it("contact question carries entityFocus=persons", () => {
-    const r = parseQuery("我有哪些联系人", { now: NOW });
-    expect(r.entityFocus).toBe("persons");
-    expect(r.intent).toBe("list");
-  });
-
-  it("full parse for footprint question", () => {
-    const r = parseQuery("去年我在高德上去过哪些地方", { now: NOW });
-    expect(r.timeWindow.since).toBe(new Date(2025, 0, 1).getTime());
-    expect(r.filters.adapter).toBe("amap");
-    expect(r.intent).toBe("list");
-  });
-
-  it("non-string question returns empty raw + nulls", () => {
-    const r = parseQuery(undefined);
-    expect(r.raw).toBe("");
-    expect(r.timeWindow).toBeNull();
-    expect(r.filters).toEqual({});
-  });
-});
-
-// ─── extractEntityTerm — FTS5 fulltext routing helper ───────────────────
-//
-// 2026-05-24 — Powers AnalysisEngine._gatherFacts intent=list augmentation:
-// when the parser pulls a probable entity-name out of the question, the
-// engine appends vault.searchEvents(q=term) hits to the FACTS pool. Wrong
-// extractions are intentionally non-fatal — they waste a few rows of
-// budget at worst, never lose existing events. Memory:
-// pdh_analysis_engine_intent_routing.md.
-
-describe("extractEntityTerm", () => {
-  it("extracts named entity from '提到 X 的消息' phrasing", () => {
-    expect(extractEntityTerm("提到王老板的消息")).toBe("王老板");
-  });
-
-  it("returns null when only stop-words remain (no entity hint)", () => {
-    expect(extractEntityTerm("上个月在淘宝总共花了多少？")).toBeNull();
-    expect(extractEntityTerm("在淘宝买了什么")).toBeNull();
-  });
-
-  it("picks the longest remaining chunk when several survive cleaning", () => {
-    // 苹果(2) vs 订单(stop) — only 苹果 left.
-    expect(extractEntityTerm("苹果的订单")).toBe("苹果");
-  });
-
-  it("strips list/search trigger words ('提到', '查找', '看一下')", () => {
-    expect(extractEntityTerm("查找王医生的订单")).toBe("王医生");
-    expect(extractEntityTerm("看一下王医生的最新消息")).toBe("王医生");
-  });
-
-  it("strips compound subtype keywords before shorter intent forms", () => {
-    // "多少钱" must clear before "多少" leaves stranded "钱". With clean
-    // stripping there is no leftover ≥2 char chunk → null.
-    expect(extractEntityTerm("我总共花了多少钱")).toBeNull();
-  });
-
-  it("ignores single-character residues (verbs leak through; 1-char names skipped first-pass)", () => {
-    // "我妈" → "我" stripped (pronoun), "妈" left as single char → filtered.
-    // Documented limitation; first-pass tradeoff for higher precision.
-    expect(extractEntityTerm("我妈最近发的微信")).toBeNull();
-  });
-
-  it("handles ASCII entity tokens (≥2 chars)", () => {
-    expect(extractEntityTerm("提到 GitHub 的消息")).toBe("GitHub");
-  });
-
-  it("returns null for non-string / empty input", () => {
-    expect(extractEntityTerm("")).toBeNull();
-    expect(extractEntityTerm(null)).toBeNull();
-    expect(extractEntityTerm(undefined)).toBeNull();
-    expect(extractEntityTerm(123)).toBeNull();
-  });
-
-  it("does not pick adapter keywords as the entity (handled by filters)", () => {
-    // "淘宝" 是 adapter，会被 parseFilters 抽走当 q.adapter；不该再被
-    // 当实体名重复 FTS 搜。
-    expect(extractEntityTerm("看下淘宝的订单")).toBeNull();
-  });
-
-  it("> 10 char tokens are dropped (probable concatenated noise)", () => {
-    // 拼出一个 12 char ASCII token，期望被 length 上限过滤掉
-    const r = extractEntityTerm("提到 abcdefghijkl 的消息");
-    expect(r).toBeNull();
-  });
-});
-
-// ─── extractPersonNameCandidate — persons-branch name search ─────────────
-//
-// 2026-05-27 — Powers AnalysisEngine entityFocus=persons name-search
-// short-circuit. Differs from extractEntityTerm in two ways: strips
-// person-FOCUS framing words first (联系人/手机号/etc.) and allows
-// single-char Chinese names from a relation whitelist (妈/爸/姐/...).
-
-describe("extractPersonNameCandidate", () => {
-  it("extracts multi-char name when present", () => {
-    expect(extractPersonNameCandidate("张三的电话号码")).toBe("张三");
-    expect(extractPersonNameCandidate("王医生手机号是多少")).toBe("王医生");
-  });
-
-  it("falls back to single-char relation word ('妈', '爸', '姐')", () => {
-    expect(extractPersonNameCandidate("妈手机号是多少")).toBe("妈");
-    expect(extractPersonNameCandidate("爸的电话")).toBe("爸");
-    expect(extractPersonNameCandidate("姐姐的号码")).toBe("姐姐");
-  });
-
-  it("multi-char wins over single-char fallback", () => {
-    // "王医生" (3 char) preferred over leaked single "医" / "生".
-    expect(extractPersonNameCandidate("王医生的手机号")).toBe("王医生");
-  });
-
-  it("returns null when no name candidate (pure framing)", () => {
-    expect(extractPersonNameCandidate("我有哪些联系人")).toBeNull();
-    expect(extractPersonNameCandidate("通讯录里有多少人")).toBeNull();
-  });
-
-  it("ignores single-char Chinese outside the relation whitelist", () => {
-    // "说" / "看" are not relation chars — should NOT slip through as names.
-    expect(extractPersonNameCandidate("说手机号")).toBeNull();
-  });
-
-  it("returns null for non-string / empty input", () => {
-    expect(extractPersonNameCandidate("")).toBeNull();
-    expect(extractPersonNameCandidate(null)).toBeNull();
-    expect(extractPersonNameCandidate(undefined)).toBeNull();
-  });
-
-  it("handles ASCII names ≥2 chars", () => {
-    expect(extractPersonNameCandidate("Alice 的电话号码")).toBe("Alice");
-  });
-});

package/__tests__/rag-derive.test.js

@@ -1,169 +0,0 @@
-"use strict";
-
-import { describe, it, expect } from "vitest";
-
-const {
-  eventToRagDoc,
-  personToRagDoc,
-  placeToRagDoc,
-  itemToRagDoc,
-  topicToRagDoc,
-  entityToRagDoc,
-  deriveBatchDocs,
-} = require("../lib/rag-derive");
-
-const sourceOk = (adapter = "test") => ({
-  adapter,
-  adapterVersion: "0.1.0",
-  capturedAt: 1700000000000,
-  capturedBy: "api",
-  originalId: "ord-42",
-});
-
-describe("eventToRagDoc", () => {
-  it("includes title + text + amount + subtype + adapter prose", () => {
-    const doc = eventToRagDoc({
-      id: "e1",
-      type: "event",
-      subtype: "order",
-      occurredAt: 1700000000000,
-      ingestedAt: 1700000000001,
-      content: {
-        title: "妈妈生日蛋白粉",
-        text: "送到妈妈家",
-        amount: { value: 288.5, currency: "CNY", direction: "out" },
-      },
-      source: sourceOk("taobao"),
-    });
-    expect(doc.id).toBe("e1");
-    expect(doc.type).toBe("event");
-    expect(doc.text).toContain("妈妈生日蛋白粉");
-    expect(doc.text).toContain("送到妈妈家");
-    expect(doc.text).toContain("-288.5 CNY");
-    expect(doc.text).toContain("type: order");
-    expect(doc.text).toContain("from: taobao");
-    expect(doc.metadata.subtype).toBe("order");
-    expect(doc.metadata.adapter).toBe("taobao");
-    expect(doc.metadata.originalId).toBe("ord-42");
-    expect(doc.metadata.occurredAt).toBe(1700000000000);
-  });
-
-  it("'in' direction renders with '+' sign", () => {
-    const doc = eventToRagDoc({
-      id: "e",
-      type: "event",
-      subtype: "income",
-      occurredAt: 1,
-      ingestedAt: 1,
-      content: { amount: { value: 5000, currency: "CNY", direction: "in" } },
-      source: sourceOk(),
-    });
-    expect(doc.text).toContain("+5000 CNY");
-  });
-
-  it("propagates topics into metadata", () => {
-    const doc = eventToRagDoc({
-      id: "e",
-      type: "event",
-      subtype: "message",
-      occurredAt: 1,
-      ingestedAt: 1,
-      content: { text: "hi" },
-      source: sourceOk(),
-      topics: ["topic-fam"],
-    });
-    expect(doc.metadata.topics).toEqual(["topic-fam"]);
-  });
-});
-
-describe("personToRagDoc", () => {
-  it("packs names + relation + identifiers into searchable text", () => {
-    const doc = personToRagDoc({
-      id: "p1",
-      type: "person",
-      subtype: "contact",
-      names: ["妈妈", "陈某某"],
-      relation: "母亲",
-      identifiers: { phone: ["13800001111"], wechatId: "wxid_xyz" },
-      ingestedAt: 1,
-      source: sourceOk(),
-    });
-    expect(doc.text).toContain("妈妈");
-    expect(doc.text).toContain("陈某某");
-    expect(doc.text).toContain("relation: 母亲");
-    expect(doc.text).toContain("13800001111");
-    expect(doc.text).toContain("wechatId: wxid_xyz");
-    expect(doc.metadata.subtype).toBe("contact");
-  });
-});
-
-describe("placeToRagDoc", () => {
-  it("emits name + alias dedup + address + category", () => {
-    const doc = placeToRagDoc({
-      id: "pl",
-      type: "place",
-      name: "妈妈家",
-      aliases: ["妈妈家", "妈家"],
-      address: "厦门思明区",
-      category: "home",
-      coordinates: { lat: 24.5, lng: 118.1 },
-      ingestedAt: 1,
-      source: sourceOk(),
-    });
-    expect(doc.text).toContain("妈妈家");
-    expect(doc.text).toContain("妈家");
-    expect(doc.text).toContain("厦门思明区");
-    expect(doc.text).toContain("category: home");
-    expect(doc.metadata.coordinates).toEqual({ lat: 24.5, lng: 118.1 });
-  });
-});
-
-describe("itemToRagDoc", () => {
-  it("includes price + category", () => {
-    const doc = itemToRagDoc({
-      id: "i",
-      type: "item",
-      subtype: "product",
-      name: "蛋白粉",
-      category: "保健品",
-      price: { value: 288, currency: "CNY" },
-      ingestedAt: 1,
-      source: sourceOk(),
-    });
-    expect(doc.text).toContain("蛋白粉");
-    expect(doc.text).toContain("category: 保健品");
-    expect(doc.text).toContain("288 CNY");
-  });
-});
-
-describe("topicToRagDoc + entityToRagDoc + deriveBatchDocs", () => {
-  it("topic doc is its name", () => {
-    const doc = topicToRagDoc({
-      id: "t",
-      type: "topic",
-      name: "母亲健康",
-      ingestedAt: 1,
-      source: sourceOk(),
-    });
-    expect(doc.text).toBe("母亲健康");
-  });
-
-  it("entityToRagDoc returns null for unknown types", () => {
-    expect(entityToRagDoc(null)).toBeNull();
-    expect(entityToRagDoc({ type: "alien" })).toBeNull();
-  });
-
-  it("deriveBatchDocs filters empty-text entities", () => {
-    const docs = deriveBatchDocs({
-      events: [
-        // empty text — should be filtered
-        { id: "empty", type: "event", subtype: "message", occurredAt: 1, ingestedAt: 1, content: {}, source: { adapter: "x", adapterVersion: "0.1.0", capturedAt: 1, capturedBy: "api" } },
-        // has text
-        { id: "kept", type: "event", subtype: "message", occurredAt: 1, ingestedAt: 1, content: { text: "hi" }, source: { adapter: "x", adapterVersion: "0.1.0", capturedAt: 1, capturedBy: "api" } },
-      ],
-    });
-    expect(docs.length).toBe(2); // 'empty' includes 'type: message' + 'from: x' so text is non-empty
-    // Both have text because structural prose is added — verify text is non-empty for both
-    expect(docs.every((d) => d.text.length > 0)).toBe(true);
-  });
-});