@zenalexa/unicli 0.219.0 → 0.220.1

package/src/adapters/bbc/topic.test.ts

@@ -0,0 +1,52 @@
+import { describe, expect, it } from "vitest";
+import {
+  bbcPubDateToIso,
+  decodeBbcEntities,
+  extractRssTag,
+  parseBbcRssItems,
+  requireBbcLimit,
+  requireBbcTopic,
+} from "./topic.js";
+
+describe("bbc agent-facing topic command", () => {
+  it("validates topics and limits", () => {
+    expect(requireBbcTopic("science-and-environment")).toBe(
+      "science_and_environment",
+    );
+    expect(requireBbcTopic(" Entertainment and Arts ")).toBe(
+      "entertainment_and_arts",
+    );
+    expect(() => requireBbcTopic("sport")).toThrow("not supported");
+    expect(requireBbcLimit(undefined)).toBe(20);
+    expect(requireBbcLimit("50")).toBe(50);
+    expect(() => requireBbcLimit("51")).toThrow("bbc limit");
+  });
+
+  it("parses RSS items with CDATA and entities", () => {
+    expect(decodeBbcEntities("A & B 'C'")).toBe("A & B 'C'");
+    const block = "<title><![CDATA[World &amp; News]]></title>";
+    expect(extractRssTag(block, "title")).toBe("World &amp; News");
+    expect(
+      parseBbcRssItems(`
+        <rss><channel>
+          <item>
+            <title><![CDATA[World &amp; News]]></title>
+            <description>Summary &lt;one&gt;</description>
+            <link>https://www.bbc.com/news/world-1</link>
+            <pubDate>Tue, 12 May 2026 10:00:00 GMT</pubDate>
+            <guid>guid-1</guid>
+          </item>
+        </channel></rss>
+      `),
+    ).toEqual([
+      {
+        title: "World & News",
+        description: "Summary <one>",
+        link: "https://www.bbc.com/news/world-1",
+        pubDate: "Tue, 12 May 2026 10:00:00 GMT",
+        guid: "guid-1",
+      },
+    ]);
+    expect(bbcPubDateToIso("Tue, 12 May 2026 10:00:00 GMT")).toBe("2026-05-12");
+  });
+});

package/src/adapters/bbc/topic.ts

@@ -0,0 +1,149 @@
+/**
+ * @owner   src/adapters/bbc/topic.ts
+ * @does    Register agent-facing BBC topic RSS command.
+ * @needs   feeds.bbci.co.uk public RSS feeds, topic whitelist, conservative RSS parsing.
+ * @feeds   surface coverage ledger, BBC section headline discovery.
+ * @breaks  BBC RSS envelope drift, loose topic parsing, or silent empty feeds hide news coverage failures.
+ */
+
+import { cli, Strategy } from "../../registry.js";
+
+const BBC_FEED_BASE = "https://feeds.bbci.co.uk/news";
+const TOPICS = [
+  "world",
+  "business",
+  "politics",
+  "health",
+  "education",
+  "science_and_environment",
+  "technology",
+  "entertainment_and_arts",
+] as const;
+
+type BbcTopic = (typeof TOPICS)[number];
+
+interface RssItem {
+  title: string;
+  description: string;
+  link: string;
+  pubDate: string;
+  guid: string;
+}
+
+export function requireBbcLimit(value: unknown, fallback = 20): number {
+  if (value === undefined || value === null || value === "") return fallback;
+  const n = Number(value);
+  if (!Number.isInteger(n) || n < 1 || n > 50) {
+    throw new Error("bbc limit must be an integer in [1, 50].");
+  }
+  return n;
+}
+
+export function requireBbcTopic(value: unknown): BbcTopic {
+  const topic = String(value ?? "")
+    .trim()
+    .toLowerCase()
+    .replace(/[\s-]+/g, "_");
+  if (!TOPICS.includes(topic as BbcTopic)) {
+    throw new Error(`bbc topic "${String(value)}" is not supported.`);
+  }
+  return topic as BbcTopic;
+}
+
+export function decodeBbcEntities(value: unknown): string {
+  return String(value ?? "")
+    .replace(/&#x([0-9a-fA-F]+);/g, (_m, hex: string) =>
+      String.fromCodePoint(Number.parseInt(hex, 16)),
+    )
+    .replace(/&#(\d+);/g, (_m, dec: string) =>
+      String.fromCodePoint(Number.parseInt(dec, 10)),
+    )
+    .replace(/&amp;/g, "&")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&apos;/g, "'")
+    .replace(/&#39;/g, "'")
+    .replace(/&nbsp;/g, " ");
+}
+
+export function extractRssTag(block: string, tag: string): string {
+  const cdata = block.match(
+    new RegExp(
+      `<${tag}[^>]*>\\s*<!\\[CDATA\\[([\\s\\S]*?)\\]\\]>\\s*<\\/${tag}>`,
+    ),
+  );
+  if (cdata) return cdata[1];
+  const plain = block.match(new RegExp(`<${tag}[^>]*>([\\s\\S]*?)<\\/${tag}>`));
+  return plain ? plain[1] : "";
+}
+
+export function parseBbcRssItems(xml: string): RssItem[] {
+  const items: RssItem[] = [];
+  const re = /<item[^>]*>([\s\S]*?)<\/item>/g;
+  let match: RegExpExecArray | null;
+  while ((match = re.exec(xml)) !== null) {
+    const block = match[1];
+    items.push({
+      title: decodeBbcEntities(extractRssTag(block, "title")).trim(),
+      description: decodeBbcEntities(
+        extractRssTag(block, "description"),
+      ).trim(),
+      link: decodeBbcEntities(extractRssTag(block, "link")).trim(),
+      pubDate: decodeBbcEntities(extractRssTag(block, "pubDate")).trim(),
+      guid: decodeBbcEntities(extractRssTag(block, "guid")).trim(),
+    });
+  }
+  return items;
+}
+
+export function bbcPubDateToIso(value: unknown): string {
+  const date = new Date(String(value ?? ""));
+  return Number.isNaN(date.getTime()) ? "" : date.toISOString().slice(0, 10);
+}
+
+async function fetchBbcRss(topic: BbcTopic): Promise<string> {
+  const response = await fetch(`${BBC_FEED_BASE}/${topic}/rss.xml`, {
+    headers: {
+      "User-Agent": "unicli-bbc/1.0 (https://github.com/olo-dot-io/Uni-CLI)",
+      Accept: "application/rss+xml, application/xml",
+    },
+  });
+  if (!response.ok)
+    throw new Error(`bbc topic ${topic} returned HTTP ${response.status}.`);
+  return response.text();
+}
+
+cli({
+  site: "bbc",
+  name: "topic",
+  description: "BBC News headlines for a specific section",
+  domain: "feeds.bbci.co.uk",
+  strategy: Strategy.PUBLIC,
+  args: [
+    {
+      name: "topic",
+      type: "str",
+      required: true,
+      positional: true,
+      choices: [...TOPICS],
+      description: "BBC section name",
+    },
+    { name: "limit", type: "int", default: 20, description: "Max headlines" },
+  ],
+  columns: ["rank", "title", "description", "pubDate", "url"],
+  func: async (_page, kwargs) => {
+    const topic = requireBbcTopic(kwargs.topic);
+    const limit = requireBbcLimit(kwargs.limit);
+    const items = parseBbcRssItems(await fetchBbcRss(topic));
+    if (items.length === 0)
+      throw new Error(`BBC ${topic} feed returned no items.`);
+    return items.slice(0, limit).map((item, index) => ({
+      rank: index + 1,
+      title: item.title,
+      description: item.description,
+      pubDate: bbcPubDateToIso(item.pubDate),
+      url: item.link,
+    }));
+  },
+});

package/src/adapters/chatgpt/web.test.ts

@@ -0,0 +1,121 @@
+import { describe, expect, it, vi } from "vitest";
+import { resolveCommand } from "../../registry.js";
+import {
+  chatGptHtmlToMarkdown,
+  mapChatGptConversations,
+  mapChatGptMessages,
+  normalizeChatGptBoolean,
+  parseChatGptConversationId,
+} from "./web.js";
+
+function pageMock(evaluateResults: unknown[]) {
+  return {
+    goto: vi.fn().mockResolvedValue(undefined),
+    wait: vi.fn().mockResolvedValue(undefined),
+    evaluate: vi.fn(async () => evaluateResults.shift()),
+  };
+}
+
+describe("chatgpt agent-facing web readers", () => {
+  it("validates conversation ids and boolean flags", () => {
+    expect(parseChatGptConversationId(" abcDEF_12345 ")).toBe("abcDEF_12345");
+    expect(
+      parseChatGptConversationId("https://chatgpt.com/c/abcDEF_12345"),
+    ).toBe("abcDEF_12345");
+    expect(() => parseChatGptConversationId("https://chatgpt.com/")).toThrow(
+      "conversation id",
+    );
+    expect(normalizeChatGptBoolean("yes")).toBe(true);
+    expect(normalizeChatGptBoolean(undefined, true)).toBe(true);
+  });
+
+  it("maps conversations and messages to stable columns", () => {
+    expect(
+      mapChatGptConversations([
+        { Id: "c1", Title: " First ", Url: "https://chatgpt.com/c/c1" },
+        { Id: "", Title: "skip", Url: "" },
+      ]),
+    ).toEqual([
+      {
+        Index: 1,
+        Id: "c1",
+        Title: "First",
+        Url: "https://chatgpt.com/c/c1",
+      },
+    ]);
+    expect(chatGptHtmlToMarkdown("<p>A&nbsp;&lt;B&gt;</p>")).toBe("A <B>");
+    expect(
+      mapChatGptMessages(
+        [
+          { role: "user", text: "Hi" },
+          { role: "assistant", text: "Hello", html: "<p>Hello&nbsp;AI</p>" },
+        ],
+        true,
+      ),
+    ).toEqual([
+      { Index: 1, Role: "User", Text: "Hi" },
+      { Index: 2, Role: "Assistant", Text: "Hello AI" },
+    ]);
+  });
+
+  it("lists visible ChatGPT web history", async () => {
+    const command = resolveCommand("chatgpt", "history")?.command;
+    expect(command?.func).toBeTypeOf("function");
+    const page = pageMock([
+      true,
+      [
+        {
+          Id: "c1",
+          Title: "Conversation",
+          Url: "https://chatgpt.com/c/c1",
+        },
+      ],
+    ]);
+    const result = await command!.func!(page, { limit: 10 });
+    expect(page.goto).toHaveBeenCalledWith("https://chatgpt.com", {
+      settleMs: 2000,
+    });
+    expect(result).toEqual([
+      {
+        Index: 1,
+        Id: "c1",
+        Title: "Conversation",
+        Url: "https://chatgpt.com/c/c1",
+      },
+    ]);
+  });
+
+  it("opens a ChatGPT conversation detail and reads messages", async () => {
+    const command = resolveCommand("chatgpt", "detail")?.command;
+    expect(command?.func).toBeTypeOf("function");
+    const page = pageMock([
+      [
+        { role: "user", text: "Prompt" },
+        { role: "assistant", text: "Answer", html: "<p>Answer</p>" },
+      ],
+    ]);
+    const result = await command!.func!(page, {
+      id: "abcDEF_12345",
+      markdown: true,
+    });
+    expect(page.goto).toHaveBeenCalledWith(
+      "https://chatgpt.com/c/abcDEF_12345",
+      { settleMs: 2000 },
+    );
+    expect(result).toEqual([
+      { Index: 1, Role: "User", Text: "Prompt" },
+      { Index: 2, Role: "Assistant", Text: "Answer" },
+    ]);
+  });
+
+  it("fails closed when history or detail has no rows", async () => {
+    const history = resolveCommand("chatgpt", "history")?.command;
+    await expect(history!.func!(pageMock([false, []]), {})).rejects.toThrow(
+      "No ChatGPT conversation links",
+    );
+    const detail = resolveCommand("chatgpt", "detail")?.command;
+    await expect(
+      detail!.func!(pageMock([[]]), { id: "abcDEF_12345" }),
+    ).rejects.toThrow("No visible ChatGPT messages");
+  });
+});

package/src/adapters/chatgpt/web.ts

@@ -0,0 +1,286 @@
+/**
+ * @owner   src/adapters/chatgpt/web.ts
+ * @does    Register agent-facing ChatGPT web history and detail readers implemented with site-specific safety checks.
+ * @needs   Logged-in chatgpt.com browser session, visible sidebar links, rendered conversation messages.
+ * @feeds   surface coverage ledger, ChatGPT conversation discovery, and browser-session read workflows.
+ * @breaks  ChatGPT DOM selector drift or login gates can hide history/messages.
+ */
+
+import { cli, Strategy } from "../../registry.js";
+import { intArg, js, str } from "../_shared/browser-tools.js";
+
+const CHATGPT_HOME = "https://chatgpt.com";
+const CONVERSATION_LINK_SELECTOR = 'a[href*="/c/"]';
+const CONVERSATION_MESSAGE_SELECTOR =
+  '[data-message-author-role], article[data-testid*="conversation-turn"]';
+
+interface RawChatGptConversation {
+  Id?: unknown;
+  Title?: unknown;
+  Url?: unknown;
+}
+
+interface RawChatGptMessage {
+  role?: unknown;
+  text?: unknown;
+  html?: unknown;
+}
+
+interface ChatGptBrowserPage {
+  goto: (url: string, opts?: Record<string, unknown>) => Promise<unknown>;
+  wait: (args: unknown) => Promise<unknown>;
+  evaluate: (script: string) => Promise<unknown>;
+}
+
+export function parseChatGptConversationId(value: unknown): string {
+  const raw = str(value).trim();
+  const match = raw.match(/(?:^|\/c\/)([A-Za-z0-9_-]{8,})(?:[/?#]|$)/);
+  if (match?.[1]) return match[1];
+  if (/^[A-Za-z0-9_-]{8,}$/.test(raw)) return raw;
+  throw new Error("chatgpt detail requires a conversation id or /c/<id> URL.");
+}
+
+export function normalizeChatGptBoolean(
+  value: unknown,
+  fallback = false,
+): boolean {
+  if (typeof value === "boolean") return value;
+  if (value === undefined || value === null || value === "") return fallback;
+  return /^(true|1|yes|on)$/i.test(str(value).trim());
+}
+
+export function chatGptHtmlToMarkdown(value: unknown): string {
+  return str(value)
+    .replace(/<br\s*\/?>/gi, "\n")
+    .replace(/<\/(p|div|li|h[1-6]|pre)>/gi, "\n")
+    .replace(/<[^>]+>/g, " ")
+    .replace(/&nbsp;/g, " ")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&amp;/g, "&")
+    .replace(/&quot;/g, '"')
+    .replace(/&#39;/g, "'")
+    .replace(/[ \t]+\n/g, "\n")
+    .replace(/\n{3,}/g, "\n\n")
+    .replace(/[ \t]{2,}/g, " ")
+    .trim();
+}
+
+export function mapChatGptConversations(
+  items: RawChatGptConversation[],
+): Record<string, unknown>[] {
+  return items
+    .map((item, index) => ({
+      Index: index + 1,
+      Id: str(item.Id),
+      Title: str(item.Title, "(untitled)").trim() || "(untitled)",
+      Url: str(item.Url),
+    }))
+    .filter((item) => item.Id);
+}
+
+export function mapChatGptMessages(
+  items: RawChatGptMessage[],
+  wantMarkdown: boolean,
+): Record<string, unknown>[] {
+  return items
+    .map((item, index) => {
+      const role = /assistant/i.test(str(item.role)) ? "Assistant" : "User";
+      const text =
+        wantMarkdown && role === "Assistant" && str(item.html).trim()
+          ? chatGptHtmlToMarkdown(item.html)
+          : str(item.text).trim();
+      return {
+        Index: index + 1,
+        Role: role,
+        Text: text,
+      };
+    })
+    .filter((item) => item.Text);
+}
+
+async function openChatGpt(page: ChatGptBrowserPage): Promise<void> {
+  await page.goto(CHATGPT_HOME, { settleMs: 2000 });
+}
+
+async function readConversationLinks(
+  page: ChatGptBrowserPage,
+): Promise<RawChatGptConversation[]> {
+  const rows = await page.evaluate(`(() => {
+    const isVisible = (el) => {
+      if (!(el instanceof HTMLElement)) return false;
+      const style = window.getComputedStyle(el);
+      if (style.display === 'none' || style.visibility === 'hidden') return false;
+      const rect = el.getBoundingClientRect();
+      return rect.width > 0 && rect.height > 0;
+    };
+    const links = Array.from(document.querySelectorAll(${js(CONVERSATION_LINK_SELECTOR)}))
+      .filter((link) => link instanceof HTMLAnchorElement && isVisible(link));
+    const seen = new Set();
+    const items = [];
+    for (const link of links) {
+      const href = link.getAttribute('href') || '';
+      const match = href.match(/\\/c\\/([^/?#]+)/);
+      if (!match || seen.has(match[1])) continue;
+      seen.add(match[1]);
+      items.push({
+        Id: match[1],
+        Title: (link.innerText || link.textContent || '').replace(/\\s+/g, ' ').trim() || '(untitled)',
+        Url: href.startsWith('http') ? href : (${js(CHATGPT_HOME)} + href),
+      });
+    }
+    return items;
+  })()`);
+  return Array.isArray(rows) ? (rows as RawChatGptConversation[]) : [];
+}
+
+async function openSidebar(page: ChatGptBrowserPage): Promise<void> {
+  const opened = await page.evaluate(`(() => {
+    const button = Array.from(document.querySelectorAll('button'))
+      .find((node) => /open sidebar|打开边栏|打开侧边栏/i.test(node.getAttribute('aria-label') || ''));
+    if (button instanceof HTMLElement) {
+      button.click();
+      return true;
+    }
+    return false;
+  })()`);
+  if (opened) {
+    try {
+      await page.wait({ selector: CONVERSATION_LINK_SELECTOR, timeout: 3 });
+    } catch {
+      return;
+    }
+  }
+}
+
+async function getConversationHistory(
+  page: ChatGptBrowserPage,
+  limit: number,
+): Promise<Record<string, unknown>[]> {
+  await openChatGpt(page);
+  await openSidebar(page);
+  let rows = mapChatGptConversations(await readConversationLinks(page));
+  if (!rows.length) {
+    await page.goto(CHATGPT_HOME, { settleMs: 2000 });
+    try {
+      await page.wait({ selector: CONVERSATION_LINK_SELECTOR, timeout: 8 });
+    } catch {
+      rows = [];
+    }
+    rows = mapChatGptConversations(await readConversationLinks(page));
+  }
+  if (!rows.length) {
+    throw new Error(
+      "No ChatGPT conversation links were visible in the sidebar.",
+    );
+  }
+  return rows.slice(0, limit);
+}
+
+async function readVisibleMessages(
+  page: ChatGptBrowserPage,
+  wantMarkdown: boolean,
+): Promise<Record<string, unknown>[]> {
+  const rows = await page.evaluate(`(() => {
+    const isVisible = (el) => {
+      if (!(el instanceof HTMLElement)) return false;
+      const style = window.getComputedStyle(el);
+      if (style.display === 'none' || style.visibility === 'hidden') return false;
+      const rect = el.getBoundingClientRect();
+      return rect.width > 0 && rect.height > 0;
+    };
+    const normalize = (value) => String(value || '').replace(/\\u00a0/g, ' ').replace(/[ \\t]+\\n/g, '\\n').replace(/\\n{3,}/g, '\\n\\n').trim();
+    const roleOf = (node) => {
+      const attr = node.getAttribute('data-message-author-role') || node.getAttribute('data-author') || '';
+      if (/assistant/i.test(attr)) return 'Assistant';
+      if (/user/i.test(attr)) return 'User';
+      const testid = node.getAttribute('data-testid') || '';
+      if (/assistant/i.test(testid)) return 'Assistant';
+      if (/user/i.test(testid)) return 'User';
+      const label = node.getAttribute('aria-label') || '';
+      if (/assistant|chatgpt/i.test(label)) return 'Assistant';
+      if (/you|user/i.test(label)) return 'User';
+      return '';
+    };
+    const nodes = Array.from(document.querySelectorAll(${js(CONVERSATION_MESSAGE_SELECTOR)}))
+      .filter((node) => node instanceof HTMLElement && isVisible(node));
+    const seen = new Set();
+    const items = [];
+    for (const node of nodes) {
+      let role = roleOf(node);
+      const roleNode = node.querySelector('[data-message-author-role], [data-author]');
+      if (!role && roleNode) role = roleOf(roleNode);
+      if (!role) continue;
+      const contentNode = node.querySelector('[data-message-author-role] .markdown')
+        || node.querySelector('.markdown')
+        || node.querySelector('[data-message-author-role]')
+        || node;
+      const html = contentNode instanceof HTMLElement ? (contentNode.innerHTML || '') : '';
+      const text = normalize(contentNode instanceof HTMLElement ? (contentNode.innerText || contentNode.textContent || '') : '');
+      if (!text) continue;
+      const key = role + '\\n' + text;
+      if (seen.has(key)) continue;
+      seen.add(key);
+      items.push({ role, text, html });
+    }
+    return items;
+  })()`);
+  const messages = Array.isArray(rows) ? (rows as RawChatGptMessage[]) : [];
+  const mapped = mapChatGptMessages(messages, wantMarkdown);
+  if (!mapped.length) {
+    throw new Error("No visible ChatGPT messages were found.");
+  }
+  return mapped;
+}
+
+cli({
+  site: "chatgpt",
+  name: "history",
+  description: "List visible ChatGPT web conversation history from the sidebar",
+  domain: "chatgpt.com",
+  strategy: Strategy.COOKIE,
+  browser: true,
+  args: [{ name: "limit", type: "int", default: 20 }],
+  columns: ["Index", "Id", "Title", "Url"],
+  func: async (page, kwargs) => {
+    const limit = intArg(kwargs.limit, 20, 100);
+    return getConversationHistory(page as ChatGptBrowserPage, limit);
+  },
+});
+
+cli({
+  site: "chatgpt",
+  name: "detail",
+  description: "Open a ChatGPT web conversation by ID and read its messages",
+  domain: "chatgpt.com",
+  strategy: Strategy.COOKIE,
+  browser: true,
+  args: [
+    {
+      name: "id",
+      type: "str",
+      required: true,
+      positional: true,
+      description: "Conversation ID or full /c/<id> URL",
+    },
+    {
+      name: "markdown",
+      type: "bool",
+      default: false,
+      description: "Emit assistant replies as markdown",
+    },
+  ],
+  columns: ["Index", "Role", "Text"],
+  func: async (page, kwargs) => {
+    const p = page as ChatGptBrowserPage;
+    const id = parseChatGptConversationId(kwargs.id);
+    const wantMarkdown = normalizeChatGptBoolean(kwargs.markdown, false);
+    await p.goto(`${CHATGPT_HOME}/c/${id}`, { settleMs: 2000 });
+    try {
+      await p.wait({ selector: CONVERSATION_MESSAGE_SELECTOR, timeout: 10 });
+    } catch {
+      await p.wait(0.5);
+    }
+    return readVisibleMessages(p, wantMarkdown);
+  },
+});