@cloud411716/fancy-webnovel 2.2.0 → 3.0.0

package/lib/projects.js

@@ -0,0 +1,400 @@
+/**
+ * Project registry, project directories, and project-scoped database users.
+ *
+ * The registry lives in one shared SurrealDB database
+ * (`fancy_registry`). A project's novel data lives in its own database
+ * (`proj_<uuid>`), and the project directory only stores connection metadata:
+ *
+ *   <workspace>/projects/<uuid>/.fancy-project.json
+ *
+ * The novel title is NOT duplicated on disk. Renaming a project updates the
+ * registry and the project's `story:main` record; the directory name (the
+ * UUID) never changes.
+ */
+
+import fs from 'fs';
+import path from 'path';
+import crypto from 'crypto';
+import { Surreal, createRemoteEngines } from 'surrealdb';
+import {
+  WorkspaceError,
+  atomicWriteJson,
+  readJsonFile,
+  localIsoString,
+  readSecrets,
+  readWorkspace,
+  resolveWorkspace,
+  setActiveProject,
+  workspacePaths,
+  getActiveProject,
+} from './workspace.js';
+import { rpcUrl, startServer } from './surreal-server.js';
+
+export const PROJECT_SCHEMA_VERSION = 1;
+export const PROJECT_CONFIG_FILE = '.fancy-project.json';
+export const REGISTRY_TABLE = 'project';
+
+export class ProjectError extends Error {
+  constructor(code, message, detail) {
+    super(message);
+    this.name = 'ProjectError';
+    this.code = code;
+    this.detail = detail;
+  }
+}
+
+export function projectRecordId(projectId) {
+  return 'project:' + String(projectId).replace(/-/g, '_');
+}
+
+export function databaseName(projectId) {
+  return 'proj_' + String(projectId).replace(/-/g, '');
+}
+
+export function databaseUser(projectId) {
+  return 'u_' + String(projectId).replace(/-/g, '').slice(0, 16);
+}
+
+export function projectConfigPath(dir) {
+  return path.join(dir, PROJECT_CONFIG_FILE);
+}
+
+export function readProjectConfig(dir) {
+  const file = projectConfigPath(dir);
+  const data = readJsonFile(file);
+  if (!data || typeof data !== 'object' || Array.isArray(data) || !data.project_id) return null;
+  return data;
+}
+
+export function writeProjectConfig(dir, config) {
+  fs.mkdirSync(dir, { recursive: true });
+  return atomicWriteJson(projectConfigPath(dir), {
+    schema_version: PROJECT_SCHEMA_VERSION,
+    ...config,
+  }, { mode: 0o600 });
+}
+
+export function projectDir(root, projectId) {
+  return workspacePaths(root).project(projectId);
+}
+
+function normalizeProject(row) {
+  if (!row) return null;
+  const topic = row.topic && typeof row.topic === 'object' ? row.topic : {};
+  return {
+    ...row,
+    project_id: String(row.project_id || ''),
+    name: String(row.name || topic.book_name || row.project_id || ''),
+    db_name: String(row.db_name || ''),
+    db_user: String(row.db_user || ''),
+    dir: String(row.dir || ''),
+    status: String(row.status || 'active'),
+    topic,
+  };
+}
+
+async function connectAdmin(root) {
+  // startServer verifies the root credentials, and moves the workspace to a
+  // free port if another (possibly stale) SurrealDB is squatting on ours.
+  await startServer(root);
+  const workspace = readWorkspace(root);
+  const secrets = readSecrets(root);
+  const db = new Surreal({ engines: { ...createRemoteEngines() } });
+  await db.connect(rpcUrl(workspace));
+  await db.signin({
+    username: secrets.surreal?.username || 'root',
+    password: secrets.surreal?.password,
+  });
+  return { db, workspace, secrets };
+}
+
+const REGISTRY_DDL = [
+  `DEFINE TABLE IF NOT EXISTS ${REGISTRY_TABLE} SCHEMAFULL;`,
+  `DEFINE FIELD IF NOT EXISTS project_id ON ${REGISTRY_TABLE} TYPE string;`,
+  `DEFINE FIELD IF NOT EXISTS name ON ${REGISTRY_TABLE} TYPE string;`,
+  `DEFINE FIELD IF NOT EXISTS db_name ON ${REGISTRY_TABLE} TYPE string;`,
+  `DEFINE FIELD IF NOT EXISTS db_user ON ${REGISTRY_TABLE} TYPE string;`,
+  `DEFINE FIELD IF NOT EXISTS dir ON ${REGISTRY_TABLE} TYPE string;`,
+  `DEFINE FIELD IF NOT EXISTS status ON ${REGISTRY_TABLE} TYPE string DEFAULT 'active';`,
+  `DEFINE FIELD IF NOT EXISTS topic ON ${REGISTRY_TABLE} TYPE object FLEXIBLE;`,
+  `DEFINE FIELD IF NOT EXISTS created_at ON ${REGISTRY_TABLE} TYPE datetime VALUE time::now();`,
+  `DEFINE FIELD IF NOT EXISTS updated_at ON ${REGISTRY_TABLE} TYPE datetime VALUE time::now();`,
+  `DEFINE INDEX IF NOT EXISTS project_id_unique ON ${REGISTRY_TABLE} FIELDS project_id UNIQUE;`,
+  `DEFINE INDEX IF NOT EXISTS project_name_unique ON ${REGISTRY_TABLE} FIELDS name UNIQUE;`,
+];
+
+export async function withRegistry(root, fn) {
+  const { db, workspace } = await connectAdmin(root);
+  try {
+    await db.use({ namespace: workspace.surreal?.namespace || 'fancy', database: workspace.surreal?.registry_database || 'fancy_registry' });
+    for (const statement of REGISTRY_DDL) await db.query(statement);
+    return await fn(db, workspace);
+  } finally {
+    try { await db.close(); } catch { /* ignore */ }
+  }
+}
+
+export async function listProjects(root) {
+  return withRegistry(root, async (db) => {
+    const rows = await db.query(`SELECT * FROM ${REGISTRY_TABLE} WHERE status != 'deleted' ORDER BY name;`);
+    const list = Array.isArray(rows) && Array.isArray(rows[0]) ? rows[0] : [];
+    return list.map(normalizeProject);
+  });
+}
+
+export async function getProject(root, nameOrId) {
+  const key = String(nameOrId || '').trim();
+  if (!key) return null;
+  const projects = await listProjects(root);
+  const lower = key.toLowerCase();
+  return projects.find((row) => row.name === key
+    || row.project_id === key
+    || row.db_name === key
+    || row.dir === key
+    || path.basename(row.dir) === key
+    || row.name.toLowerCase() === lower)
+    || projects.find((row) => row.name.includes(key) || row.project_id.startsWith(key))
+    || null;
+}
+
+export async function findProjects(root, query) {
+  const key = String(query || '').trim();
+  const projects = await listProjects(root);
+  if (!key) return projects;
+  const lower = key.toLowerCase();
+  return projects.filter((row) => row.name.toLowerCase().includes(lower)
+    || row.project_id.startsWith(key)
+    || row.db_name === key
+    || path.basename(row.dir) === key);
+}
+
+async function assertNameAvailable(db, name, exceptId = null) {
+  const rows = await db.query(
+    `SELECT project_id, name FROM ${REGISTRY_TABLE} WHERE name = $name AND status != 'deleted';`,
+    { name },
+  );
+  const list = Array.isArray(rows) && Array.isArray(rows[0]) ? rows[0] : [];
+  const conflict = list.find((row) => row.project_id !== exceptId);
+  if (conflict) {
+    throw new ProjectError('PROJECT_NAME_EXISTS', '小说名已被占用：' + name + '（已有项目 UUID：' + conflict.project_id + '）。请更换小说名后重试。');
+  }
+}
+
+export async function createProject(root, opts = {}) {
+  const title = String(opts.title || opts.name || '').trim();
+  if (!title) throw new ProjectError('PROJECT_TITLE_REQUIRED', '创建项目时必须提供小说名。');
+  const projectId = opts.projectId || crypto.randomUUID();
+  const dbName = databaseName(projectId);
+  const dbUser = databaseUser(projectId);
+  const password = crypto.randomBytes(24).toString('base64url');
+  const dir = projectDir(root, projectId);
+  const topic = opts.topic && typeof opts.topic === 'object' ? opts.topic : {};
+
+  const result = await withRegistry(root, async (db, workspace) => {
+    await assertNameAvailable(db, title);
+    const namespace = workspace.surreal?.namespace || 'fancy';
+    await db.query(`DEFINE DATABASE IF NOT EXISTS ${dbName};`);
+    // DEFINE USER ... ON DATABASE is resolved against the connection's
+    // currently selected database. Switch to the project database before
+    // defining its user, then switch back to the registry.
+    await db.use({ namespace, database: dbName });
+    await db.query(`DEFINE USER IF NOT EXISTS ${dbUser} ON DATABASE PASSWORD '${password}' ROLES VIEWER;`);
+    await db.use({ namespace, database: workspace.surreal?.registry_database || 'fancy_registry' });
+    const created = await db.query(
+      `CREATE ${REGISTRY_TABLE} SET
+         project_id = $project_id,
+         name = $name,
+         db_name = $db_name,
+         db_user = $db_user,
+         dir = $dir,
+         status = 'active',
+         topic = $topic;`,
+      { project_id: projectId, name: title, db_name: dbName, db_user: dbUser, dir, topic },
+    );
+    const row = Array.isArray(created) && Array.isArray(created[0]) ? created[0][0] : null;
+    return { row: normalizeProject(row), namespace };
+  });
+
+  fs.mkdirSync(dir, { recursive: true });
+  fs.mkdirSync(path.join(dir, '导出'), { recursive: true });
+  fs.mkdirSync(path.join(dir, '设定'), { recursive: true });
+  writeProjectConfig(dir, {
+    project_id: projectId,
+    workspace: root,
+    database: dbName,
+    namespace: result.namespace,
+    db_user: dbUser,
+    created_at: localIsoString(),
+  });
+
+  const secrets = readSecrets(root);
+  secrets.projects = secrets.projects || {};
+  secrets.projects[projectId] = { username: dbUser, password };
+  atomicWriteJson(path.join(root, '.fancy', 'secrets.json'), secrets, { mode: 0o600 });
+
+  if (opts.makeActive !== false) setActiveProject(root, projectId);
+  return getProject(root, projectId);
+}
+
+export async function updateProjectTopic(root, nameOrId, topic) {
+  const project = await getProject(root, nameOrId);
+  if (!project) throw new ProjectError('PROJECT_NOT_FOUND', '未找到匹配的项目：' + nameOrId + '。可运行 /fancy-project list 查看当前工作区的项目。');
+  const value = topic && typeof topic === 'object' ? topic : {};
+  await withRegistry(root, async (db) => {
+    await db.query(
+      `UPDATE ${REGISTRY_TABLE} SET topic = $topic, updated_at = time::now() WHERE project_id = $id;`,
+      { topic: value, id: project.project_id },
+    );
+  });
+  try {
+    const { openStoryGraph } = await import('../plugins/fancy-story-builder/storydb/client.js');
+    const sdb = await openStoryGraph(project.dir, { project, admin: true });
+    try {
+      await sdb.query(
+        `UPDATE story:main SET payload.topic = $topic, title = $title, updated_at = time::now();`,
+        { topic: value, title: value.book_name || value.title || project.name },
+      );
+    } finally {
+      await sdb.close();
+    }
+  } catch { /* registry is authoritative for lookup */ }
+  return getProject(root, project.project_id);
+}
+
+export async function renameProject(root, nameOrId, newTitle) {
+  const title = String(newTitle || '').trim();
+  if (!title) throw new ProjectError('PROJECT_TITLE_REQUIRED', '新小说名不能为空。');
+  const project = await getProject(root, nameOrId);
+  if (!project) throw new ProjectError('PROJECT_NOT_FOUND', '未找到匹配的项目：' + nameOrId + '。可运行 /fancy-project list 查看当前工作区的项目。');
+  const topic = {
+    ...(project.topic || {}),
+    book_name: title,
+    title,
+  };
+  await withRegistry(root, async (db) => {
+    await assertNameAvailable(db, title, project.project_id);
+    await db.query(
+      `UPDATE ${REGISTRY_TABLE} SET name = $name, topic = $topic, updated_at = time::now()
+       WHERE project_id = $id;`,
+      { name: title, topic, id: project.project_id },
+    );
+  });
+  // Keep the story record in sync; if the project DB is unavailable, the
+  // registry rename still wins and the next open will repair it.
+  try {
+    const { openStoryGraph } = await import('../plugins/fancy-story-builder/storydb/client.js');
+    const sdb = await openStoryGraph(project.dir, { admin: true });
+    try {
+      await sdb.query('UPDATE story:main SET title = $title, updated_at = time::now();', { title });
+    } finally {
+      await sdb.close();
+    }
+  } catch { /* registry is authoritative for lookup */ }
+  return getProject(root, project.project_id);
+}
+
+export async function deleteProject(root, nameOrId, opts = {}) {
+  const project = await getProject(root, nameOrId);
+  if (!project) throw new ProjectError('PROJECT_NOT_FOUND', '未找到匹配的项目：' + nameOrId + '。可运行 /fancy-project list 查看当前工作区的项目。');
+  if (opts.confirm !== true) {
+    throw new ProjectError('CONFIRMATION_REQUIRED', '删除项目需要明确确认（confirm=true）。', { project });
+  }
+  await withRegistry(root, async (db) => {
+    await db.query(`REMOVE DATABASE ${project.db_name};`);
+    await db.query(`UPDATE ${REGISTRY_TABLE} SET status = 'deleted', updated_at = time::now() WHERE project_id = $id;`, { id: project.project_id });
+  });
+  const dir = path.resolve(project.dir || projectDir(root, project.project_id));
+  if (opts.purge === true) {
+    try { fs.rmSync(dir, { recursive: true, force: true }); } catch { /* ignore */ }
+  } else if (fs.existsSync(dir)) {
+    const trash = path.join(workspacePaths(root).fancy, 'trash', `${project.project_id}-${Date.now()}`);
+    fs.mkdirSync(path.dirname(trash), { recursive: true });
+    try { fs.renameSync(dir, trash); } catch { /* ignore */ }
+  }
+  const active = getActiveProject(root);
+  if (active === project.project_id) setActiveProject(root, null);
+  return { project, removed: true, purged: opts.purge === true };
+}
+
+export async function setProjectActive(root, nameOrId) {
+  const project = await getProject(root, nameOrId);
+  if (!project) throw new ProjectError('PROJECT_NOT_FOUND', '未找到匹配的项目：' + nameOrId + '。可运行 /fancy-project list 查看当前工作区的项目。');
+  setActiveProject(root, project.project_id);
+  return project;
+}
+
+export function projectFromCwd(root, cwd) {
+  if (!cwd) return null;
+  let current = path.resolve(cwd);
+  const workspace = path.resolve(root);
+  for (;;) {
+    const config = readProjectConfig(current);
+    if (config) return { config, dir: current };
+    if (current === workspace) return null;
+    const parent = path.dirname(current);
+    if (parent === current) return null;
+    current = parent;
+  }
+}
+
+/**
+ * Resolve a project without requiring the registry when the cwd already points
+ * at a project directory. If only the workspace is known, use the active
+ * project or, when there is exactly one, that project.
+ */
+export async function resolveProject(opts = {}) {
+  const root = opts.workspace || resolveWorkspace({ cwd: opts.cwd, explicit: opts.explicitWorkspace, required: true });
+  const fromCwd = projectFromCwd(root, opts.cwd || process.cwd());
+  if (fromCwd) {
+    let record = null;
+    try { record = await getProject(root, fromCwd.config.project_id); } catch { /* registry may be down */ }
+    return {
+      ...(record || {}),
+      project_id: fromCwd.config.project_id,
+      name: (record && record.name) || fromCwd.config.project_id,
+      db_name: fromCwd.config.database,
+      db_user: fromCwd.config.db_user,
+      dir: fromCwd.dir,
+      config: fromCwd.config,
+      workspace: root,
+    };
+  }
+
+  if (opts.project) {
+    const project = await getProject(root, opts.project);
+    if (!project) throw new ProjectError('PROJECT_NOT_FOUND', '未找到匹配的项目：' + opts.project + '。可运行 /fancy-project list 查看当前工作区的项目。');
+    return { ...project, config: readProjectConfig(project.dir), workspace: root };
+  }
+
+  const active = getActiveProject(root);
+  if (active) {
+    const project = await getProject(root, active);
+    if (project) return { ...project, config: readProjectConfig(project.dir), workspace: root };
+  }
+
+  const projects = await listProjects(root);
+  if (projects.length === 1) {
+    return { ...projects[0], config: readProjectConfig(projects[0].dir), workspace: root };
+  }
+  if (!projects.length) {
+    throw new ProjectError('PROJECT_NOT_FOUND', '当前工作区尚未创建任何项目。请先运行 /fancy-topic setup 创建项目。');
+  }
+  throw new ProjectError(
+    'PROJECT_AMBIGUOUS',
+    '当前工作区存在多个项目，无法自动选择。请使用 --project <小说名|UUID> 指定项目，或先运行 /fancy-project use <小说名|UUID> 设置当前项目。',
+    { projects: projects.map((p) => ({ id: p.project_id, name: p.name })) },
+  );
+}
+
+export async function ensureProjectSchemaReady(project, opts = {}) {
+  const { openStoryGraph } = await import('../plugins/fancy-story-builder/storydb/client.js');
+  const sdb = await openStoryGraph(project.dir, { admin: true, ...opts });
+  try {
+    return sdb;
+  } catch (err) {
+    await sdb.close();
+    throw err;
+  }
+}
+
+export { WorkspaceError };

package/lib/scraper-utils.js

@@ -46,6 +46,126 @@ export function normalizeScanMode(raw) {
   return SCAN_MODES.includes(s) ? s : DEFAULT_SCAN_MODE;
 }
 
+/**
+ * Markdown 元字符的“保形替换表”。
+ *
+ * 扫榜数据来自平台页面，书名/作者/简介里可能出现 `#`、`*`、`[]`、`---`、
+ * `1.` 等写法。直接写进 Markdown 会变成标题、列表、分隔线或链接语法，
+ * 渲染后就不再是原文。统一替换成全角等价字符：看起来仍是同一个符号，
+ * 但不再触发 Markdown 语法。
+ */
+const MARKDOWN_REPLACEMENTS = {
+  '#': '＃', '*': '＊', '[': '［', ']': '］', '_': '＿',
+  '`': '｀', '>': '＞', '<': '＜', '|': '｜', '~': '～',
+  '\\': '＼', '-': '－', '+': '＋', '=': '＝', '&': '＆',
+};
+const MARKDOWN_INLINE_RE = /[#*\[\]_`<>|~\\\-+=&]/g;
+
+/**
+ * 处理单行/内联文本里的 Markdown 元字符。
+ * 保留换行；用于作者、书名、标签、分类名等单行字段。
+ */
+export function sanitizeMarkdownInline(value) {
+  return String(value == null ? '' : value)
+    .replace(MARKDOWN_INLINE_RE, (ch) => MARKDOWN_REPLACEMENTS[ch] || ch);
+}
+
+/**
+ * 处理一行正文：先拆掉行首的有序列表标记（`1. `），再处理其余元字符。
+ * 无序列表/标题/引用/分隔线的标记字符已在 sanitizeMarkdownInline 中替换。
+ */
+export function sanitizeMarkdownLine(value) {
+  const text = String(value == null ? '' : value)
+    .replace(/^[\s　]+|[\s　]+$/g, '')
+    // 1. / 1) 都算有序列表标记；即使点号后没有空格也要处理
+    .replace(/^(\d+)[.)](?=\s|$)/, (_m, num) => num + '．');
+  return sanitizeMarkdownInline(text);
+}
+
+/** 把一本书里要写进 Markdown 的单行字段做保形替换。 */
+export function sanitizeBookFields(book) {
+  if (!book || typeof book !== 'object') return book;
+  const out = { ...book };
+  for (const key of [
+    'title', 'author', 'genre', 'type', 'status', 'words', 'wordCount',
+    'group', 'category', 'subGenre', 'rankChange',
+  ]) {
+    if (typeof out[key] === 'string') out[key] = sanitizeMarkdownInline(out[key]);
+  }
+  if (Array.isArray(out.tags)) out.tags = out.tags.map(sanitizeMarkdownInline);
+  return out;
+}
+
+/**
+ * 把作品简介渲染成 markdown 行（四个平台共用一套格式）。
+ *
+ * ## 为什么抽出来（2026-10-29）
+ *
+ * 四个平台的简介此前各有各的写法：
+ *
+ *   起点  `**简介：** 首段` + 后续段落各自成段    ← 只有起点带冒号
+ *   番茄  空行 + `**简介**` + 简介全文整块
+ *   晋江  空行 + `**简介**` + 简介全文整块
+ *   七猫  空行 + `**简介**` + 分段
+ *
+ * 同一个插件产出的四份文件，字段格式却对不上：
+ * 番茄/晋江/七猫的"简介"没有冒号（不像 `**标签：**` 那样一眼看出是字段），
+ * 而且简介前多一个空行，跟前面的字段行断开了。
+ *
+ * 统一成一个无序列表项（2026-11-05 调整）：
+ *
+ *   - **简介：** 首段
+ *
+ *     第二段
+ *
+ * 规则：
+ *   - 简介本身是一个 `- ` 列表项，和作者/状态/在读等字段同一层级；
+ *   - 单换行 = 段内硬换行（行尾两个空格），四个平台统一；
+ *   - 空行 = 段落分隔，用缩进 + 空行留在同一个列表项里；
+ *   - **每段首尾的空白字符都去掉**（半角空格 / 全角空格 / 换行）。
+ *
+ * @param {string|string[]} raw - 简介正文，或已切好的段落数组
+ * @returns {string[]} markdown 行；没有简介时返回空数组
+ */
+export function renderIntroLines(raw) {
+  const source = Array.isArray(raw)
+    ? raw
+    : String(raw == null ? '' : raw)
+      .replace(/\r\n?/g, '\n')
+      // 只有空行（允许空行里夹空格/全角空格）才是段落分隔
+      .replace(/\n[ \t　]*\n/g, '\n\n')
+      .split(/\n{2,}/);
+  const paras = source
+    // 去掉每段首尾的空白字符：\s 已含全角空格 U+3000，这里再写一遍是防它被漏
+    .map(p => String(p).replace(/^[\s　]+|[\s　]+$/g, ''))
+    .filter(Boolean);
+  if (!paras.length) return [];
+
+  // 简介本身是列表项；段落之间用空行 + 两格缩进，段落内部的单换行
+  // （番茄/起点的 <br> 转出来的）用两个尾随空格做硬换行。
+  const lines = [];
+  let first = true;
+  for (const para of paras) {
+    const parts = String(para)
+      .split('\n')
+      .map((line) => sanitizeMarkdownLine(line.trim()))
+      .filter(Boolean);
+    for (let i = 0; i < parts.length; i += 1) {
+      const text = parts[i];
+      if (first) {
+        lines.push('- **简介：** ' + text);
+        first = false;
+      } else if (i === 0) {
+        lines.push('', '  ' + text);
+      } else {
+        lines[lines.length - 1] += '  ';
+        lines.push('  ' + text);
+      }
+    }
+  }
+  return lines;
+}
+
 /**
  * 归一化作品简介文本。
  *
@@ -77,6 +197,26 @@ export function normalizeBookIntro(raw) {
     .trim();
 }
 
+/**
+ * 去掉数字里的千分位逗号。
+ *
+ * 晋江榜单页的积分是 "1,099,899,136" 这样带逗号的（10 亿级必带），
+ * 字数也可能到百万级带逗号。采集文件里存纯数字：
+ *   - 下游才好比较 / 求和 / 算份额；
+ *   - 模型照抄数字时不会把逗号一起抄走（"327,650,784" 和 "327650784"
+ *     在模型的算术里不是一回事）。
+ *
+ * 只处理"数字 + 逗号 + 数字"的形态，其余字符原样保留 ——
+ * 英文正文里的逗号（"共 3, 本"）后面跟的不是数字，不会被动到。
+ *
+ * @param {string|number|null|undefined} raw
+ * @returns {string}
+ */
+export function stripThousands(raw) {
+  if (raw === null || raw === undefined) return '';
+  return String(raw).replace(/(\d),(?=\d)/g, '$1');
+}
+
 /**
  * 本地日期戳 YYYYMMDD（本地时区，非 UTC）。
  *