hm-doc-tool 0.3.0

package/dist/download/downloader.js

@@ -0,0 +1,251 @@
+"use strict";
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.downloadSingle = downloadSingle;
+exports.downloadSubtree = downloadSubtree;
+/**
+ * 文档下载调度模块
+ *
+ * 职责：
+ * 1. 获取目录树 → 平铺 → 过滤 → 计算路径 → 并发下载 → 写入文件
+ * 2. 支持 --dry-run（只打印树形结构）
+ * 3. 单个文档失败不中断，最后汇总输出
+ */
+const fs = __importStar(require("fs/promises"));
+const path = __importStar(require("path"));
+const api_client_js_1 = require("../api-client.js");
+const config_js_1 = require("../common/config.js");
+const markdown_converter_js_1 = require("./markdown-converter.js");
+const tree_processor_js_1 = require("../common/tree-processor.js");
+const path_generator_js_1 = require("../common/path-generator.js");
+const tree_formatter_js_1 = require("../common/tree-formatter.js");
+/**
+ * 通过 URL 下载单篇文档（不建目录，直接写入 outputDir）
+ *
+ * @param objectId - 文档 objectId
+ * @param catalogName - catalog 名称
+ * @param nodeName - 节点名称（用作文件名）
+ * @param outputDir - 输出目录
+ */
+async function downloadSingle(objectId, catalogName, nodeName, outputDir) {
+    const doc = await (0, api_client_js_1.fetchDocument)(objectId);
+    let md = (0, markdown_converter_js_1.htmlToMarkdown)(doc.html);
+    const sourceUrl = `https://developer.huawei.com/consumer/cn/doc/${doc.catalogName}/${doc.objectId}`;
+    const header = [
+        "---",
+        `title: ${nodeName}`,
+        `source: ${sourceUrl}`,
+        `category: ${(0, config_js_1.findCatalogName)(doc.catalogName) || doc.catalogName}`,
+        `downloaded_at: ${new Date().toLocaleDateString("sv-SE")}`,
+        "---",
+        "",
+    ].join("\n");
+    md = header + (md.startsWith("\n") ? md : "\n" + md);
+    // 确保输出目录存在
+    await fs.mkdir(outputDir, { recursive: true });
+    const filePath = path.join(outputDir, (0, path_generator_js_1.safeFileName)(nodeName) + ".md");
+    await fs.writeFile(filePath, md, "utf-8");
+    console.log(`下载完成: ${nodeName}.md`);
+}
+/**
+ * 并发执行任务，使用简单信号量模式控制并发数
+ * @param tasks - 任务列表
+ * @param concurrency - 最大并发数
+ * @param fn - 执行函数
+ */
+async function runConcurrent(tasks, concurrency, fn) {
+    let running = 0;
+    let nextIndex = 0;
+    return new Promise((resolve, reject) => {
+        function scheduleNext() {
+            while (running < concurrency && nextIndex < tasks.length) {
+                const idx = nextIndex++;
+                running++;
+                fn(tasks[idx], idx)
+                    .then(() => {
+                    running--;
+                    if (running === 0 && nextIndex >= tasks.length) {
+                        resolve();
+                    }
+                    else {
+                        scheduleNext();
+                    }
+                })
+                    .catch(reject);
+            }
+            if (tasks.length === 0)
+                resolve();
+        }
+        scheduleNext();
+    });
+}
+/**
+ * 下载单篇文档并写入文件
+ *
+ * 流程：fetchDocument → htmlToMarkdown → 插入元信息头 → writeFile
+ *
+ * @param task - 下载任务
+ * @param total - 任务总数
+ * @param current - 当前序号
+ */
+async function downloadOne(task, total, current) {
+    if (!task.relateDocument)
+        return;
+    const doc = await (0, api_client_js_1.fetchDocument)(task.relateDocument);
+    let md = (0, markdown_converter_js_1.htmlToMarkdown)(doc.html);
+    const sourceUrl = `https://developer.huawei.com/consumer/cn/doc/${doc.catalogName}/${doc.objectId}`;
+    // Build metadata front-matter (consistent with harmony-doc2)
+    const header = [
+        "---",
+        `title: ${task.nodeName}`,
+        `source: ${sourceUrl}`,
+        `category: ${(0, config_js_1.findCatalogName)(doc.catalogName) || doc.catalogName}`,
+        `downloaded_at: ${new Date().toLocaleDateString("sv-SE")}`,
+        "---",
+        "",
+    ].join("\n");
+    md = header + (md.startsWith("\n") ? md : "\n" + md);
+    const filePath = path.join(task.outputDir, task.fileName);
+    await fs.writeFile(filePath, md, "utf-8");
+    console.log(`[${current}/${total}] 下载: ${task.nodeName}.md`);
+}
+/**
+ * 下载文档子树
+ *
+ * 完整流程：
+ * 1. 获取目录树 → 平铺为 FlatNode[]
+ * 2. 定位目标节点（按 nodeId 或 nodeName）
+ * 3. 收集子树节点
+ * 4. C++ 过滤（--no-cpp，含后代）
+ * 5. leafOnly 过滤
+ * 6. 计算输出路径（用全量 nodes 算路径，用过滤后 subtreeNodes 生成任务）
+ * 7. --dry-run 则打印树形结构后返回
+ * 8. 创建所有目录（先建目录避免并发竞态）
+ * 9. 并发下载文档并写入文件
+ * 10. 输出汇总（成功数、失败列表）
+ *
+ * @param catalogName - catalog 名称
+ * @param rootId - 子树根节点 ID，undefined 表示整个 catalog
+ * @param nodeName - 按名称定位节点（与 nodeId 二选一）
+ * @param nodeId - 按 ID 定位节点（优先于 nodeName）
+ * @param options - 下载选项
+ */
+async function downloadSubtree(catalogName, rootId, nodeName, nodeId, options) {
+    const { outputDir, concurrency, leafOnly, noCpp, dryRun, topLevel } = options;
+    // 构建实际输出目录：{outputDir}/{topLevel}/
+    // catalog 名由 path-generator.ts 自动添加
+    let actualOutputDir = outputDir;
+    if (topLevel) {
+        actualOutputDir = path.join(outputDir, topLevel);
+    }
+    // 1. 获取目录树
+    console.log(`正在获取目录树: ${catalogName} ...`);
+    const nodes = await (0, tree_processor_js_1.fetchCatalogTree)(catalogName);
+    console.log(`目录树节点数: ${nodes.length}`);
+    // 2. 定位目标节点
+    let targetRootId = rootId;
+    if (!targetRootId && (nodeName || nodeId)) {
+        const target = (0, tree_processor_js_1.findNode)(nodes, nodeId, nodeName);
+        if (!target) {
+            throw new Error(`找不到节点: ${nodeId || nodeName}`);
+        }
+        targetRootId = target.nodeId;
+    }
+    // 3. 收集子树
+    let subtreeNodes;
+    if (targetRootId) {
+        subtreeNodes = (0, tree_processor_js_1.collectSubtree)(nodes, targetRootId);
+    }
+    else {
+        subtreeNodes = [...nodes];
+    }
+    // 4. C++ 过滤（含后代）
+    if (noCpp) {
+        const skipSet = (0, tree_processor_js_1.buildCppSkipSet)(nodes, catalogName, noCpp);
+        subtreeNodes = subtreeNodes.filter((n) => !skipSet.has(n.nodeId));
+    }
+    // 5. leafOnly 过滤
+    if (leafOnly) {
+        subtreeNodes = subtreeNodes.filter((n) => n.isLeaf || n.children.length === 0);
+    }
+    // 6. 计算输出路径
+    //    nodes（全量）用于路径计算以保证父子关系完整
+    //    subtreeNodes（过滤后）用于决定生成哪些任务
+    const tasks = (0, path_generator_js_1.buildOutputPaths)(nodes, targetRootId, catalogName, actualOutputDir, subtreeNodes);
+    const downloadTasks = tasks.filter((t) => t.relateDocument && t.relateDocument !== "_");
+    // 7. dry-run：打印树形结构后返回
+    if (dryRun) {
+        const treeOutput = (0, tree_formatter_js_1.buildTreeOutput)(nodes, targetRootId, catalogName, new Set(), leafOnly, undefined, noCpp);
+        console.log(treeOutput);
+        console.log(`\n输出目录: ${actualOutputDir}`);
+        console.log(`共 ${downloadTasks.length} 篇文档待下载`);
+        return undefined;
+    }
+    // 8. 创建所有目录
+    //    nodes（全量）用于路径计算，subtreeNodes 用于决定创建哪些目录
+    const dirs = (0, path_generator_js_1.collectDirectories)(nodes, targetRootId, catalogName, actualOutputDir, subtreeNodes);
+    for (const dir of dirs) {
+        await fs.mkdir(dir, { recursive: true });
+    }
+    // 9. 并发下载
+    console.log(`开始下载 ${downloadTasks.length} 篇文档 (并发数: ${concurrency}) ...`);
+    const failed = [];
+    let completed = 0;
+    await runConcurrent(downloadTasks, concurrency, async (task) => {
+        try {
+            completed++;
+            await downloadOne(task, downloadTasks.length, completed);
+        }
+        catch (err) {
+            failed.push({
+                nodeId: task.nodeId,
+                nodeName: task.nodeName,
+                error: err instanceof Error ? err.message : String(err),
+            });
+        }
+    });
+    // 10. 汇总
+    const successCount = downloadTasks.length - failed.length;
+    console.log(`\n完成: 成功 ${successCount} 篇, 失败 ${failed.length} 篇`);
+    if (failed.length > 0) {
+        console.log("\n失败列表:");
+        for (const f of failed) {
+            console.log(`  - ${f.nodeName} (${f.nodeId}): ${f.error}`);
+        }
+    }
+    // 返回任务列表和节点数据，供 summary 生成和链接本地化使用
+    // outputDir 是用户指定的原始目录，summary 文件放在那里
+    return { tasks: downloadTasks, nodes, outputDir };
+}

package/dist/download/index.d.ts

@@ -0,0 +1,16 @@
+/** Download 命令选项 */
+export interface DownloadOptions {
+    catalog?: string[];
+    nodeName?: string;
+    nodeId?: string;
+    url?: string;
+    output: string;
+    leafOnly: boolean;
+    cpp: boolean;
+    concurrency: string;
+    dryRun: boolean;
+    summary: boolean;
+    localize: boolean;
+}
+/** Download 命令处理函数 */
+export declare function handleDownload(opts: DownloadOptions): Promise<void>;

package/dist/download/index.js

@@ -0,0 +1,168 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.handleDownload = handleDownload;
+/**
+ * Download 命令入口
+ *
+ * 职责：
+ * 1. 处理 CLI 选项
+ * 2. 调用 downloader 执行下载
+ */
+const tree_processor_js_1 = require("../common/tree-processor.js");
+const config_js_1 = require("../common/config.js");
+const downloader_js_1 = require("./downloader.js");
+const api_client_js_1 = require("../api-client.js");
+const config_js_2 = require("../common/config.js");
+const summary_generator_js_1 = require("./summary-generator.js");
+const link_localizer_js_1 = require("./link-localizer.js");
+/** Download 命令处理函数 */
+async function handleDownload(opts) {
+    try {
+        if (opts.url && opts.catalog) {
+            console.error("错误: --url 和 --catalog 互斥，请只使用其中一个");
+            process.exit(1);
+        }
+        let catalogName;
+        let targetNodeId;
+        let targetNodeName;
+        if (opts.url) {
+            // URL 模式：单文件下载，不建目录
+            const objectId = (0, api_client_js_1.objectIdFromUrl)(opts.url);
+            if (!objectId) {
+                console.error("错误: 无法从 URL 提取 objectId");
+                process.exit(1);
+            }
+            catalogName = (0, api_client_js_1.catalogNameFromUrl)(opts.url) ?? undefined;
+            if (!catalogName) {
+                // URL 中无法提取 catalogName → 遍历所有 catalog 查找
+                console.log("正在查找文档所属分类...");
+                for (const catName of (0, config_js_2.getAllCatalogIds)()) {
+                    try {
+                        const flatNodes = await (0, tree_processor_js_1.fetchCatalogTree)(catName);
+                        const found = flatNodes.find((n) => n.relateDocument === objectId);
+                        if (found) {
+                            catalogName = catName;
+                            targetNodeName = found.nodeName;
+                            break;
+                        }
+                    }
+                    catch {
+                        // 单个 catalog 查询失败则跳过
+                    }
+                }
+                if (!catalogName) {
+                    console.error("错误: 找不到文档所属分类");
+                    process.exit(1);
+                }
+            }
+            else {
+                // URL 中有 catalogName → 从目录树中查找节点名
+                const flatNodes = await (0, tree_processor_js_1.fetchCatalogTree)(catalogName);
+                const found = flatNodes.find((n) => n.relateDocument === objectId);
+                if (found) {
+                    targetNodeName = found.nodeName;
+                }
+                else {
+                    console.error(`错误: 在 ${catalogName} 中找不到文档 ${objectId}`);
+                    process.exit(1);
+                }
+            }
+            await (0, downloader_js_1.downloadSingle)(objectId, catalogName, targetNodeName, opts.output);
+        }
+        else if (opts.catalog) {
+            // catalog 模式（支持多个 -c）
+            const catalogs = opts.catalog;
+            targetNodeName = opts.nodeName;
+            targetNodeId = opts.nodeId;
+            // --no-cpp 在 commander 中会反转为 opts.cpp（false = 用户传了 --no-cpp）
+            const noCpp = opts.cpp === false;
+            const results = [];
+            for (const cat of catalogs) {
+                const topLevel = (0, config_js_1.findTopLevelForCatalog)(cat);
+                const result = await (0, downloader_js_1.downloadSubtree)(cat, targetNodeId, targetNodeName, targetNodeId, {
+                    outputDir: opts.output,
+                    concurrency: parseInt(opts.concurrency, 10),
+                    leafOnly: opts.leafOnly,
+                    noCpp,
+                    dryRun: opts.dryRun,
+                    topLevel,
+                });
+                if (!opts.dryRun && result) {
+                    results.push({ catalogId: cat, result });
+                }
+            }
+            if (results.length > 0) {
+                // 先生成 summary（使用原始链接提取摘要）
+                if (opts.summary) {
+                    for (const { catalogId, result } of results) {
+                        await (0, summary_generator_js_1.generateSummaryFromTasks)(catalogId, result.outputDir, result.tasks, result.nodes, targetNodeName);
+                    }
+                }
+                // 链接本地化（需显式指定 -l，支持跨 catalog）
+                if (opts.localize) {
+                    const globalPathMap = new Map();
+                    for (const { catalogId, result } of results) {
+                        (0, link_localizer_js_1.buildCatalogPathMap)(catalogId, result.tasks, globalPathMap);
+                    }
+                    for (const { result } of results) {
+                        await (0, link_localizer_js_1.localizeAllLinks)(result.tasks, globalPathMap);
+                    }
+                }
+            }
+        }
+        else {
+            // 批量模式：使用 docs_catalog.json 配置文件
+            const catalogsWithTopLevel = (0, config_js_1.getAllCatalogsWithTopLevel)();
+            if (catalogsWithTopLevel.length === 0) {
+                console.error("错误: docs_catalog.json 中没有配置任何分类");
+                process.exit(1);
+            }
+            // --no-cpp 在 commander 中会反转为 opts.cpp（false = 用户传了 --no-cpp）
+            const noCpp = opts.cpp === false;
+            // 先下载所有 catalog，收集结果
+            const results = [];
+            let currentTopLevel = "";
+            for (const { topLevel, catalog } of catalogsWithTopLevel) {
+                if (topLevel !== currentTopLevel) {
+                    currentTopLevel = topLevel;
+                    console.log(`\n=== ${topLevel} ===`);
+                }
+                console.log(`\n开始处理: ${catalog.name} (${catalog.id})`);
+                const result = await (0, downloader_js_1.downloadSubtree)(catalog.id, undefined, undefined, undefined, {
+                    outputDir: opts.output,
+                    concurrency: parseInt(opts.concurrency, 10),
+                    leafOnly: opts.leafOnly,
+                    noCpp,
+                    dryRun: opts.dryRun,
+                    topLevel,
+                });
+                if (!opts.dryRun && result) {
+                    results.push({ catalogId: catalog.id, result });
+                }
+            }
+            if (results.length > 0) {
+                // 先生成 summary（使用原始链接提取摘要）
+                if (opts.summary) {
+                    for (const { catalogId, result } of results) {
+                        await (0, summary_generator_js_1.generateSummaryFromTasks)(catalogId, result.outputDir, result.tasks, result.nodes);
+                    }
+                }
+                // 链接本地化（需显式指定 -l，支持跨 catalog）
+                if (opts.localize) {
+                    console.log("\n正在执行链接本地化 ...");
+                    const globalPathMap = new Map();
+                    for (const { catalogId, result } of results) {
+                        (0, link_localizer_js_1.buildCatalogPathMap)(catalogId, result.tasks, globalPathMap);
+                    }
+                    for (const { result } of results) {
+                        await (0, link_localizer_js_1.localizeAllLinks)(result.tasks, globalPathMap);
+                    }
+                }
+            }
+        }
+    }
+    catch (err) {
+        console.error("错误:", err instanceof Error ? err.message : String(err));
+        process.exit(1);
+    }
+}

package/dist/download/link-localizer.d.ts

@@ -0,0 +1,13 @@
+import type { OutputTask } from "../common/types.js";
+/**
+ * 构建 "catalogName/relateDocument" → 本地文件路径的映射（带 catalogName）
+ */
+export declare function buildCatalogPathMap(catalogName: string, tasks: OutputTask[], map?: Map<string, string>): Map<string, string>;
+/**
+ * 本地化单个 Markdown 文件中的链接
+ */
+export declare function localizeLinks(content: string, currentFilePath: string, pathMap: Map<string, string>): string;
+/**
+ * 对已下载的 Markdown 文件执行链接本地化
+ */
+export declare function localizeAllLinks(tasks: OutputTask[], pathMap: Map<string, string>): Promise<void>;

package/dist/download/link-localizer.js

@@ -0,0 +1,116 @@
+"use strict";
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.buildCatalogPathMap = buildCatalogPathMap;
+exports.localizeLinks = localizeLinks;
+exports.localizeAllLinks = localizeAllLinks;
+/**
+ * 链接本地化模块
+ *
+ * 将下载的 Markdown 中的华为文档绝对链接替换为本地相对路径
+ * 只替换 markdown 链接语法 [text](url) 中的 URL，保留 front-matter source 等元数据
+ * 支持跨 catalog 链接替换，支持 #锚点 保留
+ */
+const fs = __importStar(require("fs/promises"));
+const path = __importStar(require("path"));
+/**
+ * 匹配 markdown 链接中的华为文档 URL
+ * 只匹配 [text](url) 格式，不匹配 front-matter 中的裸 URL
+ *
+ * 捕获组：
+ * \1 = 链接文本
+ * \2 = catalogName（如 harmonyos-references）
+ * \3 = relateDocument（如 hiai-foundation-c）
+ * \4 = 可选的 #锚点（如 #获取应用文件路径）
+ */
+const MD_LINK_PATTERN = /\[([^\]]+)\]\(https:\/\/developer\.huawei\.com\/consumer\/cn\/doc\/([^/]+)\/([^)#]+)(#[^\)]+)?\)/g;
+/**
+ * 构建 "catalogName/relateDocument" → 本地文件路径的映射（带 catalogName）
+ */
+function buildCatalogPathMap(catalogName, tasks, map = new Map()) {
+    for (const task of tasks) {
+        if (!task.relateDocument)
+            continue;
+        const filePath = path.join(task.outputDir, task.fileName);
+        map.set(`${catalogName}/${task.relateDocument}`, filePath);
+    }
+    return map;
+}
+/**
+ * 本地化单个 Markdown 文件中的链接
+ */
+function localizeLinks(content, currentFilePath, pathMap) {
+    const currentDir = path.dirname(currentFilePath);
+    return content.replace(MD_LINK_PATTERN, (match, linkText, urlCatalogName, urlDocId, hash) => {
+        const key = `${urlCatalogName}/${urlDocId}`;
+        const targetPath = pathMap.get(key);
+        if (!targetPath)
+            return match; // 找不到对应文件，不替换
+        let relPath = path.relative(currentDir, targetPath);
+        // Windows 路径分隔符统一为 /
+        relPath = relPath.replace(/\\/g, "/");
+        // 确保以 ./ 开头
+        if (!relPath.startsWith("."))
+            relPath = "./" + relPath;
+        return `[${linkText}](${relPath}${hash || ""})`;
+    });
+}
+/**
+ * 对已下载的 Markdown 文件执行链接本地化
+ */
+async function localizeAllLinks(tasks, pathMap) {
+    let replacedCount = 0;
+    let fileCount = 0;
+    for (const task of tasks) {
+        const filePath = path.join(task.outputDir, task.fileName);
+        try {
+            const content = await fs.readFile(filePath, "utf-8");
+            const newContent = localizeLinks(content, filePath, pathMap);
+            if (newContent !== content) {
+                await fs.writeFile(filePath, newContent, "utf-8");
+                fileCount++;
+                const oldMatches = content.match(MD_LINK_PATTERN);
+                const newMatches = newContent.match(MD_LINK_PATTERN);
+                replacedCount += (oldMatches?.length ?? 0) - (newMatches?.length ?? 0);
+            }
+        }
+        catch {
+            // 文件不存在或读取失败，跳过
+        }
+    }
+    if (fileCount > 0) {
+        console.log(`链接本地化: ${fileCount} 个文件, ${replacedCount} 个链接已替换`);
+    }
+}

package/dist/download/markdown-converter.d.ts

@@ -0,0 +1 @@
+export declare function htmlToMarkdown(html: string): string;

package/dist/download/markdown-converter.js

@@ -0,0 +1,96 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.htmlToMarkdown = htmlToMarkdown;
+const turndown_1 = __importDefault(require("turndown"));
+const turndown_rules_js_1 = require("./turndown-rules.js");
+let instance = null;
+function getTurndown() {
+    if (!instance) {
+        instance = new turndown_1.default({
+            headingStyle: "atx",
+            codeBlockStyle: "fenced",
+            bulletListMarker: "-",
+        });
+        (0, turndown_rules_js_1.addCustomRules)(instance);
+    }
+    return instance;
+}
+function htmlToMarkdown(html) {
+    const md = getTurndown().turndown(html);
+    return cleanupMarkdown(md);
+}
+function cleanupMarkdown(md) {
+    let out = md;
+    // Fix heading levels from Huawei docs
+    // Rule: <h4>标题</h4> → ## 标题 (h2)
+    //       <h4>[h2]标题</h4> → ### 标题 (h3)
+    // Turndown escapes [ ] as \[ \], so we match the escaped form
+    // First convert #### \[h\d+\]... to ### (with tag removed)
+    out = out.replace(/^####\s+\\\[h\d+\\\]\s*(.+)$/gm, "### $1");
+    // Also handle unescaped form just in case
+    out = out.replace(/^####\s+\[h\d+\]\s*(.+)$/gm, "### $1");
+    // Then convert remaining #### to ##
+    out = out.replace(/^####\s+(.+)$/gm, "## $1");
+    // Fix escaped chars inside fenced code blocks
+    out = out.replace(/```(\w*)\n?([\s\S]*?)```/g, (_match, lang, code) => {
+        let fixed = code;
+        fixed = fixed.replace(/\\\\/g, "\x00BS\x00");
+        fixed = fixed.replace(/&#36;/g, "$");
+        fixed = fixed.replace(/&amp;/g, "&");
+        fixed = fixed.replace(/&lt;/g, "<");
+        fixed = fixed.replace(/&gt;/g, ">");
+        // Unescape markdown special chars
+        for (const ch of ["`", "*", "_", "{", "}", "[", "]", "(", ")", "#", "+", "-", ".", "!", "|", "~", ">", "<"]) {
+            fixed = fixed.replaceAll("\\" + ch, ch);
+        }
+        fixed = fixed.replace(/\x00BS\x00/g, "\\");
+        return "```" + lang + "\n" + fixed + "```";
+    });
+    // Remove standalone UI text lines
+    for (const pattern of [
+        /^展开$/gm, /^收起$/gm, /^复制$/gm, /^深色$/gm,
+        /^自动换行$/gm, /^深色代码主题$/gm, /^代码主题$/gm, /^收起代码$/gm,
+    ]) {
+        out = out.replace(pattern, "");
+    }
+    // Remove footer content - find footer start and truncate from there
+    const footerPatterns = [
+        /华为开发者联盟 版权所有/,
+        /版权所有/,
+        /Copyright.* Huawei/,
+        /\[使用条款\]\(https:\/\/developer\.huawei\.com\/consumer\/cn\/devservice\/use\)/,
+    ];
+    for (const pattern of footerPatterns) {
+        const match = out.match(pattern);
+        if (match && match.index !== undefined) {
+            const matchIndex = match.index;
+            const contentBeforeMatch = out.substring(0, matchIndex);
+            const lines = contentBeforeMatch.split("\n");
+            // Check if there's enough context (at least 10 lines)
+            if (lines.length > 10) {
+                // Try to find a better truncation point (e.g., empty line)
+                const linesAfterMatch = out.substring(matchIndex).split("\n");
+                let trimFrom = 0;
+                for (let i = 0; i < Math.min(5, linesAfterMatch.length); i++) {
+                    if (i > 0 && linesAfterMatch[i - 1].trim() === "") {
+                        trimFrom = i;
+                        break;
+                    }
+                }
+                out = out.substring(0, matchIndex) + linesAfterMatch.slice(0, trimFrom).join("\n");
+            }
+            break;
+        }
+    }
+    // Remove trailing navigation / ad text
+    out = out.replace(/\n\n在 指南 中进行搜索.*$/s, "");
+    out = out.replace(/\n\n智能客服.*$/s, "");
+    out = out.replace(/\n\n合作咨询.*$/s, "");
+    out = out.replace(/\n\n下载APP.*$/s, "");
+    // Collapse multiple blank lines
+    out = out.replace(/\n{3,}/g, "\n\n");
+    return out.trim();
+}