@aiello/wechat-to-markdown 1.2.12 → 1.2.13

package/dist/index.cjs

@@ -27,9 +27,10 @@ var __toCommonJS = (mod) => __copyProps(__defProp({}, "__esModule", { value: tru
 var src_exports = {};
 __export(src_exports, {
   Status: () => Status,
-  default: () => transformHtml2Markdown,
   getTurnDownService: () => getTurnDownService,
-  parseHTML: () => parseHTML
+  parseHTML: () => parseHTML,
+  transformHtml2Markdown: () => transformHtml2Markdown,
+  transformUrl2Markdown: () => transformUrl2Markdown
 });
 module.exports = __toCommonJS(src_exports);
 var import_axios = __toESM(require("axios"), 1);
@@ -192,7 +193,15 @@ async function parseHTML(htmlRaw, meta) {
   }
   return getError(400 /* Fail */);
 }
-async function transformHtml2Markdown(url, options = {}) {
+async function transformHtml2Markdown(html, url) {
+  try {
+    return parseHTML(html, { url });
+  } catch (err) {
+    console.log(err);
+    return getError(400 /* Fail */);
+  }
+}
+async function transformUrl2Markdown(url, options = {}) {
   const { axiosConfig = {} } = options;
   const { headers = {}, ...restConfig } = axiosConfig;
   const u = new URL(url);
@@ -209,7 +218,7 @@ async function transformHtml2Markdown(url, options = {}) {
       },
       ...restConfig
     });
-    return parseHTML(res.data, { url: u.href });
+    return transformHtml2Markdown(res.data, url);
   } catch (err) {
     console.log(err);
     return getError(400 /* Fail */);
@@ -219,6 +228,8 @@ async function transformHtml2Markdown(url, options = {}) {
 0 && (module.exports = {
   Status,
   getTurnDownService,
-  parseHTML
+  parseHTML,
+  transformHtml2Markdown,
+  transformUrl2Markdown
 });
 //# sourceMappingURL=index.cjs.map

package/dist/index.cjs.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts","../src/error.ts","../src/type.ts","../src/turndownCode.ts","../src/formatHtml.ts"],"sourcesContent":["import axios, { AxiosRequestConfig } from 'axios'\nimport { load } from 'cheerio'\nimport { errObj } from './error'\nimport type { TurnDownResult } from './type'\nimport { Status } from './type'\nimport { getTurnDownService } from './turndownCode'\n\nconst getError = (code: number) => {\n    return {\n        code,\n        success: false,\n        msg: errObj[code],\n    }\n}\n\nexport { TurnDownResult, Status }\n\nexport async function parseHTML(htmlRaw: string, meta: { url: string }) {\n    const $ = load(htmlRaw)\n\n    let title = $('#activity-name').text()\n\n    title = title.trim() || ''\n    const author = Array.from(\n        new Set(\n            [\n                $('meta[name=\"author\"]')?.attr('content'),\n                ...$('#js_name').text().split('\\n'),\n            ]\n                .map((item) => (item ? item.trim() : ''))\n                .filter(Boolean)\n        )\n    ).join('\\n')\n\n    const htmlEl = $('#js_content')\n    const html = htmlEl.html()\n\n    if (html && html.length > 0) {\n        let res = getTurnDownService(meta).turndown(html)\n\n        res = `## ${title} \\n \\n` + `## 作者 ${author} \\n \\n` + res\n\n        return {\n            success: true,\n            code: Status.Success,\n            data: {\n                title,\n                author,\n                content: res,\n            },\n        }\n    }\n\n    return getError(Status.Fail)\n}\n\n/**\n * 支持添加代理服务器\n */\ninterface TransformHtml2MarkdownOptions {\n    axiosConfig?: AxiosRequestConfig\n}\n\nexport default async function transformHtml2Markdown(\n    url: string,\n    options: TransformHtml2MarkdownOptions = {}\n): Promise<TurnDownResult> {\n    const { axiosConfig = {} } = options\n    const { headers = {}, ...restConfig } = axiosConfig\n\n    const u = new URL(url)\n    // 移除该参数\n    // 避免出现 302 跳转\n    u.searchParams.delete('poc_token')\n\n    try {\n        const res = await axios.get(u.href, {\n            timeout: 30000,\n            maxRedirects: 5,\n            headers: {\n                DNT: '1',\n                'Upgrade-Insecure-Requests': '1',\n                'User-Agent':\n                    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36',\n                ...headers,\n            },\n            ...restConfig,\n        })\n\n        return parseHTML(res.data, { url: u.href })\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\nexport { getTurnDownService } from './turndownCode'\n","export const errObj: {\n    [key: number]: string\n} = {\n    '400': '内容解析失败',\n}\n","export interface TurnDownResult {\n    success: boolean\n    code: number\n    data?: {\n        title?: string\n        author?: string\n        content?: string\n    }\n    msg?: string\n}\n\nexport const enum Status {\n    Success = 200,\n    Fail = 400,\n}\n","/**\n *  html 转换 markdown 格式\n */\nimport turnDownService from 'turndown'\nimport TurndownPluginGfm from '@guyplusplus/turndown-plugin-gfm'\nimport { formatCode, figure2markdown } from './formatHtml'\n\ninterface Params {\n    url: string\n}\n\nfunction getTurnDownService(params: Params) {\n    const turndownService = new turnDownService({\n        codeBlockStyle: 'fenced',\n        hr: '',\n    })\n\n    TurndownPluginGfm.gfm(turndownService)\n\n    let videoCounter = 0\n\n    // 自定义配置\n    turndownService\n        .addRule('pre2Code', {\n            filter: ['pre'],\n            replacement(content, node: any) {\n                const len = content.length\n                // 微信文章获取到的 content， 会出现首尾都有 '`'\n                const isCode = content[0] === '`' && content[len - 1] === '`'\n\n                let pre_Markdown = ''\n\n                if (isCode) {\n                    pre_Markdown = formatCode(node.innerHTML)\n                }\n\n                const res = isCode ? pre_Markdown : content\n\n                return '```\\n' + res + '\\n```\\n'\n            },\n        })\n        .addRule('getImage', {\n            filter: ['img'],\n            replacement(content, node: any) {\n                const src = node.getAttribute('data-src') || ''\n\n                return src ? `\\n\\n![](${src}) \\n\\n` : ''\n            },\n        })\n        .addRule('video', {\n            filter: (node: HTMLElement) => {\n                return (\n                    node.tagName.toLowerCase() === 'iframe' &&\n                    node.className.includes('video_iframe')\n                )\n            },\n            replacement(content, _node: Node) {\n                const node = _node as HTMLIFrameElement\n\n                const cover = decodeURIComponent(\n                    node.getAttribute('data-cover') || ''\n                )\n\n                const u = new URL(params.url)\n                u.hash = `js_mp_video_container_${videoCounter++}`\n\n                return cover ? `\\n\\n[![](${cover})](${u.href}) \\n\\n` : ''\n            },\n        })\n        .addRule('lineBreaks', {\n            filter: 'br',\n            replacement: () => '\\n',\n        })\n        .addRule('img2Code', {\n            filter: ['figure'],\n            replacement(content, node: any) {\n                const res = figure2markdown(node.innerHTML)\n                return res || ''\n            },\n        })\n\n    return turndownService\n}\n\nexport { getTurnDownService }\n","import cheerio from 'cheerio'\n\n/**\n * 微信不同代码风格\n * 1. <code><span>code</span></code>\n * 2. <code><span><span>123</span><br></span></code>\n * turndown 不解析 code 下的 br 标签，需要使用正则替换 br 标签为 \\n 才可以继续解析\n * @param htmlStr\n * @returns\n */\n\nexport function formatCode(htmlStr: string) {\n    let code = htmlStr\n\n    code = code.replace(/<br>/gi, '\\n')\n\n    code = code.replace(/&nbsp;/gi, ' ')\n\n    code = code.replace(/&lt;/gi, '<')\n\n    code = code.replace(/&gt;/gi, '>')\n\n    code = code.replace(/&amp;/gi, '&')\n\n    code = code.replace(/&quot;/gi, '\"')\n\n    code = code.replace(/&apos;/gi, '‘')\n\n    code = code.replace(/&times;/gi, '*')\n\n    code = code.replace(/&divide;/gi, '%')\n\n    const $ = cheerio.load(code)\n\n    return $.text()\n}\n\n/**\n * 解决如下格式\n * <figcaption><img><figcaption></figcaption></figcaption>\n * @param figureHTML\n * @returns\n */\nexport function figure2markdown(figureHTML: string) {\n    const imgRegex = /<img.*?data-src=['\"](.*?)['\"]/\n\n    const descRegex = /\\<figcaption .*?>(.+)<\\/figcaption>/\n\n    const imgArr = figureHTML.match(imgRegex)\n\n    const descArr = figureHTML.match(descRegex)\n\n    let imgUrl = ''\n\n    let desc = ''\n\n    if (Array.isArray(imgArr)) {\n        imgUrl = imgArr[1]\n    }\n\n    if (Array.isArray(descArr)) {\n        desc = descArr[1]\n    }\n\n    // img 可能没有图片说明\n    if (imgUrl) {\n        return `\\n\\n ![${desc}](${imgUrl}) \\n\\n`\n    }\n\n    return\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,mBAA0C;AAC1C,IAAAA,kBAAqB;;;ACDd,IAAM,SAET;AAAA,EACA,OAAO;AACX;;;ACOO,IAAW,SAAX,kBAAWC,YAAX;AACH,EAAAA,gBAAA,aAAU,OAAV;AACA,EAAAA,gBAAA,UAAO,OAAP;AAFc,SAAAA;AAAA,GAAA;;;ACRlB,sBAA4B;AAC5B,iCAA8B;;;ACJ9B,qBAAoB;AAWb,SAAS,WAAW,SAAiB;AACxC,MAAI,OAAO;AAEX,SAAO,KAAK,QAAQ,UAAU,IAAI;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,WAAW,GAAG;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,YAAY,QAAG;AAEnC,SAAO,KAAK,QAAQ,aAAa,GAAG;AAEpC,SAAO,KAAK,QAAQ,cAAc,GAAG;AAErC,QAAM,IAAI,eAAAC,QAAQ,KAAK,IAAI;AAE3B,SAAO,EAAE,KAAK;AAClB;AAQO,SAAS,gBAAgB,YAAoB;AAChD,QAAM,WAAW;AAEjB,QAAM,YAAY;AAElB,QAAM,SAAS,WAAW,MAAM,QAAQ;AAExC,QAAM,UAAU,WAAW,MAAM,SAAS;AAE1C,MAAI,SAAS;AAEb,MAAI,OAAO;AAEX,MAAI,MAAM,QAAQ,MAAM,GAAG;AACvB,aAAS,OAAO;AAAA,EACpB;AAEA,MAAI,MAAM,QAAQ,OAAO,GAAG;AACxB,WAAO,QAAQ;AAAA,EACnB;AAGA,MAAI,QAAQ;AACR,WAAO;AAAA;AAAA,KAAU,SAAS;AAAA;AAAA;AAAA,EAC9B;AAEA;AACJ;;;AD3DA,SAAS,mBAAmB,QAAgB;AACxC,QAAM,kBAAkB,IAAI,gBAAAC,QAAgB;AAAA,IACxC,gBAAgB;AAAA,IAChB,IAAI;AAAA,EACR,CAAC;AAED,6BAAAC,QAAkB,IAAI,eAAe;AAErC,MAAI,eAAe;AAGnB,kBACK,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,QAAQ;AAEpB,YAAM,SAAS,QAAQ,OAAO,OAAO,QAAQ,MAAM,OAAO;AAE1D,UAAI,eAAe;AAEnB,UAAI,QAAQ;AACR,uBAAe,WAAW,KAAK,SAAS;AAAA,MAC5C;AAEA,YAAM,MAAM,SAAS,eAAe;AAEpC,aAAO,UAAU,MAAM;AAAA,IAC3B;AAAA,EACJ,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,KAAK,aAAa,UAAU,KAAK;AAE7C,aAAO,MAAM;AAAA;AAAA,MAAW;AAAA;AAAA,IAAc;AAAA,IAC1C;AAAA,EACJ,CAAC,EACA,QAAQ,SAAS;AAAA,IACd,QAAQ,CAAC,SAAsB;AAC3B,aACI,KAAK,QAAQ,YAAY,MAAM,YAC/B,KAAK,UAAU,SAAS,cAAc;AAAA,IAE9C;AAAA,IACA,YAAY,SAAS,OAAa;AAC9B,YAAM,OAAO;AAEb,YAAM,QAAQ;AAAA,QACV,KAAK,aAAa,YAAY,KAAK;AAAA,MACvC;AAEA,YAAM,IAAI,IAAI,IAAI,OAAO,GAAG;AAC5B,QAAE,OAAO,yBAAyB;AAElC,aAAO,QAAQ;AAAA;AAAA,OAAY,WAAW,EAAE;AAAA;AAAA,IAAe;AAAA,IAC3D;AAAA,EACJ,CAAC,EACA,QAAQ,cAAc;AAAA,IACnB,QAAQ;AAAA,IACR,aAAa,MAAM;AAAA,EACvB,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,QAAQ;AAAA,IACjB,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,gBAAgB,KAAK,SAAS;AAC1C,aAAO,OAAO;AAAA,IAClB;AAAA,EACJ,CAAC;AAEL,SAAO;AACX;;;AH3EA,IAAM,WAAW,CAAC,SAAiB;AAC/B,SAAO;AAAA,IACH;AAAA,IACA,SAAS;AAAA,IACT,KAAK,OAAO;AAAA,EAChB;AACJ;AAIA,eAAsB,UAAU,SAAiB,MAAuB;AAjBxE;AAkBI,QAAM,QAAI,sBAAK,OAAO;AAEtB,MAAI,QAAQ,EAAE,gBAAgB,EAAE,KAAK;AAErC,UAAQ,MAAM,KAAK,KAAK;AACxB,QAAM,SAAS,MAAM;AAAA,IACjB,IAAI;AAAA,MACA;AAAA,SACI,OAAE,qBAAqB,MAAvB,mBAA0B,KAAK;AAAA,QAC/B,GAAG,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,IAAI;AAAA,MACtC,EACK,IAAI,CAAC,SAAU,OAAO,KAAK,KAAK,IAAI,EAAG,EACvC,OAAO,OAAO;AAAA,IACvB;AAAA,EACJ,EAAE,KAAK,IAAI;AAEX,QAAM,SAAS,EAAE,aAAa;AAC9B,QAAM,OAAO,OAAO,KAAK;AAEzB,MAAI,QAAQ,KAAK,SAAS,GAAG;AACzB,QAAI,MAAM,mBAAmB,IAAI,EAAE,SAAS,IAAI;AAEhD,UAAM,MAAM;AAAA;AAAA,kBAAyB;AAAA;AAAA,IAAiB;AAEtD,WAAO;AAAA,MACH,SAAS;AAAA,MACT;AAAA,MACA,MAAM;AAAA,QACF;AAAA,QACA;AAAA,QACA,SAAS;AAAA,MACb;AAAA,IACJ;AAAA,EACJ;AAEA,SAAO,uBAAoB;AAC/B;AASA,eAAO,uBACH,KACA,UAAyC,CAAC,GACnB;AACvB,QAAM,EAAE,cAAc,CAAC,EAAE,IAAI;AAC7B,QAAM,EAAE,UAAU,CAAC,MAAM,WAAW,IAAI;AAExC,QAAM,IAAI,IAAI,IAAI,GAAG;AAGrB,IAAE,aAAa,OAAO,WAAW;AAEjC,MAAI;AACA,UAAM,MAAM,MAAM,aAAAC,QAAM,IAAI,EAAE,MAAM;AAAA,MAChC,SAAS;AAAA,MACT,cAAc;AAAA,MACd,SAAS;AAAA,QACL,KAAK;AAAA,QACL,6BAA6B;AAAA,QAC7B,cACI;AAAA,QACJ,GAAG;AAAA,MACP;AAAA,MACA,GAAG;AAAA,IACP,CAAC;AAED,WAAO,UAAU,IAAI,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC;AAAA,EAC9C,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;","names":["import_cheerio","Status","cheerio","turnDownService","TurndownPluginGfm","axios"]}
+{"version":3,"sources":["../src/index.ts","../src/error.ts","../src/type.ts","../src/turndownCode.ts","../src/formatHtml.ts"],"sourcesContent":["import axios, { AxiosRequestConfig } from 'axios'\nimport { load } from 'cheerio'\nimport { errObj } from './error'\nimport type { TurnDownResult } from './type'\nimport { Status } from './type'\nimport { getTurnDownService } from './turndownCode'\n\nconst getError = (code: number) => {\n    return {\n        code,\n        success: false,\n        msg: errObj[code],\n    }\n}\n\nexport { TurnDownResult, Status }\n\nexport async function parseHTML(htmlRaw: string, meta: { url: string }) {\n    const $ = load(htmlRaw)\n\n    let title = $('#activity-name').text()\n\n    title = title.trim() || ''\n    const author = Array.from(\n        new Set(\n            [\n                $('meta[name=\"author\"]')?.attr('content'),\n                ...$('#js_name').text().split('\\n'),\n            ]\n                .map((item) => (item ? item.trim() : ''))\n                .filter(Boolean)\n        )\n    ).join('\\n')\n\n    const htmlEl = $('#js_content')\n    const html = htmlEl.html()\n\n    if (html && html.length > 0) {\n        let res = getTurnDownService(meta).turndown(html)\n\n        res = `## ${title} \\n \\n` + `## 作者 ${author} \\n \\n` + res\n\n        return {\n            success: true,\n            code: Status.Success,\n            data: {\n                title,\n                author,\n                content: res,\n            },\n        }\n    }\n\n    return getError(Status.Fail)\n}\n\nexport async function transformHtml2Markdown(\n    html: string,\n    /**\n     * 这里的 url 是原始的 url，主要是用来映射内部跳转链接\n     */\n    url: string\n): Promise<TurnDownResult> {\n    try {\n        return parseHTML(html, { url })\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\n/**\n * 支持添加代理服务器\n */\ninterface TransformHtml2MarkdownOptions {\n    axiosConfig?: AxiosRequestConfig\n}\n\nexport async function transformUrl2Markdown(\n    url: string,\n    options: TransformHtml2MarkdownOptions = {}\n): Promise<TurnDownResult> {\n    const { axiosConfig = {} } = options\n    const { headers = {}, ...restConfig } = axiosConfig\n\n    const u = new URL(url)\n    // 移除该参数\n    // 避免出现 302 跳转\n    u.searchParams.delete('poc_token')\n\n    try {\n        const res = await axios.get(u.href, {\n            timeout: 30000,\n            maxRedirects: 5,\n            headers: {\n                DNT: '1',\n                'Upgrade-Insecure-Requests': '1',\n                'User-Agent':\n                    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36',\n                ...headers,\n            },\n            ...restConfig,\n        })\n\n        return transformHtml2Markdown(res.data, url)\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\nexport { getTurnDownService } from './turndownCode'\n","export const errObj: {\n    [key: number]: string\n} = {\n    '400': '内容解析失败',\n}\n","export interface TurnDownResult {\n    success: boolean\n    code: number\n    data?: {\n        title?: string\n        author?: string\n        content?: string\n    }\n    msg?: string\n}\n\nexport const enum Status {\n    Success = 200,\n    Fail = 400,\n}\n","/**\n *  html 转换 markdown 格式\n */\nimport turnDownService from 'turndown'\nimport TurndownPluginGfm from '@guyplusplus/turndown-plugin-gfm'\nimport { formatCode, figure2markdown } from './formatHtml'\n\ninterface Params {\n    url: string\n}\n\nfunction getTurnDownService(params: Params) {\n    const turndownService = new turnDownService({\n        codeBlockStyle: 'fenced',\n        hr: '',\n    })\n\n    TurndownPluginGfm.gfm(turndownService)\n\n    let videoCounter = 0\n\n    // 自定义配置\n    turndownService\n        .addRule('pre2Code', {\n            filter: ['pre'],\n            replacement(content, node: any) {\n                const len = content.length\n                // 微信文章获取到的 content， 会出现首尾都有 '`'\n                const isCode = content[0] === '`' && content[len - 1] === '`'\n\n                let pre_Markdown = ''\n\n                if (isCode) {\n                    pre_Markdown = formatCode(node.innerHTML)\n                }\n\n                const res = isCode ? pre_Markdown : content\n\n                return '```\\n' + res + '\\n```\\n'\n            },\n        })\n        .addRule('getImage', {\n            filter: ['img'],\n            replacement(content, node: any) {\n                const src = node.getAttribute('data-src') || ''\n\n                return src ? `\\n\\n![](${src}) \\n\\n` : ''\n            },\n        })\n        .addRule('video', {\n            filter: (node: HTMLElement) => {\n                return (\n                    node.tagName.toLowerCase() === 'iframe' &&\n                    node.className.includes('video_iframe')\n                )\n            },\n            replacement(content, _node: Node) {\n                const node = _node as HTMLIFrameElement\n\n                const cover = decodeURIComponent(\n                    node.getAttribute('data-cover') || ''\n                )\n\n                const u = new URL(params.url)\n                u.hash = `js_mp_video_container_${videoCounter++}`\n\n                return cover ? `\\n\\n[![](${cover})](${u.href}) \\n\\n` : ''\n            },\n        })\n        .addRule('lineBreaks', {\n            filter: 'br',\n            replacement: () => '\\n',\n        })\n        .addRule('img2Code', {\n            filter: ['figure'],\n            replacement(content, node: any) {\n                const res = figure2markdown(node.innerHTML)\n                return res || ''\n            },\n        })\n\n    return turndownService\n}\n\nexport { getTurnDownService }\n","import cheerio from 'cheerio'\n\n/**\n * 微信不同代码风格\n * 1. <code><span>code</span></code>\n * 2. <code><span><span>123</span><br></span></code>\n * turndown 不解析 code 下的 br 标签，需要使用正则替换 br 标签为 \\n 才可以继续解析\n * @param htmlStr\n * @returns\n */\n\nexport function formatCode(htmlStr: string) {\n    let code = htmlStr\n\n    code = code.replace(/<br>/gi, '\\n')\n\n    code = code.replace(/&nbsp;/gi, ' ')\n\n    code = code.replace(/&lt;/gi, '<')\n\n    code = code.replace(/&gt;/gi, '>')\n\n    code = code.replace(/&amp;/gi, '&')\n\n    code = code.replace(/&quot;/gi, '\"')\n\n    code = code.replace(/&apos;/gi, '‘')\n\n    code = code.replace(/&times;/gi, '*')\n\n    code = code.replace(/&divide;/gi, '%')\n\n    const $ = cheerio.load(code)\n\n    return $.text()\n}\n\n/**\n * 解决如下格式\n * <figcaption><img><figcaption></figcaption></figcaption>\n * @param figureHTML\n * @returns\n */\nexport function figure2markdown(figureHTML: string) {\n    const imgRegex = /<img.*?data-src=['\"](.*?)['\"]/\n\n    const descRegex = /\\<figcaption .*?>(.+)<\\/figcaption>/\n\n    const imgArr = figureHTML.match(imgRegex)\n\n    const descArr = figureHTML.match(descRegex)\n\n    let imgUrl = ''\n\n    let desc = ''\n\n    if (Array.isArray(imgArr)) {\n        imgUrl = imgArr[1]\n    }\n\n    if (Array.isArray(descArr)) {\n        desc = descArr[1]\n    }\n\n    // img 可能没有图片说明\n    if (imgUrl) {\n        return `\\n\\n ![${desc}](${imgUrl}) \\n\\n`\n    }\n\n    return\n}\n"],"mappings":";;;;;;;;;;;;;;;;;;;;;;;;;;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,mBAA0C;AAC1C,IAAAA,kBAAqB;;;ACDd,IAAM,SAET;AAAA,EACA,OAAO;AACX;;;ACOO,IAAW,SAAX,kBAAWC,YAAX;AACH,EAAAA,gBAAA,aAAU,OAAV;AACA,EAAAA,gBAAA,UAAO,OAAP;AAFc,SAAAA;AAAA,GAAA;;;ACRlB,sBAA4B;AAC5B,iCAA8B;;;ACJ9B,qBAAoB;AAWb,SAAS,WAAW,SAAiB;AACxC,MAAI,OAAO;AAEX,SAAO,KAAK,QAAQ,UAAU,IAAI;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,WAAW,GAAG;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,YAAY,QAAG;AAEnC,SAAO,KAAK,QAAQ,aAAa,GAAG;AAEpC,SAAO,KAAK,QAAQ,cAAc,GAAG;AAErC,QAAM,IAAI,eAAAC,QAAQ,KAAK,IAAI;AAE3B,SAAO,EAAE,KAAK;AAClB;AAQO,SAAS,gBAAgB,YAAoB;AAChD,QAAM,WAAW;AAEjB,QAAM,YAAY;AAElB,QAAM,SAAS,WAAW,MAAM,QAAQ;AAExC,QAAM,UAAU,WAAW,MAAM,SAAS;AAE1C,MAAI,SAAS;AAEb,MAAI,OAAO;AAEX,MAAI,MAAM,QAAQ,MAAM,GAAG;AACvB,aAAS,OAAO;AAAA,EACpB;AAEA,MAAI,MAAM,QAAQ,OAAO,GAAG;AACxB,WAAO,QAAQ;AAAA,EACnB;AAGA,MAAI,QAAQ;AACR,WAAO;AAAA;AAAA,KAAU,SAAS;AAAA;AAAA;AAAA,EAC9B;AAEA;AACJ;;;AD3DA,SAAS,mBAAmB,QAAgB;AACxC,QAAM,kBAAkB,IAAI,gBAAAC,QAAgB;AAAA,IACxC,gBAAgB;AAAA,IAChB,IAAI;AAAA,EACR,CAAC;AAED,6BAAAC,QAAkB,IAAI,eAAe;AAErC,MAAI,eAAe;AAGnB,kBACK,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,QAAQ;AAEpB,YAAM,SAAS,QAAQ,OAAO,OAAO,QAAQ,MAAM,OAAO;AAE1D,UAAI,eAAe;AAEnB,UAAI,QAAQ;AACR,uBAAe,WAAW,KAAK,SAAS;AAAA,MAC5C;AAEA,YAAM,MAAM,SAAS,eAAe;AAEpC,aAAO,UAAU,MAAM;AAAA,IAC3B;AAAA,EACJ,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,KAAK,aAAa,UAAU,KAAK;AAE7C,aAAO,MAAM;AAAA;AAAA,MAAW;AAAA;AAAA,IAAc;AAAA,IAC1C;AAAA,EACJ,CAAC,EACA,QAAQ,SAAS;AAAA,IACd,QAAQ,CAAC,SAAsB;AAC3B,aACI,KAAK,QAAQ,YAAY,MAAM,YAC/B,KAAK,UAAU,SAAS,cAAc;AAAA,IAE9C;AAAA,IACA,YAAY,SAAS,OAAa;AAC9B,YAAM,OAAO;AAEb,YAAM,QAAQ;AAAA,QACV,KAAK,aAAa,YAAY,KAAK;AAAA,MACvC;AAEA,YAAM,IAAI,IAAI,IAAI,OAAO,GAAG;AAC5B,QAAE,OAAO,yBAAyB;AAElC,aAAO,QAAQ;AAAA;AAAA,OAAY,WAAW,EAAE;AAAA;AAAA,IAAe;AAAA,IAC3D;AAAA,EACJ,CAAC,EACA,QAAQ,cAAc;AAAA,IACnB,QAAQ;AAAA,IACR,aAAa,MAAM;AAAA,EACvB,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,QAAQ;AAAA,IACjB,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,gBAAgB,KAAK,SAAS;AAC1C,aAAO,OAAO;AAAA,IAClB;AAAA,EACJ,CAAC;AAEL,SAAO;AACX;;;AH3EA,IAAM,WAAW,CAAC,SAAiB;AAC/B,SAAO;AAAA,IACH;AAAA,IACA,SAAS;AAAA,IACT,KAAK,OAAO;AAAA,EAChB;AACJ;AAIA,eAAsB,UAAU,SAAiB,MAAuB;AAjBxE;AAkBI,QAAM,QAAI,sBAAK,OAAO;AAEtB,MAAI,QAAQ,EAAE,gBAAgB,EAAE,KAAK;AAErC,UAAQ,MAAM,KAAK,KAAK;AACxB,QAAM,SAAS,MAAM;AAAA,IACjB,IAAI;AAAA,MACA;AAAA,SACI,OAAE,qBAAqB,MAAvB,mBAA0B,KAAK;AAAA,QAC/B,GAAG,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,IAAI;AAAA,MACtC,EACK,IAAI,CAAC,SAAU,OAAO,KAAK,KAAK,IAAI,EAAG,EACvC,OAAO,OAAO;AAAA,IACvB;AAAA,EACJ,EAAE,KAAK,IAAI;AAEX,QAAM,SAAS,EAAE,aAAa;AAC9B,QAAM,OAAO,OAAO,KAAK;AAEzB,MAAI,QAAQ,KAAK,SAAS,GAAG;AACzB,QAAI,MAAM,mBAAmB,IAAI,EAAE,SAAS,IAAI;AAEhD,UAAM,MAAM;AAAA;AAAA,kBAAyB;AAAA;AAAA,IAAiB;AAEtD,WAAO;AAAA,MACH,SAAS;AAAA,MACT;AAAA,MACA,MAAM;AAAA,QACF;AAAA,QACA;AAAA,QACA,SAAS;AAAA,MACb;AAAA,IACJ;AAAA,EACJ;AAEA,SAAO,uBAAoB;AAC/B;AAEA,eAAsB,uBAClB,MAIA,KACuB;AACvB,MAAI;AACA,WAAO,UAAU,MAAM,EAAE,IAAI,CAAC;AAAA,EAClC,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;AASA,eAAsB,sBAClB,KACA,UAAyC,CAAC,GACnB;AACvB,QAAM,EAAE,cAAc,CAAC,EAAE,IAAI;AAC7B,QAAM,EAAE,UAAU,CAAC,MAAM,WAAW,IAAI;AAExC,QAAM,IAAI,IAAI,IAAI,GAAG;AAGrB,IAAE,aAAa,OAAO,WAAW;AAEjC,MAAI;AACA,UAAM,MAAM,MAAM,aAAAC,QAAM,IAAI,EAAE,MAAM;AAAA,MAChC,SAAS;AAAA,MACT,cAAc;AAAA,MACd,SAAS;AAAA,QACL,KAAK;AAAA,QACL,6BAA6B;AAAA,QAC7B,cACI;AAAA,QACJ,GAAG;AAAA,MACP;AAAA,MACA,GAAG;AAAA,IACP,CAAC;AAED,WAAO,uBAAuB,IAAI,MAAM,GAAG;AAAA,EAC/C,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;","names":["import_cheerio","Status","cheerio","turnDownService","TurndownPluginGfm","axios"]}

package/dist/index.d.ts

@@ -40,12 +40,17 @@ declare function parseHTML(htmlRaw: string, meta: {
         content: string;
     };
 }>;
+declare function transformHtml2Markdown(html: string, 
+/**
+ * 这里的 url 是原始的 url，主要是用来映射内部跳转链接
+ */
+url: string): Promise<TurnDownResult>;
 /**
  * 支持添加代理服务器
  */
 interface TransformHtml2MarkdownOptions {
     axiosConfig?: AxiosRequestConfig;
 }
-declare function transformHtml2Markdown(url: string, options?: TransformHtml2MarkdownOptions): Promise<TurnDownResult>;
+declare function transformUrl2Markdown(url: string, options?: TransformHtml2MarkdownOptions): Promise<TurnDownResult>;
 
-export { Status, TurnDownResult, transformHtml2Markdown as default, getTurnDownService, parseHTML };
+export { Status, TurnDownResult, getTurnDownService, parseHTML, transformHtml2Markdown, transformUrl2Markdown };

package/dist/index.js

@@ -159,7 +159,15 @@ async function parseHTML(htmlRaw, meta) {
   }
   return getError(400 /* Fail */);
 }
-async function transformHtml2Markdown(url, options = {}) {
+async function transformHtml2Markdown(html, url) {
+  try {
+    return parseHTML(html, { url });
+  } catch (err) {
+    console.log(err);
+    return getError(400 /* Fail */);
+  }
+}
+async function transformUrl2Markdown(url, options = {}) {
   const { axiosConfig = {} } = options;
   const { headers = {}, ...restConfig } = axiosConfig;
   const u = new URL(url);
@@ -176,7 +184,7 @@ async function transformHtml2Markdown(url, options = {}) {
       },
       ...restConfig
     });
-    return parseHTML(res.data, { url: u.href });
+    return transformHtml2Markdown(res.data, url);
   } catch (err) {
     console.log(err);
     return getError(400 /* Fail */);
@@ -184,8 +192,9 @@ async function transformHtml2Markdown(url, options = {}) {
 }
 export {
   Status,
-  transformHtml2Markdown as default,
   getTurnDownService,
-  parseHTML
+  parseHTML,
+  transformHtml2Markdown,
+  transformUrl2Markdown
 };
 //# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -1 +1 @@
-{"version":3,"sources":["../src/index.ts","../src/error.ts","../src/type.ts","../src/turndownCode.ts","../src/formatHtml.ts"],"sourcesContent":["import axios, { AxiosRequestConfig } from 'axios'\nimport { load } from 'cheerio'\nimport { errObj } from './error'\nimport type { TurnDownResult } from './type'\nimport { Status } from './type'\nimport { getTurnDownService } from './turndownCode'\n\nconst getError = (code: number) => {\n    return {\n        code,\n        success: false,\n        msg: errObj[code],\n    }\n}\n\nexport { TurnDownResult, Status }\n\nexport async function parseHTML(htmlRaw: string, meta: { url: string }) {\n    const $ = load(htmlRaw)\n\n    let title = $('#activity-name').text()\n\n    title = title.trim() || ''\n    const author = Array.from(\n        new Set(\n            [\n                $('meta[name=\"author\"]')?.attr('content'),\n                ...$('#js_name').text().split('\\n'),\n            ]\n                .map((item) => (item ? item.trim() : ''))\n                .filter(Boolean)\n        )\n    ).join('\\n')\n\n    const htmlEl = $('#js_content')\n    const html = htmlEl.html()\n\n    if (html && html.length > 0) {\n        let res = getTurnDownService(meta).turndown(html)\n\n        res = `## ${title} \\n \\n` + `## 作者 ${author} \\n \\n` + res\n\n        return {\n            success: true,\n            code: Status.Success,\n            data: {\n                title,\n                author,\n                content: res,\n            },\n        }\n    }\n\n    return getError(Status.Fail)\n}\n\n/**\n * 支持添加代理服务器\n */\ninterface TransformHtml2MarkdownOptions {\n    axiosConfig?: AxiosRequestConfig\n}\n\nexport default async function transformHtml2Markdown(\n    url: string,\n    options: TransformHtml2MarkdownOptions = {}\n): Promise<TurnDownResult> {\n    const { axiosConfig = {} } = options\n    const { headers = {}, ...restConfig } = axiosConfig\n\n    const u = new URL(url)\n    // 移除该参数\n    // 避免出现 302 跳转\n    u.searchParams.delete('poc_token')\n\n    try {\n        const res = await axios.get(u.href, {\n            timeout: 30000,\n            maxRedirects: 5,\n            headers: {\n                DNT: '1',\n                'Upgrade-Insecure-Requests': '1',\n                'User-Agent':\n                    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36',\n                ...headers,\n            },\n            ...restConfig,\n        })\n\n        return parseHTML(res.data, { url: u.href })\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\nexport { getTurnDownService } from './turndownCode'\n","export const errObj: {\n    [key: number]: string\n} = {\n    '400': '内容解析失败',\n}\n","export interface TurnDownResult {\n    success: boolean\n    code: number\n    data?: {\n        title?: string\n        author?: string\n        content?: string\n    }\n    msg?: string\n}\n\nexport const enum Status {\n    Success = 200,\n    Fail = 400,\n}\n","/**\n *  html 转换 markdown 格式\n */\nimport turnDownService from 'turndown'\nimport TurndownPluginGfm from '@guyplusplus/turndown-plugin-gfm'\nimport { formatCode, figure2markdown } from './formatHtml'\n\ninterface Params {\n    url: string\n}\n\nfunction getTurnDownService(params: Params) {\n    const turndownService = new turnDownService({\n        codeBlockStyle: 'fenced',\n        hr: '',\n    })\n\n    TurndownPluginGfm.gfm(turndownService)\n\n    let videoCounter = 0\n\n    // 自定义配置\n    turndownService\n        .addRule('pre2Code', {\n            filter: ['pre'],\n            replacement(content, node: any) {\n                const len = content.length\n                // 微信文章获取到的 content， 会出现首尾都有 '`'\n                const isCode = content[0] === '`' && content[len - 1] === '`'\n\n                let pre_Markdown = ''\n\n                if (isCode) {\n                    pre_Markdown = formatCode(node.innerHTML)\n                }\n\n                const res = isCode ? pre_Markdown : content\n\n                return '```\\n' + res + '\\n```\\n'\n            },\n        })\n        .addRule('getImage', {\n            filter: ['img'],\n            replacement(content, node: any) {\n                const src = node.getAttribute('data-src') || ''\n\n                return src ? `\\n\\n![](${src}) \\n\\n` : ''\n            },\n        })\n        .addRule('video', {\n            filter: (node: HTMLElement) => {\n                return (\n                    node.tagName.toLowerCase() === 'iframe' &&\n                    node.className.includes('video_iframe')\n                )\n            },\n            replacement(content, _node: Node) {\n                const node = _node as HTMLIFrameElement\n\n                const cover = decodeURIComponent(\n                    node.getAttribute('data-cover') || ''\n                )\n\n                const u = new URL(params.url)\n                u.hash = `js_mp_video_container_${videoCounter++}`\n\n                return cover ? `\\n\\n[![](${cover})](${u.href}) \\n\\n` : ''\n            },\n        })\n        .addRule('lineBreaks', {\n            filter: 'br',\n            replacement: () => '\\n',\n        })\n        .addRule('img2Code', {\n            filter: ['figure'],\n            replacement(content, node: any) {\n                const res = figure2markdown(node.innerHTML)\n                return res || ''\n            },\n        })\n\n    return turndownService\n}\n\nexport { getTurnDownService }\n","import cheerio from 'cheerio'\n\n/**\n * 微信不同代码风格\n * 1. <code><span>code</span></code>\n * 2. <code><span><span>123</span><br></span></code>\n * turndown 不解析 code 下的 br 标签，需要使用正则替换 br 标签为 \\n 才可以继续解析\n * @param htmlStr\n * @returns\n */\n\nexport function formatCode(htmlStr: string) {\n    let code = htmlStr\n\n    code = code.replace(/<br>/gi, '\\n')\n\n    code = code.replace(/&nbsp;/gi, ' ')\n\n    code = code.replace(/&lt;/gi, '<')\n\n    code = code.replace(/&gt;/gi, '>')\n\n    code = code.replace(/&amp;/gi, '&')\n\n    code = code.replace(/&quot;/gi, '\"')\n\n    code = code.replace(/&apos;/gi, '‘')\n\n    code = code.replace(/&times;/gi, '*')\n\n    code = code.replace(/&divide;/gi, '%')\n\n    const $ = cheerio.load(code)\n\n    return $.text()\n}\n\n/**\n * 解决如下格式\n * <figcaption><img><figcaption></figcaption></figcaption>\n * @param figureHTML\n * @returns\n */\nexport function figure2markdown(figureHTML: string) {\n    const imgRegex = /<img.*?data-src=['\"](.*?)['\"]/\n\n    const descRegex = /\\<figcaption .*?>(.+)<\\/figcaption>/\n\n    const imgArr = figureHTML.match(imgRegex)\n\n    const descArr = figureHTML.match(descRegex)\n\n    let imgUrl = ''\n\n    let desc = ''\n\n    if (Array.isArray(imgArr)) {\n        imgUrl = imgArr[1]\n    }\n\n    if (Array.isArray(descArr)) {\n        desc = descArr[1]\n    }\n\n    // img 可能没有图片说明\n    if (imgUrl) {\n        return `\\n\\n ![${desc}](${imgUrl}) \\n\\n`\n    }\n\n    return\n}\n"],"mappings":";AAAA,OAAO,WAAmC;AAC1C,SAAS,YAAY;;;ACDd,IAAM,SAET;AAAA,EACA,OAAO;AACX;;;ACOO,IAAW,SAAX,kBAAWA,YAAX;AACH,EAAAA,gBAAA,aAAU,OAAV;AACA,EAAAA,gBAAA,UAAO,OAAP;AAFc,SAAAA;AAAA,GAAA;;;ACRlB,OAAO,qBAAqB;AAC5B,OAAO,uBAAuB;;;ACJ9B,OAAO,aAAa;AAWb,SAAS,WAAW,SAAiB;AACxC,MAAI,OAAO;AAEX,SAAO,KAAK,QAAQ,UAAU,IAAI;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,WAAW,GAAG;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,YAAY,QAAG;AAEnC,SAAO,KAAK,QAAQ,aAAa,GAAG;AAEpC,SAAO,KAAK,QAAQ,cAAc,GAAG;AAErC,QAAM,IAAI,QAAQ,KAAK,IAAI;AAE3B,SAAO,EAAE,KAAK;AAClB;AAQO,SAAS,gBAAgB,YAAoB;AAChD,QAAM,WAAW;AAEjB,QAAM,YAAY;AAElB,QAAM,SAAS,WAAW,MAAM,QAAQ;AAExC,QAAM,UAAU,WAAW,MAAM,SAAS;AAE1C,MAAI,SAAS;AAEb,MAAI,OAAO;AAEX,MAAI,MAAM,QAAQ,MAAM,GAAG;AACvB,aAAS,OAAO;AAAA,EACpB;AAEA,MAAI,MAAM,QAAQ,OAAO,GAAG;AACxB,WAAO,QAAQ;AAAA,EACnB;AAGA,MAAI,QAAQ;AACR,WAAO;AAAA;AAAA,KAAU,SAAS;AAAA;AAAA;AAAA,EAC9B;AAEA;AACJ;;;AD3DA,SAAS,mBAAmB,QAAgB;AACxC,QAAM,kBAAkB,IAAI,gBAAgB;AAAA,IACxC,gBAAgB;AAAA,IAChB,IAAI;AAAA,EACR,CAAC;AAED,oBAAkB,IAAI,eAAe;AAErC,MAAI,eAAe;AAGnB,kBACK,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,QAAQ;AAEpB,YAAM,SAAS,QAAQ,OAAO,OAAO,QAAQ,MAAM,OAAO;AAE1D,UAAI,eAAe;AAEnB,UAAI,QAAQ;AACR,uBAAe,WAAW,KAAK,SAAS;AAAA,MAC5C;AAEA,YAAM,MAAM,SAAS,eAAe;AAEpC,aAAO,UAAU,MAAM;AAAA,IAC3B;AAAA,EACJ,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,KAAK,aAAa,UAAU,KAAK;AAE7C,aAAO,MAAM;AAAA;AAAA,MAAW;AAAA;AAAA,IAAc;AAAA,IAC1C;AAAA,EACJ,CAAC,EACA,QAAQ,SAAS;AAAA,IACd,QAAQ,CAAC,SAAsB;AAC3B,aACI,KAAK,QAAQ,YAAY,MAAM,YAC/B,KAAK,UAAU,SAAS,cAAc;AAAA,IAE9C;AAAA,IACA,YAAY,SAAS,OAAa;AAC9B,YAAM,OAAO;AAEb,YAAM,QAAQ;AAAA,QACV,KAAK,aAAa,YAAY,KAAK;AAAA,MACvC;AAEA,YAAM,IAAI,IAAI,IAAI,OAAO,GAAG;AAC5B,QAAE,OAAO,yBAAyB;AAElC,aAAO,QAAQ;AAAA;AAAA,OAAY,WAAW,EAAE;AAAA;AAAA,IAAe;AAAA,IAC3D;AAAA,EACJ,CAAC,EACA,QAAQ,cAAc;AAAA,IACnB,QAAQ;AAAA,IACR,aAAa,MAAM;AAAA,EACvB,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,QAAQ;AAAA,IACjB,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,gBAAgB,KAAK,SAAS;AAC1C,aAAO,OAAO;AAAA,IAClB;AAAA,EACJ,CAAC;AAEL,SAAO;AACX;;;AH3EA,IAAM,WAAW,CAAC,SAAiB;AAC/B,SAAO;AAAA,IACH;AAAA,IACA,SAAS;AAAA,IACT,KAAK,OAAO;AAAA,EAChB;AACJ;AAIA,eAAsB,UAAU,SAAiB,MAAuB;AAjBxE;AAkBI,QAAM,IAAI,KAAK,OAAO;AAEtB,MAAI,QAAQ,EAAE,gBAAgB,EAAE,KAAK;AAErC,UAAQ,MAAM,KAAK,KAAK;AACxB,QAAM,SAAS,MAAM;AAAA,IACjB,IAAI;AAAA,MACA;AAAA,SACI,OAAE,qBAAqB,MAAvB,mBAA0B,KAAK;AAAA,QAC/B,GAAG,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,IAAI;AAAA,MACtC,EACK,IAAI,CAAC,SAAU,OAAO,KAAK,KAAK,IAAI,EAAG,EACvC,OAAO,OAAO;AAAA,IACvB;AAAA,EACJ,EAAE,KAAK,IAAI;AAEX,QAAM,SAAS,EAAE,aAAa;AAC9B,QAAM,OAAO,OAAO,KAAK;AAEzB,MAAI,QAAQ,KAAK,SAAS,GAAG;AACzB,QAAI,MAAM,mBAAmB,IAAI,EAAE,SAAS,IAAI;AAEhD,UAAM,MAAM;AAAA;AAAA,kBAAyB;AAAA;AAAA,IAAiB;AAEtD,WAAO;AAAA,MACH,SAAS;AAAA,MACT;AAAA,MACA,MAAM;AAAA,QACF;AAAA,QACA;AAAA,QACA,SAAS;AAAA,MACb;AAAA,IACJ;AAAA,EACJ;AAEA,SAAO,uBAAoB;AAC/B;AASA,eAAO,uBACH,KACA,UAAyC,CAAC,GACnB;AACvB,QAAM,EAAE,cAAc,CAAC,EAAE,IAAI;AAC7B,QAAM,EAAE,UAAU,CAAC,MAAM,WAAW,IAAI;AAExC,QAAM,IAAI,IAAI,IAAI,GAAG;AAGrB,IAAE,aAAa,OAAO,WAAW;AAEjC,MAAI;AACA,UAAM,MAAM,MAAM,MAAM,IAAI,EAAE,MAAM;AAAA,MAChC,SAAS;AAAA,MACT,cAAc;AAAA,MACd,SAAS;AAAA,QACL,KAAK;AAAA,QACL,6BAA6B;AAAA,QAC7B,cACI;AAAA,QACJ,GAAG;AAAA,MACP;AAAA,MACA,GAAG;AAAA,IACP,CAAC;AAED,WAAO,UAAU,IAAI,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC;AAAA,EAC9C,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;","names":["Status"]}
+{"version":3,"sources":["../src/index.ts","../src/error.ts","../src/type.ts","../src/turndownCode.ts","../src/formatHtml.ts"],"sourcesContent":["import axios, { AxiosRequestConfig } from 'axios'\nimport { load } from 'cheerio'\nimport { errObj } from './error'\nimport type { TurnDownResult } from './type'\nimport { Status } from './type'\nimport { getTurnDownService } from './turndownCode'\n\nconst getError = (code: number) => {\n    return {\n        code,\n        success: false,\n        msg: errObj[code],\n    }\n}\n\nexport { TurnDownResult, Status }\n\nexport async function parseHTML(htmlRaw: string, meta: { url: string }) {\n    const $ = load(htmlRaw)\n\n    let title = $('#activity-name').text()\n\n    title = title.trim() || ''\n    const author = Array.from(\n        new Set(\n            [\n                $('meta[name=\"author\"]')?.attr('content'),\n                ...$('#js_name').text().split('\\n'),\n            ]\n                .map((item) => (item ? item.trim() : ''))\n                .filter(Boolean)\n        )\n    ).join('\\n')\n\n    const htmlEl = $('#js_content')\n    const html = htmlEl.html()\n\n    if (html && html.length > 0) {\n        let res = getTurnDownService(meta).turndown(html)\n\n        res = `## ${title} \\n \\n` + `## 作者 ${author} \\n \\n` + res\n\n        return {\n            success: true,\n            code: Status.Success,\n            data: {\n                title,\n                author,\n                content: res,\n            },\n        }\n    }\n\n    return getError(Status.Fail)\n}\n\nexport async function transformHtml2Markdown(\n    html: string,\n    /**\n     * 这里的 url 是原始的 url，主要是用来映射内部跳转链接\n     */\n    url: string\n): Promise<TurnDownResult> {\n    try {\n        return parseHTML(html, { url })\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\n/**\n * 支持添加代理服务器\n */\ninterface TransformHtml2MarkdownOptions {\n    axiosConfig?: AxiosRequestConfig\n}\n\nexport async function transformUrl2Markdown(\n    url: string,\n    options: TransformHtml2MarkdownOptions = {}\n): Promise<TurnDownResult> {\n    const { axiosConfig = {} } = options\n    const { headers = {}, ...restConfig } = axiosConfig\n\n    const u = new URL(url)\n    // 移除该参数\n    // 避免出现 302 跳转\n    u.searchParams.delete('poc_token')\n\n    try {\n        const res = await axios.get(u.href, {\n            timeout: 30000,\n            maxRedirects: 5,\n            headers: {\n                DNT: '1',\n                'Upgrade-Insecure-Requests': '1',\n                'User-Agent':\n                    'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/134.0.0.0 Safari/537.36',\n                ...headers,\n            },\n            ...restConfig,\n        })\n\n        return transformHtml2Markdown(res.data, url)\n    } catch (err) {\n        console.log(err)\n        return getError(Status.Fail)\n    }\n}\n\nexport { getTurnDownService } from './turndownCode'\n","export const errObj: {\n    [key: number]: string\n} = {\n    '400': '内容解析失败',\n}\n","export interface TurnDownResult {\n    success: boolean\n    code: number\n    data?: {\n        title?: string\n        author?: string\n        content?: string\n    }\n    msg?: string\n}\n\nexport const enum Status {\n    Success = 200,\n    Fail = 400,\n}\n","/**\n *  html 转换 markdown 格式\n */\nimport turnDownService from 'turndown'\nimport TurndownPluginGfm from '@guyplusplus/turndown-plugin-gfm'\nimport { formatCode, figure2markdown } from './formatHtml'\n\ninterface Params {\n    url: string\n}\n\nfunction getTurnDownService(params: Params) {\n    const turndownService = new turnDownService({\n        codeBlockStyle: 'fenced',\n        hr: '',\n    })\n\n    TurndownPluginGfm.gfm(turndownService)\n\n    let videoCounter = 0\n\n    // 自定义配置\n    turndownService\n        .addRule('pre2Code', {\n            filter: ['pre'],\n            replacement(content, node: any) {\n                const len = content.length\n                // 微信文章获取到的 content， 会出现首尾都有 '`'\n                const isCode = content[0] === '`' && content[len - 1] === '`'\n\n                let pre_Markdown = ''\n\n                if (isCode) {\n                    pre_Markdown = formatCode(node.innerHTML)\n                }\n\n                const res = isCode ? pre_Markdown : content\n\n                return '```\\n' + res + '\\n```\\n'\n            },\n        })\n        .addRule('getImage', {\n            filter: ['img'],\n            replacement(content, node: any) {\n                const src = node.getAttribute('data-src') || ''\n\n                return src ? `\\n\\n![](${src}) \\n\\n` : ''\n            },\n        })\n        .addRule('video', {\n            filter: (node: HTMLElement) => {\n                return (\n                    node.tagName.toLowerCase() === 'iframe' &&\n                    node.className.includes('video_iframe')\n                )\n            },\n            replacement(content, _node: Node) {\n                const node = _node as HTMLIFrameElement\n\n                const cover = decodeURIComponent(\n                    node.getAttribute('data-cover') || ''\n                )\n\n                const u = new URL(params.url)\n                u.hash = `js_mp_video_container_${videoCounter++}`\n\n                return cover ? `\\n\\n[![](${cover})](${u.href}) \\n\\n` : ''\n            },\n        })\n        .addRule('lineBreaks', {\n            filter: 'br',\n            replacement: () => '\\n',\n        })\n        .addRule('img2Code', {\n            filter: ['figure'],\n            replacement(content, node: any) {\n                const res = figure2markdown(node.innerHTML)\n                return res || ''\n            },\n        })\n\n    return turndownService\n}\n\nexport { getTurnDownService }\n","import cheerio from 'cheerio'\n\n/**\n * 微信不同代码风格\n * 1. <code><span>code</span></code>\n * 2. <code><span><span>123</span><br></span></code>\n * turndown 不解析 code 下的 br 标签，需要使用正则替换 br 标签为 \\n 才可以继续解析\n * @param htmlStr\n * @returns\n */\n\nexport function formatCode(htmlStr: string) {\n    let code = htmlStr\n\n    code = code.replace(/<br>/gi, '\\n')\n\n    code = code.replace(/&nbsp;/gi, ' ')\n\n    code = code.replace(/&lt;/gi, '<')\n\n    code = code.replace(/&gt;/gi, '>')\n\n    code = code.replace(/&amp;/gi, '&')\n\n    code = code.replace(/&quot;/gi, '\"')\n\n    code = code.replace(/&apos;/gi, '‘')\n\n    code = code.replace(/&times;/gi, '*')\n\n    code = code.replace(/&divide;/gi, '%')\n\n    const $ = cheerio.load(code)\n\n    return $.text()\n}\n\n/**\n * 解决如下格式\n * <figcaption><img><figcaption></figcaption></figcaption>\n * @param figureHTML\n * @returns\n */\nexport function figure2markdown(figureHTML: string) {\n    const imgRegex = /<img.*?data-src=['\"](.*?)['\"]/\n\n    const descRegex = /\\<figcaption .*?>(.+)<\\/figcaption>/\n\n    const imgArr = figureHTML.match(imgRegex)\n\n    const descArr = figureHTML.match(descRegex)\n\n    let imgUrl = ''\n\n    let desc = ''\n\n    if (Array.isArray(imgArr)) {\n        imgUrl = imgArr[1]\n    }\n\n    if (Array.isArray(descArr)) {\n        desc = descArr[1]\n    }\n\n    // img 可能没有图片说明\n    if (imgUrl) {\n        return `\\n\\n ![${desc}](${imgUrl}) \\n\\n`\n    }\n\n    return\n}\n"],"mappings":";AAAA,OAAO,WAAmC;AAC1C,SAAS,YAAY;;;ACDd,IAAM,SAET;AAAA,EACA,OAAO;AACX;;;ACOO,IAAW,SAAX,kBAAWA,YAAX;AACH,EAAAA,gBAAA,aAAU,OAAV;AACA,EAAAA,gBAAA,UAAO,OAAP;AAFc,SAAAA;AAAA,GAAA;;;ACRlB,OAAO,qBAAqB;AAC5B,OAAO,uBAAuB;;;ACJ9B,OAAO,aAAa;AAWb,SAAS,WAAW,SAAiB;AACxC,MAAI,OAAO;AAEX,SAAO,KAAK,QAAQ,UAAU,IAAI;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,UAAU,GAAG;AAEjC,SAAO,KAAK,QAAQ,WAAW,GAAG;AAElC,SAAO,KAAK,QAAQ,YAAY,GAAG;AAEnC,SAAO,KAAK,QAAQ,YAAY,QAAG;AAEnC,SAAO,KAAK,QAAQ,aAAa,GAAG;AAEpC,SAAO,KAAK,QAAQ,cAAc,GAAG;AAErC,QAAM,IAAI,QAAQ,KAAK,IAAI;AAE3B,SAAO,EAAE,KAAK;AAClB;AAQO,SAAS,gBAAgB,YAAoB;AAChD,QAAM,WAAW;AAEjB,QAAM,YAAY;AAElB,QAAM,SAAS,WAAW,MAAM,QAAQ;AAExC,QAAM,UAAU,WAAW,MAAM,SAAS;AAE1C,MAAI,SAAS;AAEb,MAAI,OAAO;AAEX,MAAI,MAAM,QAAQ,MAAM,GAAG;AACvB,aAAS,OAAO;AAAA,EACpB;AAEA,MAAI,MAAM,QAAQ,OAAO,GAAG;AACxB,WAAO,QAAQ;AAAA,EACnB;AAGA,MAAI,QAAQ;AACR,WAAO;AAAA;AAAA,KAAU,SAAS;AAAA;AAAA;AAAA,EAC9B;AAEA;AACJ;;;AD3DA,SAAS,mBAAmB,QAAgB;AACxC,QAAM,kBAAkB,IAAI,gBAAgB;AAAA,IACxC,gBAAgB;AAAA,IAChB,IAAI;AAAA,EACR,CAAC;AAED,oBAAkB,IAAI,eAAe;AAErC,MAAI,eAAe;AAGnB,kBACK,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,QAAQ;AAEpB,YAAM,SAAS,QAAQ,OAAO,OAAO,QAAQ,MAAM,OAAO;AAE1D,UAAI,eAAe;AAEnB,UAAI,QAAQ;AACR,uBAAe,WAAW,KAAK,SAAS;AAAA,MAC5C;AAEA,YAAM,MAAM,SAAS,eAAe;AAEpC,aAAO,UAAU,MAAM;AAAA,IAC3B;AAAA,EACJ,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,KAAK;AAAA,IACd,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,KAAK,aAAa,UAAU,KAAK;AAE7C,aAAO,MAAM;AAAA;AAAA,MAAW;AAAA;AAAA,IAAc;AAAA,IAC1C;AAAA,EACJ,CAAC,EACA,QAAQ,SAAS;AAAA,IACd,QAAQ,CAAC,SAAsB;AAC3B,aACI,KAAK,QAAQ,YAAY,MAAM,YAC/B,KAAK,UAAU,SAAS,cAAc;AAAA,IAE9C;AAAA,IACA,YAAY,SAAS,OAAa;AAC9B,YAAM,OAAO;AAEb,YAAM,QAAQ;AAAA,QACV,KAAK,aAAa,YAAY,KAAK;AAAA,MACvC;AAEA,YAAM,IAAI,IAAI,IAAI,OAAO,GAAG;AAC5B,QAAE,OAAO,yBAAyB;AAElC,aAAO,QAAQ;AAAA;AAAA,OAAY,WAAW,EAAE;AAAA;AAAA,IAAe;AAAA,IAC3D;AAAA,EACJ,CAAC,EACA,QAAQ,cAAc;AAAA,IACnB,QAAQ;AAAA,IACR,aAAa,MAAM;AAAA,EACvB,CAAC,EACA,QAAQ,YAAY;AAAA,IACjB,QAAQ,CAAC,QAAQ;AAAA,IACjB,YAAY,SAAS,MAAW;AAC5B,YAAM,MAAM,gBAAgB,KAAK,SAAS;AAC1C,aAAO,OAAO;AAAA,IAClB;AAAA,EACJ,CAAC;AAEL,SAAO;AACX;;;AH3EA,IAAM,WAAW,CAAC,SAAiB;AAC/B,SAAO;AAAA,IACH;AAAA,IACA,SAAS;AAAA,IACT,KAAK,OAAO;AAAA,EAChB;AACJ;AAIA,eAAsB,UAAU,SAAiB,MAAuB;AAjBxE;AAkBI,QAAM,IAAI,KAAK,OAAO;AAEtB,MAAI,QAAQ,EAAE,gBAAgB,EAAE,KAAK;AAErC,UAAQ,MAAM,KAAK,KAAK;AACxB,QAAM,SAAS,MAAM;AAAA,IACjB,IAAI;AAAA,MACA;AAAA,SACI,OAAE,qBAAqB,MAAvB,mBAA0B,KAAK;AAAA,QAC/B,GAAG,EAAE,UAAU,EAAE,KAAK,EAAE,MAAM,IAAI;AAAA,MACtC,EACK,IAAI,CAAC,SAAU,OAAO,KAAK,KAAK,IAAI,EAAG,EACvC,OAAO,OAAO;AAAA,IACvB;AAAA,EACJ,EAAE,KAAK,IAAI;AAEX,QAAM,SAAS,EAAE,aAAa;AAC9B,QAAM,OAAO,OAAO,KAAK;AAEzB,MAAI,QAAQ,KAAK,SAAS,GAAG;AACzB,QAAI,MAAM,mBAAmB,IAAI,EAAE,SAAS,IAAI;AAEhD,UAAM,MAAM;AAAA;AAAA,kBAAyB;AAAA;AAAA,IAAiB;AAEtD,WAAO;AAAA,MACH,SAAS;AAAA,MACT;AAAA,MACA,MAAM;AAAA,QACF;AAAA,QACA;AAAA,QACA,SAAS;AAAA,MACb;AAAA,IACJ;AAAA,EACJ;AAEA,SAAO,uBAAoB;AAC/B;AAEA,eAAsB,uBAClB,MAIA,KACuB;AACvB,MAAI;AACA,WAAO,UAAU,MAAM,EAAE,IAAI,CAAC;AAAA,EAClC,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;AASA,eAAsB,sBAClB,KACA,UAAyC,CAAC,GACnB;AACvB,QAAM,EAAE,cAAc,CAAC,EAAE,IAAI;AAC7B,QAAM,EAAE,UAAU,CAAC,MAAM,WAAW,IAAI;AAExC,QAAM,IAAI,IAAI,IAAI,GAAG;AAGrB,IAAE,aAAa,OAAO,WAAW;AAEjC,MAAI;AACA,UAAM,MAAM,MAAM,MAAM,IAAI,EAAE,MAAM;AAAA,MAChC,SAAS;AAAA,MACT,cAAc;AAAA,MACd,SAAS;AAAA,QACL,KAAK;AAAA,QACL,6BAA6B;AAAA,QAC7B,cACI;AAAA,QACJ,GAAG;AAAA,MACP;AAAA,MACA,GAAG;AAAA,IACP,CAAC;AAED,WAAO,uBAAuB,IAAI,MAAM,GAAG;AAAA,EAC/C,SAAS,KAAP;AACE,YAAQ,IAAI,GAAG;AACf,WAAO,uBAAoB;AAAA,EAC/B;AACJ;","names":["Status"]}

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "@aiello/wechat-to-markdown",
-    "version": "1.2.12",
+    "version": "1.2.13",
     "description": "解析微信文章 URL 为 markdown",
     "author": "Aiello Chan<aiello.chan@gmail.com>",
     "keywords": [