@isdk/mdast-plus 0.2.4 → 0.3.0

package/README.cn.md

@@ -185,7 +185,19 @@ const result = await mdast('Hello').to('reverse');
 | `extract-code-meta` | normalize | 从代码块元数据中解析 `title="foo"`。 |
 | `image-size` | normalize | 从图片 URL 中解析 `#=WxH`。 |
 | `normalize-inline-styles` | normalize | 标准化 `==mark==`、`~sub~` 和 `^sup^`。 |
-| `html-readability` | parse | 使用 Mozilla 的 Readability 从 HTML 中提取主体内容。支持 `frontmatter` 注入和 `sourceLink` (原文链接) 页脚。使用 `htmlReadabilityPlugins` 数组可以简化配置。 |
+| `html-readability` | parse | 使用 Mozilla 的 Readability 从 HTML 中提取主体内容。支持 `frontmatter` 注入、`sourceLink` (原文链接) 页脚以及 `smartExcerpt` 智能摘要管理。使用 `htmlReadabilityPlugins` 数组可以简化配置。 |
+
+### html-readability 选项
+
+- `url`: (string) HTML 文档的 URL。
+- `frontmatter`: (boolean | 'yaml' | 'toml') 是否将元数据注入为 Frontmatter。默认值: `false`。
+- `sourceLink`: (boolean) 是否在底部添加原文链接。即使在 `fields` 选项中过滤或重命名了 title/url 字段，该链接仍将根据原始信息正确生成。默认值: `false`。
+- `fields`: (string[] | object) 控制保留哪些元数据字段或如何重命名它们。
+  - 如果是数组：作为白名单（例如 `['title', 'excerpt']`）。
+  - 如果是对象：将原始键映射到新名称（例如 `{ title: 'headline' }`）。只有映射中存在的键才会被保留（投影）。
+- `smartExcerpt`: (boolean | object) 是否在摘要与正文内容重复或近乎重复时移除摘要。默认值: `true`。
+  - `threshold`: (number) 摘要长度与正文长度的比率阈值 (0.0 到 1.0)。默认值: `0.6`。
+  - `minContentLength`: (number) 保留摘要所需的正文最小长度。默认值: `300`。
 
 ## 贡献
 

package/README.md

@@ -185,7 +185,19 @@ Each stage can have one "main" plugin. If a plugin is marked with `main: true`,
 | `extract-code-meta` | normalize | Parses `title="foo"` from code block meta. |
 | `image-size` | normalize | Parses `#=WxH` from image URLs. |
 | `normalize-inline-styles` | normalize | Standardizes `==mark==`, `~sub~`, and `^sup^`. |
-| `html-readability` | parse | Uses Mozilla's Readability to extract main content from HTML. Supports `frontmatter` injection and `sourceLink` (source link) footer. Use `htmlReadabilityPlugins` array for easier setup. |
+| `html-readability` | parse | Uses Mozilla's Readability to extract main content from HTML. Supports `frontmatter` injection, `sourceLink` (source link) footer, and `smartExcerpt` for intelligent summary management. Use `htmlReadabilityPlugins` array for easier setup. |
+
+### html-readability Options
+
+- `url`: (string) The URL of the HTML document.
+- `frontmatter`: (boolean | 'yaml' | 'toml') Whether to inject metadata as frontmatter. Default: `false`.
+- `sourceLink`: (boolean) Whether to append source link at the bottom. The link will be generated based on the original title/URL even if they are filtered or renamed in the `fields` option. Default: `false`.
+- `fields`: (string[] | object) Control which metadata fields are kept or how they are renamed. 
+  - If an array: acts as an allowlist (e.g., `['title', 'excerpt']`).
+  - If an object: maps original keys to new names (e.g., `{ title: 'headline' }`). Only keys in the map are kept (Projection).
+- `smartExcerpt`: (boolean | object) Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content. Default: `true`.
+  - `threshold`: (number) The ratio of excerpt length to content length (0.0 to 1.0). Default: `0.6`.
+  - `minContentLength`: (number) Minimum length of the main content required to keep the excerpt. Default: `300`.
 
 ## Contributing
 

package/dist/index.d.mts

@@ -1741,6 +1741,20 @@ declare function jsonParser(this: any): void;
  */
 declare const astFormat: MdastFormat;
 
+interface SmartExcerptOptions {
+    /**
+     * The threshold ratio of excerpt length to content length.
+     * If (excerptLength / contentLength) > threshold, the excerpt is considered redundant.
+     * @default 0.6
+     */
+    threshold?: number;
+    /**
+     * The minimum length of the main content required to keep the excerpt.
+     * If content length is less than this value, the excerpt is considered redundant (if it is contained in the content).
+     * @default 300
+     */
+    minContentLength?: number;
+}
 interface ReadabilityOptions {
     url?: string;
     readability?: Record<string, any> | false;
@@ -1757,6 +1771,18 @@ interface ReadabilityOptions {
      * @default false
      */
     sourceLink?: boolean;
+    /**
+     * Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content.
+     * Useful when the content is short or the excerpt is just a subset of the content.
+     * @default true
+     */
+    smartExcerpt?: boolean | SmartExcerptOptions;
+    /**
+     * Control the fields and names in metadata.
+     * - If an array of strings, it acts as an allowlist (only these fields are kept).
+     * - If an object, it maps original field names to new names. Only the keys present in the map are kept (Projection).
+     */
+    fields?: string[] | Record<string, string>;
 }
 /**
  * A unified/rehype plugin that uses Mozilla's Readability to parse the input HTML.
@@ -1795,5 +1821,6 @@ declare const htmlReadabilityPlugins: ({
     after: string;
 })[];
 declare function checkUrlExists(tree: any, url: string): boolean;
+declare function checkHtmlUrlExists(tree: any, url: string): boolean;
 
-export { DefaultPipelineStage, MdastBasePipeline, type MdastDataOrigin, type MdastFormat, type MdastMark, MdastPipeline, type MdastPlugin, type MdastSub, type MdastSup, type PipelineRunOptions, PipelineStage, type PipelineStageName, type ReadabilityOptions, astCompiler, astFormat, checkUrlExists, htmlFormat, htmlReadability, htmlReadabilityPlugin, htmlReadabilityPlugins, jsonParser, markdownFormat, mdast, restoreReadabilityMetaPlugin };
+export { DefaultPipelineStage, MdastBasePipeline, type MdastDataOrigin, type MdastFormat, type MdastMark, MdastPipeline, type MdastPlugin, type MdastSub, type MdastSup, type PipelineRunOptions, PipelineStage, type PipelineStageName, type ReadabilityOptions, type SmartExcerptOptions, astCompiler, astFormat, checkHtmlUrlExists, checkUrlExists, htmlFormat, htmlReadability, htmlReadabilityPlugin, htmlReadabilityPlugins, jsonParser, markdownFormat, mdast, restoreReadabilityMetaPlugin };

package/dist/index.d.ts

@@ -1741,6 +1741,20 @@ declare function jsonParser(this: any): void;
  */
 declare const astFormat: MdastFormat;
 
+interface SmartExcerptOptions {
+    /**
+     * The threshold ratio of excerpt length to content length.
+     * If (excerptLength / contentLength) > threshold, the excerpt is considered redundant.
+     * @default 0.6
+     */
+    threshold?: number;
+    /**
+     * The minimum length of the main content required to keep the excerpt.
+     * If content length is less than this value, the excerpt is considered redundant (if it is contained in the content).
+     * @default 300
+     */
+    minContentLength?: number;
+}
 interface ReadabilityOptions {
     url?: string;
     readability?: Record<string, any> | false;
@@ -1757,6 +1771,18 @@ interface ReadabilityOptions {
      * @default false
      */
     sourceLink?: boolean;
+    /**
+     * Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content.
+     * Useful when the content is short or the excerpt is just a subset of the content.
+     * @default true
+     */
+    smartExcerpt?: boolean | SmartExcerptOptions;
+    /**
+     * Control the fields and names in metadata.
+     * - If an array of strings, it acts as an allowlist (only these fields are kept).
+     * - If an object, it maps original field names to new names. Only the keys present in the map are kept (Projection).
+     */
+    fields?: string[] | Record<string, string>;
 }
 /**
  * A unified/rehype plugin that uses Mozilla's Readability to parse the input HTML.
@@ -1795,5 +1821,6 @@ declare const htmlReadabilityPlugins: ({
     after: string;
 })[];
 declare function checkUrlExists(tree: any, url: string): boolean;
+declare function checkHtmlUrlExists(tree: any, url: string): boolean;
 
-export { DefaultPipelineStage, MdastBasePipeline, type MdastDataOrigin, type MdastFormat, type MdastMark, MdastPipeline, type MdastPlugin, type MdastSub, type MdastSup, type PipelineRunOptions, PipelineStage, type PipelineStageName, type ReadabilityOptions, astCompiler, astFormat, checkUrlExists, htmlFormat, htmlReadability, htmlReadabilityPlugin, htmlReadabilityPlugins, jsonParser, markdownFormat, mdast, restoreReadabilityMetaPlugin };
+export { DefaultPipelineStage, MdastBasePipeline, type MdastDataOrigin, type MdastFormat, type MdastMark, MdastPipeline, type MdastPlugin, type MdastSub, type MdastSup, type PipelineRunOptions, PipelineStage, type PipelineStageName, type ReadabilityOptions, type SmartExcerptOptions, astCompiler, astFormat, checkHtmlUrlExists, checkUrlExists, htmlFormat, htmlReadability, htmlReadabilityPlugin, htmlReadabilityPlugins, jsonParser, markdownFormat, mdast, restoreReadabilityMetaPlugin };

package/dist/index.js

@@ -1 +1 @@
-"use strict";var t,e=Object.create,r=Object.defineProperty,i=Object.getOwnPropertyDescriptor,n=Object.getOwnPropertyNames,s=Object.getPrototypeOf,o=Object.prototype.hasOwnProperty,a=(t,e,s,a)=>{if(e&&"object"==typeof e||"function"==typeof e)for(let u of n(e))o.call(t,u)||u===s||r(t,u,{get:()=>e[u],enumerable:!(a=i(e,u))||a.enumerable});return t},u=(t,i,n)=>(n=null!=t?e(s(t)):{},a(!i&&t&&t.__esModule?n:r(n,"default",{value:t,enumerable:!0}),t)),l={};((t,e)=>{for(var i in e)r(t,i,{get:e[i],enumerable:!0})})(l,{DefaultPipelineStage:()=>p,MdastBasePipeline:()=>Q,MdastPipeline:()=>K,PipelineStage:()=>c,astCompiler:()=>U,astFormat:()=>_,checkUrlExists:()=>ot,htmlFormat:()=>I,htmlReadability:()=>rt,htmlReadabilityPlugin:()=>it,htmlReadabilityPlugins:()=>st,jsonParser:()=>V,markdownFormat:()=>E,mdast:()=>W,restoreReadabilityMetaPlugin:()=>nt}),module.exports=(t=l,a(r({},"__esModule",{value:!0}),t));var c=(t=>(t[t.parse=0]="parse",t[t.normalize=100]="normalize",t[t.compile=200]="compile",t[t.finalize=300]="finalize",t[t.stringify=400]="stringify",t))(c||{}),p=200,f=require("unified"),m=require("vfile"),h=u(require("remark-parse")),g=u(require("remark-stringify")),d=u(require("remark-gfm")),y=u(require("remark-directive")),b=u(require("remark-math")),v=u(require("remark-frontmatter")),w=require("unist-util-visit");function k(t,e){return{type:t,children:e,data:{hName:t}}}var q={mark:(t,e,r)=>"=="+r.containerPhrasing(t,{before:"==",after:"=="})+"==",sub:(t,e,r)=>"~"+r.containerPhrasing(t,{before:"~",after:"~"})+"~",sup:(t,e,r)=>"^"+r.containerPhrasing(t,{before:"^",after:"^"})+"^"},j={plugin:()=>t=>{!function(t){(0,w.visit)(t,"text",(t,e,r)=>{if(!r||void 0===e)return;const i=t.value;let n=0;const s=[];let o=!1;const a=/(==[^=]+==|~[^~]+~|\^[^^]+\^)/g;let u;for(;null!==(u=a.exec(i));){o=!0;const t=u[0],e=u.index;e>n&&s.push({type:"text",value:i.slice(n,e)});let r="mark",l="";t.startsWith("==")?(r="mark",l=t.slice(2,-2)):t.startsWith("~")?(r="sub",l=t.slice(1,-1)):t.startsWith("^")&&(r="sup",l=t.slice(1,-1)),s.push(k(r,[{type:"text",value:l}])),n=a.lastIndex}return o?(n<i.length&&s.push({type:"text",value:i.slice(n)}),r.children.splice(e,1,...s),e+s.length):void 0})}(t)},stage:100},x=require("unist-util-visit"),P={error:"danger",warn:"warning",success:"tip",important:"important",caution:"caution",note:"note"},M={plugin:()=>async t=>{(0,x.visit)(t,["containerDirective","leafDirective","textDirective"],t=>{const e=t,r=e.name.toLowerCase();if(e.name=P[r]||r,e.children&&e.children.length>0){const t=e.children[0];if(t.data?.directiveLabel||"directiveLabel"===t.type){const r=t;let i="";(0,x.visit)(r,"text",t=>{i+=t.value}),i&&!e.attributes?.title&&(e.attributes=e.attributes||{},e.attributes.title=i.trim()),e.children.shift()}}e.attributes?.title&&(e.attributes.title=String(e.attributes.title).trim()),e.data=e.data||{},e.data.hName=e.data.hName||("containerDirective"===e.type?"div":"span"),e.data.hProperties={...e.data.hProperties||{},...e.attributes,className:[e.name,e.data.hProperties?.className].filter(Boolean).join(" ")}})},stage:100,order:10},S=require("unist-util-visit"),T={plugin:()=>async t=>{(0,S.visit)(t,"tableCell",t=>{if(t.data){const{rowspan:e,colspan:r}=t.data;t.data.hProperties=t.data.hProperties||{},void 0!==e&&(t.data.hProperties.rowSpan=e,delete t.data.rowspan),void 0!==r&&(t.data.hProperties.colSpan=r,delete t.data.colspan)}})},stage:100,order:20},F=require("unist-util-visit"),N=require("shell-quote");var O={plugin:()=>async t=>{(0,F.visit)(t,"code",t=>{if(t.meta){const e=function(t){const e={},r=(0,N.parse)(t);for(const t of r)if("string"==typeof t){const r=t.split("=",2);2===r.length?e[r[0]]=r[1]:e[t]="true"}return e}(t.meta),r=t.data=t.data||{};e.title&&(r.title=e.title),e.filename&&(r.filename=e.filename),r.kv={...r.kv||{},...e}}})},stage:100,order:30},A=require("unist-util-visit"),D={plugin:()=>async t=>{(0,A.visit)(t,"image",t=>{const e=t.data=t.data||{},r=e.hProperties=e.hProperties||{},i=/[#?&](?:width=([0-9]+))?(?:&?height=([0-9]+))?(?:=([0-9]+)x([0-9]+))?$/,n=t.url.match(i);if(n){const e=n[1]||n[3],s=n[2]||n[4];e&&!r.width&&(r.width=parseInt(e,10)),s&&!r.height&&(r.height=parseInt(s,10)),t.url=t.url.replace(i,"")}e.width&&!r.width&&(r.width=e.width),e.height&&!r.height&&(r.height=e.height)})},stage:100,order:40},R=[{plugin:h.default,stage:0},{plugin:d.default,options:[{singleTilde:!1}],stage:0},{plugin:y.default,stage:0},{plugin:b.default,stage:0},{plugin:v.default,options:[["yaml","toml"]],stage:0},M,T,O,D,j],z=[{plugin:g.default,options:[{handlers:q}],stage:400},{plugin:d.default,options:[{singleTilde:!1}],stage:400},{plugin:y.default,stage:400},{plugin:b.default,stage:400},{plugin:v.default,options:[["yaml","toml"]],stage:400}];z.forEach(t=>{t.plugin===g.default?t.order=100:t.order=10});var E={id:"markdown",title:"Markdown (GFM + Directives)",extensions:["md","markdown","mdown","mkdn"],mediaTypes:["text/markdown"],input:R,output:z},L=u(require("rehype-parse")),$=u(require("rehype-remark")),B=u(require("remark-rehype")),C=u(require("rehype-sanitize")),H=u(require("rehype-stringify")),I={id:"html",title:"HTML",extensions:["html","htm"],mediaTypes:["text/html"],input:[{name:"rehype-parse",plugin:L.default,stage:0},{name:"rehype-remark",plugin:$.default,options:[{handlers:{mark:(t,e)=>{const r={type:"mark",children:t.all(e)};return t.patch(e,r),r},sub:(t,e)=>{const r={type:"sub",children:t.all(e)};return t.patch(e,r),r},sup:(t,e)=>{const r={type:"sup",children:t.all(e)};return t.patch(e,r),r}}}],stage:0}],output:[{plugin:B.default,stage:300,order:10},{plugin:C.default,options:[{...C.defaultSchema,tagNames:[...C.defaultSchema.tagNames||[],"mark","sub","sup"],attributes:{...C.defaultSchema.attributes,"*":[...C.defaultSchema.attributes?.["*"]||[],"className","id","style"],td:[...C.defaultSchema.attributes?.td||[],"rowSpan","colSpan","rowspan","colspan"],th:[...C.defaultSchema.attributes?.th||[],"rowSpan","colSpan","rowspan","colspan"],img:[...C.defaultSchema.attributes?.img||[],"width","height"]}}],stage:300,order:20},{plugin:H.default,stage:400}]};function U(){this.Compiler=t=>t}function V(){this.Parser=t=>JSON.parse(t)}var _={id:"ast",title:"MDAST",input:[{plugin:V,stage:0},M,T,O,D,j],output:[{plugin:U,options:[],stage:400}]};function G(t){return"object"==typeof t&&null!==t&&"string"==typeof t.type}var J=class t{constructor(t){this.queue=[],this._data={},this.input=t}static register(t){this.registry.set(t.id,t)}static getFormat(t){return this.registry.get(t)}data(t,e){return"string"==typeof t?this._data[t]=e:Object.assign(this._data,t),this}getFormat(t){return this.constructor.getFormat(t)}resolveFormat(t){if("string"==typeof t){const e=this.getFormat(t);if(!e)throw new Error(`[MdastPlus] Format '${t}' is not registered.`);return e}return t}toRuntimeEntry(t,e,r){let i=e;void 0!==t.stage&&(i="string"==typeof t.stage?c[t.stage]??e:t.stage);let n=t.options||[];const s=t.name||t.plugin.name;if(r&&s&&s in r){const e=r[s];"object"!=typeof e||null===e||Array.isArray(e)?n=Array.isArray(e)?e:[e]:("main"in e&&(t.main=!!e.main),"before"in e&&(t.before=e.before),"after"in e&&(t.after=e.after),n=[e])}return{name:s,plugin:t.plugin,options:n,stage:i,order:t.order||0,main:t.main,before:t.before,after:t.after}}ensureInputPlugins(e,r,i=400){const n=e.some(t=>0===(t.stage??p)),s=G(this.input);if(!n){let n=[];if(s){const e=t.getFormat("ast");e&&e.input&&(n=e.input)}else{const e=t.getFormat("markdown");e&&e.input&&(n=e.input)}for(const t of n){const n=this.toRuntimeEntry(t,0,r);(n.stage??p)<=i&&e.push(n)}}}from(t,e){const r=this.resolveFormat(t);if(!r.input||0===r.input.length)throw new Error(`[MdastPlus] Format '${r.id}' does not support input.`);for(const t of r.input)this.queue.push(this.toRuntimeEntry(t,0,e));return this}resolveRunQueue(t,e,r,i){let n=[];if(n=e?this.queue.map(t=>this.toRuntimeEntry(t,t.stage??p,e)):[...this.queue],void 0!==r){const t=r;n=n.filter(e=>(e.stage??p)<=t);const s=n.findIndex(e=>(e.stage??p)===t),o=-1!==s?s+(i??0):n.length-1,a=n.slice(0,o+1),u=n.filter(t=>t.main&&!a.includes(t));n=a.concat(u),this.ensureInputPlugins(n,e,t),n.push({plugin:U,options:[],stage:400,order:0})}else if(this.ensureInputPlugins(n,e,400),t.output)for(const r of t.output)n.push(this.toRuntimeEntry(r,300,e));return n}async to(t,e){const r=this.resolveFormat(t);if(!r.output)throw new Error(`[MdastPlus] Format '${r.id}' does not support output.`);let i,n,s;if(e){const t=e;"overrides"in t||"stage"in t||"stopAtIndex"in t?(i=t.overrides,n="string"==typeof t.stage?c[t.stage]:t.stage,s=t.stopAtIndex):i=e}const o=this.resolveRunQueue(r,i,n,s),a=this.assembleProcessor(o);if(G(this.input)){const t=await a.run(this.input),e=a.stringify(t),r=new m.VFile;return"string"==typeof e||Buffer.isBuffer(e)?r.value=e:r.result=e,r}return a.process(this.input)}use(t,...e){return this.useAt("compile",t,...e)}normalizePluggable(t,e=[]){const r=[];if(Array.isArray(t)){let i=!1;if(t.length>0&&"function"==typeof t[0]){t.slice(1).some(t=>!function(t){if("function"==typeof t)return!0;if(Array.isArray(t))return!0;if("object"==typeof t&&null!==t){if("plugin"in t&&"function"==typeof t.plugin)return!0;if("plugins"in t)return!0}return!1}(t))&&(i=!0)}if(i){const[e,...i]=t;r.push({plugin:e,options:i})}else for(const i of t)r.push(...this.normalizePluggable(i,e))}else if("function"==typeof t)r.push({plugin:t,options:e});else if("object"==typeof t&&null!==t)if("plugin"in t&&"function"==typeof t.plugin){const i=t,n=e.length>0?e:i.options||[];r.push({...i,options:n})}else if("plugins"in t){const e=t;e.plugins&&r.push(...this.normalizePluggable(e.plugins))}return r}useAt(t,e,...r){let i,n,s;"string"==typeof t||"number"==typeof t?(i=("string"==typeof t?c[t]:t)??p,n=e,s=r):(i=void 0,n=t,s=[e,...r].filter(t=>void 0!==t));const o=this.normalizePluggable(n,s);for(const t of o){const e=i??p;this.queue.push(this.toRuntimeEntry(t,e))}return this}priority(t){const e=this.queue[this.queue.length-1];return e&&(e.order=t),this}configure(t,...e){for(let r=this.queue.length-1;r>=0;r--){const i=this.queue[r];if((i.name||i.plugin.name)===t){i.options=e;break}}return this}assembleProcessor(t){const e={};for(const r of t){const t=r.stage??p;e[t]||(e[t]=[]),e[t].push(r)}const r=[],i=Object.keys(e).map(Number).sort((t,e)=>t-e);for(const t of i){const i=e[t].sort((t,e)=>(t.order||0)-(e.order||0)),n=i.findIndex(t=>t.main);if(-1!==n){const t=i[n];!1===t.options?.[0]?console.warn(`Main Plugin "${t.name}" is disabled. Skipping.`):0!==n&&(i.splice(n,1),i[0]=t)}let s=!0,o=0;for(;s&&o<i.length;){s=!1,o++;for(let t=0;t<i.length;t++){const e=i[t];if(e.after){const r=i.findIndex(t=>t.name===e.after);if(-1!==r&&r>t){i.splice(t,1),i.splice(r,0,e),s=!0;break}}if(e.before){const r=i.findIndex(t=>t.name===e.before);if(-1!==r&&r<t){i.splice(t,1),i.splice(r,0,e),s=!0;break}}}}r.push(...i)}const n=(0,f.unified)();Object.keys(this._data).length>0&&n.data(this._data);for(const t of r)n.use(t.plugin,...t.options||[]);return n}};J.registry=new Map;var Q=J,K=class extends Q{async toMarkdown(){const t=await this.to("markdown");return String(t)}toMarkdownVFile(){return this.to("markdown")}async toHtml(){const t=await this.to("html");return String(t)}toHtmlVFile(){return this.to("html")}async toAst(t){return(await this.to("ast",t)).result}toHTML(){return this.toHtml()}toAST(t){return this.toAst(t)}};function W(t){return new K(t)}K.register(E),K.register(I),K.register(_);var X=require("unist-util-visit"),Y=require("hast-util-from-html"),Z=require("hast-util-from-dom"),tt=require("yaml"),et=require("lodash-es"),rt=function(t){const{readability:e,jsdom:r,hast:i,url:n}=t||{};this.parser=function(t,s){if(!1===e)return(0,Y.fromHtml)(t,{fragment:!0,...i});let o,a;try{o=require("jsdom").JSDOM;a=require("@mozilla/readability").Readability}catch(t){throw new Error("[html-readability] Dependency missing. Please install 'jsdom' and '@mozilla/readability'.")}const u=new a(new o(t,{url:n,pretendToBeVisual:!0,...r,includeNodeLocations:!0}).window.document,{maxElemsToParse:1e5,nbTopCandidates:5,charThreshold:500,keepClasses:!0,...e,serializer:t=>t}).parse();if(!u||!u.content)return(0,Y.fromHtml)(t,{fragment:!0,...i});const l=u.content;let c=(0,Z.fromDom)(l,{afterTransform:i?.afterTransform});const p=!1!==i?.fragment,f=(0,et.omitBy)(u,et.isNil);return n&&(f.url=n),delete f.content,delete f.textContent,s&&(s.data=s.data||{},s.data.readability=f),c&&(c=p?{type:"root",children:[c]}:{type:"root",children:[{type:"element",tagName:"html",properties:{},children:[{type:"element",tagName:"head",properties:{},children:[]},{type:"element",tagName:"body",properties:{},children:[c]}]}]},c.data=c.data||{},c.data.readability=f),c}},it={name:"readability",plugin:rt,stage:0,main:!0},nt={name:"restore-readability-meta",plugin:t=>(e,r)=>{if(r.data?.readability){e.data=e.data||{},e.data.readability=r.data.readability;const{frontmatter:i,sourceLink:n}=t||{};if(i){const t="toml"===i?"toml":"yaml",n=(0,tt.stringify)(r.data.readability).trim();e.children.unshift({type:t,value:n})}if(n&&r.data.readability.url){const{url:t,title:i}=r.data.readability;ot(e,t)||e.children.push({type:"blockquote",children:[{type:"paragraph",children:[{type:"text",value:"Source: "},{type:"link",url:t,children:[{type:"text",value:i||t}]}]}]})}}},stage:0,after:"rehype-remark"},st=[it,nt];function ot(t,e){if(!e)return!1;let r=!1;return(0,X.visit)(t,"link",t=>{if(function(t,e){if(!t||!e)return!1;if(t===e)return!0;try{const r=new URL(t),i=new URL(e);return r.href===i.href}catch{return t.replace(/\/$/,"")===e.replace(/\/$/,"")}}(t.url,e))return r=!0,!1}),r}
+"use strict";var t,e=Object.create,r=Object.defineProperty,i=Object.getOwnPropertyDescriptor,n=Object.getOwnPropertyNames,s=Object.getPrototypeOf,o=Object.prototype.hasOwnProperty,a=(t,e,s,a)=>{if(e&&"object"==typeof e||"function"==typeof e)for(let l of n(e))o.call(t,l)||l===s||r(t,l,{get:()=>e[l],enumerable:!(a=i(e,l))||a.enumerable});return t},l=(t,i,n)=>(n=null!=t?e(s(t)):{},a(!i&&t&&t.__esModule?n:r(n,"default",{value:t,enumerable:!0}),t)),u={};((t,e)=>{for(var i in e)r(t,i,{get:e[i],enumerable:!0})})(u,{DefaultPipelineStage:()=>p,MdastBasePipeline:()=>Q,MdastPipeline:()=>K,PipelineStage:()=>c,astCompiler:()=>U,astFormat:()=>_,checkHtmlUrlExists:()=>lt,checkUrlExists:()=>at,htmlFormat:()=>I,htmlReadability:()=>rt,htmlReadabilityPlugin:()=>it,htmlReadabilityPlugins:()=>st,jsonParser:()=>V,markdownFormat:()=>R,mdast:()=>W,restoreReadabilityMetaPlugin:()=>nt}),module.exports=(t=u,a(r({},"__esModule",{value:!0}),t));var c=(t=>(t[t.parse=0]="parse",t[t.normalize=100]="normalize",t[t.compile=200]="compile",t[t.finalize=300]="finalize",t[t.stringify=400]="stringify",t))(c||{}),p=200,f=require("unified"),m=require("vfile"),h=l(require("remark-parse")),d=l(require("remark-stringify")),g=l(require("remark-gfm")),y=l(require("remark-directive")),b=l(require("remark-math")),v=l(require("remark-frontmatter")),k=require("unist-util-visit");function w(t,e){return{type:t,children:e,data:{hName:t}}}var q={mark:(t,e,r)=>"=="+r.containerPhrasing(t,{before:"==",after:"=="})+"==",sub:(t,e,r)=>"~"+r.containerPhrasing(t,{before:"~",after:"~"})+"~",sup:(t,e,r)=>"^"+r.containerPhrasing(t,{before:"^",after:"^"})+"^"},x={plugin:()=>t=>{!function(t){(0,k.visit)(t,"text",(t,e,r)=>{if(!r||void 0===e)return;const i=t.value;let n=0;const s=[];let o=!1;const a=/(==[^=]+==|~[^~]+~|\^[^^]+\^)/g;let l;for(;null!==(l=a.exec(i));){o=!0;const t=l[0],e=l.index;e>n&&s.push({type:"text",value:i.slice(n,e)});let r="mark",u="";t.startsWith("==")?(r="mark",u=t.slice(2,-2)):t.startsWith("~")?(r="sub",u=t.slice(1,-1)):t.startsWith("^")&&(r="sup",u=t.slice(1,-1)),s.push(w(r,[{type:"text",value:u}])),n=a.lastIndex}return o?(n<i.length&&s.push({type:"text",value:i.slice(n)}),r.children.splice(e,1,...s),e+s.length):void 0})}(t)},stage:100},j=require("unist-util-visit"),P={error:"danger",warn:"warning",success:"tip",important:"important",caution:"caution",note:"note"},M={plugin:()=>async t=>{(0,j.visit)(t,["containerDirective","leafDirective","textDirective"],t=>{const e=t,r=e.name.toLowerCase();if(e.name=P[r]||r,e.children&&e.children.length>0){const t=e.children[0];if(t.data?.directiveLabel||"directiveLabel"===t.type){const r=t;let i="";(0,j.visit)(r,"text",t=>{i+=t.value}),i&&!e.attributes?.title&&(e.attributes=e.attributes||{},e.attributes.title=i.trim()),e.children.shift()}}e.attributes?.title&&(e.attributes.title=String(e.attributes.title).trim()),e.data=e.data||{},e.data.hName=e.data.hName||("containerDirective"===e.type?"div":"span"),e.data.hProperties={...e.data.hProperties||{},...e.attributes,className:[e.name,e.data.hProperties?.className].filter(Boolean).join(" ")}})},stage:100,order:10},S=require("unist-util-visit"),N={plugin:()=>async t=>{(0,S.visit)(t,"tableCell",t=>{if(t.data){const{rowspan:e,colspan:r}=t.data;t.data.hProperties=t.data.hProperties||{},void 0!==e&&(t.data.hProperties.rowSpan=e,delete t.data.rowspan),void 0!==r&&(t.data.hProperties.colSpan=r,delete t.data.colspan)}})},stage:100,order:20},T=require("unist-util-visit"),F=require("shell-quote");var O={plugin:()=>async t=>{(0,T.visit)(t,"code",t=>{if(t.meta){const e=function(t){const e={},r=(0,F.parse)(t);for(const t of r)if("string"==typeof t){const r=t.split("=",2);2===r.length?e[r[0]]=r[1]:e[t]="true"}return e}(t.meta),r=t.data=t.data||{};e.title&&(r.title=e.title),e.filename&&(r.filename=e.filename),r.kv={...r.kv||{},...e}}})},stage:100,order:30},A=require("unist-util-visit"),E={plugin:()=>async t=>{(0,A.visit)(t,"image",t=>{const e=t.data=t.data||{},r=e.hProperties=e.hProperties||{},i=/[#?&](?:width=([0-9]+))?(?:&?height=([0-9]+))?(?:=([0-9]+)x([0-9]+))?$/,n=t.url.match(i);if(n){const e=n[1]||n[3],s=n[2]||n[4];e&&!r.width&&(r.width=parseInt(e,10)),s&&!r.height&&(r.height=parseInt(s,10)),t.url=t.url.replace(i,"")}e.width&&!r.width&&(r.width=e.width),e.height&&!r.height&&(r.height=e.height)})},stage:100,order:40},L=[{plugin:h.default,stage:0},{plugin:g.default,options:[{singleTilde:!1}],stage:0},{plugin:y.default,stage:0},{plugin:b.default,stage:0},{plugin:v.default,options:[["yaml","toml"]],stage:0},M,N,O,E,x],D=[{plugin:d.default,options:[{handlers:q}],stage:400},{plugin:g.default,options:[{singleTilde:!1}],stage:400},{plugin:y.default,stage:400},{plugin:b.default,stage:400},{plugin:v.default,options:[["yaml","toml"]],stage:400}];D.forEach(t=>{t.plugin===d.default?t.order=100:t.order=10});var R={id:"markdown",title:"Markdown (GFM + Directives)",extensions:["md","markdown","mdown","mkdn"],mediaTypes:["text/markdown"],input:L,output:D},z=l(require("rehype-parse")),$=l(require("rehype-remark")),C=l(require("remark-rehype")),H=l(require("rehype-sanitize")),B=l(require("rehype-stringify")),I={id:"html",title:"HTML",extensions:["html","htm"],mediaTypes:["text/html"],input:[{name:"rehype-parse",plugin:z.default,stage:0},{name:"rehype-remark",plugin:$.default,options:[{handlers:{mark:(t,e)=>{const r={type:"mark",children:t.all(e)};return t.patch(e,r),r},sub:(t,e)=>{const r={type:"sub",children:t.all(e)};return t.patch(e,r),r},sup:(t,e)=>{const r={type:"sup",children:t.all(e)};return t.patch(e,r),r}}}],stage:0}],output:[{plugin:C.default,stage:300,order:10},{plugin:H.default,options:[{...H.defaultSchema,tagNames:[...H.defaultSchema.tagNames||[],"mark","sub","sup"],attributes:{...H.defaultSchema.attributes,"*":[...H.defaultSchema.attributes?.["*"]||[],"className","id","style"],td:[...H.defaultSchema.attributes?.td||[],"rowSpan","colSpan","rowspan","colspan"],th:[...H.defaultSchema.attributes?.th||[],"rowSpan","colSpan","rowspan","colspan"],img:[...H.defaultSchema.attributes?.img||[],"width","height"]}}],stage:300,order:20},{plugin:B.default,stage:400}]};function U(){this.Compiler=t=>t}function V(){this.Parser=t=>JSON.parse(t)}var _={id:"ast",title:"MDAST",input:[{plugin:V,stage:0},M,N,O,E,x],output:[{plugin:U,options:[],stage:400}]};function G(t){return"object"==typeof t&&null!==t&&"string"==typeof t.type}var J=class t{constructor(t){this.queue=[],this._data={},this.input=t}static register(t){this.registry.set(t.id,t)}static getFormat(t){return this.registry.get(t)}data(t,e){return"string"==typeof t?this._data[t]=e:Object.assign(this._data,t),this}getFormat(t){return this.constructor.getFormat(t)}resolveFormat(t){if("string"==typeof t){const e=this.getFormat(t);if(!e)throw new Error(`[MdastPlus] Format '${t}' is not registered.`);return e}return t}toRuntimeEntry(t,e,r){let i=e;void 0!==t.stage&&(i="string"==typeof t.stage?c[t.stage]??e:t.stage);let n=t.options||[];const s=t.name||t.plugin.name;if(r&&s&&s in r){const e=r[s];"object"!=typeof e||null===e||Array.isArray(e)?n=Array.isArray(e)?e:[e]:("main"in e&&(t.main=!!e.main),"before"in e&&(t.before=e.before),"after"in e&&(t.after=e.after),n=[e])}return{name:s,plugin:t.plugin,options:n,stage:i,order:t.order||0,main:t.main,before:t.before,after:t.after}}ensureInputPlugins(e,r,i=400){const n=e.some(t=>0===(t.stage??p)),s=G(this.input);if(!n){let n=[];if(s){const e=t.getFormat("ast");e&&e.input&&(n=e.input)}else{const e=t.getFormat("markdown");e&&e.input&&(n=e.input)}for(const t of n){const n=this.toRuntimeEntry(t,0,r);(n.stage??p)<=i&&e.push(n)}}}from(t,e){const r=this.resolveFormat(t);if(!r.input||0===r.input.length)throw new Error(`[MdastPlus] Format '${r.id}' does not support input.`);for(const t of r.input)this.queue.push(this.toRuntimeEntry(t,0,e));return this}resolveRunQueue(t,e,r,i){let n=[];if(n=e?this.queue.map(t=>this.toRuntimeEntry(t,t.stage??p,e)):[...this.queue],void 0!==r){const t=r;n=n.filter(e=>(e.stage??p)<=t);const s=n.findIndex(e=>(e.stage??p)===t),o=-1!==s?s+(i??0):n.length-1,a=n.slice(0,o+1),l=n.filter(t=>t.main&&!a.includes(t));n=a.concat(l),this.ensureInputPlugins(n,e,t),n.push({plugin:U,options:[],stage:400,order:0})}else if(this.ensureInputPlugins(n,e,400),t.output)for(const r of t.output)n.push(this.toRuntimeEntry(r,300,e));return n}async to(t,e){const r=this.resolveFormat(t);if(!r.output)throw new Error(`[MdastPlus] Format '${r.id}' does not support output.`);let i,n,s;if(e){const t=e;"overrides"in t||"stage"in t||"stopAtIndex"in t?(i=t.overrides,n="string"==typeof t.stage?c[t.stage]:t.stage,s=t.stopAtIndex):i=e}const o=this.resolveRunQueue(r,i,n,s),a=this.assembleProcessor(o);if(G(this.input)){const t=await a.run(this.input),e=a.stringify(t),r=new m.VFile;return"string"==typeof e||Buffer.isBuffer(e)?r.value=e:r.result=e,r}return a.process(this.input)}use(t,...e){return this.useAt("compile",t,...e)}normalizePluggable(t,e=[]){const r=[];if(Array.isArray(t)){let i=!1;if(t.length>0&&"function"==typeof t[0]){t.slice(1).some(t=>!function(t){if("function"==typeof t)return!0;if(Array.isArray(t))return!0;if("object"==typeof t&&null!==t){if("plugin"in t&&"function"==typeof t.plugin)return!0;if("plugins"in t)return!0}return!1}(t))&&(i=!0)}if(i){const[e,...i]=t;r.push({plugin:e,options:i})}else for(const i of t)r.push(...this.normalizePluggable(i,e))}else if("function"==typeof t)r.push({plugin:t,options:e});else if("object"==typeof t&&null!==t)if("plugin"in t&&"function"==typeof t.plugin){const i=t,n=e.length>0?e:i.options||[];r.push({...i,options:n})}else if("plugins"in t){const e=t;e.plugins&&r.push(...this.normalizePluggable(e.plugins))}return r}useAt(t,e,...r){let i,n,s;"string"==typeof t||"number"==typeof t?(i=("string"==typeof t?c[t]:t)??p,n=e,s=r):(i=void 0,n=t,s=[e,...r].filter(t=>void 0!==t));const o=this.normalizePluggable(n,s);for(const t of o){const e=i??p;this.queue.push(this.toRuntimeEntry(t,e))}return this}priority(t){const e=this.queue[this.queue.length-1];return e&&(e.order=t),this}configure(t,...e){for(let r=this.queue.length-1;r>=0;r--){const i=this.queue[r];if((i.name||i.plugin.name)===t){i.options=e;break}}return this}assembleProcessor(t){const e={};for(const r of t){const t=r.stage??p;e[t]||(e[t]=[]),e[t].push(r)}const r=[],i=Object.keys(e).map(Number).sort((t,e)=>t-e);for(const t of i){const i=e[t].sort((t,e)=>(t.order||0)-(e.order||0)),n=i.findIndex(t=>t.main);if(-1!==n){const t=i[n];!1===t.options?.[0]?console.warn(`Main Plugin "${t.name}" is disabled. Skipping.`):0!==n&&(i.splice(n,1),i[0]=t)}let s=!0,o=0;for(;s&&o<i.length;){s=!1,o++;for(let t=0;t<i.length;t++){const e=i[t];if(e.after){const r=i.findIndex(t=>t.name===e.after);if(-1!==r&&r>t){i.splice(t,1),i.splice(r,0,e),s=!0;break}}if(e.before){const r=i.findIndex(t=>t.name===e.before);if(-1!==r&&r<t){i.splice(t,1),i.splice(r,0,e),s=!0;break}}}}r.push(...i)}const n=(0,f.unified)();Object.keys(this._data).length>0&&n.data(this._data);for(const t of r)n.use(t.plugin,...t.options||[]);return n}};J.registry=new Map;var Q=J,K=class extends Q{async toMarkdown(){const t=await this.to("markdown");return String(t)}toMarkdownVFile(){return this.to("markdown")}async toHtml(){const t=await this.to("html");return String(t)}toHtmlVFile(){return this.to("html")}async toAst(t){return(await this.to("ast",t)).result}toHTML(){return this.toHtml()}toAST(t){return this.toAst(t)}};function W(t){return new K(t)}K.register(R),K.register(I),K.register(_);var X=require("unist-util-visit"),Y=require("hast-util-from-html"),Z=require("hast-util-from-dom"),tt=require("yaml"),et=require("lodash-es"),rt=function(t){const{readability:e,jsdom:r,hast:i,url:n,fields:s}=t||{};this.parser=function(o,a){if(!1===e)return(0,Y.fromHtml)(o,{fragment:!0,...i});let l,u;try{l=require("jsdom").JSDOM;u=require("@mozilla/readability").Readability}catch(t){throw new Error("[html-readability] Dependency missing. Please install 'jsdom' and '@mozilla/readability'.")}const c=new u(new l(o,{url:n,pretendToBeVisual:!0,...r,includeNodeLocations:!0}).window.document,{maxElemsToParse:1e5,nbTopCandidates:5,charThreshold:500,keepClasses:!0,...e,serializer:t=>t}).parse();if(!c||!c.content)return(0,Y.fromHtml)(o,{fragment:!0,...i});const p=c.content;let f=(0,Z.fromDom)(p,{afterTransform:i?.afterTransform});const m=!1!==i?.fragment,{smartExcerpt:h=!0,sourceLink:d}=t||{};let g=(0,et.omitBy)(c,t=>null==t||""===t);const y=n||g.url,b=g.title;if(h&&g.excerpt&&g.textContent){const{threshold:t=.6,minContentLength:e=300}=!0===h?{}:h,r=g.excerpt.trim().toLowerCase().replace(/\s+/g," "),i=g.textContent.trim().toLowerCase().replace(/\s+/g," ");if(i.includes(r)){(r.length/i.length>t||i.length<e)&&delete g.excerpt}}if(n&&(g.url=n),delete g.content,delete g.textContent,s)if(Array.isArray(s))g=(0,et.pick)(g,s);else{const t={};for(const[e,r]of Object.entries(s))void 0!==g[e]&&(t[r]=g[e]);g=t}if(a&&(a.data=a.data||{},a.data.readability=g),f){const t=[f];d&&y&&!lt(f,y)&&t.push({type:"element",tagName:"blockquote",properties:{},children:[{type:"element",tagName:"p",properties:{},children:[{type:"text",value:"Source: "},{type:"element",tagName:"a",properties:{href:y},children:[{type:"text",value:b||y}]}]}]}),f=m?{type:"root",children:t}:{type:"root",children:[{type:"element",tagName:"html",properties:{},children:[{type:"element",tagName:"head",properties:{},children:[]},{type:"element",tagName:"body",properties:{},children:t}]}]},f.data=f.data||{},f.data.readability=g}return f}},it={name:"readability",plugin:rt,stage:0,main:!0},nt={name:"restore-readability-meta",plugin:t=>(e,r)=>{if(r.data?.readability){e.data=e.data||{},e.data.readability=r.data.readability;const{frontmatter:i,sourceLink:n}=t||{};if(i){const t="toml"===i?"toml":"yaml",n=(0,tt.stringify)(r.data.readability).trim();e.children.unshift({type:t,value:n})}if(n&&r.data.readability.url){const{url:t,title:i}=r.data.readability;at(e,t)||e.children.push({type:"blockquote",children:[{type:"paragraph",children:[{type:"text",value:"Source: "},{type:"link",url:t,children:[{type:"text",value:i||t}]}]}]})}}},stage:0,after:"rehype-remark"},st=[it,nt];function ot(t,e){if(!t||!e)return!1;if(t===e)return!0;try{const r=new URL(t),i=new URL(e);return r.href===i.href}catch{return t.replace(/\/$/,"")===e.replace(/\/$/,"")}}function at(t,e){if(!e)return!1;let r=!1;return(0,X.visit)(t,"link",t=>{if(ot(t.url,e))return r=!0,!1}),r}function lt(t,e){if(!e)return!1;let r=!1;return(0,X.visit)(t,"element",t=>{if("a"===t.tagName&&t.properties&&ot(t.properties.href,e))return r=!0,!1}),r}

package/dist/index.mjs

@@ -1 +1 @@
-var t=(t=>"undefined"!=typeof require?require:"undefined"!=typeof Proxy?new Proxy(t,{get:(t,e)=>("undefined"!=typeof require?require:t)[e]}):t)(function(t){if("undefined"!=typeof require)return require.apply(this,arguments);throw Error('Dynamic require of "'+t+'" is not supported')}),e=(t=>(t[t.parse=0]="parse",t[t.normalize=100]="normalize",t[t.compile=200]="compile",t[t.finalize=300]="finalize",t[t.stringify=400]="stringify",t))(e||{}),r=200;import{unified as i}from"unified";import{VFile as n}from"vfile";import o from"remark-parse";import s from"remark-stringify";import a from"remark-gfm";import u from"remark-directive";import l from"remark-math";import p from"remark-frontmatter";import{visit as f}from"unist-util-visit";function c(t,e){return{type:t,children:e,data:{hName:t}}}var m={plugin:()=>t=>{!function(t){f(t,"text",(t,e,r)=>{if(!r||void 0===e)return;const i=t.value;let n=0;const o=[];let s=!1;const a=/(==[^=]+==|~[^~]+~|\^[^^]+\^)/g;let u;for(;null!==(u=a.exec(i));){s=!0;const t=u[0],e=u.index;e>n&&o.push({type:"text",value:i.slice(n,e)});let r="mark",l="";t.startsWith("==")?(r="mark",l=t.slice(2,-2)):t.startsWith("~")?(r="sub",l=t.slice(1,-1)):t.startsWith("^")&&(r="sup",l=t.slice(1,-1)),o.push(c(r,[{type:"text",value:l}])),n=a.lastIndex}return s?(n<i.length&&o.push({type:"text",value:i.slice(n)}),r.children.splice(e,1,...o),e+o.length):void 0})}(t)},stage:100};import{visit as h}from"unist-util-visit";var d={error:"danger",warn:"warning",success:"tip",important:"important",caution:"caution",note:"note"},g={plugin:()=>async t=>{h(t,["containerDirective","leafDirective","textDirective"],t=>{const e=t,r=e.name.toLowerCase();if(e.name=d[r]||r,e.children&&e.children.length>0){const t=e.children[0];if(t.data?.directiveLabel||"directiveLabel"===t.type){let r="";h(t,"text",t=>{r+=t.value}),r&&!e.attributes?.title&&(e.attributes=e.attributes||{},e.attributes.title=r.trim()),e.children.shift()}}e.attributes?.title&&(e.attributes.title=String(e.attributes.title).trim()),e.data=e.data||{},e.data.hName=e.data.hName||("containerDirective"===e.type?"div":"span"),e.data.hProperties={...e.data.hProperties||{},...e.attributes,className:[e.name,e.data.hProperties?.className].filter(Boolean).join(" ")}})},stage:100,order:10};import{visit as y}from"unist-util-visit";var v={plugin:()=>async t=>{y(t,"tableCell",t=>{if(t.data){const{rowspan:e,colspan:r}=t.data;t.data.hProperties=t.data.hProperties||{},void 0!==e&&(t.data.hProperties.rowSpan=e,delete t.data.rowspan),void 0!==r&&(t.data.hProperties.colSpan=r,delete t.data.colspan)}})},stage:100,order:20};import{visit as w}from"unist-util-visit";import{parse as b}from"shell-quote";var k={plugin:()=>async t=>{w(t,"code",t=>{if(t.meta){const e=function(t){const e={},r=b(t);for(const t of r)if("string"==typeof t){const r=t.split("=",2);2===r.length?e[r[0]]=r[1]:e[t]="true"}return e}(t.meta),r=t.data=t.data||{};e.title&&(r.title=e.title),e.filename&&(r.filename=e.filename),r.kv={...r.kv||{},...e}}})},stage:100,order:30};import{visit as x}from"unist-util-visit";var T={plugin:()=>async t=>{x(t,"image",t=>{const e=t.data=t.data||{},r=e.hProperties=e.hProperties||{},i=/[#?&](?:width=([0-9]+))?(?:&?height=([0-9]+))?(?:=([0-9]+)x([0-9]+))?$/,n=t.url.match(i);if(n){const e=n[1]||n[3],o=n[2]||n[4];e&&!r.width&&(r.width=parseInt(e,10)),o&&!r.height&&(r.height=parseInt(o,10)),t.url=t.url.replace(i,"")}e.width&&!r.width&&(r.width=e.width),e.height&&!r.height&&(r.height=e.height)})},stage:100,order:40},M=[{plugin:o,stage:0},{plugin:a,options:[{singleTilde:!1}],stage:0},{plugin:u,stage:0},{plugin:l,stage:0},{plugin:p,options:[["yaml","toml"]],stage:0},g,v,k,T,m],S=[{plugin:s,options:[{handlers:{mark:(t,e,r)=>"=="+r.containerPhrasing(t,{before:"==",after:"=="})+"==",sub:(t,e,r)=>"~"+r.containerPhrasing(t,{before:"~",after:"~"})+"~",sup:(t,e,r)=>"^"+r.containerPhrasing(t,{before:"^",after:"^"})+"^"}}],stage:400},{plugin:a,options:[{singleTilde:!1}],stage:400},{plugin:u,stage:400},{plugin:l,stage:400},{plugin:p,options:[["yaml","toml"]],stage:400}];S.forEach(t=>{t.plugin===s?t.order=100:t.order=10});var P={id:"markdown",title:"Markdown (GFM + Directives)",extensions:["md","markdown","mdown","mkdn"],mediaTypes:["text/markdown"],input:M,output:S};import j from"rehype-parse";import N from"rehype-remark";import q from"remark-rehype";import A,{defaultSchema as F}from"rehype-sanitize";import D from"rehype-stringify";var z={id:"html",title:"HTML",extensions:["html","htm"],mediaTypes:["text/html"],input:[{name:"rehype-parse",plugin:j,stage:0},{name:"rehype-remark",plugin:N,options:[{handlers:{mark:(t,e)=>{const r={type:"mark",children:t.all(e)};return t.patch(e,r),r},sub:(t,e)=>{const r={type:"sub",children:t.all(e)};return t.patch(e,r),r},sup:(t,e)=>{const r={type:"sup",children:t.all(e)};return t.patch(e,r),r}}}],stage:0}],output:[{plugin:q,stage:300,order:10},{plugin:A,options:[{...F,tagNames:[...F.tagNames||[],"mark","sub","sup"],attributes:{...F.attributes,"*":[...F.attributes?.["*"]||[],"className","id","style"],td:[...F.attributes?.td||[],"rowSpan","colSpan","rowspan","colspan"],th:[...F.attributes?.th||[],"rowSpan","colSpan","rowspan","colspan"],img:[...F.attributes?.img||[],"width","height"]}}],stage:300,order:20},{plugin:D,stage:400}]};function E(){this.Compiler=t=>t}function L(){this.Parser=t=>JSON.parse(t)}var $={id:"ast",title:"MDAST",input:[{plugin:L,stage:0},g,v,k,T,m],output:[{plugin:E,options:[],stage:400}]};function H(t){return"object"==typeof t&&null!==t&&"string"==typeof t.type}var I=class t{constructor(t){this.queue=[],this._data={},this.input=t}static register(t){this.registry.set(t.id,t)}static getFormat(t){return this.registry.get(t)}data(t,e){return"string"==typeof t?this._data[t]=e:Object.assign(this._data,t),this}getFormat(t){return this.constructor.getFormat(t)}resolveFormat(t){if("string"==typeof t){const e=this.getFormat(t);if(!e)throw new Error(`[MdastPlus] Format '${t}' is not registered.`);return e}return t}toRuntimeEntry(t,r,i){let n=r;void 0!==t.stage&&(n="string"==typeof t.stage?e[t.stage]??r:t.stage);let o=t.options||[];const s=t.name||t.plugin.name;if(i&&s&&s in i){const e=i[s];"object"!=typeof e||null===e||Array.isArray(e)?o=Array.isArray(e)?e:[e]:("main"in e&&(t.main=!!e.main),"before"in e&&(t.before=e.before),"after"in e&&(t.after=e.after),o=[e])}return{name:s,plugin:t.plugin,options:o,stage:n,order:t.order||0,main:t.main,before:t.before,after:t.after}}ensureInputPlugins(e,i,n=400){const o=e.some(t=>0===(t.stage??r)),s=H(this.input);if(!o){let o=[];if(s){const e=t.getFormat("ast");e&&e.input&&(o=e.input)}else{const e=t.getFormat("markdown");e&&e.input&&(o=e.input)}for(const t of o){const o=this.toRuntimeEntry(t,0,i);(o.stage??r)<=n&&e.push(o)}}}from(t,e){const r=this.resolveFormat(t);if(!r.input||0===r.input.length)throw new Error(`[MdastPlus] Format '${r.id}' does not support input.`);for(const t of r.input)this.queue.push(this.toRuntimeEntry(t,0,e));return this}resolveRunQueue(t,e,i,n){let o=[];if(o=e?this.queue.map(t=>this.toRuntimeEntry(t,t.stage??r,e)):[...this.queue],void 0!==i){const t=i;o=o.filter(e=>(e.stage??r)<=t);const s=o.findIndex(e=>(e.stage??r)===t),a=-1!==s?s+(n??0):o.length-1,u=o.slice(0,a+1),l=o.filter(t=>t.main&&!u.includes(t));o=u.concat(l),this.ensureInputPlugins(o,e,t),o.push({plugin:E,options:[],stage:400,order:0})}else if(this.ensureInputPlugins(o,e,400),t.output)for(const r of t.output)o.push(this.toRuntimeEntry(r,300,e));return o}async to(t,r){const i=this.resolveFormat(t);if(!i.output)throw new Error(`[MdastPlus] Format '${i.id}' does not support output.`);let o,s,a;if(r){const t=r;"overrides"in t||"stage"in t||"stopAtIndex"in t?(o=t.overrides,s="string"==typeof t.stage?e[t.stage]:t.stage,a=t.stopAtIndex):o=r}const u=this.resolveRunQueue(i,o,s,a),l=this.assembleProcessor(u);if(H(this.input)){const t=await l.run(this.input),e=l.stringify(t),r=new n;return"string"==typeof e||Buffer.isBuffer(e)?r.value=e:r.result=e,r}return l.process(this.input)}use(t,...e){return this.useAt("compile",t,...e)}normalizePluggable(t,e=[]){const r=[];if(Array.isArray(t)){let i=!1;if(t.length>0&&"function"==typeof t[0]){t.slice(1).some(t=>!function(t){if("function"==typeof t)return!0;if(Array.isArray(t))return!0;if("object"==typeof t&&null!==t){if("plugin"in t&&"function"==typeof t.plugin)return!0;if("plugins"in t)return!0}return!1}(t))&&(i=!0)}if(i){const[e,...i]=t;r.push({plugin:e,options:i})}else for(const i of t)r.push(...this.normalizePluggable(i,e))}else if("function"==typeof t)r.push({plugin:t,options:e});else if("object"==typeof t&&null!==t)if("plugin"in t&&"function"==typeof t.plugin){const i=t,n=e.length>0?e:i.options||[];r.push({...i,options:n})}else if("plugins"in t){const e=t;e.plugins&&r.push(...this.normalizePluggable(e.plugins))}return r}useAt(t,i,...n){let o,s,a;"string"==typeof t||"number"==typeof t?(o=("string"==typeof t?e[t]:t)??r,s=i,a=n):(o=void 0,s=t,a=[i,...n].filter(t=>void 0!==t));const u=this.normalizePluggable(s,a);for(const t of u){const e=o??r;this.queue.push(this.toRuntimeEntry(t,e))}return this}priority(t){const e=this.queue[this.queue.length-1];return e&&(e.order=t),this}configure(t,...e){for(let r=this.queue.length-1;r>=0;r--){const i=this.queue[r];if((i.name||i.plugin.name)===t){i.options=e;break}}return this}assembleProcessor(t){const e={};for(const i of t){const t=i.stage??r;e[t]||(e[t]=[]),e[t].push(i)}const n=[],o=Object.keys(e).map(Number).sort((t,e)=>t-e);for(const t of o){const r=e[t].sort((t,e)=>(t.order||0)-(e.order||0)),i=r.findIndex(t=>t.main);if(-1!==i){const t=r[i];!1===t.options?.[0]?console.warn(`Main Plugin "${t.name}" is disabled. Skipping.`):0!==i&&(r.splice(i,1),r[0]=t)}let o=!0,s=0;for(;o&&s<r.length;){o=!1,s++;for(let t=0;t<r.length;t++){const e=r[t];if(e.after){const i=r.findIndex(t=>t.name===e.after);if(-1!==i&&i>t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}if(e.before){const i=r.findIndex(t=>t.name===e.before);if(-1!==i&&i<t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}}}n.push(...r)}const s=i();Object.keys(this._data).length>0&&s.data(this._data);for(const t of n)s.use(t.plugin,...t.options||[]);return s}};I.registry=new Map;var O=I,R=class extends O{async toMarkdown(){const t=await this.to("markdown");return String(t)}toMarkdownVFile(){return this.to("markdown")}async toHtml(){const t=await this.to("html");return String(t)}toHtmlVFile(){return this.to("html")}async toAst(t){return(await this.to("ast",t)).result}toHTML(){return this.toHtml()}toAST(t){return this.toAst(t)}};function B(t){return new R(t)}R.register(P),R.register(z),R.register($);import{visit as C}from"unist-util-visit";import{fromHtml as V}from"hast-util-from-html";import{fromDom as U}from"hast-util-from-dom";import{stringify as G}from"yaml";import{isNil as J,omitBy as Q}from"lodash-es";var K=function(e){const{readability:r,jsdom:i,hast:n,url:o}=e||{};this.parser=function(e,s){if(!1===r)return V(e,{fragment:!0,...n});let a,u;try{a=t("jsdom").JSDOM;u=t("@mozilla/readability").Readability}catch(t){throw new Error("[html-readability] Dependency missing. Please install 'jsdom' and '@mozilla/readability'.")}const l=new u(new a(e,{url:o,pretendToBeVisual:!0,...i,includeNodeLocations:!0}).window.document,{maxElemsToParse:1e5,nbTopCandidates:5,charThreshold:500,keepClasses:!0,...r,serializer:t=>t}).parse();if(!l||!l.content)return V(e,{fragment:!0,...n});const p=l.content;let f=U(p,{afterTransform:n?.afterTransform});const c=!1!==n?.fragment,m=Q(l,J);return o&&(m.url=o),delete m.content,delete m.textContent,s&&(s.data=s.data||{},s.data.readability=m),f&&(f=c?{type:"root",children:[f]}:{type:"root",children:[{type:"element",tagName:"html",properties:{},children:[{type:"element",tagName:"head",properties:{},children:[]},{type:"element",tagName:"body",properties:{},children:[f]}]}]},f.data=f.data||{},f.data.readability=m),f}},W={name:"readability",plugin:K,stage:0,main:!0},X={name:"restore-readability-meta",plugin:t=>(e,r)=>{if(r.data?.readability){e.data=e.data||{},e.data.readability=r.data.readability;const{frontmatter:i,sourceLink:n}=t||{};if(i){const t="toml"===i?"toml":"yaml",n=G(r.data.readability).trim();e.children.unshift({type:t,value:n})}if(n&&r.data.readability.url){const{url:t,title:i}=r.data.readability;Z(e,t)||e.children.push({type:"blockquote",children:[{type:"paragraph",children:[{type:"text",value:"Source: "},{type:"link",url:t,children:[{type:"text",value:i||t}]}]}]})}}},stage:0,after:"rehype-remark"},Y=[W,X];function Z(t,e){if(!e)return!1;let r=!1;return C(t,"link",t=>{if(function(t,e){if(!t||!e)return!1;if(t===e)return!0;try{const r=new URL(t),i=new URL(e);return r.href===i.href}catch{return t.replace(/\/$/,"")===e.replace(/\/$/,"")}}(t.url,e))return r=!0,!1}),r}export{r as DefaultPipelineStage,O as MdastBasePipeline,R as MdastPipeline,e as PipelineStage,E as astCompiler,$ as astFormat,Z as checkUrlExists,z as htmlFormat,K as htmlReadability,W as htmlReadabilityPlugin,Y as htmlReadabilityPlugins,L as jsonParser,P as markdownFormat,B as mdast,X as restoreReadabilityMetaPlugin};
+var t=(t=>"undefined"!=typeof require?require:"undefined"!=typeof Proxy?new Proxy(t,{get:(t,e)=>("undefined"!=typeof require?require:t)[e]}):t)(function(t){if("undefined"!=typeof require)return require.apply(this,arguments);throw Error('Dynamic require of "'+t+'" is not supported')}),e=(t=>(t[t.parse=0]="parse",t[t.normalize=100]="normalize",t[t.compile=200]="compile",t[t.finalize=300]="finalize",t[t.stringify=400]="stringify",t))(e||{}),r=200;import{unified as i}from"unified";import{VFile as n}from"vfile";import o from"remark-parse";import s from"remark-stringify";import a from"remark-gfm";import l from"remark-directive";import u from"remark-math";import p from"remark-frontmatter";import{visit as f}from"unist-util-visit";function c(t,e){return{type:t,children:e,data:{hName:t}}}var m={plugin:()=>t=>{!function(t){f(t,"text",(t,e,r)=>{if(!r||void 0===e)return;const i=t.value;let n=0;const o=[];let s=!1;const a=/(==[^=]+==|~[^~]+~|\^[^^]+\^)/g;let l;for(;null!==(l=a.exec(i));){s=!0;const t=l[0],e=l.index;e>n&&o.push({type:"text",value:i.slice(n,e)});let r="mark",u="";t.startsWith("==")?(r="mark",u=t.slice(2,-2)):t.startsWith("~")?(r="sub",u=t.slice(1,-1)):t.startsWith("^")&&(r="sup",u=t.slice(1,-1)),o.push(c(r,[{type:"text",value:u}])),n=a.lastIndex}return s?(n<i.length&&o.push({type:"text",value:i.slice(n)}),r.children.splice(e,1,...o),e+o.length):void 0})}(t)},stage:100};import{visit as h}from"unist-util-visit";var d={error:"danger",warn:"warning",success:"tip",important:"important",caution:"caution",note:"note"},g={plugin:()=>async t=>{h(t,["containerDirective","leafDirective","textDirective"],t=>{const e=t,r=e.name.toLowerCase();if(e.name=d[r]||r,e.children&&e.children.length>0){const t=e.children[0];if(t.data?.directiveLabel||"directiveLabel"===t.type){let r="";h(t,"text",t=>{r+=t.value}),r&&!e.attributes?.title&&(e.attributes=e.attributes||{},e.attributes.title=r.trim()),e.children.shift()}}e.attributes?.title&&(e.attributes.title=String(e.attributes.title).trim()),e.data=e.data||{},e.data.hName=e.data.hName||("containerDirective"===e.type?"div":"span"),e.data.hProperties={...e.data.hProperties||{},...e.attributes,className:[e.name,e.data.hProperties?.className].filter(Boolean).join(" ")}})},stage:100,order:10};import{visit as y}from"unist-util-visit";var v={plugin:()=>async t=>{y(t,"tableCell",t=>{if(t.data){const{rowspan:e,colspan:r}=t.data;t.data.hProperties=t.data.hProperties||{},void 0!==e&&(t.data.hProperties.rowSpan=e,delete t.data.rowspan),void 0!==r&&(t.data.hProperties.colSpan=r,delete t.data.colspan)}})},stage:100,order:20};import{visit as w}from"unist-util-visit";import{parse as b}from"shell-quote";var k={plugin:()=>async t=>{w(t,"code",t=>{if(t.meta){const e=function(t){const e={},r=b(t);for(const t of r)if("string"==typeof t){const r=t.split("=",2);2===r.length?e[r[0]]=r[1]:e[t]="true"}return e}(t.meta),r=t.data=t.data||{};e.title&&(r.title=e.title),e.filename&&(r.filename=e.filename),r.kv={...r.kv||{},...e}}})},stage:100,order:30};import{visit as x}from"unist-util-visit";var N={plugin:()=>async t=>{x(t,"image",t=>{const e=t.data=t.data||{},r=e.hProperties=e.hProperties||{},i=/[#?&](?:width=([0-9]+))?(?:&?height=([0-9]+))?(?:=([0-9]+)x([0-9]+))?$/,n=t.url.match(i);if(n){const e=n[1]||n[3],o=n[2]||n[4];e&&!r.width&&(r.width=parseInt(e,10)),o&&!r.height&&(r.height=parseInt(o,10)),t.url=t.url.replace(i,"")}e.width&&!r.width&&(r.width=e.width),e.height&&!r.height&&(r.height=e.height)})},stage:100,order:40},S=[{plugin:o,stage:0},{plugin:a,options:[{singleTilde:!1}],stage:0},{plugin:l,stage:0},{plugin:u,stage:0},{plugin:p,options:[["yaml","toml"]],stage:0},g,v,k,N,m],T=[{plugin:s,options:[{handlers:{mark:(t,e,r)=>"=="+r.containerPhrasing(t,{before:"==",after:"=="})+"==",sub:(t,e,r)=>"~"+r.containerPhrasing(t,{before:"~",after:"~"})+"~",sup:(t,e,r)=>"^"+r.containerPhrasing(t,{before:"^",after:"^"})+"^"}}],stage:400},{plugin:a,options:[{singleTilde:!1}],stage:400},{plugin:l,stage:400},{plugin:u,stage:400},{plugin:p,options:[["yaml","toml"]],stage:400}];T.forEach(t=>{t.plugin===s?t.order=100:t.order=10});var M={id:"markdown",title:"Markdown (GFM + Directives)",extensions:["md","markdown","mdown","mkdn"],mediaTypes:["text/markdown"],input:S,output:T};import j from"rehype-parse";import P from"rehype-remark";import q from"remark-rehype";import A,{defaultSchema as F}from"rehype-sanitize";import L from"rehype-stringify";var D={id:"html",title:"HTML",extensions:["html","htm"],mediaTypes:["text/html"],input:[{name:"rehype-parse",plugin:j,stage:0},{name:"rehype-remark",plugin:P,options:[{handlers:{mark:(t,e)=>{const r={type:"mark",children:t.all(e)};return t.patch(e,r),r},sub:(t,e)=>{const r={type:"sub",children:t.all(e)};return t.patch(e,r),r},sup:(t,e)=>{const r={type:"sup",children:t.all(e)};return t.patch(e,r),r}}}],stage:0}],output:[{plugin:q,stage:300,order:10},{plugin:A,options:[{...F,tagNames:[...F.tagNames||[],"mark","sub","sup"],attributes:{...F.attributes,"*":[...F.attributes?.["*"]||[],"className","id","style"],td:[...F.attributes?.td||[],"rowSpan","colSpan","rowspan","colspan"],th:[...F.attributes?.th||[],"rowSpan","colSpan","rowspan","colspan"],img:[...F.attributes?.img||[],"width","height"]}}],stage:300,order:20},{plugin:L,stage:400}]};function E(){this.Compiler=t=>t}function z(){this.Parser=t=>JSON.parse(t)}var $={id:"ast",title:"MDAST",input:[{plugin:z,stage:0},g,v,k,N,m],output:[{plugin:E,options:[],stage:400}]};function O(t){return"object"==typeof t&&null!==t&&"string"==typeof t.type}var C=class t{constructor(t){this.queue=[],this._data={},this.input=t}static register(t){this.registry.set(t.id,t)}static getFormat(t){return this.registry.get(t)}data(t,e){return"string"==typeof t?this._data[t]=e:Object.assign(this._data,t),this}getFormat(t){return this.constructor.getFormat(t)}resolveFormat(t){if("string"==typeof t){const e=this.getFormat(t);if(!e)throw new Error(`[MdastPlus] Format '${t}' is not registered.`);return e}return t}toRuntimeEntry(t,r,i){let n=r;void 0!==t.stage&&(n="string"==typeof t.stage?e[t.stage]??r:t.stage);let o=t.options||[];const s=t.name||t.plugin.name;if(i&&s&&s in i){const e=i[s];"object"!=typeof e||null===e||Array.isArray(e)?o=Array.isArray(e)?e:[e]:("main"in e&&(t.main=!!e.main),"before"in e&&(t.before=e.before),"after"in e&&(t.after=e.after),o=[e])}return{name:s,plugin:t.plugin,options:o,stage:n,order:t.order||0,main:t.main,before:t.before,after:t.after}}ensureInputPlugins(e,i,n=400){const o=e.some(t=>0===(t.stage??r)),s=O(this.input);if(!o){let o=[];if(s){const e=t.getFormat("ast");e&&e.input&&(o=e.input)}else{const e=t.getFormat("markdown");e&&e.input&&(o=e.input)}for(const t of o){const o=this.toRuntimeEntry(t,0,i);(o.stage??r)<=n&&e.push(o)}}}from(t,e){const r=this.resolveFormat(t);if(!r.input||0===r.input.length)throw new Error(`[MdastPlus] Format '${r.id}' does not support input.`);for(const t of r.input)this.queue.push(this.toRuntimeEntry(t,0,e));return this}resolveRunQueue(t,e,i,n){let o=[];if(o=e?this.queue.map(t=>this.toRuntimeEntry(t,t.stage??r,e)):[...this.queue],void 0!==i){const t=i;o=o.filter(e=>(e.stage??r)<=t);const s=o.findIndex(e=>(e.stage??r)===t),a=-1!==s?s+(n??0):o.length-1,l=o.slice(0,a+1),u=o.filter(t=>t.main&&!l.includes(t));o=l.concat(u),this.ensureInputPlugins(o,e,t),o.push({plugin:E,options:[],stage:400,order:0})}else if(this.ensureInputPlugins(o,e,400),t.output)for(const r of t.output)o.push(this.toRuntimeEntry(r,300,e));return o}async to(t,r){const i=this.resolveFormat(t);if(!i.output)throw new Error(`[MdastPlus] Format '${i.id}' does not support output.`);let o,s,a;if(r){const t=r;"overrides"in t||"stage"in t||"stopAtIndex"in t?(o=t.overrides,s="string"==typeof t.stage?e[t.stage]:t.stage,a=t.stopAtIndex):o=r}const l=this.resolveRunQueue(i,o,s,a),u=this.assembleProcessor(l);if(O(this.input)){const t=await u.run(this.input),e=u.stringify(t),r=new n;return"string"==typeof e||Buffer.isBuffer(e)?r.value=e:r.result=e,r}return u.process(this.input)}use(t,...e){return this.useAt("compile",t,...e)}normalizePluggable(t,e=[]){const r=[];if(Array.isArray(t)){let i=!1;if(t.length>0&&"function"==typeof t[0]){t.slice(1).some(t=>!function(t){if("function"==typeof t)return!0;if(Array.isArray(t))return!0;if("object"==typeof t&&null!==t){if("plugin"in t&&"function"==typeof t.plugin)return!0;if("plugins"in t)return!0}return!1}(t))&&(i=!0)}if(i){const[e,...i]=t;r.push({plugin:e,options:i})}else for(const i of t)r.push(...this.normalizePluggable(i,e))}else if("function"==typeof t)r.push({plugin:t,options:e});else if("object"==typeof t&&null!==t)if("plugin"in t&&"function"==typeof t.plugin){const i=t,n=e.length>0?e:i.options||[];r.push({...i,options:n})}else if("plugins"in t){const e=t;e.plugins&&r.push(...this.normalizePluggable(e.plugins))}return r}useAt(t,i,...n){let o,s,a;"string"==typeof t||"number"==typeof t?(o=("string"==typeof t?e[t]:t)??r,s=i,a=n):(o=void 0,s=t,a=[i,...n].filter(t=>void 0!==t));const l=this.normalizePluggable(s,a);for(const t of l){const e=o??r;this.queue.push(this.toRuntimeEntry(t,e))}return this}priority(t){const e=this.queue[this.queue.length-1];return e&&(e.order=t),this}configure(t,...e){for(let r=this.queue.length-1;r>=0;r--){const i=this.queue[r];if((i.name||i.plugin.name)===t){i.options=e;break}}return this}assembleProcessor(t){const e={};for(const i of t){const t=i.stage??r;e[t]||(e[t]=[]),e[t].push(i)}const n=[],o=Object.keys(e).map(Number).sort((t,e)=>t-e);for(const t of o){const r=e[t].sort((t,e)=>(t.order||0)-(e.order||0)),i=r.findIndex(t=>t.main);if(-1!==i){const t=r[i];!1===t.options?.[0]?console.warn(`Main Plugin "${t.name}" is disabled. Skipping.`):0!==i&&(r.splice(i,1),r[0]=t)}let o=!0,s=0;for(;o&&s<r.length;){o=!1,s++;for(let t=0;t<r.length;t++){const e=r[t];if(e.after){const i=r.findIndex(t=>t.name===e.after);if(-1!==i&&i>t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}if(e.before){const i=r.findIndex(t=>t.name===e.before);if(-1!==i&&i<t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}}}n.push(...r)}const s=i();Object.keys(this._data).length>0&&s.data(this._data);for(const t of n)s.use(t.plugin,...t.options||[]);return s}};C.registry=new Map;var H=C,I=class extends H{async toMarkdown(){const t=await this.to("markdown");return String(t)}toMarkdownVFile(){return this.to("markdown")}async toHtml(){const t=await this.to("html");return String(t)}toHtmlVFile(){return this.to("html")}async toAst(t){return(await this.to("ast",t)).result}toHTML(){return this.toHtml()}toAST(t){return this.toAst(t)}};function R(t){return new I(t)}I.register(M),I.register(D),I.register($);import{visit as B}from"unist-util-visit";import{fromHtml as V}from"hast-util-from-html";import{fromDom as U}from"hast-util-from-dom";import{stringify as G}from"yaml";import{omitBy as J,pick as Q}from"lodash-es";var K=function(e){const{readability:r,jsdom:i,hast:n,url:o,fields:s}=e||{};this.parser=function(a,l){if(!1===r)return V(a,{fragment:!0,...n});let u,p;try{u=t("jsdom").JSDOM;p=t("@mozilla/readability").Readability}catch(t){throw new Error("[html-readability] Dependency missing. Please install 'jsdom' and '@mozilla/readability'.")}const f=new p(new u(a,{url:o,pretendToBeVisual:!0,...i,includeNodeLocations:!0}).window.document,{maxElemsToParse:1e5,nbTopCandidates:5,charThreshold:500,keepClasses:!0,...r,serializer:t=>t}).parse();if(!f||!f.content)return V(a,{fragment:!0,...n});const c=f.content;let m=U(c,{afterTransform:n?.afterTransform});const h=!1!==n?.fragment,{smartExcerpt:d=!0,sourceLink:g}=e||{};let y=J(f,t=>null==t||""===t);const v=o||y.url,w=y.title;if(d&&y.excerpt&&y.textContent){const{threshold:t=.6,minContentLength:e=300}=!0===d?{}:d,r=y.excerpt.trim().toLowerCase().replace(/\s+/g," "),i=y.textContent.trim().toLowerCase().replace(/\s+/g," ");if(i.includes(r)){(r.length/i.length>t||i.length<e)&&delete y.excerpt}}if(o&&(y.url=o),delete y.content,delete y.textContent,s)if(Array.isArray(s))y=Q(y,s);else{const t={};for(const[e,r]of Object.entries(s))void 0!==y[e]&&(t[r]=y[e]);y=t}if(l&&(l.data=l.data||{},l.data.readability=y),m){const t=[m];g&&v&&!tt(m,v)&&t.push({type:"element",tagName:"blockquote",properties:{},children:[{type:"element",tagName:"p",properties:{},children:[{type:"text",value:"Source: "},{type:"element",tagName:"a",properties:{href:v},children:[{type:"text",value:w||v}]}]}]}),m=h?{type:"root",children:t}:{type:"root",children:[{type:"element",tagName:"html",properties:{},children:[{type:"element",tagName:"head",properties:{},children:[]},{type:"element",tagName:"body",properties:{},children:t}]}]},m.data=m.data||{},m.data.readability=y}return m}},W={name:"readability",plugin:K,stage:0,main:!0},X={name:"restore-readability-meta",plugin:t=>(e,r)=>{if(r.data?.readability){e.data=e.data||{},e.data.readability=r.data.readability;const{frontmatter:i,sourceLink:n}=t||{};if(i){const t="toml"===i?"toml":"yaml",n=G(r.data.readability).trim();e.children.unshift({type:t,value:n})}if(n&&r.data.readability.url){const{url:t,title:i}=r.data.readability;_(e,t)||e.children.push({type:"blockquote",children:[{type:"paragraph",children:[{type:"text",value:"Source: "},{type:"link",url:t,children:[{type:"text",value:i||t}]}]}]})}}},stage:0,after:"rehype-remark"},Y=[W,X];function Z(t,e){if(!t||!e)return!1;if(t===e)return!0;try{const r=new URL(t),i=new URL(e);return r.href===i.href}catch{return t.replace(/\/$/,"")===e.replace(/\/$/,"")}}function _(t,e){if(!e)return!1;let r=!1;return B(t,"link",t=>{if(Z(t.url,e))return r=!0,!1}),r}function tt(t,e){if(!e)return!1;let r=!1;return B(t,"element",t=>{if("a"===t.tagName&&t.properties&&Z(t.properties.href,e))return r=!0,!1}),r}export{r as DefaultPipelineStage,H as MdastBasePipeline,I as MdastPipeline,e as PipelineStage,E as astCompiler,$ as astFormat,tt as checkHtmlUrlExists,_ as checkUrlExists,D as htmlFormat,K as htmlReadability,W as htmlReadabilityPlugin,Y as htmlReadabilityPlugins,z as jsonParser,M as markdownFormat,R as mdast,X as restoreReadabilityMetaPlugin};

package/docs/README.md

@@ -189,7 +189,19 @@ Each stage can have one "main" plugin. If a plugin is marked with `main: true`,
 | `extract-code-meta` | normalize | Parses `title="foo"` from code block meta. |
 | `image-size` | normalize | Parses `#=WxH` from image URLs. |
 | `normalize-inline-styles` | normalize | Standardizes `==mark==`, `~sub~`, and `^sup^`. |
-| `html-readability` | parse | Uses Mozilla's Readability to extract main content from HTML. Supports `frontmatter` injection and `sourceLink` (source link) footer. Use `htmlReadabilityPlugins` array for easier setup. |
+| `html-readability` | parse | Uses Mozilla's Readability to extract main content from HTML. Supports `frontmatter` injection, `sourceLink` (source link) footer, and `smartExcerpt` for intelligent summary management. Use `htmlReadabilityPlugins` array for easier setup. |
+
+### html-readability Options
+
+- `url`: (string) The URL of the HTML document.
+- `frontmatter`: (boolean | 'yaml' | 'toml') Whether to inject metadata as frontmatter. Default: `false`.
+- `sourceLink`: (boolean) Whether to append source link at the bottom. The link will be generated based on the original title/URL even if they are filtered or renamed in the `fields` option. Default: `false`.
+- `fields`: (string[] | object) Control which metadata fields are kept or how they are renamed. 
+  - If an array: acts as an allowlist (e.g., `['title', 'excerpt']`).
+  - If an object: maps original keys to new names (e.g., `{ title: 'headline' }`). Only keys in the map are kept (Projection).
+- `smartExcerpt`: (boolean | object) Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content. Default: `true`.
+  - `threshold`: (number) The ratio of excerpt length to content length (0.0 to 1.0). Default: `0.6`.
+  - `minContentLength`: (number) Minimum length of the main content required to keep the excerpt. Default: `300`.
 
 ## Contributing
 

package/docs/_media/CONTRIBUTING.md

@@ -145,6 +145,16 @@ const myPlugin = function(options) {
 const myPlugin = function(options, context) { ... }
 ```
 
+#### 4. Implementation Considerations: Tree Wrapping
+
+When a plugin needs to wrap the AST tree (e.g., adding `<html>`/`<body>` tags or appending footers), follow this pattern:
+
+1. **Prepare Content First**: Collect all content nodes (main content, footers, etc.) into an array.
+2. **Check Existence**: Always check if a node (like a source link) already exists before appending to avoid duplicates.
+3. **Wrap Once**: Perform the final wrapping logic (e.g., based on `isFragment` option) using the prepared array.
+
+This approach ensures robustness across different output configurations and avoids redundant logic.
+
 ### Main Plugins
 
 Each stage supports a **single** main plugin. When a plugin is marked with `main: true`, it will replace the first plugin that was originally registered for that stage. This is primarily used by formats to allow users to override the default parser or stringifier by injecting a different one at the same stage.

package/docs/_media/README.cn.md

@@ -185,7 +185,19 @@ const result = await mdast('Hello').to('reverse');
 | `extract-code-meta` | normalize | 从代码块元数据中解析 `title="foo"`。 |
 | `image-size` | normalize | 从图片 URL 中解析 `#=WxH`。 |
 | `normalize-inline-styles` | normalize | 标准化 `==mark==`、`~sub~` 和 `^sup^`。 |
-| `html-readability` | parse | 使用 Mozilla 的 Readability 从 HTML 中提取主体内容。支持 `frontmatter` 注入和 `sourceLink` (原文链接) 页脚。使用 `htmlReadabilityPlugins` 数组可以简化配置。 |
+| `html-readability` | parse | 使用 Mozilla 的 Readability 从 HTML 中提取主体内容。支持 `frontmatter` 注入、`sourceLink` (原文链接) 页脚以及 `smartExcerpt` 智能摘要管理。使用 `htmlReadabilityPlugins` 数组可以简化配置。 |
+
+### html-readability 选项
+
+- `url`: (string) HTML 文档的 URL。
+- `frontmatter`: (boolean | 'yaml' | 'toml') 是否将元数据注入为 Frontmatter。默认值: `false`。
+- `sourceLink`: (boolean) 是否在底部添加原文链接。即使在 `fields` 选项中过滤或重命名了 title/url 字段，该链接仍将根据原始信息正确生成。默认值: `false`。
+- `fields`: (string[] | object) 控制保留哪些元数据字段或如何重命名它们。
+  - 如果是数组：作为白名单（例如 `['title', 'excerpt']`）。
+  - 如果是对象：将原始键映射到新名称（例如 `{ title: 'headline' }`）。只有映射中存在的键才会被保留（投影）。
+- `smartExcerpt`: (boolean | object) 是否在摘要与正文内容重复或近乎重复时移除摘要。默认值: `true`。
+  - `threshold`: (number) 摘要长度与正文长度的比率阈值 (0.0 到 1.0)。默认值: `0.6`。
+  - `minContentLength`: (number) 保留摘要所需的正文最小长度。默认值: `300`。
 
 ## 贡献
 

package/docs/classes/MdastBasePipeline.md

@@ -6,7 +6,7 @@
 
 # Class: MdastBasePipeline
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:42](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L42)
+Defined in: [packages/mdast-plus/src/pipeline.ts:42](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L42)
 
 Base implementation of the fluent mdast processing pipeline.
 Manages the plugin registry and the execution queue.
@@ -21,7 +21,7 @@ Manages the plugin registry and the execution queue.
 
 > **new MdastBasePipeline**(`input`): `MdastBasePipeline`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:70](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L70)
+Defined in: [packages/mdast-plus/src/pipeline.ts:70](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L70)
 
 Initializes a new pipeline instance with the given input.
 
@@ -43,7 +43,7 @@ Content to process (string, Buffer, VFile, or AST Node).
 
 > `protected` **\_data**: `Record`\<`string`, `any`\> = `{}`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L64)
+Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L64)
 
 ***
 
@@ -51,7 +51,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mda
 
 > `protected` **input**: `Compatible`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L62)
+Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L62)
 
 ***
 
@@ -59,7 +59,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mda
 
 > `protected` **queue**: [`MdastPlugin`](../interfaces/MdastPlugin.md)[] = `[]`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L63)
+Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L63)
 
 ## Methods
 
@@ -67,7 +67,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mda
 
 > `protected` **assembleProcessor**(`queue`): `Processor`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:494](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L494)
+Defined in: [packages/mdast-plus/src/pipeline.ts:494](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L494)
 
 Assembles a unified processor based on the sorted plugin queue.
 
@@ -87,7 +87,7 @@ Assembles a unified processor based on the sorted plugin queue.
 
 > **configure**(`pluginName`, ...`options`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:478](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L478)
+Defined in: [packages/mdast-plus/src/pipeline.ts:478](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L478)
 
 Modifies the options of a plugin that is already in the pipeline queue.
 Searches from the end of the queue and updates the first match found.
@@ -119,7 +119,7 @@ The pipeline instance for chaining.
 
 > **data**(`key`, `value?`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:80](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L80)
+Defined in: [packages/mdast-plus/src/pipeline.ts:80](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L80)
 
 Configures global data for the pipeline, which will be available to all plugins via this.data().
 
@@ -149,7 +149,7 @@ The pipeline instance for chaining.
 
 > `protected` **ensureInputPlugins**(`queue`, `overrides?`, `maxStage?`): `void`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:165](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L165)
+Defined in: [packages/mdast-plus/src/pipeline.ts:165](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L165)
 
 Ensures that input plugins (parser, normalizers) are present in the queue.
 Adds implicit plugins if no parser is detected.
@@ -178,7 +178,7 @@ Adds implicit plugins if no parser is detected.
 
 > **from**(`fmt`, `overrides?`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:194](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L194)
+Defined in: [packages/mdast-plus/src/pipeline.ts:194](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L194)
 
 Configures the input format and adds its associated plugins to the pipeline.
 
@@ -208,7 +208,7 @@ The pipeline instance for chaining.
 
 > **getFormat**(`id`): `undefined` \| [`MdastFormat`](../interfaces/MdastFormat.md)
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:92](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L92)
+Defined in: [packages/mdast-plus/src/pipeline.ts:92](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L92)
 
 Instance-level access to the global format registry.
 
@@ -228,7 +228,7 @@ Instance-level access to the global format registry.
 
 > **priority**(`order`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:461](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L461)
+Defined in: [packages/mdast-plus/src/pipeline.ts:461](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L461)
 
 Sets the priority order for the most recently added plugin.
 Plugins with lower order run earlier within the same stage.
@@ -253,7 +253,7 @@ The pipeline instance for chaining.
 
 > `protected` **resolveRunQueue**(`format`, `overrides?`, `stage?`, `stopAtIndex?`): [`MdastPlugin`](../interfaces/MdastPlugin.md)[]
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:214](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L214)
+Defined in: [packages/mdast-plus/src/pipeline.ts:214](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L214)
 
 Resolves the final plugin queue for execution based on the target format and run options.
 Calculates the effective plugin list by applying overrides, handling partial execution (stage/stopAtIndex),
@@ -287,7 +287,7 @@ and injecting necessary input/output plugins.
 
 > **to**(`fmt`, `optionsOrOverrides?`): `Promise`\<`VFile`\>
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:290](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L290)
+Defined in: [packages/mdast-plus/src/pipeline.ts:290](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L290)
 
 Processes the pipeline and serializes the result into the specified format.
 
@@ -317,7 +317,7 @@ A promise resolving to a VFile containing the result.
 
 > `protected` **toRuntimeEntry**(`entry`, `defaultStage`, `overrides?`): [`MdastPlugin`](../interfaces/MdastPlugin.md) & `object`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:116](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L116)
+Defined in: [packages/mdast-plus/src/pipeline.ts:116](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L116)
 
 Normalizes a plugin entry for runtime execution.
 
@@ -345,7 +345,7 @@ Normalizes a plugin entry for runtime execution.
 
 > **use**(`plugin`, ...`options`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:336](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L336)
+Defined in: [packages/mdast-plus/src/pipeline.ts:336](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L336)
 
 Adds a plugin or an array of plugins to the pipeline's compile stage.
 
@@ -377,7 +377,7 @@ The pipeline instance for chaining.
 
 > **useAt**(`stage`, `plugin`, ...`options`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:418](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L418)
+Defined in: [packages/mdast-plus/src/pipeline.ts:418](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L418)
 
 Adds a plugin or an array of plugins to the pipeline at a specific stage.
 
@@ -411,7 +411,7 @@ The pipeline instance for chaining.
 
 > **useAt**(`plugin`, ...`options`): `this`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:425](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L425)
+Defined in: [packages/mdast-plus/src/pipeline.ts:425](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L425)
 
 Adds a plugin or an array of plugins to the pipeline. The stage is taken from the plugin object(s) or defaults to 'compile'.
 
@@ -441,7 +441,7 @@ The pipeline instance for chaining.
 
 > `static` **getFormat**(`id`): `undefined` \| [`MdastFormat`](../interfaces/MdastFormat.md)
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:58](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L58)
+Defined in: [packages/mdast-plus/src/pipeline.ts:58](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L58)
 
 Retrieves a registered format by its ID.
 
@@ -465,7 +465,7 @@ The format definition or undefined if not found.
 
 > `static` **register**(`format`): `void`
 
-Defined in: [packages/mdast-plus/src/pipeline.ts:49](https://github.com/isdk/mdast-plus.js/blob/fb690626024343866805f4e55a09e901ff9d4a09/src/pipeline.ts#L49)
+Defined in: [packages/mdast-plus/src/pipeline.ts:49](https://github.com/isdk/mdast-plus.js/blob/fc889343b0a625d8edbf2036c3e25b676d1bf163/src/pipeline.ts#L49)
 
 Registers a global document format.