@isdk/mdast-plus 0.3.0 → 0.3.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.cn.md +1 -0
- package/README.md +1 -0
- package/dist/index.d.mts +5 -0
- package/dist/index.d.ts +5 -0
- package/dist/index.js +1 -1
- package/dist/index.mjs +1 -1
- package/docs/README.md +1 -0
- package/docs/_media/README.cn.md +1 -0
- package/docs/classes/MdastBasePipeline.md +20 -20
- package/docs/classes/MdastPipeline.md +27 -27
- package/docs/enumerations/PipelineStage.md +6 -6
- package/docs/functions/astCompiler.md +1 -1
- package/docs/functions/checkHtmlUrlExists.md +1 -1
- package/docs/functions/checkUrlExists.md +1 -1
- package/docs/functions/jsonParser.md +1 -1
- package/docs/functions/mdast.md +1 -1
- package/docs/interfaces/MdastDataOrigin.md +4 -4
- package/docs/interfaces/MdastFormat.md +7 -7
- package/docs/interfaces/MdastMark.md +3 -3
- package/docs/interfaces/MdastPlugin.md +9 -9
- package/docs/interfaces/MdastSub.md +3 -3
- package/docs/interfaces/MdastSup.md +3 -3
- package/docs/interfaces/PipelineRunOptions.md +4 -4
- package/docs/interfaces/ReadabilityOptions.md +21 -10
- package/docs/interfaces/SmartExcerptOptions.md +3 -3
- package/docs/type-aliases/PipelineStageName.md +1 -1
- package/docs/variables/DefaultPipelineStage.md +1 -1
- package/docs/variables/astFormat.md +1 -1
- package/docs/variables/htmlFormat.md +1 -1
- package/docs/variables/htmlReadability.md +1 -1
- package/docs/variables/htmlReadabilityPlugin.md +1 -1
- package/docs/variables/htmlReadabilityPlugins.md +1 -1
- package/docs/variables/markdownFormat.md +1 -1
- package/docs/variables/restoreReadabilityMetaPlugin.md +1 -1
- package/package.json +2 -1
package/README.cn.md
CHANGED
|
@@ -195,6 +195,7 @@ const result = await mdast('Hello').to('reverse');
|
|
|
195
195
|
- `fields`: (string[] | object) 控制保留哪些元数据字段或如何重命名它们。
|
|
196
196
|
- 如果是数组:作为白名单(例如 `['title', 'excerpt']`)。
|
|
197
197
|
- 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)。
|
|
198
|
+
- `extraMetadata`: (object) 注入到 Frontmatter 中的额外键值对。这些数据将与 readability 提取的元数据合并。
|
|
198
199
|
- `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`。
|
|
199
200
|
- `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`。
|
|
200
201
|
- `minContentLength`: (number) 保留摘要所需的正文最小长度。默认值: `300`。
|
package/README.md
CHANGED
|
@@ -195,6 +195,7 @@ Each stage can have one "main" plugin. If a plugin is marked with `main: true`,
|
|
|
195
195
|
- `fields`: (string[] | object) Control which metadata fields are kept or how they are renamed.
|
|
196
196
|
- If an array: acts as an allowlist (e.g., `['title', 'excerpt']`).
|
|
197
197
|
- If an object: maps original keys to new names (e.g., `{ title: 'headline' }`). Only keys in the map are kept (Projection).
|
|
198
|
+
- `extraMetadata`: (object) Extra key-value pairs to inject into the frontmatter. These will be merged with the readability metadata.
|
|
198
199
|
- `smartExcerpt`: (boolean | object) Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content. Default: `true`.
|
|
199
200
|
- `threshold`: (number) The ratio of excerpt length to content length (0.0 to 1.0). Default: `0.6`.
|
|
200
201
|
- `minContentLength`: (number) Minimum length of the main content required to keep the excerpt. Default: `300`.
|
package/dist/index.d.mts
CHANGED
|
@@ -1783,6 +1783,11 @@ interface ReadabilityOptions {
|
|
|
1783
1783
|
* - If an object, it maps original field names to new names. Only the keys present in the map are kept (Projection).
|
|
1784
1784
|
*/
|
|
1785
1785
|
fields?: string[] | Record<string, string>;
|
|
1786
|
+
/**
|
|
1787
|
+
* Extra key-value pairs to inject into the frontmatter.
|
|
1788
|
+
* These will be merged with the readability metadata.
|
|
1789
|
+
*/
|
|
1790
|
+
extraMetadata?: Record<string, any>;
|
|
1786
1791
|
}
|
|
1787
1792
|
/**
|
|
1788
1793
|
* A unified/rehype plugin that uses Mozilla's Readability to parse the input HTML.
|
package/dist/index.d.ts
CHANGED
|
@@ -1783,6 +1783,11 @@ interface ReadabilityOptions {
|
|
|
1783
1783
|
* - If an object, it maps original field names to new names. Only the keys present in the map are kept (Projection).
|
|
1784
1784
|
*/
|
|
1785
1785
|
fields?: string[] | Record<string, string>;
|
|
1786
|
+
/**
|
|
1787
|
+
* Extra key-value pairs to inject into the frontmatter.
|
|
1788
|
+
* These will be merged with the readability metadata.
|
|
1789
|
+
*/
|
|
1790
|
+
extraMetadata?: Record<string, any>;
|
|
1786
1791
|
}
|
|
1787
1792
|
/**
|
|
1788
1793
|
* A unified/rehype plugin that uses Mozilla's Readability to parse the input HTML.
|
package/dist/index.js
CHANGED
|
@@ -1 +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 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}
|
|
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 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:()=>ut,checkUrlExists:()=>lt,htmlFormat:()=>I,htmlReadability:()=>it,htmlReadabilityPlugin:()=>nt,htmlReadabilityPlugins:()=>ot,jsonParser:()=>V,markdownFormat:()=>R,mdast:()=>W,restoreReadabilityMetaPlugin:()=>st}),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"),M={error:"danger",warn:"warning",success:"tip",important:"important",caution:"caution",note:"note"},P={plugin:()=>async t=>{(0,j.visit)(t,["containerDirective","leafDirective","textDirective"],t=>{const e=t,r=e.name.toLowerCase();if(e.name=M[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},P,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},P,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("smol-toml"),rt=require("lodash-es"),it=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,rt.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,rt.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(t?.extraMetadata&&(g={...g,...t.extraMetadata}),a&&(a.data=a.data||{},a.data.readability=g),f){const t=[f];d&&y&&!ut(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}},nt={name:"readability",plugin:it,stage:0,main:!0},st={name:"restore-readability-meta",plugin:t=>(e,r)=>{const{frontmatter:i,sourceLink:n,extraMetadata:s}=t||{};if(r.data?.readability||s){if(r.data=r.data||{},r.data.readability={...r.data.readability,...s},e.data=e.data||{},e.data.readability=r.data.readability,i){const t="toml"===i?"toml":"yaml",n=("toml"===t?(0,et.stringify)(r.data.readability).trim():(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;lt(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"},ot=[nt,st];function at(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 lt(t,e){if(!e)return!1;let r=!1;return(0,X.visit)(t,"link",t=>{if(at(t.url,e))return r=!0,!1}),r}function ut(t,e){if(!e)return!1;let r=!1;return(0,X.visit)(t,"element",t=>{if("a"===t.tagName&&t.properties&&at(t.properties.href,e))return r=!0,!1}),r}
|
package/dist/index.mjs
CHANGED
|
@@ -1 +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 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};
|
|
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 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 M={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},N=[{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,M,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:l,stage:400},{plugin:u,stage:400},{plugin:p,options:[["yaml","toml"]],stage:400}];S.forEach(t=>{t.plugin===s?t.order=100:t.order=10});var T={id:"markdown",title:"Markdown (GFM + Directives)",extensions:["md","markdown","mdown","mkdn"],mediaTypes:["text/markdown"],input:N,output:S};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 $={id:"ast",title:"MDAST",input:[{plugin:z,stage:0},g,v,k,M,m],output:[{plugin:E,options:[],stage:400}]};function O(t){return"object"==typeof t&&null!==t&&"string"==typeof t.type}var C=class t{constructor(t){this.queue=[],this._data={},this.input=t}static register(t){this.registry.set(t.id,t)}static getFormat(t){return this.registry.get(t)}data(t,e){return"string"==typeof t?this._data[t]=e:Object.assign(this._data,t),this}getFormat(t){return this.constructor.getFormat(t)}resolveFormat(t){if("string"==typeof t){const e=this.getFormat(t);if(!e)throw new Error(`[MdastPlus] Format '${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 "${t.name}" is disabled. Skipping.`):0!==i&&(r.splice(i,1),r[0]=t)}let o=!0,s=0;for(;o&&s<r.length;){o=!1,s++;for(let t=0;t<r.length;t++){const e=r[t];if(e.after){const i=r.findIndex(t=>t.name===e.after);if(-1!==i&&i>t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}if(e.before){const i=r.findIndex(t=>t.name===e.before);if(-1!==i&&i<t){r.splice(t,1),r.splice(i,0,e),o=!0;break}}}}n.push(...r)}const s=i();Object.keys(this._data).length>0&&s.data(this._data);for(const t of n)s.use(t.plugin,...t.options||[]);return s}};C.registry=new Map;var H=C,I=class extends H{async toMarkdown(){const t=await this.to("markdown");return String(t)}toMarkdownVFile(){return this.to("markdown")}async toHtml(){const t=await this.to("html");return String(t)}toHtmlVFile(){return this.to("html")}async toAst(t){return(await this.to("ast",t)).result}toHTML(){return this.toHtml()}toAST(t){return this.toAst(t)}};function R(t){return new I(t)}I.register(T),I.register(D),I.register($);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{stringify as J}from"smol-toml";import{omitBy as Q,pick as K}from"lodash-es";var W=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=Q(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=K(y,s);else{const t={};for(const[e,r]of Object.entries(s))void 0!==y[e]&&(t[r]=y[e]);y=t}if(e?.extraMetadata&&(y={...y,...e.extraMetadata}),l&&(l.data=l.data||{},l.data.readability=y),m){const t=[m];g&&v&&!et(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}},X={name:"readability",plugin:W,stage:0,main:!0},Y={name:"restore-readability-meta",plugin:t=>(e,r)=>{const{frontmatter:i,sourceLink:n,extraMetadata:o}=t||{};if(r.data?.readability||o){if(r.data=r.data||{},r.data.readability={...r.data.readability,...o},e.data=e.data||{},e.data.readability=r.data.readability,i){const t="toml"===i?"toml":"yaml",n=("toml"===t?J(r.data.readability).trim():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;tt(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"},Z=[X,Y];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(/\/$/,"")===e.replace(/\/$/,"")}}function tt(t,e){if(!e)return!1;let r=!1;return B(t,"link",t=>{if(_(t.url,e))return r=!0,!1}),r}function et(t,e){if(!e)return!1;let r=!1;return B(t,"element",t=>{if("a"===t.tagName&&t.properties&&_(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,et as checkHtmlUrlExists,tt as checkUrlExists,D as htmlFormat,W as htmlReadability,X as htmlReadabilityPlugin,Z as htmlReadabilityPlugins,z as jsonParser,T as markdownFormat,R as mdast,Y as restoreReadabilityMetaPlugin};
|
package/docs/README.md
CHANGED
|
@@ -199,6 +199,7 @@ Each stage can have one "main" plugin. If a plugin is marked with `main: true`,
|
|
|
199
199
|
- `fields`: (string[] | object) Control which metadata fields are kept or how they are renamed.
|
|
200
200
|
- If an array: acts as an allowlist (e.g., `['title', 'excerpt']`).
|
|
201
201
|
- If an object: maps original keys to new names (e.g., `{ title: 'headline' }`). Only keys in the map are kept (Projection).
|
|
202
|
+
- `extraMetadata`: (object) Extra key-value pairs to inject into the frontmatter. These will be merged with the readability metadata.
|
|
202
203
|
- `smartExcerpt`: (boolean | object) Whether to remove the excerpt if it is a duplicate or near-duplicate of the main content. Default: `true`.
|
|
203
204
|
- `threshold`: (number) The ratio of excerpt length to content length (0.0 to 1.0). Default: `0.6`.
|
|
204
205
|
- `minContentLength`: (number) Minimum length of the main content required to keep the excerpt. Default: `300`.
|
package/docs/_media/README.cn.md
CHANGED
|
@@ -195,6 +195,7 @@ const result = await mdast('Hello').to('reverse');
|
|
|
195
195
|
- `fields`: (string[] | object) 控制保留哪些元数据字段或如何重命名它们。
|
|
196
196
|
- 如果是数组:作为白名单(例如 `['title', 'excerpt']`)。
|
|
197
197
|
- 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)。
|
|
198
|
+
- `extraMetadata`: (object) 注入到 Frontmatter 中的额外键值对。这些数据将与 readability 提取的元数据合并。
|
|
198
199
|
- `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`。
|
|
199
200
|
- `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`。
|
|
200
201
|
- `minContentLength`: (number) 保留摘要所需的正文最小长度。默认值: `300`。
|
|
@@ -6,7 +6,7 @@
|
|
|
6
6
|
|
|
7
7
|
# Class: MdastBasePipeline
|
|
8
8
|
|
|
9
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:42](https://github.com/isdk/mdast-plus.js/blob/
|
|
9
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:42](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L42)
|
|
10
10
|
|
|
11
11
|
Base implementation of the fluent mdast processing pipeline.
|
|
12
12
|
Manages the plugin registry and the execution queue.
|
|
@@ -21,7 +21,7 @@ Manages the plugin registry and the execution queue.
|
|
|
21
21
|
|
|
22
22
|
> **new MdastBasePipeline**(`input`): `MdastBasePipeline`
|
|
23
23
|
|
|
24
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:70](https://github.com/isdk/mdast-plus.js/blob/
|
|
24
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:70](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L70)
|
|
25
25
|
|
|
26
26
|
Initializes a new pipeline instance with the given input.
|
|
27
27
|
|
|
@@ -43,7 +43,7 @@ Content to process (string, Buffer, VFile, or AST Node).
|
|
|
43
43
|
|
|
44
44
|
> `protected` **\_data**: `Record`\<`string`, `any`\> = `{}`
|
|
45
45
|
|
|
46
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mdast-plus.js/blob/
|
|
46
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L64)
|
|
47
47
|
|
|
48
48
|
***
|
|
49
49
|
|
|
@@ -51,7 +51,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:64](https://github.com/isdk/mda
|
|
|
51
51
|
|
|
52
52
|
> `protected` **input**: `Compatible`
|
|
53
53
|
|
|
54
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mdast-plus.js/blob/
|
|
54
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L62)
|
|
55
55
|
|
|
56
56
|
***
|
|
57
57
|
|
|
@@ -59,7 +59,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:62](https://github.com/isdk/mda
|
|
|
59
59
|
|
|
60
60
|
> `protected` **queue**: [`MdastPlugin`](../interfaces/MdastPlugin.md)[] = `[]`
|
|
61
61
|
|
|
62
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mdast-plus.js/blob/
|
|
62
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L63)
|
|
63
63
|
|
|
64
64
|
## Methods
|
|
65
65
|
|
|
@@ -67,7 +67,7 @@ Defined in: [packages/mdast-plus/src/pipeline.ts:63](https://github.com/isdk/mda
|
|
|
67
67
|
|
|
68
68
|
> `protected` **assembleProcessor**(`queue`): `Processor`
|
|
69
69
|
|
|
70
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:494](https://github.com/isdk/mdast-plus.js/blob/
|
|
70
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:494](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L494)
|
|
71
71
|
|
|
72
72
|
Assembles a unified processor based on the sorted plugin queue.
|
|
73
73
|
|
|
@@ -87,7 +87,7 @@ Assembles a unified processor based on the sorted plugin queue.
|
|
|
87
87
|
|
|
88
88
|
> **configure**(`pluginName`, ...`options`): `this`
|
|
89
89
|
|
|
90
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:478](https://github.com/isdk/mdast-plus.js/blob/
|
|
90
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:478](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L478)
|
|
91
91
|
|
|
92
92
|
Modifies the options of a plugin that is already in the pipeline queue.
|
|
93
93
|
Searches from the end of the queue and updates the first match found.
|
|
@@ -119,7 +119,7 @@ The pipeline instance for chaining.
|
|
|
119
119
|
|
|
120
120
|
> **data**(`key`, `value?`): `this`
|
|
121
121
|
|
|
122
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:80](https://github.com/isdk/mdast-plus.js/blob/
|
|
122
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:80](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L80)
|
|
123
123
|
|
|
124
124
|
Configures global data for the pipeline, which will be available to all plugins via this.data().
|
|
125
125
|
|
|
@@ -149,7 +149,7 @@ The pipeline instance for chaining.
|
|
|
149
149
|
|
|
150
150
|
> `protected` **ensureInputPlugins**(`queue`, `overrides?`, `maxStage?`): `void`
|
|
151
151
|
|
|
152
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:165](https://github.com/isdk/mdast-plus.js/blob/
|
|
152
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:165](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L165)
|
|
153
153
|
|
|
154
154
|
Ensures that input plugins (parser, normalizers) are present in the queue.
|
|
155
155
|
Adds implicit plugins if no parser is detected.
|
|
@@ -178,7 +178,7 @@ Adds implicit plugins if no parser is detected.
|
|
|
178
178
|
|
|
179
179
|
> **from**(`fmt`, `overrides?`): `this`
|
|
180
180
|
|
|
181
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:194](https://github.com/isdk/mdast-plus.js/blob/
|
|
181
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:194](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L194)
|
|
182
182
|
|
|
183
183
|
Configures the input format and adds its associated plugins to the pipeline.
|
|
184
184
|
|
|
@@ -208,7 +208,7 @@ The pipeline instance for chaining.
|
|
|
208
208
|
|
|
209
209
|
> **getFormat**(`id`): `undefined` \| [`MdastFormat`](../interfaces/MdastFormat.md)
|
|
210
210
|
|
|
211
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:92](https://github.com/isdk/mdast-plus.js/blob/
|
|
211
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:92](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L92)
|
|
212
212
|
|
|
213
213
|
Instance-level access to the global format registry.
|
|
214
214
|
|
|
@@ -228,7 +228,7 @@ Instance-level access to the global format registry.
|
|
|
228
228
|
|
|
229
229
|
> **priority**(`order`): `this`
|
|
230
230
|
|
|
231
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:461](https://github.com/isdk/mdast-plus.js/blob/
|
|
231
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:461](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L461)
|
|
232
232
|
|
|
233
233
|
Sets the priority order for the most recently added plugin.
|
|
234
234
|
Plugins with lower order run earlier within the same stage.
|
|
@@ -253,7 +253,7 @@ The pipeline instance for chaining.
|
|
|
253
253
|
|
|
254
254
|
> `protected` **resolveRunQueue**(`format`, `overrides?`, `stage?`, `stopAtIndex?`): [`MdastPlugin`](../interfaces/MdastPlugin.md)[]
|
|
255
255
|
|
|
256
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:214](https://github.com/isdk/mdast-plus.js/blob/
|
|
256
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:214](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L214)
|
|
257
257
|
|
|
258
258
|
Resolves the final plugin queue for execution based on the target format and run options.
|
|
259
259
|
Calculates the effective plugin list by applying overrides, handling partial execution (stage/stopAtIndex),
|
|
@@ -287,7 +287,7 @@ and injecting necessary input/output plugins.
|
|
|
287
287
|
|
|
288
288
|
> **to**(`fmt`, `optionsOrOverrides?`): `Promise`\<`VFile`\>
|
|
289
289
|
|
|
290
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:290](https://github.com/isdk/mdast-plus.js/blob/
|
|
290
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:290](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L290)
|
|
291
291
|
|
|
292
292
|
Processes the pipeline and serializes the result into the specified format.
|
|
293
293
|
|
|
@@ -317,7 +317,7 @@ A promise resolving to a VFile containing the result.
|
|
|
317
317
|
|
|
318
318
|
> `protected` **toRuntimeEntry**(`entry`, `defaultStage`, `overrides?`): [`MdastPlugin`](../interfaces/MdastPlugin.md) & `object`
|
|
319
319
|
|
|
320
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:116](https://github.com/isdk/mdast-plus.js/blob/
|
|
320
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:116](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L116)
|
|
321
321
|
|
|
322
322
|
Normalizes a plugin entry for runtime execution.
|
|
323
323
|
|
|
@@ -345,7 +345,7 @@ Normalizes a plugin entry for runtime execution.
|
|
|
345
345
|
|
|
346
346
|
> **use**(`plugin`, ...`options`): `this`
|
|
347
347
|
|
|
348
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:336](https://github.com/isdk/mdast-plus.js/blob/
|
|
348
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:336](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L336)
|
|
349
349
|
|
|
350
350
|
Adds a plugin or an array of plugins to the pipeline's compile stage.
|
|
351
351
|
|
|
@@ -377,7 +377,7 @@ The pipeline instance for chaining.
|
|
|
377
377
|
|
|
378
378
|
> **useAt**(`stage`, `plugin`, ...`options`): `this`
|
|
379
379
|
|
|
380
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:418](https://github.com/isdk/mdast-plus.js/blob/
|
|
380
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:418](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L418)
|
|
381
381
|
|
|
382
382
|
Adds a plugin or an array of plugins to the pipeline at a specific stage.
|
|
383
383
|
|
|
@@ -411,7 +411,7 @@ The pipeline instance for chaining.
|
|
|
411
411
|
|
|
412
412
|
> **useAt**(`plugin`, ...`options`): `this`
|
|
413
413
|
|
|
414
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:425](https://github.com/isdk/mdast-plus.js/blob/
|
|
414
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:425](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L425)
|
|
415
415
|
|
|
416
416
|
Adds a plugin or an array of plugins to the pipeline. The stage is taken from the plugin object(s) or defaults to 'compile'.
|
|
417
417
|
|
|
@@ -441,7 +441,7 @@ The pipeline instance for chaining.
|
|
|
441
441
|
|
|
442
442
|
> `static` **getFormat**(`id`): `undefined` \| [`MdastFormat`](../interfaces/MdastFormat.md)
|
|
443
443
|
|
|
444
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:58](https://github.com/isdk/mdast-plus.js/blob/
|
|
444
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:58](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L58)
|
|
445
445
|
|
|
446
446
|
Retrieves a registered format by its ID.
|
|
447
447
|
|
|
@@ -465,7 +465,7 @@ The format definition or undefined if not found.
|
|
|
465
465
|
|
|
466
466
|
> `static` **register**(`format`): `void`
|
|
467
467
|
|
|
468
|
-
Defined in: [packages/mdast-plus/src/pipeline.ts:49](https://github.com/isdk/mdast-plus.js/blob/
|
|
468
|
+
Defined in: [packages/mdast-plus/src/pipeline.ts:49](https://github.com/isdk/mdast-plus.js/blob/6a398ff355b39ed17183d5afad29bd4ca7551ec6/src/pipeline.ts#L49)
|
|
469
469
|
|
|
470
470
|
Registers a global document format.
|
|
471
471
|
|