@aiquants/html-to-markdown 1.0.0 → 1.2.0

package/dist/index.cjs

@@ -1 +1 @@
-const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e}),Object.defineProperty(exports,Symbol.toStringTag,{value:"Module"});const r=require("./core--XIBQvpW.cjs"),t=require("./mcp-server-eznfa4gh.cjs"),c=require("./mcp-server-streamable-DJ-R_Z_E.cjs");exports.getMessage=r.getMessage,exports.htmlToMarkdown=r.htmlToMarkdown,exports.createMcpServer=t.createMcpServer,exports.runMcpServer=t.runMcpServer,exports.createStreamableMcpServer=c.createStreamableMcpServer,exports.runStreamableMcpServer=c.runStreamableMcpServer;
+const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e}),Object.defineProperty(exports,Symbol.toStringTag,{value:"Module"});const r=require("./core-diGadcw4.cjs"),t=require("./mcp-server-DSLDlYZX.cjs"),c=require("./mcp-server-streamable-DZ7VyWAg.cjs");exports.getMessage=r.getMessage,exports.htmlToMarkdown=r.htmlToMarkdown,exports.createMcpServer=t.createMcpServer,exports.runMcpServer=t.runMcpServer,exports.createStreamableMcpServer=c.createStreamableMcpServer,exports.runStreamableMcpServer=c.runStreamableMcpServer;

package/dist/index.js

@@ -1 +1 @@
-import{Window as e}from"happy-dom";const r=new e;Object.assign(global,{document:r.document,window:r,self:r});import{g as a,h as s}from"./core-ChRwTJHy.js";import{c as o,r as m}from"./mcp-server-CvLGX0Fe.js";import{c as t,r as c}from"./mcp-server-streamable-CjY9xfH0.js";export{o as createMcpServer,t as createStreamableMcpServer,a as getMessage,s as htmlToMarkdown,m as runMcpServer,c as runStreamableMcpServer};
+import{Window as e}from"happy-dom";const r=new e;Object.assign(global,{document:r.document,window:r,self:r});import{g as a,h as s}from"./core-CTBCnkuY.js";import{c as o,r as m}from"./mcp-server-DmJXoVRZ.js";import{c as t,r as c}from"./mcp-server-streamable-CgJP5t9P.js";export{o as createMcpServer,t as createStreamableMcpServer,a as getMessage,s as htmlToMarkdown,m as runMcpServer,c as runStreamableMcpServer};

package/dist/main.cjs

@@ -1,2 +1,2 @@
 #!/usr/bin/env node
-Object.create,Object.defineProperty,Object.getOwnPropertyDescriptor,Object.getOwnPropertyNames,Object.getPrototypeOf,Object.prototype.hasOwnProperty;const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./core--XIBQvpW.cjs");var r="undefined"!=typeof document?document.currentScript:null;(async()=>{(("undefined"==typeof document?require("url").pathToFileURL(__filename).href:r&&"SCRIPT"===r.tagName.toUpperCase()&&r.src||new URL("main.cjs",document.baseURI).href).endsWith(process.argv[1])||process.argv[1]&&process.argv[1].includes("vite-node"))&&(async()=>{const e=(await import("yargs-parser")).default,r=await import("node:fs"),o=await import("node:path"),c=e(process.argv.slice(2),{string:["locale","output","html-content"],alias:{output:["o"],"html-content":["h"]},default:{locale:"en-US"}}),n=c._[0],s=c.locale,a=c.output,i=c["html-content"];if(n||i||process.exit(1),n&&!i)try{new URL(n)}catch(p){process.exit(1)}try{const{markdown:e,html:c}=await t.htmlToMarkdown(n||"direct-html-input",{locale:s,htmlContent:i});let p;if(a){p=o.resolve(process.cwd(),a);const e=o.dirname(p);r.existsSync(e)||r.mkdirSync(e,{recursive:!0})}else{let e;e=i?"html_content":n.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_");const t=`${e}_${Date.now()}.md`,c=o.join(process.cwd(),".outputs","raw");r.existsSync(c)||r.mkdirSync(c,{recursive:!0}),p=o.join(c,t)}r.writeFileSync(p,e);const l=o.parse(p),d=o.join(l.dir,`${l.name}.html`);r.writeFileSync(d,c)}catch(p){process.exit(1)}})()})();
+Object.create,Object.defineProperty,Object.getOwnPropertyDescriptor,Object.getOwnPropertyNames,Object.getPrototypeOf,Object.prototype.hasOwnProperty;const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./core-diGadcw4.cjs");var r="undefined"!=typeof document?document.currentScript:null;(async()=>{(("undefined"==typeof document?require("url").pathToFileURL(__filename).href:r&&"SCRIPT"===r.tagName.toUpperCase()&&r.src||new URL("main.cjs",document.baseURI).href).endsWith(process.argv[1])||process.argv[1]&&process.argv[1].includes("vite-node"))&&(async()=>{const e=(await import("yargs-parser")).default,r=await import("node:fs"),o=await import("node:path"),c=e(process.argv.slice(2),{string:["locale","output","html-content"],alias:{output:["o"],"html-content":["h"]},default:{locale:"en-US"}}),n=c._[0],s=c.locale,a=c.output,i=c["html-content"];if(n||i||process.exit(1),n&&!i)try{new URL(n)}catch(p){process.exit(1)}try{const{markdown:e,html:c}=await t.htmlToMarkdown(n||"direct-html-input",{locale:s,htmlContent:i});let p;if(a){p=o.resolve(process.cwd(),a);const e=o.dirname(p);r.existsSync(e)||r.mkdirSync(e,{recursive:!0})}else{let e;e=i?"html_content":n.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_");const t=`${e}_${Date.now()}.md`,c=o.join(process.cwd(),".outputs","raw");r.existsSync(c)||r.mkdirSync(c,{recursive:!0}),p=o.join(c,t)}r.writeFileSync(p,e);const l=o.parse(p),d=o.join(l.dir,`${l.name}.html`);r.writeFileSync(d,c)}catch(p){process.exit(1)}})()})();

package/dist/main.js

@@ -1,2 +1,2 @@
 #!/usr/bin/env node
-import{Window as t}from"happy-dom";const e=new t;Object.assign(global,{document:e.document,window:e,self:e});import{g as o,h as s}from"./core-ChRwTJHy.js";(async()=>{(import.meta.url.endsWith(process.argv[1])||process.argv[1]&&process.argv[1].includes("vite-node"))&&(async()=>{const t=(await import("yargs-parser")).default,e=await import("node:fs"),o=await import("node:path"),c=t(process.argv.slice(2),{string:["locale","output","html-content"],alias:{output:["o"],"html-content":["h"]},default:{locale:"en-US"}}),r=c._[0],n=c.locale,a=c.output,i=c["html-content"];if(r||i||process.exit(1),r&&!i)try{new URL(r)}catch(l){process.exit(1)}try{const{markdown:t,html:c}=await s(r||"direct-html-input",{locale:n,htmlContent:i});let l;if(a){l=o.resolve(process.cwd(),a);const t=o.dirname(l);e.existsSync(t)||e.mkdirSync(t,{recursive:!0})}else{let t;t=i?"html_content":r.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_");const s=`${t}_${Date.now()}.md`,c=o.join(process.cwd(),".outputs","raw");e.existsSync(c)||e.mkdirSync(c,{recursive:!0}),l=o.join(c,s)}e.writeFileSync(l,t);const p=o.parse(l),m=o.join(p.dir,`${p.name}.html`);e.writeFileSync(m,c)}catch(l){process.exit(1)}})()})();
+import{Window as t}from"happy-dom";const e=new t;Object.assign(global,{document:e.document,window:e,self:e});import{g as o,h as s}from"./core-CTBCnkuY.js";(async()=>{(import.meta.url.endsWith(process.argv[1])||process.argv[1]&&process.argv[1].includes("vite-node"))&&(async()=>{const t=(await import("yargs-parser")).default,e=await import("node:fs"),o=await import("node:path"),c=t(process.argv.slice(2),{string:["locale","output","html-content"],alias:{output:["o"],"html-content":["h"]},default:{locale:"en-US"}}),r=c._[0],n=c.locale,a=c.output,i=c["html-content"];if(r||i||process.exit(1),r&&!i)try{new URL(r)}catch(l){process.exit(1)}try{const{markdown:t,html:c}=await s(r||"direct-html-input",{locale:n,htmlContent:i});let l;if(a){l=o.resolve(process.cwd(),a);const t=o.dirname(l);e.existsSync(t)||e.mkdirSync(t,{recursive:!0})}else{let t;t=i?"html_content":r.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_");const s=`${t}_${Date.now()}.md`,c=o.join(process.cwd(),".outputs","raw");e.existsSync(c)||e.mkdirSync(c,{recursive:!0}),l=o.join(c,s)}e.writeFileSync(l,t);const p=o.parse(l),m=o.join(p.dir,`${p.name}.html`);e.writeFileSync(m,c)}catch(l){process.exit(1)}})()})();

package/dist/mcp-server-DSLDlYZX.cjs

@@ -0,0 +1 @@
+const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./version-DAQkOvqo.cjs"),r=require("node:process"),a=require("./core-diGadcw4.cjs");class n{append(e){this._buffer=this._buffer?Buffer.concat([this._buffer,e]):e}readMessage(){if(!this._buffer)return null;const e=this._buffer.indexOf("\n");if(-1===e)return null;const r=this._buffer.toString("utf8",0,e).replace(/\r$/,"");return this._buffer=this._buffer.subarray(e+1),function(e){return t.JSONRPCMessageSchema.parse(JSON.parse(e))}(r)}clear(){this._buffer=void 0}}class s{constructor(e=r.stdin,t=r.stdout){this._stdin=e,this._stdout=t,this._readBuffer=new n,this._started=!1,this._ondata=e=>{this._readBuffer.append(e),this.processReadBuffer()},this._onerror=e=>{var t;null===(t=this.onerror)||void 0===t||t.call(this,e)}}async start(){if(this._started)throw new Error("StdioServerTransport already started! If using Server class, note that connect() calls start() automatically.");this._started=!0,this._stdin.on("data",this._ondata),this._stdin.on("error",this._onerror)}processReadBuffer(){for(var e,t;;)try{const t=this._readBuffer.readMessage();if(null===t)break;null===(e=this.onmessage)||void 0===e||e.call(this,t)}catch(r){null===(t=this.onerror)||void 0===t||t.call(this,r)}}async close(){var e;this._stdin.off("data",this._ondata),this._stdin.off("error",this._onerror);0===this._stdin.listenerCount("data")&&this._stdin.pause(),this._readBuffer.clear(),null===(e=this.onclose)||void 0===e||e.call(this)}send(e){return new Promise(t=>{const r=function(e){return JSON.stringify(e)+"\n"}(e);this._stdout.write(r)?t():this._stdout.once("drain",t)})}}class o{constructor(){this.version=t.getPackageVersion(),this.server=new t.Server({name:"@aiquants/html-to-markdown",version:this.version},{capabilities:{tools:{}}}),this.setupToolHandlers()}setupToolHandlers(){this.server.setRequestHandler(t.ListToolsRequestSchema,async()=>({tools:[{name:"html_to_markdown",description:"Convert web pages or HTML strings to Markdown format. Handles JavaScript-rendered dynamic content and preserves table structures.",inputSchema:t.MCP_TOOL_SCHEMAS.htmlToMarkdown},{name:"save_content_to_file",description:"Save text content to a file with specified path or directory. Supports automatic filename generation and format detection.",inputSchema:t.MCP_TOOL_SCHEMAS.saveContentToFile},{name:"url_to_markdown_file",description:"Convert web pages or HTML strings directly to Markdown files. Combines HTML-to-Markdown conversion and file saving in one step. Handles JavaScript-rendered dynamic content.",inputSchema:t.MCP_TOOL_SCHEMAS.urlToMarkdownFile}]})),this.server.setRequestHandler(t.CallToolRequestSchema,async e=>{switch(e.params.name){case"html_to_markdown":return await this.handleHtmlToMarkdown(e.params.arguments);case"save_content_to_file":return await this.handleSaveContentToFile(e.params.arguments);case"url_to_markdown_file":return await this.handleUrlToMarkdownFile(e.params.arguments);default:throw new Error(`Unknown tool: ${e.params.name}`)}})}async handleHtmlToMarkdown(e){try{const r=t.parseHtmlToMarkdownArgs(e),{url:n,html_content:s,locale:o="en-US"}=r;if(n)try{new URL(n)}catch{throw new Error(`Invalid URL: ${n}`)}const i=(await a.htmlToMarkdown(n||"direct-html-input",{locale:o,htmlContent:s})).markdown;return t.createSuccessResult(i)}catch(r){return t.createErrorResult(r)}}async handleSaveContentToFile(e){try{const r=t.parseSaveContentArgs(e),{content:a,save_path:n,save_directory:s,filename:o}=r,i=`File saved successfully to: ${await t.saveContentToFile(a,n,s,o)}`;return t.createSuccessResult(i)}catch(r){return t.createErrorResult(r)}}async handleUrlToMarkdownFile(e){try{const r=t.parseUrlToMarkdownFileArgs(e),{url:n,html_content:s,locale:o="en-US",save_path:i,save_directory:c,filename:l}=r;if(n)try{new URL(n)}catch{throw new Error(`Invalid URL: ${n}`)}const d=(await a.htmlToMarkdown(n||"direct-html-input",{locale:o,htmlContent:s})).markdown;let u=l;!u&&n&&(u=this.generateFilenameFromUrl(n));const h=`URL successfully converted to Markdown and saved to: ${await t.saveContentToFile(d,i,c,u)}`;return t.createSuccessResult(h)}catch(r){return t.createErrorResult(r)}}generateFilenameFromUrl(e){try{let t=new URL(e).pathname.split("/").pop()||"page";return t=t.replace(/\.[^.]*$/,""),t=t.replace(/[<>:"/\\|?*]/g,"_"),t&&!t.match(/^\.+$/)||(t="page"),t}catch{return"page"}}async start(){const e=new s;await this.server.connect(e)}}const i=()=>new o;exports.createMcpServer=i,exports.runMcpServer=async()=>{const e=i();await e.start()};

package/dist/mcp-server-DmJXoVRZ.js

@@ -0,0 +1 @@
+import{Window as t}from"happy-dom";const e=new t;Object.assign(global,{document:e.document,window:e,self:e});import{J as r,g as a,S as n,L as s,M as o,C as i,p as c,c as l,a as d,b as h,s as u,d as f}from"./version-yDYKn_vW.js";import m from"node:process";import{h as p}from"./core-CTBCnkuY.js";class _{append(t){this._buffer=this._buffer?Buffer.concat([this._buffer,t]):t}readMessage(){if(!this._buffer)return null;const t=this._buffer.indexOf("\n");if(-1===t)return null;const e=this._buffer.toString("utf8",0,t).replace(/\r$/,"");return this._buffer=this._buffer.subarray(t+1),function(t){return r.parse(JSON.parse(t))}(e)}clear(){this._buffer=void 0}}class w{constructor(t=m.stdin,e=m.stdout){this._stdin=t,this._stdout=e,this._readBuffer=new _,this._started=!1,this._ondata=t=>{this._readBuffer.append(t),this.processReadBuffer()},this._onerror=t=>{var e;null===(e=this.onerror)||void 0===e||e.call(this,t)}}async start(){if(this._started)throw new Error("StdioServerTransport already started! If using Server class, note that connect() calls start() automatically.");this._started=!0,this._stdin.on("data",this._ondata),this._stdin.on("error",this._onerror)}processReadBuffer(){for(var t,e;;)try{const e=this._readBuffer.readMessage();if(null===e)break;null===(t=this.onmessage)||void 0===t||t.call(this,e)}catch(r){null===(e=this.onerror)||void 0===e||e.call(this,r)}}async close(){var t;this._stdin.off("data",this._ondata),this._stdin.off("error",this._onerror);0===this._stdin.listenerCount("data")&&this._stdin.pause(),this._readBuffer.clear(),null===(t=this.onclose)||void 0===t||t.call(this)}send(t){return new Promise(e=>{const r=function(t){return JSON.stringify(t)+"\n"}(t);this._stdout.write(r)?e():this._stdout.once("drain",e)})}}class v{constructor(){this.version=a(),this.server=new n({name:"@aiquants/html-to-markdown",version:this.version},{capabilities:{tools:{}}}),this.setupToolHandlers()}setupToolHandlers(){this.server.setRequestHandler(s,async()=>({tools:[{name:"html_to_markdown",description:"Convert web pages or HTML strings to Markdown format. Handles JavaScript-rendered dynamic content and preserves table structures.",inputSchema:o.htmlToMarkdown},{name:"save_content_to_file",description:"Save text content to a file with specified path or directory. Supports automatic filename generation and format detection.",inputSchema:o.saveContentToFile},{name:"url_to_markdown_file",description:"Convert web pages or HTML strings directly to Markdown files. Combines HTML-to-Markdown conversion and file saving in one step. Handles JavaScript-rendered dynamic content.",inputSchema:o.urlToMarkdownFile}]})),this.server.setRequestHandler(i,async t=>{switch(t.params.name){case"html_to_markdown":return await this.handleHtmlToMarkdown(t.params.arguments);case"save_content_to_file":return await this.handleSaveContentToFile(t.params.arguments);case"url_to_markdown_file":return await this.handleUrlToMarkdownFile(t.params.arguments);default:throw new Error(`Unknown tool: ${t.params.name}`)}})}async handleHtmlToMarkdown(t){try{const e=c(t),{url:r,html_content:a,locale:n="en-US"}=e;if(r)try{new URL(r)}catch{throw new Error(`Invalid URL: ${r}`)}const s=(await p(r||"direct-html-input",{locale:n,htmlContent:a})).markdown;return l(s)}catch(e){return d(e)}}async handleSaveContentToFile(t){try{const e=h(t),{content:r,save_path:a,save_directory:n,filename:s}=e,o=await u(r,a,n,s);return l(`File saved successfully to: ${o}`)}catch(e){return d(e)}}async handleUrlToMarkdownFile(t){try{const e=f(t),{url:r,html_content:a,locale:n="en-US",save_path:s,save_directory:o,filename:i}=e;if(r)try{new URL(r)}catch{throw new Error(`Invalid URL: ${r}`)}const c=(await p(r||"direct-html-input",{locale:n,htmlContent:a})).markdown;let d=i;!d&&r&&(d=this.generateFilenameFromUrl(r));const h=await u(c,s,o,d);return l(`URL successfully converted to Markdown and saved to: ${h}`)}catch(e){return d(e)}}generateFilenameFromUrl(t){try{let e=new URL(t).pathname.split("/").pop()||"page";return e=e.replace(/\.[^.]*$/,""),e=e.replace(/[<>:"/\\|?*]/g,"_"),e&&!e.match(/^\.+$/)||(e="page"),e}catch{return"page"}}async start(){const t=new w;await this.server.connect(t)}}const y=()=>new v,g=async()=>{const t=y();await t.start()};export{y as c,g as r};