@aiquants/html-to-markdown 0.4.0 → 0.5.0

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-D2dlEt1l.cjs"),t=require("./mcp-server-Co1kuNGX.cjs"),c=require("./mcp-server-streamable-BNn6Kd9b.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-Dq4LfiYy.cjs"),t=require("./mcp-server-CMBUHOmc.cjs"),c=require("./mcp-server-streamable-IQlimU2H.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

@@ -5,9 +5,9 @@ Object.assign(global, {
   window,
   self: window
 });
-import { g, h } from "./core-CouDTni-.js";
-import { c, r } from "./mcp-server-DzB_5cTs.js";
-import { c as c2, r as r2 } from "./mcp-server-streamable-CcZ7Lsel.js";
+import { g, h } from "./core-DJV8t_Qj.js";
+import { c, r } from "./mcp-server-DD8CqdNw.js";
+import { c as c2, r as r2 } from "./mcp-server-streamable-DODolSbj.js";
 export {
   c as createMcpServer,
   c2 as createStreamableMcpServer,

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-D2dlEt1l.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"],alias:{output:["o"]},default:{locale:"en-US"}}),s=c._[0],n=c.locale,a=c.output;s||process.exit(1);try{new URL(s)}catch(i){process.exit(1)}try{const{markdown:e,html:c}=await t.htmlToMarkdown(s,{locale:n});let i;if(a){i=o.resolve(process.cwd(),a);const e=o.dirname(i);r.existsSync(e)||r.mkdirSync(e,{recursive:!0})}else{const e=`${s.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_")}_${Date.now()}.md`,t=o.join(process.cwd(),".outputs","raw");r.existsSync(t)||r.mkdirSync(t,{recursive:!0}),i=o.join(t,e)}r.writeFileSync(i,e);const p=o.parse(i),l=o.join(p.dir,`${p.name}.html`);r.writeFileSync(l,c)}catch(i){process.exit(1)}})()})();
+Object.create,Object.defineProperty,Object.getOwnPropertyDescriptor,Object.getOwnPropertyNames,Object.getPrototypeOf,Object.prototype.hasOwnProperty;const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./core-Dq4LfiYy.cjs");var r="undefined"!=typeof document?document.currentScript:null;(async()=>{(("undefined"==typeof document?require("url").pathToFileURL(__filename).href:r&&"SCRIPT"===r.tagName.toUpperCase()&&r.src||new URL("main.cjs",document.baseURI).href).endsWith(process.argv[1])||process.argv[1]&&process.argv[1].includes("vite-node"))&&(async()=>{const e=(await import("yargs-parser")).default,r=await import("node:fs"),o=await import("node:path"),c=e(process.argv.slice(2),{string:["locale","output","html-content"],alias:{output:["o"],"html-content":["h"]},default:{locale:"en-US"}}),n=c._[0],s=c.locale,a=c.output,i=c["html-content"];if(n||i||process.exit(1),n&&!i)try{new URL(n)}catch(p){process.exit(1)}try{const{markdown:e,html:c}=await t.htmlToMarkdown(n||"direct-html-input",{locale:s,htmlContent:i});let p;if(a){p=o.resolve(process.cwd(),a);const e=o.dirname(p);r.existsSync(e)||r.mkdirSync(e,{recursive:!0})}else{let e;e=i?"html_content":n.replace(/https?:\/\//,"").replace(/[^a-zA-Z0-9]/g,"_");const t=`${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

@@ -6,15 +6,16 @@ Object.assign(global, {
   window,
   self: window
 });
-import { g as getMessage, h as htmlToMarkdown } from "./core-CouDTni-.js";
+import { g as getMessage, h as htmlToMarkdown } from "./core-DJV8t_Qj.js";
 const runCli = async () => {
   const yargsParser = (await import("yargs-parser")).default;
   const fs = await import("node:fs");
   const path = await import("node:path");
   const argv = yargsParser(process.argv.slice(2), {
-    string: ["locale", "output"],
+    string: ["locale", "output", "html-content"],
     alias: {
-      output: ["o"]
+      output: ["o"],
+      "html-content": ["h"]
     },
     default: {
       locale: "en-US"
@@ -24,27 +25,38 @@ const runCli = async () => {
   const targetUrl = argv._[0];
   const locale = argv.locale;
   const outputFile = argv.output;
-  if (!targetUrl) {
-    console.error(getMessage(locale, "error_url_required"));
+  const htmlContent = argv["html-content"];
+  if (!(targetUrl || htmlContent)) {
+    console.error(getMessage(locale, "error_url_or_html_required"));
     console.info(getMessage(locale, "usage"));
     console.info(getMessage(locale, "example"));
     process.exit(1);
   }
-  try {
-    new URL(targetUrl);
-  } catch (error) {
-    console.error(getMessage(locale, "error_app"), error);
-    process.exit(1);
+  if (targetUrl && !htmlContent) {
+    try {
+      new URL(targetUrl);
+    } catch (error) {
+      console.error(getMessage(locale, "error_app"), error);
+      process.exit(1);
+    }
   }
   try {
     console.info(`
 ${getMessage(locale, "cli_start")}`);
-    console.info(getMessage(locale, "cli_url", { url: targetUrl }));
+    if (targetUrl) {
+      console.info(getMessage(locale, "cli_url", { url: targetUrl }));
+    }
+    if (htmlContent) {
+      console.info(getMessage(locale, "cli_html_content", { length: htmlContent.length.toString() }));
+    }
     console.info(getMessage(locale, "cli_locale", { locale }));
     if (outputFile) {
       console.info(getMessage(locale, "cli_output", { path: outputFile }));
     }
-    const { markdown, html } = await htmlToMarkdown(targetUrl, { locale });
+    const { markdown, html } = await htmlToMarkdown(targetUrl || "direct-html-input", {
+      locale,
+      htmlContent
+    });
     let filePath;
     if (outputFile) {
       filePath = path.resolve(process.cwd(), outputFile);
@@ -53,7 +65,12 @@ ${getMessage(locale, "cli_start")}`);
         fs.mkdirSync(outputDir, { recursive: true });
       }
     } else {
-      const safeFileName = targetUrl.replace(/https?:\/\//, "").replace(/[^a-zA-Z0-9]/g, "_");
+      let safeFileName;
+      if (htmlContent) {
+        safeFileName = "html_content";
+      } else {
+        safeFileName = targetUrl.replace(/https?:\/\//, "").replace(/[^a-zA-Z0-9]/g, "_");
+      }
       const timestamp = Date.now();
       const fileName = `${safeFileName}_${timestamp}.md`;
       const outputDir = path.join(process.cwd(), ".outputs", "raw");

package/dist/{mcp-server-Co1kuNGX.cjs → mcp-server-CMBUHOmc.cjs}

@@ -1 +1 @@
-const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./version-dpS3SSul.cjs"),r=require("node:process"),a=require("./core-D2dlEt1l.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,description:"Convert web pages and HTML to Markdown with JavaScript dynamic content support and table structure preservation. Includes separate file saving tool."},{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(/^\.+$/)||(t="page"),t}catch{return"page"}}async start(){const e=new s;await this.server.connect(e)}}const i=()=>new o;exports.createMcpServer=i,exports.runMcpServer=async()=>{const e=i();await e.start()};
+const e=new(require("happy-dom").Window);Object.assign(global,{document:e.document,window:e,self:e});const t=require("./version-BkgGdAYt.cjs"),r=require("node:process"),a=require("./core-Dq4LfiYy.cjs");class n{append(e){this._buffer=this._buffer?Buffer.concat([this._buffer,e]):e}readMessage(){if(!this._buffer)return null;const e=this._buffer.indexOf("\n");if(-1===e)return null;const r=this._buffer.toString("utf8",0,e).replace(/\r$/,"");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,description:"Convert web pages and HTML to Markdown with JavaScript dynamic content support and table structure preservation. Includes separate file saving tool."},{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-DzB_5cTs.js → mcp-server-DD8CqdNw.js}

@@ -5,9 +5,9 @@ Object.assign(global, {
   window,
   self: window
 });
-import { J as JSONRPCMessageSchema, g as getPackageVersion, S as Server, L as ListToolsRequestSchema, M as MCP_TOOL_SCHEMAS, C as CallToolRequestSchema, p as parseHtmlToMarkdownArgs, c as createSuccessResult, a as createErrorResult, b as parseSaveContentArgs, s as saveContentToFile, d as parseUrlToMarkdownFileArgs } from "./version-BSh9aXhF.js";
+import { J as JSONRPCMessageSchema, g as getPackageVersion, S as Server, L as ListToolsRequestSchema, M as MCP_TOOL_SCHEMAS, C as CallToolRequestSchema, p as parseHtmlToMarkdownArgs, c as createSuccessResult, a as createErrorResult, b as parseSaveContentArgs, s as saveContentToFile, d as parseUrlToMarkdownFileArgs } from "./version-CTdrYd1G.js";
 import process from "node:process";
-import { h as htmlToMarkdown } from "./core-CouDTni-.js";
+import { h as htmlToMarkdown } from "./core-DJV8t_Qj.js";
 class ReadBuffer {
   append(chunk) {
     this._buffer = this._buffer ? Buffer.concat([this._buffer, chunk]) : chunk;

package/dist/{mcp-server-streamable-CcZ7Lsel.js → mcp-server-streamable-DODolSbj.js}

@@ -1,5 +1,5 @@
-import { J as JSONRPCMessageSchema, i as isInitializeRequest, e as isJSONRPCRequest, D as DEFAULT_NEGOTIATED_PROTOCOL_VERSION, f as SUPPORTED_PROTOCOL_VERSIONS, h as isJSONRPCResponse, j as isJSONRPCError, g as getPackageVersion, S as Server, I as InitializeRequestSchema, L as ListToolsRequestSchema, C as CallToolRequestSchema, p as parseHtmlToMarkdownArgs, b as parseSaveContentArgs, s as saveContentToFile, d as parseUrlToMarkdownFileArgs } from "./version-BSh9aXhF.js";
-import { a as getAugmentedNamespace, c as commonjsGlobal, b as getDefaultExportFromCjs, h as htmlToMarkdown } from "./core-CouDTni-.js";
+import { J as JSONRPCMessageSchema, i as isInitializeRequest, e as isJSONRPCRequest, D as DEFAULT_NEGOTIATED_PROTOCOL_VERSION, f as SUPPORTED_PROTOCOL_VERSIONS, h as isJSONRPCResponse, j as isJSONRPCError, g as getPackageVersion, S as Server, I as InitializeRequestSchema, L as ListToolsRequestSchema, C as CallToolRequestSchema, p as parseHtmlToMarkdownArgs, b as parseSaveContentArgs, s as saveContentToFile, d as parseUrlToMarkdownFileArgs } from "./version-CTdrYd1G.js";
+import { a as getAugmentedNamespace, c as commonjsGlobal, b as getDefaultExportFromCjs, h as htmlToMarkdown } from "./core-DJV8t_Qj.js";
 import { randomUUID } from "node:crypto";
 import cors from "cors";
 import express from "express";
@@ -7551,7 +7551,7 @@ class StreamableHtmlToMarkdownMcpServer {
       try {
         res.setHeader("Content-Type", "application/json");
         const request = req.body;
-        if (!request || !request.method) {
+        if (!(request && request.method)) {
           res.status(400).json({
             jsonrpc: "2.0",
             error: {