mdream 0.2.3 → 0.2.4

package/README.md

@@ -49,8 +49,6 @@ Other LLM specific convertors focus on supporting _all_ document formats, result
 Mdream is an ultra-performant HTML to Markdown converter built specifically for producing high-quality Markdown for LLMs as quickly as possible. It provides
 a powerful plugin system to customize the conversion process, allowing you to extract, transform, and filter content as needed.
 
-Perfect for: RAG systems, web scraping, content extraction, ChatGPT/Claude integration, and large-scale document processing.
-
 ## CLI Usage
 
 The Mdream CLI is designed to work exclusively with Unix pipes, providing flexibility and freedom to integrate with other tools.

package/dist/cli.mjs

@@ -2,7 +2,7 @@ import { Readable } from 'node:stream';
 import { cac } from 'cac';
 import { f as frontmatterPlugin } from './shared/mdream.C6Z2rfeq.mjs';
 import { r as readabilityPlugin } from './shared/mdream.DMUbnRbh.mjs';
-import { s as streamHtmlToMarkdown } from './shared/mdream.CRBi8vE8.mjs';
+import { s as streamHtmlToMarkdown } from './shared/mdream.CsDVbUMp.mjs';
 import './shared/mdream.Ch6B8TEB.mjs';
 
 async function streamingConvert(options = {}) {

package/dist/index.d.mts

@@ -1,5 +1,5 @@
-import { H as HTMLToMarkdownOptions } from './shared/mdream.C9ruFMrk.mjs';
-export { B as BufferRegion, E as ELEMENT_NODE, b as ElementNode, f as HandlerContext, M as MdreamProcessingState, d as MdreamRuntimeState, N as Node, e as NodeEvent, P as Plugin, a as PluginCreationOptions, T as TEXT_NODE, g as TagHandler, c as TextNode } from './shared/mdream.C9ruFMrk.mjs';
+import { H as HTMLToMarkdownOptions } from './shared/mdream.-SGj02be.mjs';
+export { B as BufferRegion, E as ELEMENT_NODE, b as ElementNode, f as HandlerContext, M as MdreamProcessingState, d as MdreamRuntimeState, N as Node, e as NodeEvent, P as Plugin, a as PluginCreationOptions, T as TEXT_NODE, g as TagHandler, c as TextNode } from './shared/mdream.-SGj02be.mjs';
 import { ReadableStream } from 'node:stream/web';
 
 /**

package/dist/index.d.ts

@@ -1,5 +1,5 @@
-import { H as HTMLToMarkdownOptions } from './shared/mdream.C9ruFMrk.js';
-export { B as BufferRegion, E as ELEMENT_NODE, b as ElementNode, f as HandlerContext, M as MdreamProcessingState, d as MdreamRuntimeState, N as Node, e as NodeEvent, P as Plugin, a as PluginCreationOptions, T as TEXT_NODE, g as TagHandler, c as TextNode } from './shared/mdream.C9ruFMrk.js';
+import { H as HTMLToMarkdownOptions } from './shared/mdream.-SGj02be.js';
+export { B as BufferRegion, E as ELEMENT_NODE, b as ElementNode, f as HandlerContext, M as MdreamProcessingState, d as MdreamRuntimeState, N as Node, e as NodeEvent, P as Plugin, a as PluginCreationOptions, T as TEXT_NODE, g as TagHandler, c as TextNode } from './shared/mdream.-SGj02be.js';
 import { ReadableStream } from 'node:stream/web';
 
 /**

package/dist/index.mjs

@@ -1,5 +1,5 @@
-import { p as processPartialHTMLToMarkdown } from './shared/mdream.CRBi8vE8.mjs';
-export { s as streamHtmlToMarkdown } from './shared/mdream.CRBi8vE8.mjs';
+import { p as processPartialHTMLToMarkdown } from './shared/mdream.CsDVbUMp.mjs';
+export { s as streamHtmlToMarkdown } from './shared/mdream.CsDVbUMp.mjs';
 import './shared/mdream.Ch6B8TEB.mjs';
 
 function htmlToMarkdown(html, options = {}) {

package/dist/plugins.d.mts

@@ -1,4 +1,4 @@
-import { P as Plugin, b as ElementNode } from './shared/mdream.C9ruFMrk.mjs';
+import { P as Plugin, b as ElementNode } from './shared/mdream.-SGj02be.mjs';
 
 /**
  * Create a plugin that implements the Plugin interface with improved type inference

package/dist/plugins.d.ts

@@ -1,4 +1,4 @@
-import { P as Plugin, b as ElementNode } from './shared/mdream.C9ruFMrk.js';
+import { P as Plugin, b as ElementNode } from './shared/mdream.-SGj02be.js';
 
 /**
  * Create a plugin that implements the Plugin interface with improved type inference

package/dist/preset/minimal.d.mts

@@ -1,4 +1,4 @@
-import { H as HTMLToMarkdownOptions } from '../shared/mdream.C9ruFMrk.mjs';
+import { H as HTMLToMarkdownOptions } from '../shared/mdream.-SGj02be.mjs';
 
 /**
  * Creates a configurable minimal preset with advanced options

package/dist/preset/minimal.d.ts

@@ -1,4 +1,4 @@
-import { H as HTMLToMarkdownOptions } from '../shared/mdream.C9ruFMrk.js';
+import { H as HTMLToMarkdownOptions } from '../shared/mdream.-SGj02be.js';
 
 /**
  * Creates a configurable minimal preset with advanced options

package/dist/shared/{mdream.C9ruFMrk.d.mts → mdream.-SGj02be.d.mts}

@@ -147,6 +147,12 @@ interface MdreamProcessingState {
     isFirstTextInElement?: boolean;
     /** Reference to the last processed text node - for context tracking */
     lastTextNode?: Node;
+    /** Quote state tracking for non-nesting tags - avoids backward scanning */
+    inSingleQuote?: boolean;
+    inDoubleQuote?: boolean;
+    inBacktick?: boolean;
+    /** Backslash escaping state tracking - avoids checking previous character */
+    lastCharWasBackslash?: boolean;
     /** Plugin instances array for efficient iteration */
     plugins?: Plugin[];
     /** Configuration options for conversion */

package/dist/shared/{mdream.C9ruFMrk.d.ts → mdream.-SGj02be.d.ts}

@@ -147,6 +147,12 @@ interface MdreamProcessingState {
     isFirstTextInElement?: boolean;
     /** Reference to the last processed text node - for context tracking */
     lastTextNode?: Node;
+    /** Quote state tracking for non-nesting tags - avoids backward scanning */
+    inSingleQuote?: boolean;
+    inDoubleQuote?: boolean;
+    inBacktick?: boolean;
+    /** Backslash escaping state tracking - avoids checking previous character */
+    lastCharWasBackslash?: boolean;
     /** Plugin instances array for efficient iteration */
     plugins?: Plugin[];
     /** Configuration options for conversion */

package/dist/shared/{mdream.CRBi8vE8.mjs → mdream.CsDVbUMp.mjs}

@@ -872,6 +872,10 @@ const SPACE_CHAR = 32;
 const TAB_CHAR = 9;
 const NEWLINE_CHAR = 10;
 const CARRIAGE_RETURN_CHAR = 13;
+const BACKTICK_CHAR = 96;
+const PIPE_CHAR = 124;
+const OPEN_BRACKET_CHAR = 91;
+const CLOSE_BRACKET_CHAR = 93;
 const EMPTY_ATTRIBUTES = Object.freeze({});
 function copyDepthMap(depthMap) {
   return new Uint8Array(depthMap);
@@ -895,6 +899,7 @@ function parseHTML(htmlChunk, state, handleEvent) {
   state.lastCharWasWhitespace ??= true;
   state.justClosedTag ??= false;
   state.isFirstTextInElement ??= false;
+  state.lastCharWasBackslash ??= false;
   let i = 0;
   const chunkLength = htmlChunk.length;
   while (i < chunkLength) {
@@ -922,23 +927,36 @@ function parseHTML(htmlChunk, state, handleEvent) {
         }
         state.lastCharWasWhitespace = true;
         state.textBufferContainsWhitespace = true;
+        state.lastCharWasBackslash = false;
       } else {
         state.textBufferContainsNonWhitespace = true;
         state.lastCharWasWhitespace = false;
         state.justClosedTag = false;
-        if (currentCharCode === 124 && state.depthMap[TAG_TABLE]) {
+        if (currentCharCode === PIPE_CHAR && state.depthMap[TAG_TABLE]) {
           textBuffer += "\\|";
-        } else if (currentCharCode === 96 && (state.depthMap[TAG_CODE] || state.depthMap[TAG_PRE])) {
+        } else if (currentCharCode === BACKTICK_CHAR && (state.depthMap[TAG_CODE] || state.depthMap[TAG_PRE])) {
           textBuffer += "\\`";
-        } else if (currentCharCode === 91 && state.depthMap[TAG_A]) {
+        } else if (currentCharCode === OPEN_BRACKET_CHAR && state.depthMap[TAG_A]) {
           textBuffer += "\\[";
-        } else if (currentCharCode === 93 && state.depthMap[TAG_A]) {
+        } else if (currentCharCode === CLOSE_BRACKET_CHAR && state.depthMap[TAG_A]) {
           textBuffer += "\\]";
-        } else if (currentCharCode === 62 && state.depthMap[TAG_BLOCKQUOTE]) {
+        } else if (currentCharCode === GT_CHAR && state.depthMap[TAG_BLOCKQUOTE]) {
           textBuffer += "\\>";
         } else {
           textBuffer += htmlChunk[i];
         }
+        if (state.currentNode?.tagHandler?.isNonNesting) {
+          if (!state.lastCharWasBackslash) {
+            if (currentCharCode === APOS_CHAR && !state.inDoubleQuote && !state.inBacktick) {
+              state.inSingleQuote = !state.inSingleQuote;
+            } else if (currentCharCode === QUOTE_CHAR && !state.inSingleQuote && !state.inBacktick) {
+              state.inDoubleQuote = !state.inDoubleQuote;
+            } else if (currentCharCode === BACKTICK_CHAR && !state.inSingleQuote && !state.inDoubleQuote) {
+              state.inBacktick = !state.inBacktick;
+            }
+          }
+        }
+        state.lastCharWasBackslash = currentCharCode === BACKSLASH_CHAR;
       }
       i++;
       continue;
@@ -961,6 +979,12 @@ function parseHTML(htmlChunk, state, handleEvent) {
         break;
       }
     } else if (nextCharCode === SLASH_CHAR) {
+      const inQuotes = state.inSingleQuote || state.inDoubleQuote || state.inBacktick;
+      if (state.currentNode?.tagHandler?.isNonNesting && inQuotes) {
+        textBuffer += htmlChunk[i];
+        i++;
+        continue;
+      }
       if (textBuffer.length > 0) {
         processTextBuffer(textBuffer, state, handleEvent);
         textBuffer = "";
@@ -1043,7 +1067,7 @@ function processTextBuffer(textBuffer, state, handleEvent) {
   const firstBlockParent = parentsToIncrement[parentsToIncrement.length - 1];
   if (containsWhitespace && !firstBlockParent?.childTextNodeIndex) {
     let start = 0;
-    while (start < text.length && (inPreTag ? text.charCodeAt(start) === 10 || text.charCodeAt(start) === 13 : isWhitespace(text.charCodeAt(start)))) {
+    while (start < text.length && (inPreTag ? text.charCodeAt(start) === NEWLINE_CHAR || text.charCodeAt(start) === CARRIAGE_RETURN_CHAR : isWhitespace(text.charCodeAt(start)))) {
       start++;
     }
     if (start > 0) {
@@ -1109,7 +1133,7 @@ function processClosingTag(htmlChunk, position, state, handleEvent) {
     }
   }
   if (curr) {
-    closeNode(state.currentNode, state, handleEvent);
+    closeNode(curr, state, handleEvent);
   }
   state.justClosedTag = true;
   return {
@@ -1143,6 +1167,12 @@ function closeNode(node, state, handleEvent) {
   if (node.tagId) {
     state.depthMap[node.tagId] = Math.max(0, state.depthMap[node.tagId] - 1);
   }
+  if (node.tagHandler?.isNonNesting) {
+    state.inSingleQuote = false;
+    state.inDoubleQuote = false;
+    state.inBacktick = false;
+    state.lastCharWasBackslash = false;
+  }
   state.depth--;
   handleEvent({ type: NodeEventExit, node });
   state.currentNode = state.currentNode.parent;
@@ -1233,6 +1263,12 @@ function processOpeningTag(tagName, tagId, htmlChunk, i, state, handleEvent) {
   parentNode.currentWalkIndex = 0;
   state.currentNode = parentNode;
   state.hasEncodedHtmlEntity = false;
+  if (tagHandler?.isNonNesting && !result.selfClosing) {
+    state.inSingleQuote = false;
+    state.inDoubleQuote = false;
+    state.inBacktick = false;
+    state.lastCharWasBackslash = false;
+  }
   if (result.selfClosing) {
     closeNode(tag, state, handleEvent);
     state.justClosedTag = true;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "mdream",
   "type": "module",
-  "version": "0.2.3",
+  "version": "0.2.4",
   "description": "Ultra-performant JavaScript HTML to Markdown converter optimized for LLMs.",
   "author": {
     "name": "Harlan Wilton",