mdream 0.15.2 → 0.16.0

package/README.md

@@ -201,7 +201,6 @@ Mdream includes several built-in plugins that can be used individually or combin
 - **[`frontmatterPlugin`](./src/plugins/frontmatter.ts)**: Generate YAML frontmatter from HTML head elements (title, meta tags)
 - **[`isolateMainPlugin`](./src/plugins/isolate-main.ts)**: Isolate main content using `<main>` elements or header-to-footer boundaries
 - **[`tailwindPlugin`](./src/plugins/tailwind.ts)**: Convert Tailwind CSS classes to Markdown formatting (bold, italic, etc.)
-- **[`readabilityPlugin`](./src/plugins/readability.ts)**: Content scoring and extraction (experimental)
 
 ```ts
 import { filterPlugin, frontmatterPlugin, isolateMainPlugin } from 'mdream/plugins'
@@ -215,6 +214,26 @@ const markdown = htmlToMarkdown(html, {
 })
 ```
 
+### Content Extraction with Readability
+
+For advanced content extraction (article detection, boilerplate removal), use [@mozilla/readability](https://github.com/mozilla/readability) before mdream:
+
+```ts
+import { Readability } from '@mozilla/readability'
+import { JSDOM } from 'jsdom'
+import { htmlToMarkdown } from 'mdream'
+
+const dom = new JSDOM(html, { url: 'https://example.com' })
+const article = new Readability(dom.window.document).parse()
+
+if (article) {
+  const markdown = htmlToMarkdown(article.content)
+  // article.title, article.excerpt, article.byline also available
+}
+```
+
+This pipeline gives you battle-tested content extraction + fast markdown conversion.
+
 ### Plugin Hooks
 
 - `beforeNodeProcess`: Called before any node processing, can skip nodes

package/dist/_chunks/{const-Bf_XN9U9.mjs → const.mjs}

@@ -1,42 +1,3 @@
-//#region src/buffer-region.ts
-/**
-* Creates a new buffer region
-* Returns null if node already has a region assigned
-*/
-function createBufferRegion(node, state, include) {
-	if (node.regionId) return null;
-	const id = state.regionToggles.size + 1;
-	node.regionId = id;
-	state.regionToggles.set(id, include);
-	state.regionContentBuffers.set(id, []);
-	return id;
-}
-/**
-* Collects content for a node into appropriate buffer (optimized)
-*/
-function collectNodeContent(node, content, state) {
-	if (!content) return;
-	const regionId = node.regionId || 0;
-	const targetBuffer = state.regionContentBuffers.get(regionId);
-	if (targetBuffer) {
-		targetBuffer.push(content);
-		state.lastContentCache = content;
-	}
-}
-/**
-* Assembles final content from buffer regions and clears them after use
-* Ensures frontmatter (regionId -1) appears first, followed by other included regions
-*/
-function assembleBufferedContent(state) {
-	const fragments = [];
-	for (const [regionId, content] of Array.from(state.regionContentBuffers.entries())) if (state.regionToggles.get(regionId)) fragments.push(...content);
-	state.regionToggles.clear();
-	state.regionContentBuffers.clear();
-	return fragments.join("").trimStart();
-}
-
-//#endregion
-//#region src/const.ts
 const TAG_HTML = 0;
 const TAG_HEAD = 1;
 const TAG_DETAILS = 2;
@@ -280,6 +241,4 @@ const DEFAULT_BLOCK_SPACING = [2, 2];
 const BLOCKQUOTE_SPACING = [1, 1];
 const LIST_ITEM_SPACING = [1, 0];
 const TABLE_ROW_SPACING = [0, 1];
-
-//#endregion
-export { TAG_H2 as $, TAG_TBODY as $t, TAG_BUTTON as A, TAG_P as At, TAG_DFN as B, TAG_SCRIPT as Bt, TAG_AUDIO as C, TAG_METER as Ct, TAG_BLOCKQUOTE as D, TAG_OBJECT as Dt, TAG_BDO as E, TAG_NOSCRIPT as Et, TAG_CODE as F, TAG_Q as Ft, TAG_EM as G, TAG_SPAN as Gt, TAG_DIV as H, TAG_SELECT as Ht, TAG_COL as I, TAG_RP as It, TAG_FIGCAPTION as J, TAG_SUB as Jt, TAG_EMBED as K, TAG_STRONG as Kt, TAG_DD as L, TAG_RT as Lt, TAG_CAPTION as M, TAG_PLAINTEXT as Mt, TAG_CENTER as N, TAG_PRE as Nt, TAG_BODY as O, TAG_OL as Ot, TAG_CITE as P, TAG_PROGRESS as Pt, TAG_H1 as Q, TAG_TABLE as Qt, TAG_DEL as R, TAG_RUBY as Rt, TAG_ASIDE as S, TAG_META as St, TAG_BASE as T, TAG_NOFRAMES as Tt, TAG_DL as U, TAG_SMALL as Ut, TAG_DIALOG as V, TAG_SECTION as Vt, TAG_DT as W, TAG_SOURCE as Wt, TAG_FOOTER as X, TAG_SUP as Xt, TAG_FIGURE as Y, TAG_SUMMARY as Yt, TAG_FORM as Z, TAG_SVG as Zt, TAG_A as _, TagIdMap as _n, TAG_LI as _t, LIST_ITEM_SPACING as a, TAG_THEAD as an, TAG_HEADER as at, TAG_AREA as b, createBufferRegion as bn, TAG_MAP as bt, MARKDOWN_HORIZONTAL_RULE as c, TAG_TR as cn, TAG_I as ct, MARKDOWN_STRONG as d, TAG_UL as dn, TAG_INPUT as dt, TAG_TD as en, TAG_H3 as et, MAX_TAG_ID as f, TAG_VAR as fn, TAG_INS as ft, TABLE_ROW_SPACING as g, TEXT_NODE as gn, TAG_LEGEND as gt, NodeEventExit as h, TAG_XMP as hn, TAG_LABEL as ht, HTML_ENTITIES as i, TAG_TH as in, TAG_HEAD as it, TAG_CANVAS as j, TAG_PARAM as jt, TAG_BR as k, TAG_OPTION as kt, MARKDOWN_INLINE_CODE as l, TAG_TRACK as ln, TAG_IFRAME as lt, NodeEventEnter as m, TAG_WBR as mn, TAG_KEYGEN as mt, DEFAULT_BLOCK_SPACING as n, TAG_TEXTAREA as nn, TAG_H5 as nt, MARKDOWN_CODE_BLOCK as o, TAG_TIME as on, TAG_HR as ot, NO_SPACING as p, TAG_VIDEO as pn, TAG_KBD as pt, TAG_FIELDSET as q, TAG_STYLE as qt, ELEMENT_NODE as r, TAG_TFOOT as rn, TAG_H6 as rt, MARKDOWN_EMPHASIS as s, TAG_TITLE as sn, TAG_HTML as st, BLOCKQUOTE_SPACING as t, TAG_TEMPLATE as tn, TAG_H4 as tt, MARKDOWN_STRIKETHROUGH as u, TAG_U as un, TAG_IMG as ut, TAG_ABBR as v, assembleBufferedContent as vn, TAG_LINK as vt, TAG_B as w, TAG_NAV as wt, TAG_ARTICLE as x, TAG_MARK as xt, TAG_ADDRESS as y, collectNodeContent as yn, TAG_MAIN as yt, TAG_DETAILS as z, TAG_SAMP as zt };
+export { TAG_H5 as $, TAG_TH as $t, TAG_CANVAS as A, TAG_PROGRESS as At, TAG_DIV as B, TAG_SPAN as Bt, TAG_B as C, TAG_OBJECT as Ct, TAG_BODY as D, TAG_PARAM as Dt, TAG_BLOCKQUOTE as E, TAG_P as Et, TAG_DD as F, TAG_SAMP as Ft, TAG_FIELDSET as G, TAG_SUP as Gt, TAG_DT as H, TAG_STYLE as Ht, TAG_DEL as I, TAG_SCRIPT as It, TAG_FORM as J, TAG_TBODY as Jt, TAG_FIGURE as K, TAG_SVG as Kt, TAG_DETAILS as L, TAG_SELECT as Lt, TAG_CITE as M, TAG_RP as Mt, TAG_CODE as N, TAG_RT as Nt, TAG_BR as O, TAG_PLAINTEXT as Ot, TAG_COL as P, TAG_RUBY as Pt, TAG_H4 as Q, TAG_TFOOT as Qt, TAG_DFN as R, TAG_SMALL as Rt, TAG_AUDIO as S, TAG_NOSCRIPT as St, TAG_BDO as T, TAG_OPTION as Tt, TAG_EM as U, TAG_SUB as Ut, TAG_DL as V, TAG_STRONG as Vt, TAG_EMBED as W, TAG_SUMMARY as Wt, TAG_H2 as X, TAG_TEMPLATE as Xt, TAG_H1 as Y, TAG_TD as Yt, TAG_H3 as Z, TAG_TEXTAREA as Zt, TAG_A as _, TAG_MARK as _t, LIST_ITEM_SPACING as a, TAG_U as an, TAG_IFRAME as at, TAG_AREA as b, TAG_NAV as bt, MARKDOWN_HORIZONTAL_RULE as c, TAG_VIDEO as cn, TAG_INS as ct, MARKDOWN_STRONG as d, TEXT_NODE as dn, TAG_LABEL as dt, TAG_THEAD as en, TAG_H6 as et, MAX_TAG_ID as f, TagIdMap as fn, TAG_LEGEND as ft, TABLE_ROW_SPACING as g, TAG_MAP as gt, NodeEventExit as h, TAG_MAIN as ht, HTML_ENTITIES as i, TAG_TRACK as in, TAG_I as it, TAG_CENTER as j, TAG_Q as jt, TAG_BUTTON as k, TAG_PRE as kt, MARKDOWN_INLINE_CODE as l, TAG_WBR as ln, TAG_KBD as lt, NodeEventEnter as m, TAG_LINK as mt, DEFAULT_BLOCK_SPACING as n, TAG_TITLE as nn, TAG_HEADER as nt, MARKDOWN_CODE_BLOCK as o, TAG_UL as on, TAG_IMG as ot, NO_SPACING as p, TAG_LI as pt, TAG_FOOTER as q, TAG_TABLE as qt, ELEMENT_NODE as r, TAG_TR as rn, TAG_HR as rt, MARKDOWN_EMPHASIS as s, TAG_VAR as sn, TAG_INPUT as st, BLOCKQUOTE_SPACING as t, TAG_TIME as tn, TAG_HEAD as tt, MARKDOWN_STRIKETHROUGH as u, TAG_XMP as un, TAG_KEYGEN as ut, TAG_ABBR as v, TAG_META as vt, TAG_BASE as w, TAG_OL as wt, TAG_ASIDE as x, TAG_NOFRAMES as xt, TAG_ADDRESS as y, TAG_METER as yt, TAG_DIALOG as z, TAG_SOURCE as zt };

package/dist/_chunks/{extraction-BA9MDtq3.mjs → extraction.mjs}

@@ -1,18 +1,10 @@
-import { t as createPlugin } from "./plugin-CjWWQTuL.mjs";
-
-//#region src/libs/query-selector.ts
-/**
-* Creates a tag selector matcher (e.g., 'div', 'p', 'h1')
-*/
+import { t as createPlugin } from "./plugin.mjs";
 function createTagSelector(tagName) {
 	return {
 		matches: (element) => element.name === tagName,
 		toString: () => tagName
 	};
 }
-/**
-* Creates an ID selector matcher (e.g., '#main', '#content')
-*/
 function createIdSelector(selector) {
 	const id = selector.slice(1);
 	return {
@@ -20,9 +12,6 @@ function createIdSelector(selector) {
 		toString: () => `#${id}`
 	};
 }
-/**
-* Creates a class selector matcher (e.g., '.container', '.header')
-*/
 function createClassSelector(selector) {
 	const className = selector.slice(1);
 	return {
@@ -33,9 +22,6 @@ function createClassSelector(selector) {
 		toString: () => `.${className}`
 	};
 }
-/**
-* Creates an attribute selector matcher (e.g., '[data-id]', '[href="https://example.com"]')
-*/
 function createAttributeSelector(selector) {
 	const match = selector.match(/\[([^\]=~|^$*]+)(?:([=~|^$*]+)["']?([^"'\]]+)["']?)?\]/);
 	const attrName = match ? match[1] : selector.slice(1, -1);
@@ -62,18 +48,12 @@ function createAttributeSelector(selector) {
 		}
 	};
 }
-/**
-* Creates a compound selector that combines multiple selectors (e.g., 'div.container', 'h1#title')
-*/
 function createCompoundSelector(selectors) {
 	return {
 		matches: (element) => selectors.every((selector) => selector.matches(element)),
 		toString: () => selectors.map((s) => s.toString()).join("")
 	};
 }
-/**
-* Parses a CSS selector into a matcher
-*/
 function parseSelector(selector) {
 	selector = selector.trim();
 	if (!selector) throw new Error("Empty selector");
@@ -100,9 +80,6 @@ function parseSelector(selector) {
 	if (selectorParts.length === 1) return selectorParts[0];
 	return createCompoundSelector(selectorParts);
 }
-
-//#endregion
-//#region src/plugins/extraction.ts
 function extractionPlugin(selectors) {
 	const matcherCallbacks = Object.entries(selectors).map(([selector, callback]) => ({
 		matcher: parseSelector(selector),
@@ -139,6 +116,4 @@ function extractionPlugin(selectors) {
 		}
 	});
 }
-
-//#endregion
-export { parseSelector as n, extractionPlugin as t };
+export { parseSelector as n, extractionPlugin as t };

package/dist/_chunks/{markdown-processor-D26Uo5td.mjs → markdown-processor.mjs}

@@ -1,6 +1,4 @@
-import { $ as TAG_H2, $t as TAG_TBODY, A as TAG_BUTTON, At as TAG_P, B as TAG_DFN, Bt as TAG_SCRIPT, C as TAG_AUDIO, Ct as TAG_METER, D as TAG_BLOCKQUOTE, E as TAG_BDO, Et as TAG_NOSCRIPT, F as TAG_CODE, Ft as TAG_Q, G as TAG_EM, Gt as TAG_SPAN, H as TAG_DIV, Ht as TAG_SELECT, I as TAG_COL, It as TAG_RP, Jt as TAG_SUB, K as TAG_EMBED, Kt as TAG_STRONG, L as TAG_DD, Lt as TAG_RT, Mt as TAG_PLAINTEXT, N as TAG_CENTER, Nt as TAG_PRE, O as TAG_BODY, Ot as TAG_OL, P as TAG_CITE, Pt as TAG_PROGRESS, Q as TAG_H1, Qt as TAG_TABLE, R as TAG_DEL, Rt as TAG_RUBY, S as TAG_ASIDE, St as TAG_META, T as TAG_BASE, Tt as TAG_NOFRAMES, U as TAG_DL, Ut as TAG_SMALL, V as TAG_DIALOG, W as TAG_DT, Wt as TAG_SOURCE, X as TAG_FOOTER, Xt as TAG_SUP, Yt as TAG_SUMMARY, Z as TAG_FORM, Zt as TAG_SVG, _ as TAG_A, _n as TagIdMap, _t as TAG_LI, a as LIST_ITEM_SPACING, an as TAG_THEAD, b as TAG_AREA, bt as TAG_MAP, c as MARKDOWN_HORIZONTAL_RULE, cn as TAG_TR, ct as TAG_I, d as MARKDOWN_STRONG, dn as TAG_UL, dt as TAG_INPUT, en as TAG_TD, et as TAG_H3, f as MAX_TAG_ID, fn as TAG_VAR, ft as TAG_INS, g as TABLE_ROW_SPACING, gn as TEXT_NODE, gt as TAG_LEGEND, h as NodeEventExit, hn as TAG_XMP, ht as TAG_LABEL, i as HTML_ENTITIES, in as TAG_TH, it as TAG_HEAD, j as TAG_CANVAS, jt as TAG_PARAM, k as TAG_BR, kt as TAG_OPTION, l as MARKDOWN_INLINE_CODE, ln as TAG_TRACK, lt as TAG_IFRAME, m as NodeEventEnter, mn as TAG_WBR, mt as TAG_KEYGEN, n as DEFAULT_BLOCK_SPACING, nn as TAG_TEXTAREA, nt as TAG_H5, o as MARKDOWN_CODE_BLOCK, on as TAG_TIME, ot as TAG_HR, p as NO_SPACING, pn as TAG_VIDEO, pt as TAG_KBD, q as TAG_FIELDSET, qt as TAG_STYLE, r as ELEMENT_NODE, rn as TAG_TFOOT, rt as TAG_H6, s as MARKDOWN_EMPHASIS, sn as TAG_TITLE, t as BLOCKQUOTE_SPACING, tn as TAG_TEMPLATE, tt as TAG_H4, u as MARKDOWN_STRIKETHROUGH, un as TAG_U, ut as TAG_IMG, v as TAG_ABBR, vn as assembleBufferedContent, vt as TAG_LINK, w as TAG_B, wt as TAG_NAV, xt as TAG_MARK, y as TAG_ADDRESS, yn as collectNodeContent, z as TAG_DETAILS, zt as TAG_SAMP } from "./const-Bf_XN9U9.mjs";
-
-//#region src/tags.ts
+import { $ as TAG_H5, $t as TAG_TH, A as TAG_CANVAS, At as TAG_PROGRESS, B as TAG_DIV, Bt as TAG_SPAN, C as TAG_B, D as TAG_BODY, Dt as TAG_PARAM, E as TAG_BLOCKQUOTE, Et as TAG_P, F as TAG_DD, Ft as TAG_SAMP, G as TAG_FIELDSET, Gt as TAG_SUP, H as TAG_DT, Ht as TAG_STYLE, I as TAG_DEL, It as TAG_SCRIPT, J as TAG_FORM, Jt as TAG_TBODY, Kt as TAG_SVG, L as TAG_DETAILS, Lt as TAG_SELECT, M as TAG_CITE, Mt as TAG_RP, N as TAG_CODE, Nt as TAG_RT, O as TAG_BR, Ot as TAG_PLAINTEXT, P as TAG_COL, Pt as TAG_RUBY, Q as TAG_H4, Qt as TAG_TFOOT, R as TAG_DFN, Rt as TAG_SMALL, S as TAG_AUDIO, St as TAG_NOSCRIPT, T as TAG_BDO, Tt as TAG_OPTION, U as TAG_EM, Ut as TAG_SUB, V as TAG_DL, Vt as TAG_STRONG, W as TAG_EMBED, Wt as TAG_SUMMARY, X as TAG_H2, Xt as TAG_TEMPLATE, Y as TAG_H1, Yt as TAG_TD, Z as TAG_H3, Zt as TAG_TEXTAREA, _ as TAG_A, _t as TAG_MARK, a as LIST_ITEM_SPACING, an as TAG_U, at as TAG_IFRAME, b as TAG_AREA, bt as TAG_NAV, c as MARKDOWN_HORIZONTAL_RULE, cn as TAG_VIDEO, ct as TAG_INS, d as MARKDOWN_STRONG, dn as TEXT_NODE, dt as TAG_LABEL, en as TAG_THEAD, et as TAG_H6, f as MAX_TAG_ID, fn as TagIdMap, ft as TAG_LEGEND, g as TABLE_ROW_SPACING, gt as TAG_MAP, h as NodeEventExit, i as HTML_ENTITIES, in as TAG_TRACK, it as TAG_I, j as TAG_CENTER, jt as TAG_Q, k as TAG_BUTTON, kt as TAG_PRE, l as MARKDOWN_INLINE_CODE, ln as TAG_WBR, lt as TAG_KBD, m as NodeEventEnter, mt as TAG_LINK, n as DEFAULT_BLOCK_SPACING, nn as TAG_TITLE, o as MARKDOWN_CODE_BLOCK, on as TAG_UL, ot as TAG_IMG, p as NO_SPACING, pt as TAG_LI, q as TAG_FOOTER, qt as TAG_TABLE, r as ELEMENT_NODE, rn as TAG_TR, rt as TAG_HR, s as MARKDOWN_EMPHASIS, sn as TAG_VAR, st as TAG_INPUT, t as BLOCKQUOTE_SPACING, tn as TAG_TIME, tt as TAG_HEAD, u as MARKDOWN_STRIKETHROUGH, un as TAG_XMP, ut as TAG_KEYGEN, v as TAG_ABBR, vt as TAG_META, w as TAG_BASE, wt as TAG_OL, x as TAG_ASIDE, xt as TAG_NOFRAMES, y as TAG_ADDRESS, yt as TAG_METER, z as TAG_DIALOG, zt as TAG_SOURCE } from "./const.mjs";
 function resolveUrl(url, origin) {
 	if (!url) return url;
 	if (url.startsWith("//")) return `https:${url}`;
@@ -529,12 +527,6 @@ const tagHandlers = {
 		spacing: [0, 1]
 	}
 };
-
-//#endregion
-//#region src/utils.ts
-/**
-* Decode HTML entities - optimized version with single pass
-*/
 function decodeHTMLEntities(text) {
 	let result = "";
 	let i = 0;
@@ -585,9 +577,6 @@ function traverseUpToFirstBlockNode(node) {
 	}
 	return parentsToIncrement;
 }
-
-//#endregion
-//#region src/parse.ts
 const LT_CHAR = 60;
 const GT_CHAR = 62;
 const SLASH_CHAR = 47;
@@ -610,16 +599,9 @@ const EMPTY_ATTRIBUTES = Object.freeze({});
 function copyDepthMap(depthMap) {
 	return new Uint8Array(depthMap);
 }
-/**
-* Fast whitespace check using direct character code comparison
-*/
 function isWhitespace(charCode) {
 	return charCode === SPACE_CHAR || charCode === TAB_CHAR || charCode === NEWLINE_CHAR || charCode === CARRIAGE_RETURN_CHAR;
 }
-/**
-* Pure HTML parser that emits DOM events
-* Completely decoupled from markdown generation
-*/
 function parseHtml(html, options = {}) {
 	const events = [];
 	return {
@@ -633,15 +615,9 @@ function parseHtml(html, options = {}) {
 		})
 	};
 }
-/**
-* Streaming HTML parser - calls onEvent for each DOM event
-*/
 function parseHtmlStream(html, state, onEvent) {
 	return parseHtmlInternal(html, state, onEvent);
 }
-/**
-* Internal parsing function - extracted from original parseHTML
-*/
 function parseHtmlInternal(htmlChunk, state, handleEvent) {
 	let textBuffer = "";
 	state.depthMap ??= new Uint8Array(MAX_TAG_ID);
@@ -772,9 +748,6 @@ function parseHtmlInternal(htmlChunk, state, handleEvent) {
 	}
 	return textBuffer;
 }
-/**
-* Process accumulated text buffer and create text node event
-*/
 function processTextBuffer(textBuffer, state, handleEvent) {
 	const containsNonWhitespace = state.textBufferContainsNonWhitespace;
 	const containsWhitespace = state.textBufferContainsWhitespace;
@@ -801,7 +774,6 @@ function processTextBuffer(textBuffer, state, handleEvent) {
 		type: TEXT_NODE,
 		value: text,
 		parent: state.currentNode,
-		regionId: state.currentNode?.regionId,
 		index: state.currentNode.currentWalkIndex++,
 		depth: state.depth,
 		containsWhitespace,
@@ -814,9 +786,6 @@ function processTextBuffer(textBuffer, state, handleEvent) {
 	});
 	state.lastTextNode = textNode;
 }
-/**
-* Process HTML closing tag
-*/
 function processClosingTag(htmlChunk, position, state, handleEvent) {
 	let i = position + 2;
 	const tagNameStart = i;
@@ -857,9 +826,6 @@ function processClosingTag(htmlChunk, position, state, handleEvent) {
 		remainingText: ""
 	};
 }
-/**
-* Close a node and emit exit event
-*/
 function closeNode(node, state, handleEvent) {
 	if (!node) return;
 	if (node.tagId === TAG_A && !node.childTextNodeIndex) {
@@ -895,9 +861,6 @@ function closeNode(node, state, handleEvent) {
 	state.hasEncodedHtmlEntity = false;
 	state.justClosedTag = true;
 }
-/**
-* Process HTML comment or doctype
-*/
 function processCommentOrDoctype(htmlChunk, position) {
 	let i = position;
 	const chunkLength = htmlChunk.length;
@@ -939,9 +902,6 @@ function processCommentOrDoctype(htmlChunk, position) {
 		};
 	}
 }
-/**
-* Process HTML opening tag
-*/
 function processOpeningTag(tagName, tagId, htmlChunk, i, state, handleEvent) {
 	if (state.currentNode?.tagHandler?.isNonNesting) closeNode(state.currentNode, state, handleEvent);
 	const tagHandler = tagHandlers[tagId];
@@ -966,7 +926,6 @@ function processOpeningTag(tagName, tagId, htmlChunk, i, state, handleEvent) {
 		depthMap: copyDepthMap(state.depthMap),
 		depth: state.depth,
 		index: currentWalkIndex,
-		regionId: state.currentNode?.regionId,
 		tagId,
 		tagHandler
 	};
@@ -996,9 +955,6 @@ function processOpeningTag(tagName, tagId, htmlChunk, i, state, handleEvent) {
 		selfClosing: result.selfClosing
 	};
 }
-/**
-* Extract and process HTML tag attributes
-*/
 function processTagAttributes(htmlChunk, position, tagHandler) {
 	let i = position;
 	const chunkLength = htmlChunk.length;
@@ -1046,9 +1002,6 @@ function processTagAttributes(htmlChunk, position, tagHandler) {
 		attrBuffer: htmlChunk.substring(attrStartPos, i)
 	};
 }
-/**
-* Parse HTML attributes string into key-value object
-*/
 function parseAttributes(attrStr) {
 	if (!attrStr) return EMPTY_ATTRIBUTES;
 	const result = {};
@@ -1128,19 +1081,6 @@ function parseAttributes(attrStr) {
 	}
 	return result;
 }
-
-//#endregion
-//#region src/plugin-processor.ts
-/**
-* Processes plugins for a given node event
-* Shared logic between markdown-processor.ts and stream.ts
-*
-* @param event - The node event to process
-* @param plugins - Array of plugins to apply
-* @param state - The current runtime state
-* @param processEvent - Callback to process the event after plugin processing
-* @returns true if the event should be skipped, false to continue processing
-*/
 function processPluginsForEvent(event, plugins, state, processEvent) {
 	if (plugins?.length) {
 		for (const plugin of plugins) {
@@ -1173,12 +1113,6 @@ function processPluginsForEvent(event, plugins, state, processEvent) {
 	processEvent(event);
 	return false;
 }
-
-//#endregion
-//#region src/markdown-processor.ts
-/**
-* Determines if spacing is needed between two characters
-*/
 function needsSpacing(lastChar, firstChar, state) {
 	if (lastChar === " " || lastChar === "\n" || lastChar === "	") return false;
 	if (firstChar === " " || firstChar === "\n" || firstChar === "	") return false;
@@ -1208,15 +1142,9 @@ function needsSpacing(lastChar, firstChar, state) {
 	if (noSpaceAfter.has(lastChar) || noSpaceBefore.has(firstChar)) return false;
 	return true;
 }
-/**
-* Determines if spacing should be added before text content
-*/
 function shouldAddSpacingBeforeText(lastChar, lastNode, textNode) {
 	return !!lastChar && lastChar !== "\n" && lastChar !== " " && lastChar !== "[" && lastChar !== ">" && !lastNode?.tagHandler?.isInline && textNode.value[0] !== " ";
 }
-/**
-* Calculate newline configuration based on tag handler spacing config
-*/
 function calculateNewLineConfig(node) {
 	const tagId = node.tagId;
 	const depthMap = node.depthMap;
@@ -1236,28 +1164,19 @@ function calculateNewLineConfig(node) {
 	if (node.tagHandler?.spacing) return node.tagHandler?.spacing;
 	return DEFAULT_BLOCK_SPACING;
 }
-/**
-* Creates a markdown processor that consumes DOM events and generates markdown
-*/
 function createMarkdownProcessor(options = {}) {
 	const state = {
 		options,
-		regionToggles: /* @__PURE__ */ new Map(),
-		regionContentBuffers: /* @__PURE__ */ new Map(),
+		buffer: [],
 		depthMap: new Uint8Array(MAX_TAG_ID)
 	};
-	state.regionToggles.set(0, true);
-	state.regionContentBuffers.set(0, []);
 	let lastYieldedLength = 0;
-	/**
-	* Process a DOM event and generate markdown
-	*/
 	function processEvent(event) {
 		const { type: eventType, node } = event;
 		const lastNode = state.lastNode;
 		state.lastNode = event.node;
 		state.depth = node.depth;
-		const buff = state.regionContentBuffers.get(node.regionId || 0) || [];
+		const buff = state.buffer;
 		const lastBuffEntry = buff[buff.length - 1];
 		const lastChar = lastBuffEntry?.charAt(lastBuffEntry.length - 1) || "";
 		let secondLastChar;
@@ -1269,7 +1188,8 @@ function createMarkdownProcessor(options = {}) {
 				if (textNode.excludedFromMarkdown) return;
 				if (textNode.value === " " && lastChar === "\n") return;
 				if (shouldAddSpacingBeforeText(lastChar, lastNode, textNode)) textNode.value = ` ${textNode.value}`;
-				collectNodeContent(textNode, textNode.value, state);
+				state.buffer.push(textNode.value);
+				state.lastContentCache = textNode.value;
 			}
 			state.lastTextNode = textNode;
 			return;
@@ -1299,7 +1219,10 @@ function createMarkdownProcessor(options = {}) {
 		const newLines = Math.max(0, configuredNewLines - lastNewLines);
 		if (newLines > 0) {
 			if (!buff.length) {
-				for (const fragment of output) collectNodeContent(node, fragment, state);
+				for (const fragment of output) if (fragment) {
+					state.buffer.push(fragment);
+					state.lastContentCache = fragment;
+				}
 				return;
 			}
 			const newlinesStr = "\n".repeat(newLines);
@@ -1321,12 +1244,15 @@ function createMarkdownProcessor(options = {}) {
 				state.lastTextNode = void 0;
 			}
 		}
-		if (output[0]?.[0] && eventType === NodeEventEnter && lastChar && needsSpacing(lastChar, output[0][0], state)) collectNodeContent(node, " ", state);
-		for (const fragment of output) collectNodeContent(node, fragment, state);
+		if (output[0]?.[0] && eventType === NodeEventEnter && lastChar && needsSpacing(lastChar, output[0][0], state)) {
+			state.buffer.push(" ");
+			state.lastContentCache = " ";
+		}
+		for (const fragment of output) if (fragment) {
+			state.buffer.push(fragment);
+			state.lastContentCache = fragment;
+		}
 	}
-	/**
-	* Process HTML string and generate events
-	*/
 	function processHtml(html) {
 		parseHtmlStream(html, {
 			depthMap: state.depthMap,
@@ -1336,19 +1262,13 @@ function createMarkdownProcessor(options = {}) {
 			processPluginsForEvent(event, state.options?.plugins, state, processEvent);
 		});
 	}
-	/**
-	* Get the final markdown output
-	*/
 	function getMarkdown() {
-		return assembleBufferedContent(state).trimEnd();
+		const result = state.buffer.join("").trimStart();
+		state.buffer.length = 0;
+		return result.trimEnd();
 	}
-	/**
-	* Get new markdown content since the last call (for streaming)
-	*/
 	function getMarkdownChunk() {
-		const fragments = [];
-		for (const [regionId, content] of Array.from(state.regionContentBuffers.entries())) if (state.regionToggles.get(regionId)) fragments.push(...content);
-		const currentContent = fragments.join("").trimStart();
+		const currentContent = state.buffer.join("").trimStart();
 		const newContent = currentContent.slice(lastYieldedLength);
 		lastYieldedLength = currentContent.length;
 		return newContent;
@@ -1362,6 +1282,4 @@ function createMarkdownProcessor(options = {}) {
 	};
 }
 const MarkdownProcessor = createMarkdownProcessor;
-
-//#endregion
-export { parseHtmlStream as a, parseHtml as i, createMarkdownProcessor as n, processPluginsForEvent as r, MarkdownProcessor as t };
+export { parseHtmlStream as a, parseHtml as i, createMarkdownProcessor as n, processPluginsForEvent as r, MarkdownProcessor as t };

package/dist/_chunks/{plugin-D5soyEXm.d.mts → plugin.d.mts}

@@ -1,7 +1,6 @@
-import { d as Plugin } from "./types-CT4ZxeOH.mjs";
+import { u as Plugin } from "./types.mjs";
 
 //#region src/pluggable/plugin.d.ts
-
 /**
  * Create a plugin that implements the Plugin interface with improved type inference
  *

package/dist/_chunks/plugin.mjs

@@ -0,0 +1,4 @@
+function createPlugin(plugin) {
+	return plugin;
+}
+export { createPlugin as t };

package/dist/_chunks/{src-BJpipdul.mjs → stream.mjs}

@@ -1,12 +1,4 @@
-import { a as parseHtmlStream, n as createMarkdownProcessor, r as processPluginsForEvent } from "./markdown-processor-D26Uo5td.mjs";
-
-//#region src/stream.ts
-/**
-* Creates a markdown stream from an HTML stream
-* @param htmlStream - ReadableStream of HTML content (as Uint8Array or string)
-* @param options - Configuration options for conversion
-* @returns An async generator yielding markdown chunks
-*/
+import { a as parseHtmlStream, n as createMarkdownProcessor, r as processPluginsForEvent } from "./markdown-processor.mjs";
 async function* streamHtmlToMarkdown(htmlStream, options = {}) {
 	if (!htmlStream) throw new Error("Invalid HTML stream provided");
 	const decoder = new TextDecoder();
@@ -38,14 +30,4 @@ async function* streamHtmlToMarkdown(htmlStream, options = {}) {
 		reader.releaseLock();
 	}
 }
-
-//#endregion
-//#region src/index.ts
-function htmlToMarkdown(html, options = {}) {
-	const processor = createMarkdownProcessor(options);
-	processor.processHtml(html);
-	return processor.getMarkdown();
-}
-
-//#endregion
-export { streamHtmlToMarkdown as n, htmlToMarkdown as t };
+export { streamHtmlToMarkdown as t };