@kreuzberg/html-to-markdown-wasm 3.4.0-rc.8 → 3.4.0

package/package.json

@@ -1,65 +1,35 @@
 {
-	"name": "@kreuzberg/html-to-markdown-wasm",
-	"version": "3.4.0-rc.8",
-	"description": "High-performance HTML to Markdown converter - WebAssembly bindings",
-	"main": "dist/html_to_markdown_wasm.js",
-	"types": "dist/html_to_markdown_wasm.d.ts",
-	"exports": {
-		".": {
-			"import": "./dist/html_to_markdown_wasm.js",
-			"types": "./dist/html_to_markdown_wasm.d.ts",
-			"default": "./dist/html_to_markdown_wasm.js"
-		},
-		"./dist-node": {
-			"import": "./dist-node/html_to_markdown_wasm.js",
-			"require": "./dist-node/html_to_markdown_wasm.js",
-			"types": "./dist-node/html_to_markdown_wasm.d.ts"
-		},
-		"./dist-node/*": "./dist-node/*",
-		"./dist-web": {
-			"import": "./dist-web/html_to_markdown_wasm.js",
-			"types": "./dist-web/html_to_markdown_wasm.d.ts"
-		},
-		"./dist-web/*": "./dist-web/*"
-	},
-	"repository": "https://github.com/kreuzberg-dev/html-to-markdown",
-	"homepage": "https://github.com/kreuzberg-dev/html-to-markdown",
-	"license": "MIT",
-	"author": "Na'aman Hirschfeld <naaman@kreuzberg.dev>",
-	"bugs": "https://github.com/kreuzberg-dev/html-to-markdown/issues",
-	"keywords": [
-		"html",
-		"markdown",
-		"converter",
-		"rust",
-		"wasm",
-		"webassembly"
-	],
-	"files": [
-		"dist",
-		"dist-node",
-		"dist-web",
-		"README.md"
-	],
-	"scripts": {
-		"build": "wasm-pack build --target bundler --out-dir dist && node ./scripts/patch-bundler-entry.js",
-		"build:nodejs": "wasm-pack build --target nodejs --out-dir dist-node && node ./scripts/patch-bundler-entry.js dist-node --types-only",
-		"build:web": "wasm-pack build --target web --out-dir dist-web && node ./scripts/patch-bundler-entry.js dist-web --types-only",
-		"build:all": "pnpm run build && pnpm run build:nodejs && pnpm run build:web && pnpm run cleanup:gitignore",
-		"cleanup:gitignore": "node ./scripts/cleanup-gitignore.js",
-		"test": "vitest run",
-		"test:watch": "vitest",
-		"test:wasm-pack": "wasm-pack test --headless --chrome",
-		"clean": "rm -rf dist dist-node dist-web node_modules pkg"
-	},
-	"devDependencies": {
-		"@types/node": "^25.6.0",
-		"tsx": "^4.21.0",
-		"vitest": "^4.1.5",
-		"wasm-pack": "^0.14.0"
-	},
-	"publishConfig": {
-		"registry": "https://registry.npmjs.org/",
-		"access": "public"
-	}
+  "name": "@kreuzberg/html-to-markdown-wasm",
+  "version": "3.4.0",
+  "private": false,
+  "description": "High-performance HTML to Markdown converter",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/kreuzberg-dev/html-to-markdown",
+    "directory": "crates/html-to-markdown-wasm"
+  },
+  "files": [
+    "pkg",
+    "*.wasm",
+    "*.d.ts",
+    "README.md"
+  ],
+  "type": "module",
+  "main": "pkg/nodejs/html-to-markdown_wasm.js",
+  "module": "pkg/web/html-to-markdown_wasm.js",
+  "types": "pkg/nodejs/html-to-markdown_wasm.d.ts",
+  "scripts": {
+    "build": "wasm-pack build --target nodejs --out-dir pkg/nodejs",
+    "build:ci": "wasm-pack build --release --target nodejs --out-dir pkg/nodejs",
+    "build:wasm:web": "wasm-pack build --release --target web --out-dir pkg/web",
+    "build:wasm:bundler": "wasm-pack build --release --target bundler --out-dir pkg/bundler",
+    "build:wasm:nodejs": "wasm-pack build --release --target nodejs --out-dir pkg/nodejs",
+    "build:wasm:deno": "wasm-pack build --release --target deno --out-dir pkg/deno",
+    "build:all": "npm run build:wasm:web && npm run build:wasm:bundler && npm run build:wasm:nodejs && npm run build:wasm:deno && find pkg -name .gitignore -delete",
+    "test": "vitest run",
+    "test:watch": "vitest watch",
+    "test:coverage": "vitest run --coverage",
+    "clean": "rm -rf pkg dist"
+  }
 }

package/{dist-node → pkg/bundler}/html_to_markdown_wasm.d.ts

@@ -63,7 +63,7 @@ export class WasmConversionOptions {
      * Create from a partial update, applying to defaults.
      */
     static fromUpdate(update: WasmConversionOptionsUpdate): WasmConversionOptions;
-    constructor(heading_style?: WasmHeadingStyle | null, list_indent_type?: WasmListIndentType | null, list_indent_width?: number | null, bullets?: string | null, strong_em_symbol?: string | null, escape_asterisks?: boolean | null, escape_underscores?: boolean | null, escape_misc?: boolean | null, escape_ascii?: boolean | null, code_language?: string | null, autolinks?: boolean | null, default_title?: boolean | null, br_in_tables?: boolean | null, highlight_style?: WasmHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: WasmWhitespaceMode | null, strip_newlines?: boolean | null, wrap?: boolean | null, wrap_width?: number | null, convert_as_inline?: boolean | null, sub_symbol?: string | null, sup_symbol?: string | null, newline_style?: WasmNewlineStyle | null, code_block_style?: WasmCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: WasmPreprocessingOptions | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: WasmLinkStyle | null, output_format?: WasmOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null, exclude_selectors?: string[] | null, max_depth?: number | null);
+    constructor(heading_style?: WasmHeadingStyle | null, list_indent_type?: WasmListIndentType | null, list_indent_width?: number | null, bullets?: string | null, strong_em_symbol?: string | null, escape_asterisks?: boolean | null, escape_underscores?: boolean | null, escape_misc?: boolean | null, escape_ascii?: boolean | null, code_language?: string | null, autolinks?: boolean | null, default_title?: boolean | null, br_in_tables?: boolean | null, highlight_style?: WasmHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: WasmWhitespaceMode | null, strip_newlines?: boolean | null, wrap?: boolean | null, wrap_width?: number | null, convert_as_inline?: boolean | null, sub_symbol?: string | null, sup_symbol?: string | null, newline_style?: WasmNewlineStyle | null, code_block_style?: WasmCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: WasmPreprocessingOptions | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: WasmLinkStyle | null, output_format?: WasmOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null, exclude_selectors?: string[] | null, max_depth?: number | null, visitor?: WasmVisitorHandle | null);
     autolinks: boolean;
     brInTables: boolean;
     bullets: string;
@@ -102,6 +102,8 @@ export class WasmConversionOptions {
     strongEmSymbol: string;
     subSymbol: string;
     supSymbol: string;
+    get visitor(): WasmVisitorHandle | undefined;
+    set visitor(value: WasmVisitorHandle | null | undefined);
     whitespaceMode: WasmWhitespaceMode;
     wrap: boolean;
     wrapWidth: number;
@@ -140,6 +142,10 @@ export class WasmConversionOptionsBuilder {
      * Set the list of HTML tag names whose content is stripped from output.
      */
     stripTags(tags: string[]): WasmConversionOptionsBuilder;
+    /**
+     * Set the visitor used during conversion.
+     */
+    visitor(visitor?: WasmVisitorHandle | null): WasmConversionOptionsBuilder;
 }
 
 /**
@@ -151,7 +157,7 @@ export class WasmConversionOptionsBuilder {
 export class WasmConversionOptionsUpdate {
     free(): void;
     [Symbol.dispose](): void;
-    constructor(heading_style?: WasmHeadingStyle | null, list_indent_type?: WasmListIndentType | null, list_indent_width?: number | null, bullets?: string | null, strong_em_symbol?: string | null, escape_asterisks?: boolean | null, escape_underscores?: boolean | null, escape_misc?: boolean | null, escape_ascii?: boolean | null, code_language?: string | null, autolinks?: boolean | null, default_title?: boolean | null, br_in_tables?: boolean | null, highlight_style?: WasmHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: WasmWhitespaceMode | null, strip_newlines?: boolean | null, wrap?: boolean | null, wrap_width?: number | null, convert_as_inline?: boolean | null, sub_symbol?: string | null, sup_symbol?: string | null, newline_style?: WasmNewlineStyle | null, code_block_style?: WasmCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: WasmPreprocessingOptionsUpdate | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: WasmLinkStyle | null, output_format?: WasmOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null, max_depth?: number | null, exclude_selectors?: string[] | null);
+    constructor(heading_style?: WasmHeadingStyle | null, list_indent_type?: WasmListIndentType | null, list_indent_width?: number | null, bullets?: string | null, strong_em_symbol?: string | null, escape_asterisks?: boolean | null, escape_underscores?: boolean | null, escape_misc?: boolean | null, escape_ascii?: boolean | null, code_language?: string | null, autolinks?: boolean | null, default_title?: boolean | null, br_in_tables?: boolean | null, highlight_style?: WasmHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: WasmWhitespaceMode | null, strip_newlines?: boolean | null, wrap?: boolean | null, wrap_width?: number | null, convert_as_inline?: boolean | null, sub_symbol?: string | null, sup_symbol?: string | null, newline_style?: WasmNewlineStyle | null, code_block_style?: WasmCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: WasmPreprocessingOptionsUpdate | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: WasmLinkStyle | null, output_format?: WasmOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null, max_depth?: number | null, exclude_selectors?: string[] | null, visitor?: WasmVisitorHandle | null);
     get autolinks(): boolean | undefined;
     set autolinks(value: boolean | null | undefined);
     get brInTables(): boolean | undefined;
@@ -226,6 +232,8 @@ export class WasmConversionOptionsUpdate {
     set subSymbol(value: string | null | undefined);
     get supSymbol(): string | undefined;
     set supSymbol(value: string | null | undefined);
+    get visitor(): WasmVisitorHandle | undefined;
+    set visitor(value: WasmVisitorHandle | null | undefined);
     get whitespaceMode(): WasmWhitespaceMode | undefined;
     set whitespaceMode(value: WasmWhitespaceMode | null | undefined);
     get wrap(): boolean | undefined;
@@ -273,7 +281,6 @@ export class WasmConversionResult {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::DocumentMetadata;
  * let doc = DocumentMetadata {
  *     title: Some("My Article".to_string()),
  *     description: Some("A great article about Rust".to_string()),
@@ -363,7 +370,6 @@ export class WasmGridCell {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::HeaderMetadata;
  * let header = HeaderMetadata {
  *     level: 1,
  *     text: "Main Title".to_string(),
@@ -389,7 +395,6 @@ export class WasmHeaderMetadata {
      * # Examples
      *
      * ```
-     * # use html_to_markdown_rs::metadata::HeaderMetadata;
      * let valid = HeaderMetadata {
      *     level: 3,
      *     text: "Title".to_string(),
@@ -451,7 +456,6 @@ export enum WasmHighlightStyle {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::HtmlMetadata;
  * let metadata = HtmlMetadata {
  *     document: Default::default(),
  *     headers: Vec::new(),
@@ -483,7 +487,6 @@ export class WasmHtmlMetadata {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::{ImageMetadata, ImageType};
  * let img = ImageMetadata {
  *     src: "https://example.com/image.jpg".to_string(),
  *     alt: Some("An example image".to_string()),
@@ -531,7 +534,6 @@ export enum WasmImageType {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::{LinkMetadata, LinkType};
  * let link = LinkMetadata {
  *     href: "https://example.com".to_string(),
  *     text: "Example".to_string(),
@@ -562,7 +564,6 @@ export class WasmLinkMetadata {
      * # Examples
      *
      * ```
-     * # use html_to_markdown_rs::metadata::{LinkMetadata, LinkType};
      * assert_eq!(LinkMetadata::classify_link("#section"), LinkType::Anchor);
      * assert_eq!(LinkMetadata::classify_link("mailto:test@example.com"), LinkType::Email);
      * assert_eq!(LinkMetadata::classify_link("tel:+1234567890"), LinkType::Phone);
@@ -867,7 +868,6 @@ export class WasmProcessingWarning {
  * # Examples
  *
  * ```
- * # use html_to_markdown_rs::metadata::{StructuredData, StructuredDataType};
  * let schema = StructuredData {
  *     data_type: StructuredDataType::JsonLd,
  *     raw_json: r#"{"@context":"https://schema.org","@type":"Article"}"#.to_string(),
@@ -961,6 +961,17 @@ export enum WasmVisitResult {
     Error = 4,
 }
 
+/**
+ * Type alias for a visitor handle (Rc-wrapped `RefCell` for interior mutability).
+ *
+ * This allows visitors to be passed around and shared while still being mutable.
+ */
+export class WasmVisitorHandle {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(visitor: any);
+}
+
 /**
  * Categories of processing warnings.
  */
@@ -983,129 +994,27 @@ export enum WasmWhitespaceMode {
     Strict = 1,
 }
 
-export function convert(html: string, options?: WasmConversionOptions | null, visitor?: any | null): WasmConversionResult;
-
-
-export type WasmHeadingStyle = "underlined" | "atx" | "atxClosed";
-export type WasmListIndentType = "spaces" | "tabs";
-export type WasmWhitespaceMode = "normalized" | "strict";
-export type WasmNewlineStyle = "spaces" | "backslash";
-export type WasmCodeBlockStyle = "indented" | "backticks" | "tildes";
-export type WasmHighlightStyle = "doubleEqual" | "html" | "bold" | "none";
-export type WasmPreprocessingPreset = "minimal" | "standard" | "aggressive";
-export type WasmOutputFormat = "markdown" | "djot" | "plain";
-
-export interface WasmPreprocessingOptions {
-  enabled?: boolean;
-  preset?: WasmPreprocessingPreset;
-  removeNavigation?: boolean;
-  removeForms?: boolean;
-}
-
-export interface WasmConversionOptions {
-  headingStyle?: WasmHeadingStyle;
-  listIndentType?: WasmListIndentType;
-  listIndentWidth?: number;
-  bullets?: string;
-  strongEmSymbol?: string;
-  escapeAsterisks?: boolean;
-  escapeUnderscores?: boolean;
-  escapeMisc?: boolean;
-  escapeAscii?: boolean;
-  codeLanguage?: string;
-  autolinks?: boolean;
-  defaultTitle?: boolean;
-  brInTables?: boolean;
-  hocrSpatialTables?: boolean;
-  highlightStyle?: WasmHighlightStyle;
-  extractMetadata?: boolean;
-  whitespaceMode?: WasmWhitespaceMode;
-  stripNewlines?: boolean;
-  wrap?: boolean;
-  wrapWidth?: number;
-  convertAsInline?: boolean;
-  subSymbol?: string;
-  supSymbol?: string;
-  newlineStyle?: WasmNewlineStyle;
-  codeBlockStyle?: WasmCodeBlockStyle;
-  keepInlineImagesIn?: string[];
-  preprocessing?: WasmPreprocessingOptions | null;
-  encoding?: string;
-  debug?: boolean;
-  stripTags?: string[];
-  preserveTags?: string[];
-  skipImages?: boolean;
-  outputFormat?: WasmOutputFormat;
-  includeDocumentStructure?: boolean;
-  extractImages?: boolean;
-  maxImageSize?: number;
-  captureSvg?: boolean;
-  inferDimensions?: boolean;
-}
-
-/** A single cell in a structured table grid. */
-export interface WasmGridCell {
-  content: string;
-  row: number;
-  col: number;
-  rowSpan: number;
-  colSpan: number;
-  isHeader: boolean;
-}
-
-/** Structured table grid with cell-level data. */
-export interface WasmTableGrid {
-  rows: number;
-  cols: number;
-  cells: WasmGridCell[];
-}
-
-/** A table extracted during conversion. */
-export interface WasmConversionTable {
-  grid: WasmTableGrid;
-  markdown: string;
-}
-
-/** Non-fatal warning emitted during conversion. */
-export interface WasmConversionWarning {
-  /** Human-readable warning message. */
-  message: string;
-  /** Warning kind identifier. */
-  kind: string;
-}
-
-/** An extracted inline image from the HTML document. */
-export interface WasmInlineImage {
-  /** Raw image data as a Uint8Array. */
-  data: Uint8Array;
-  /** Image format (png, jpeg, gif, svg, etc.). */
-  format: string;
-  /** Generated or provided filename, or null. */
-  filename: string | null;
-  /** Alt text or description, or null. */
-  description: string | null;
-  /** Image width in pixels, or null if not available. */
-  width: number | null;
-  /** Image height in pixels, or null if not available. */
-  height: number | null;
-  /** Source type ("img_data_uri" or "svg_element"). */
-  source: string;
-  /** HTML attributes from the source element. */
-  attributes: Record<string, string>;
-}
-
-/** Result of the convert() API. */
-export interface WasmConversionResult {
-  /** Converted text output (markdown, djot, or plain text), or null. */
-  content: string | null;
-  /** Structured document tree serialized as a JSON value, or null. */
-  document: unknown | null;
-  /** Extracted HTML metadata serialized as a JSON value, or null. */
-  metadata: unknown | null;
-  /** All tables found in the HTML, in document order. */
-  tables: WasmConversionTable[];
-  /** Extracted inline images (data URIs and SVGs). */
-  images: WasmInlineImage[];
-  /** Non-fatal processing warnings. */
-  warnings: WasmConversionWarning[];
-}
+/**
+ * Convert HTML to Markdown, returning a [`ConversionResult`] with content, metadata, images,
+ * and warnings.
+ *
+ * # Arguments
+ *
+ * * `html` — the HTML string to convert.
+ * * `options` — optional conversion options. Defaults to [`ConversionOptions::default`].
+ *
+ * # Example
+ *
+ * ```
+ * use html_to_markdown_rs::convert;
+ *
+ * let html = "<h1>Hello World</h1>";
+ * let result = convert(html, None).unwrap();
+ * assert!(result.content.as_deref().unwrap_or("").contains("Hello World"));
+ * ```
+ *
+ * # Errors
+ *
+ * Returns an error if HTML parsing fails or if the input contains invalid UTF-8.
+ */
+export function convert(html: string, options?: WasmConversionOptions | null): WasmConversionResult;

package/pkg/bundler/html_to_markdown_wasm.js

@@ -0,0 +1,9 @@
+/* @ts-self-types="./html_to_markdown_wasm.d.ts" */
+import * as wasm from "./html_to_markdown_wasm_bg.wasm";
+import { __wbg_set_wasm } from "./html_to_markdown_wasm_bg.js";
+
+__wbg_set_wasm(wasm);
+
+export {
+    WasmAnnotationKind, WasmCodeBlockStyle, WasmConversionOptions, WasmConversionOptionsBuilder, WasmConversionOptionsUpdate, WasmConversionResult, WasmDocumentMetadata, WasmDocumentNode, WasmDocumentStructure, WasmGridCell, WasmHeaderMetadata, WasmHeadingStyle, WasmHighlightStyle, WasmHtmlMetadata, WasmImageMetadata, WasmImageType, WasmLinkMetadata, WasmLinkStyle, WasmLinkType, WasmListIndentType, WasmNewlineStyle, WasmNodeContent, WasmNodeContext, WasmNodeType, WasmOutputFormat, WasmPreprocessingOptions, WasmPreprocessingOptionsUpdate, WasmPreprocessingPreset, WasmProcessingWarning, WasmStructuredData, WasmStructuredDataType, WasmTableData, WasmTableGrid, WasmTextAnnotation, WasmTextDirection, WasmVisitResult, WasmVisitorHandle, WasmWarningKind, WasmWhitespaceMode, convert
+} from "./html_to_markdown_wasm_bg.js";