npm - @kreuzberg/html-to-markdown-node - Versions diffs - 2.19.0-rc.1 - Mend

@kreuzberg/html-to-markdown-node 2.19.0-rc.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (5) hide show

package/README.md +350 -0
package/html-to-markdown-node.darwin-arm64.node +0 -0
package/index.d.ts +455 -0
package/index.js +608 -0
package/package.json +83 -0

package/index.d.ts ADDED Viewed

@@ -0,0 +1,455 @@
+/* auto-generated by NAPI-RS */
+/* eslint-disable */
+export declare class ExternalObject<T> {
+  readonly '': {
+    readonly '': unique symbol
+    [K: symbol]: T
+  }
+}
+/**
+ * * `html` - The HTML string to convert
+ * * `options` - Optional conversion options
+ *
+ * # Example
+ *
+ * ```javascript
+ * const { convert } = require('html-to-markdown');
+ *
+ * const html = '<h1>Hello World</h1>';
+ * const markdown = convert(html);
+ * console.log(markdown); // # Hello World
+ * ```
+ */
+export declare function convert(html: string, options?: JsConversionOptions | undefined | null): string
+/** Convert HTML to Markdown from a Buffer/Uint8Array without creating intermediate JS strings. */
+export declare function convertBuffer(html: Buffer, options?: JsConversionOptions | undefined | null): string
+export declare function convertBufferJson(html: Buffer, optionsJson?: string | undefined | null): string
+/** Convert HTML Buffer data using a previously-created ConversionOptions handle. */
+export declare function convertBufferWithOptionsHandle(html: Buffer, options: ExternalObject<RustConversionOptions>): string
+/** Convert inline images from Buffer/Uint8Array input without an intermediate string allocation. */
+export declare function convertInlineImagesBuffer(html: Buffer, options?: JsConversionOptions | undefined | null, imageConfig?: JsInlineImageConfig | undefined | null): JsHtmlExtraction
+export declare function convertInlineImagesBufferJson(html: Buffer, optionsJson?: string | undefined | null, imageConfigJson?: string | undefined | null): JsHtmlExtraction
+/** Convert inline images from Buffer/Uint8Array input using a pre-created options handle. */
+export declare function convertInlineImagesBufferWithOptionsHandle(html: Buffer, options: ExternalObject<RustConversionOptions>, imageConfig?: JsInlineImageConfig | undefined | null): JsHtmlExtraction
+export declare function convertJson(html: string, optionsJson?: string | undefined | null): string
+/**
+ * Convert HTML to Markdown while collecting inline images
+ *
+ * # Arguments
+ *
+ * * `html` - The HTML string to convert
+ * * `options` - Optional conversion options
+ * * `image_config` - Configuration for inline image extraction
+ */
+export declare function convertWithInlineImages(html: string, options?: JsConversionOptions | undefined | null, imageConfig?: JsInlineImageConfig | undefined | null): JsHtmlExtraction
+/** Convert HTML to Markdown while collecting inline images using a pre-created options handle. */
+export declare function convertWithInlineImagesHandle(html: string, options: ExternalObject<RustConversionOptions>, imageConfig?: JsInlineImageConfig | undefined | null): JsHtmlExtraction
+export declare function convertWithInlineImagesJson(html: string, optionsJson?: string | undefined | null, imageConfigJson?: string | undefined | null): JsHtmlExtraction
+/**
+ * Convert HTML to Markdown with metadata extraction.
+ *
+ * # Arguments
+ *
+ * * `html` - The HTML string to convert
+ * * `options` - Optional conversion options
+ * * `metadata_config` - Optional metadata extraction configuration
+ *
+ * # Example
+ *
+ * ```javascript
+ * const { convertWithMetadata } = require('html-to-markdown');
+ *
+ * const html = '<html lang="en"><head><title>Test</title></head><body><h1>Hello</h1></body></html>';
+ * const config = { extractHeaders: true, extractLinks: true };
+ * const result = convertWithMetadata(html, undefined, config);
+ * console.log(result.markdown);
+ * console.log(result.metadata.document.title);
+ * ```
+ */
+export declare function convertWithMetadata(html: string, options?: JsConversionOptions | undefined | null, metadataConfig?: JsMetadataConfig | undefined | null): JsMetadataExtraction
+/** Convert HTML from Buffer/Uint8Array with metadata extraction without intermediate string allocation. */
+export declare function convertWithMetadataBuffer(html: Buffer, options?: JsConversionOptions | undefined | null, metadataConfig?: JsMetadataConfig | undefined | null): JsMetadataExtraction
+/** Convert HTML from Buffer/Uint8Array with metadata extraction using JSON config. */
+export declare function convertWithMetadataBufferJson(html: Buffer, optionsJson?: string | undefined | null, metadataConfigJson?: string | undefined | null): JsMetadataExtraction
+/** Convert HTML from Buffer/Uint8Array with metadata extraction using a metadata handle. */
+export declare function convertWithMetadataBufferWithMetadataHandle(html: Buffer, metadataConfig: ExternalObject<RustMetadataConfig>): JsMetadataExtraction
+/** Convert HTML from Buffer/Uint8Array with metadata extraction using options + metadata handles. */
+export declare function convertWithMetadataBufferWithOptionsAndMetadataHandle(html: Buffer, options: ExternalObject<RustConversionOptions>, metadataConfig: ExternalObject<RustMetadataConfig>): JsMetadataExtraction
+/** Convert HTML from Buffer/Uint8Array with metadata extraction using a pre-created options handle. */
+export declare function convertWithMetadataBufferWithOptionsHandle(html: Buffer, options: ExternalObject<RustConversionOptions>, metadataConfig?: JsMetadataConfig | undefined | null): JsMetadataExtraction
+/** Convert HTML to Markdown with metadata extraction using a pre-created options handle. */
+export declare function convertWithMetadataHandle(html: string, options: ExternalObject<RustConversionOptions>, metadataConfig?: JsMetadataConfig | undefined | null): JsMetadataExtraction
+export declare function convertWithMetadataJson(html: string, optionsJson?: string | undefined | null, metadataConfigJson?: string | undefined | null): JsMetadataExtraction
+/** Convert HTML using a previously-created ConversionOptions handle. */
+export declare function convertWithOptionsHandle(html: string, options: ExternalObject<RustConversionOptions>): string
+/**
+ * Convert HTML to Markdown with an async visitor object.
+ *
+ * # Async Visitor Support
+ *
+ * This function enables full async visitor pattern support for Node.js:
+ * - JavaScript visitor callbacks are invoked asynchronously via NAPI ThreadsafeFunction
+ * - All 30+ visitor methods are supported (links, images, headings, code, lists, tables, etc.)
+ * - Callback errors gracefully default to VisitResult::Continue
+ * - Powered by tokio async runtime for seamless JS-Rust cooperation
+ *
+ * # Visitor Methods
+ *
+ * Implement any combination of these optional async methods in your visitor:
+ * - `visitText(ctx, text) -> { type: string, output?: string }`
+ * - `visitLink(ctx, href, text, title) -> VisitResult`
+ * - `visitImage(ctx, src, alt, title) -> VisitResult`
+ * - `visitHeading(ctx, level, text, id) -> VisitResult`
+ * - `visitCodeBlock(ctx, lang, code) -> VisitResult`
+ * - `visitCodeInline(ctx, code) -> VisitResult`
+ * - `visitListItem(ctx, ordered, marker, text) -> VisitResult`
+ * - `visitTableRow(ctx, cells, isHeader) -> VisitResult`
+ * - `visitBlockquote(ctx, content, depth) -> VisitResult`
+ * - And 20+ more semantic and inline element callbacks
+ *
+ * # VisitResult Types
+ *
+ * Each callback should return an object with:
+ * - `type: 'continue' | 'skip' | 'custom' | 'preservehtml' | 'error'`
+ * - `output?: string` (required for 'custom' and 'error' types)
+ *
+ * # Arguments
+ *
+ * * `html` - The HTML string to convert
+ * * `options` - Optional conversion options
+ * * `visitor` - Visitor object with optional async callback methods
+ *
+ * # Example
+ *
+ * ```javascript
+ * const { convertWithVisitor } = require('@kreuzberg/html-to-markdown-node');
+ *
+ * const html = '<a href="https://example.com">Click me</a>';
+ * const visitor = {
+ *   visitLink: async (ctx, href, text, title) => {
+ *     console.log(`Found link: ${href}`);
+ *     return { type: 'continue' };  // Use default markdown conversion
+ *   }
+ * };
+ *
+ * const markdown = await convertWithVisitor(html, undefined, visitor);
+ * console.log(markdown); // [Click me](https://example.com)
+ * ```
+ */
+export declare function convertWithVisitor(html: string, options: JsConversionOptions | undefined | null, visitor: object): string
+/** Create a reusable ConversionOptions handle. */
+export declare function createConversionOptionsHandle(options?: JsConversionOptions | undefined | null): ExternalObject<RustConversionOptions>
+export declare function createConversionOptionsHandleJson(optionsJson?: string | undefined | null): ExternalObject<RustConversionOptions>
+/** Create a reusable MetadataConfig handle. */
+export declare function createMetadataConfigHandle(metadataConfig?: JsMetadataConfig | undefined | null): ExternalObject<RustMetadataConfig>
+export declare function createMetadataConfigHandleJson(metadataConfigJson?: string | undefined | null): ExternalObject<RustMetadataConfig>
+/** Code block style */
+export declare const enum JsCodeBlockStyle {
+  /** Indented code blocks (4 spaces) - CommonMark default */
+  Indented = 'Indented',
+  /** Fenced code blocks with backticks (```) */
+  Backticks = 'Backticks',
+  /** Fenced code blocks with tildes (~~~) */
+  Tildes = 'Tildes'
+}
+/** Main conversion options */
+export interface JsConversionOptions {
+  /** Heading style */
+  headingStyle?: JsHeadingStyle
+  /** List indentation type */
+  listIndentType?: JsListIndentType
+  /** List indentation width (spaces) */
+  listIndentWidth?: number
+  /** Bullet characters for unordered lists */
+  bullets?: string
+  /** Symbol for strong/emphasis (* or _) */
+  strongEmSymbol?: string
+  /** Escape asterisks in text */
+  escapeAsterisks?: boolean
+  /** Escape underscores in text */
+  escapeUnderscores?: boolean
+  /** Escape misc markdown characters */
+  escapeMisc?: boolean
+  /** Escape all ASCII punctuation */
+  escapeAscii?: boolean
+  /** Default code language */
+  codeLanguage?: string
+  /** Use autolinks for bare URLs */
+  autolinks?: boolean
+  /** Add default title if none exists */
+  defaultTitle?: boolean
+  /** Use <br> in tables instead of spaces */
+  brInTables?: boolean
+  /** Enable spatial table reconstruction in hOCR documents */
+  hocrSpatialTables?: boolean
+  /** Highlight style for <mark> elements */
+  highlightStyle?: JsHighlightStyle
+  /** Extract metadata from HTML */
+  extractMetadata?: boolean
+  /** Whitespace handling mode */
+  whitespaceMode?: JsWhitespaceMode
+  /** Strip newlines from HTML before processing */
+  stripNewlines?: boolean
+  /** Enable text wrapping */
+  wrap?: boolean
+  /** Text wrap width */
+  wrapWidth?: number
+  /** Treat block elements as inline */
+  convertAsInline?: boolean
+  /** Subscript symbol */
+  subSymbol?: string
+  /** Superscript symbol */
+  supSymbol?: string
+  /** Newline style */
+  newlineStyle?: JsNewlineStyle
+  /** Code block style */
+  codeBlockStyle?: JsCodeBlockStyle
+  /** Elements where images should remain as markdown */
+  keepInlineImagesIn?: Array<string>
+  /** Preprocessing options */
+  preprocessing?: JsPreprocessingOptions
+  /** Source encoding (informational) */
+  encoding?: string
+  /** Enable debug mode with diagnostic warnings */
+  debug?: boolean
+  /** List of HTML tags to strip */
+  stripTags?: Array<string>
+  /** List of HTML tags to preserve as-is in the output */
+  preserveTags?: Array<string>
+}
+/** Document-level metadata */
+export interface JsDocumentMetadata {
+  title?: string
+  description?: string
+  keywords: Array<string>
+  author?: string
+  canonical_url?: string
+  base_href?: string
+  language?: string
+  text_direction?: string
+  open_graph: Record<string, string>
+  twitter_card: Record<string, string>
+  meta_tags: Record<string, string>
+}
+/** Complete extracted metadata */
+export interface JsExtendedMetadata {
+  document: JsDocumentMetadata
+  headers: Array<JsHeaderMetadata>
+  links: Array<JsLinkMetadata>
+  images: Array<JsImageMetadata>
+  structuredData: Array<JsStructuredData>
+}
+/** Header element metadata */
+export interface JsHeaderMetadata {
+  level: number
+  text: string
+  id?: string
+  depth: number
+  html_offset: number
+}
+/** Heading style options */
+export declare const enum JsHeadingStyle {
+  /** Underlined style (=== for h1, --- for h2) */
+  Underlined = 'Underlined',
+  /** ATX style (# for h1, ## for h2, etc.) */
+  Atx = 'Atx',
+  /** ATX closed style (# title #) */
+  AtxClosed = 'AtxClosed'
+}
+/** Highlight style for `<mark>` elements */
+export declare const enum JsHighlightStyle {
+  /** ==text== */
+  DoubleEqual = 'DoubleEqual',
+  /** <mark>text</mark> */
+  Html = 'Html',
+  /** **text** */
+  Bold = 'Bold',
+  /** Plain text (no formatting) */
+  None = 'None'
+}
+/** Result of HTML extraction with inline images */
+export interface JsHtmlExtraction {
+  /** Converted markdown */
+  markdown: string
+  /** Extracted inline images */
+  inlineImages: Array<JsInlineImage>
+  /** Warnings encountered during extraction */
+  warnings: Array<JsInlineImageWarning>
+}
+/** Image metadata */
+export interface JsImageMetadata {
+  src: string
+  alt?: string
+  title?: string
+  dimensions?: Array<number>
+  image_type: string
+  attributes: Record<string, string>
+}
+/** Inline image data */
+export interface JsInlineImage {
+  /** Raw image data */
+  data: Buffer
+  /** Image format (png, jpeg, gif, etc.) */
+  format: string
+  /** Generated or provided filename */
+  filename?: string
+  /** Alt text / description */
+  description?: string
+  /** Image dimensions (width, height) if available */
+  dimensions?: Array<number>
+  /** Source type (img_data_uri or svg_element) */
+  source: string
+  /** HTML attributes from the source element */
+  attributes: Record<string, string>
+}
+/** Inline image configuration */
+export interface JsInlineImageConfig {
+  /** Maximum decoded size in bytes (default: 5MB) */
+  maxDecodedSizeBytes?: bigint
+  /** Filename prefix for generated filenames */
+  filenamePrefix?: string
+  /** Capture inline SVG elements (default: true) */
+  captureSvg?: boolean
+  /** Infer image dimensions (default: false) */
+  inferDimensions?: boolean
+}
+/** Warning about inline image processing */
+export interface JsInlineImageWarning {
+  /** Index of the image that caused the warning */
+  index: number
+  /** Warning message */
+  message: string
+}
+/** Hyperlink metadata */
+export interface JsLinkMetadata {
+  href: string
+  text: string
+  title?: string
+  link_type: string
+  rel: Array<string>
+  attributes: Record<string, string>
+}
+/** List indentation type */
+export declare const enum JsListIndentType {
+  Spaces = 'Spaces',
+  Tabs = 'Tabs'
+}
+/** Metadata extraction configuration */
+export interface JsMetadataConfig {
+  extract_document?: boolean
+  extract_headers?: boolean
+  extract_links?: boolean
+  extract_images?: boolean
+  extract_structured_data?: boolean
+  max_structured_data_size?: number
+}
+/** Result of conversion with metadata extraction */
+export interface JsMetadataExtraction {
+  markdown: string
+  metadata: JsExtendedMetadata
+}
+/** Newline style */
+export declare const enum JsNewlineStyle {
+  /** Two spaces at end of line */
+  Spaces = 'Spaces',
+  /** Backslash at end of line */
+  Backslash = 'Backslash'
+}
+/**
+ * Convert HTML to Markdown
+ *
+ * # Arguments
+ */
+export interface JsNodeContext {
+  nodeType: string
+  tagName: string
+  attributes: Record<string, string>
+  depth: number
+  indexInParent: number
+  parentTag?: string
+  isInline: boolean
+}
+/** HTML preprocessing options */
+export interface JsPreprocessingOptions {
+  /** Enable preprocessing */
+  enabled?: boolean
+  /** Preprocessing preset */
+  preset?: JsPreprocessingPreset
+  /** Remove navigation elements */
+  removeNavigation?: boolean
+  /** Remove form elements */
+  removeForms?: boolean
+}
+/** Preprocessing preset levels */
+export declare const enum JsPreprocessingPreset {
+  Minimal = 'Minimal',
+  Standard = 'Standard',
+  Aggressive = 'Aggressive'
+}
+/** Structured data (JSON-LD, Microdata, RDFa) */
+export interface JsStructuredData {
+  data_type: string
+  raw_json: string
+  schema_type?: string
+}
+export interface JsVisitResult {
+  type: string
+  output?: string
+}
+/** Whitespace handling mode */
+export declare const enum JsWhitespaceMode {
+  Normalized = 'Normalized',
+  Strict = 'Strict'
+}
+export declare function startProfiling(outputPath: string, frequency?: number | undefined | null): void
+export declare function stopProfiling(): void