@kreuzberg/html-to-markdown-wasm 3.1.0 → 3.2.0

package/dist/html_to_markdown_wasm.d.ts

@@ -1,40 +1,488 @@
 /* tslint:disable */
 /* eslint-disable */
 
-/**
- * Convert HTML to Markdown, returning a JavaScript object with structured content, metadata,
- * images, and warnings in a single pass.
- *
- * This is the primary API entry point. Returns a JavaScript object with:
- * - `content`: converted text (string or null)
- * - `document`: structured document tree (object or null)
- * - `metadata`: extracted HTML metadata (object or null)
- * - `tables`: array of extracted table data
- * - `warnings`: array of non-fatal processing warnings
- *
- * # Arguments
- *
- * * `html` - The HTML string to convert
- * * `options` - Optional conversion options (as a JavaScript object)
- *
- * # Example
- *
- * ```javascript
- * import { convert } from 'html-to-markdown-wasm';
- *
- * const html = '<h1>Hello World</h1><p>Some text.</p>';
- * const result = convert(html, null);
- * console.log(result.content);   // '# Hello World\n\nSome text.'
- * console.log(result.tables);    // []
- * console.log(result.warnings);  // []
- * ```
- */
-export function convert(html: string, options?: WasmConversionOptions | null): WasmConversionResult;
-
-/**
- * Initialize panic hook for better error messages in the browser
- */
-export function init(): void;
+export enum JsAnnotationKind {
+    Bold = 0,
+    Italic = 1,
+    Underline = 2,
+    Strikethrough = 3,
+    Code = 4,
+    Subscript = 5,
+    Superscript = 6,
+    Highlight = 7,
+    Link = 8,
+}
+
+export enum JsCodeBlockStyle {
+    Indented = 0,
+    Backticks = 1,
+    Tildes = 2,
+}
+
+export class JsConversionOptions {
+    free(): void;
+    [Symbol.dispose](): void;
+    static builder(): JsConversionOptionsBuilder;
+    static default(): JsConversionOptions;
+    constructor(heading_style?: JsHeadingStyle | null, list_indent_type?: JsListIndentType | 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?: JsHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: JsWhitespaceMode | 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?: JsNewlineStyle | null, code_block_style?: JsCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: JsPreprocessingOptions | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: JsLinkStyle | null, output_format?: JsOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null);
+    autolinks: boolean;
+    brInTables: boolean;
+    bullets: string;
+    captureSvg: boolean;
+    codeBlockStyle: JsCodeBlockStyle;
+    codeLanguage: string;
+    convertAsInline: boolean;
+    debug: boolean;
+    defaultTitle: boolean;
+    encoding: string;
+    escapeAscii: boolean;
+    escapeAsterisks: boolean;
+    escapeMisc: boolean;
+    escapeUnderscores: boolean;
+    extractImages: boolean;
+    extractMetadata: boolean;
+    headingStyle: JsHeadingStyle;
+    highlightStyle: JsHighlightStyle;
+    includeDocumentStructure: boolean;
+    inferDimensions: boolean;
+    keepInlineImagesIn: string[];
+    linkStyle: JsLinkStyle;
+    listIndentType: JsListIndentType;
+    listIndentWidth: number;
+    maxImageSize: bigint;
+    newlineStyle: JsNewlineStyle;
+    outputFormat: JsOutputFormat;
+    preprocessing: JsPreprocessingOptions;
+    preserveTags: string[];
+    skipImages: boolean;
+    stripNewlines: boolean;
+    stripTags: string[];
+    strongEmSymbol: string;
+    subSymbol: string;
+    supSymbol: string;
+    whitespaceMode: JsWhitespaceMode;
+    wrap: boolean;
+    wrapWidth: number;
+}
+
+export class JsConversionOptionsBuilder {
+    private constructor();
+    free(): void;
+    [Symbol.dispose](): void;
+    build(): JsConversionOptions;
+    keepInlineImagesIn(tags: string[]): JsConversionOptionsBuilder;
+    preprocessing(preprocessing: JsPreprocessingOptions): JsConversionOptionsBuilder;
+    preserveTags(tags: string[]): JsConversionOptionsBuilder;
+    stripTags(tags: string[]): JsConversionOptionsBuilder;
+}
+
+export class JsConversionOptionsUpdate {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(heading_style?: JsHeadingStyle | null, list_indent_type?: JsListIndentType | 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?: JsHighlightStyle | null, extract_metadata?: boolean | null, whitespace_mode?: JsWhitespaceMode | 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?: JsNewlineStyle | null, code_block_style?: JsCodeBlockStyle | null, keep_inline_images_in?: string[] | null, preprocessing?: JsPreprocessingOptionsUpdate | null, encoding?: string | null, debug?: boolean | null, strip_tags?: string[] | null, preserve_tags?: string[] | null, skip_images?: boolean | null, link_style?: JsLinkStyle | null, output_format?: JsOutputFormat | null, include_document_structure?: boolean | null, extract_images?: boolean | null, max_image_size?: bigint | null, capture_svg?: boolean | null, infer_dimensions?: boolean | null);
+    get autolinks(): boolean | undefined;
+    set autolinks(value: boolean | null | undefined);
+    get brInTables(): boolean | undefined;
+    set brInTables(value: boolean | null | undefined);
+    get bullets(): string | undefined;
+    set bullets(value: string | null | undefined);
+    get captureSvg(): boolean | undefined;
+    set captureSvg(value: boolean | null | undefined);
+    get codeBlockStyle(): JsCodeBlockStyle | undefined;
+    set codeBlockStyle(value: JsCodeBlockStyle | null | undefined);
+    get codeLanguage(): string | undefined;
+    set codeLanguage(value: string | null | undefined);
+    get convertAsInline(): boolean | undefined;
+    set convertAsInline(value: boolean | null | undefined);
+    get debug(): boolean | undefined;
+    set debug(value: boolean | null | undefined);
+    get defaultTitle(): boolean | undefined;
+    set defaultTitle(value: boolean | null | undefined);
+    get encoding(): string | undefined;
+    set encoding(value: string | null | undefined);
+    get escapeAscii(): boolean | undefined;
+    set escapeAscii(value: boolean | null | undefined);
+    get escapeAsterisks(): boolean | undefined;
+    set escapeAsterisks(value: boolean | null | undefined);
+    get escapeMisc(): boolean | undefined;
+    set escapeMisc(value: boolean | null | undefined);
+    get escapeUnderscores(): boolean | undefined;
+    set escapeUnderscores(value: boolean | null | undefined);
+    get extractImages(): boolean | undefined;
+    set extractImages(value: boolean | null | undefined);
+    get extractMetadata(): boolean | undefined;
+    set extractMetadata(value: boolean | null | undefined);
+    get headingStyle(): JsHeadingStyle | undefined;
+    set headingStyle(value: JsHeadingStyle | null | undefined);
+    get highlightStyle(): JsHighlightStyle | undefined;
+    set highlightStyle(value: JsHighlightStyle | null | undefined);
+    get includeDocumentStructure(): boolean | undefined;
+    set includeDocumentStructure(value: boolean | null | undefined);
+    get inferDimensions(): boolean | undefined;
+    set inferDimensions(value: boolean | null | undefined);
+    get keepInlineImagesIn(): string[] | undefined;
+    set keepInlineImagesIn(value: string[] | null | undefined);
+    get linkStyle(): JsLinkStyle | undefined;
+    set linkStyle(value: JsLinkStyle | null | undefined);
+    get listIndentType(): JsListIndentType | undefined;
+    set listIndentType(value: JsListIndentType | null | undefined);
+    get listIndentWidth(): number | undefined;
+    set listIndentWidth(value: number | null | undefined);
+    get maxImageSize(): bigint | undefined;
+    set maxImageSize(value: bigint | null | undefined);
+    get newlineStyle(): JsNewlineStyle | undefined;
+    set newlineStyle(value: JsNewlineStyle | null | undefined);
+    get outputFormat(): JsOutputFormat | undefined;
+    set outputFormat(value: JsOutputFormat | null | undefined);
+    get preprocessing(): JsPreprocessingOptionsUpdate | undefined;
+    set preprocessing(value: JsPreprocessingOptionsUpdate | null | undefined);
+    get preserveTags(): string[] | undefined;
+    set preserveTags(value: string[] | null | undefined);
+    get skipImages(): boolean | undefined;
+    set skipImages(value: boolean | null | undefined);
+    get stripNewlines(): boolean | undefined;
+    set stripNewlines(value: boolean | null | undefined);
+    get stripTags(): string[] | undefined;
+    set stripTags(value: string[] | null | undefined);
+    get strongEmSymbol(): string | undefined;
+    set strongEmSymbol(value: string | null | undefined);
+    get subSymbol(): string | undefined;
+    set subSymbol(value: string | null | undefined);
+    get supSymbol(): string | undefined;
+    set supSymbol(value: string | null | undefined);
+    get whitespaceMode(): JsWhitespaceMode | undefined;
+    set whitespaceMode(value: JsWhitespaceMode | null | undefined);
+    get wrap(): boolean | undefined;
+    set wrap(value: boolean | null | undefined);
+    get wrapWidth(): number | undefined;
+    set wrapWidth(value: number | null | undefined);
+}
+
+export class JsConversionResult {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(metadata?: JsHtmlMetadata | null, tables?: JsTableData[] | null, images?: string[] | null, warnings?: JsProcessingWarning[] | null, content?: string | null, document?: JsDocumentStructure | null);
+    get content(): string | undefined;
+    set content(value: string | null | undefined);
+    get document(): JsDocumentStructure | undefined;
+    set document(value: JsDocumentStructure | null | undefined);
+    images: string[];
+    metadata: JsHtmlMetadata;
+    tables: JsTableData[];
+    warnings: JsProcessingWarning[];
+}
+
+export class JsDocumentMetadata {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(keywords?: string[] | null, open_graph?: any | null, twitter_card?: any | null, meta_tags?: any | null, title?: string | null, description?: string | null, author?: string | null, canonical_url?: string | null, base_href?: string | null, language?: string | null, text_direction?: JsTextDirection | null);
+    get author(): string | undefined;
+    set author(value: string | null | undefined);
+    get baseHref(): string | undefined;
+    set baseHref(value: string | null | undefined);
+    get canonicalUrl(): string | undefined;
+    set canonicalUrl(value: string | null | undefined);
+    get description(): string | undefined;
+    set description(value: string | null | undefined);
+    keywords: string[];
+    get language(): string | undefined;
+    set language(value: string | null | undefined);
+    metaTags: any;
+    openGraph: any;
+    get textDirection(): JsTextDirection | undefined;
+    set textDirection(value: JsTextDirection | null | undefined);
+    get title(): string | undefined;
+    set title(value: string | null | undefined);
+    twitterCard: any;
+}
+
+export class JsDocumentNode {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(id: string, content: JsNodeContent, children: Uint32Array, annotations: JsTextAnnotation[], parent?: number | null, attributes?: any | null);
+    annotations: JsTextAnnotation[];
+    get attributes(): any | undefined;
+    set attributes(value: any | null | undefined);
+    children: Uint32Array;
+    content: JsNodeContent;
+    id: string;
+    get parent(): number | undefined;
+    set parent(value: number | null | undefined);
+}
+
+export class JsDocumentStructure {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(nodes: JsDocumentNode[], source_format?: string | null);
+    nodes: JsDocumentNode[];
+    get sourceFormat(): string | undefined;
+    set sourceFormat(value: string | null | undefined);
+}
+
+export class JsGridCell {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(content: string, row: number, col: number, row_span: number, col_span: number, is_header: boolean);
+    col: number;
+    colSpan: number;
+    content: string;
+    isHeader: boolean;
+    row: number;
+    rowSpan: number;
+}
+
+export class JsHeaderMetadata {
+    free(): void;
+    [Symbol.dispose](): void;
+    isValid(): boolean;
+    constructor(level: number, text: string, depth: number, html_offset: number, id?: string | null);
+    depth: number;
+    htmlOffset: number;
+    get id(): string | undefined;
+    set id(value: string | null | undefined);
+    level: number;
+    text: string;
+}
+
+export enum JsHeadingStyle {
+    Underlined = 0,
+    Atx = 1,
+    AtxClosed = 2,
+}
+
+export enum JsHighlightStyle {
+    DoubleEqual = 0,
+    Html = 1,
+    Bold = 2,
+    None = 3,
+}
+
+export class JsHtmlMetadata {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(document?: JsDocumentMetadata | null, headers?: JsHeaderMetadata[] | null, links?: JsLinkMetadata[] | null, images?: JsImageMetadata[] | null, structured_data?: JsStructuredData[] | null);
+    document: JsDocumentMetadata;
+    headers: JsHeaderMetadata[];
+    images: JsImageMetadata[];
+    links: JsLinkMetadata[];
+    structuredData: JsStructuredData[];
+}
+
+export class JsImageMetadata {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(src: string, image_type: JsImageType, attributes: any, alt?: string | null, title?: string | null, dimensions?: string | null);
+    get alt(): string | undefined;
+    set alt(value: string | null | undefined);
+    attributes: any;
+    get dimensions(): string | undefined;
+    set dimensions(value: string | null | undefined);
+    imageType: JsImageType;
+    src: string;
+    get title(): string | undefined;
+    set title(value: string | null | undefined);
+}
+
+export enum JsImageType {
+    DataUri = 0,
+    InlineSvg = 1,
+    External = 2,
+    Relative = 3,
+}
+
+export class JsLinkMetadata {
+    free(): void;
+    [Symbol.dispose](): void;
+    static classifyLink(href: string): JsLinkType;
+    constructor(href: string, text: string, link_type: JsLinkType, rel: string[], attributes: any, title?: string | null);
+    attributes: any;
+    href: string;
+    linkType: JsLinkType;
+    rel: string[];
+    text: string;
+    get title(): string | undefined;
+    set title(value: string | null | undefined);
+}
+
+export enum JsLinkStyle {
+    Inline = 0,
+    Reference = 1,
+}
+
+export enum JsLinkType {
+    Anchor = 0,
+    Internal = 1,
+    External = 2,
+    Email = 3,
+    Phone = 4,
+    Other = 5,
+}
+
+export enum JsListIndentType {
+    Spaces = 0,
+    Tabs = 1,
+}
+
+export class JsMetadataConfig {
+    free(): void;
+    [Symbol.dispose](): void;
+    anyEnabled(): boolean;
+    static default(): JsMetadataConfig;
+    constructor(extract_document?: boolean | null, extract_headers?: boolean | null, extract_links?: boolean | null, extract_images?: boolean | null, extract_structured_data?: boolean | null, max_structured_data_size?: number | null);
+    extractDocument: boolean;
+    extractHeaders: boolean;
+    extractImages: boolean;
+    extractLinks: boolean;
+    extractStructuredData: boolean;
+    maxStructuredDataSize: number;
+}
+
+export class JsMetadataConfigUpdate {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(extract_document?: boolean | null, extract_headers?: boolean | null, extract_links?: boolean | null, extract_images?: boolean | null, extract_structured_data?: boolean | null, max_structured_data_size?: number | null);
+    get extractDocument(): boolean | undefined;
+    set extractDocument(value: boolean | null | undefined);
+    get extractHeaders(): boolean | undefined;
+    set extractHeaders(value: boolean | null | undefined);
+    get extractImages(): boolean | undefined;
+    set extractImages(value: boolean | null | undefined);
+    get extractLinks(): boolean | undefined;
+    set extractLinks(value: boolean | null | undefined);
+    get extractStructuredData(): boolean | undefined;
+    set extractStructuredData(value: boolean | null | undefined);
+    get maxStructuredDataSize(): number | undefined;
+    set maxStructuredDataSize(value: number | null | undefined);
+}
+
+export enum JsNewlineStyle {
+    Spaces = 0,
+    Backslash = 1,
+}
+
+export enum JsNodeContent {
+    Heading = 0,
+    Paragraph = 1,
+    List = 2,
+    ListItem = 3,
+    Table = 4,
+    Image = 5,
+    Code = 6,
+    Quote = 7,
+    DefinitionList = 8,
+    DefinitionItem = 9,
+    RawBlock = 10,
+    MetadataBlock = 11,
+    Group = 12,
+}
+
+export enum JsOutputFormat {
+    Markdown = 0,
+    Djot = 1,
+    Plain = 2,
+}
+
+export class JsPreprocessingOptions {
+    free(): void;
+    [Symbol.dispose](): void;
+    static default(): JsPreprocessingOptions;
+    constructor(enabled?: boolean | null, preset?: JsPreprocessingPreset | null, remove_navigation?: boolean | null, remove_forms?: boolean | null);
+    enabled: boolean;
+    preset: JsPreprocessingPreset;
+    removeForms: boolean;
+    removeNavigation: boolean;
+}
+
+export class JsPreprocessingOptionsUpdate {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(enabled?: boolean | null, preset?: JsPreprocessingPreset | null, remove_navigation?: boolean | null, remove_forms?: boolean | null);
+    get enabled(): boolean | undefined;
+    set enabled(value: boolean | null | undefined);
+    get preset(): JsPreprocessingPreset | undefined;
+    set preset(value: JsPreprocessingPreset | null | undefined);
+    get removeForms(): boolean | undefined;
+    set removeForms(value: boolean | null | undefined);
+    get removeNavigation(): boolean | undefined;
+    set removeNavigation(value: boolean | null | undefined);
+}
+
+export enum JsPreprocessingPreset {
+    Minimal = 0,
+    Standard = 1,
+    Aggressive = 2,
+}
+
+export class JsProcessingWarning {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(message: string, kind: JsWarningKind);
+    kind: JsWarningKind;
+    message: string;
+}
+
+export class JsStructuredData {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(data_type: JsStructuredDataType, raw_json: string, schema_type?: string | null);
+    dataType: JsStructuredDataType;
+    rawJson: string;
+    get schemaType(): string | undefined;
+    set schemaType(value: string | null | undefined);
+}
+
+export enum JsStructuredDataType {
+    JsonLd = 0,
+    Microdata = 1,
+    RDFa = 2,
+}
+
+export class JsTableData {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(grid: JsTableGrid, markdown: string);
+    grid: JsTableGrid;
+    markdown: string;
+}
+
+export class JsTableGrid {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(rows?: number | null, cols?: number | null, cells?: JsGridCell[] | null);
+    cells: JsGridCell[];
+    cols: number;
+    rows: number;
+}
+
+export class JsTextAnnotation {
+    free(): void;
+    [Symbol.dispose](): void;
+    constructor(start: number, end: number, kind: JsAnnotationKind);
+    end: number;
+    kind: JsAnnotationKind;
+    start: number;
+}
+
+export enum JsTextDirection {
+    LeftToRight = 0,
+    RightToLeft = 1,
+    Auto = 2,
+}
+
+export enum JsWarningKind {
+    ImageExtractionFailed = 0,
+    EncodingFallback = 1,
+    TruncatedInput = 2,
+    MalformedHtml = 3,
+    SanitizationApplied = 4,
+}
+
+export enum JsWhitespaceMode {
+    Normalized = 0,
+    Strict = 1,
+}
+
+export function convert(html: string, options?: JsConversionOptions | null): JsConversionResult;
 
 export declare function initWasm(): Promise<void>;
 export declare const wasmReady: Promise<void>;