@akira108sys/html-rewriter-readability/dist/types.d.ts

A library to extract readable content with Mozilla/Readability algorithm using Cloudflare HTMLRewriter.

/** Interface for storing HTML element attributes */
export interface ElementAttributes {
    [key: string]: string;
}
/** Interface for element information collected in Phase 1 */
export interface ElementInfo {
    id: number;
    parentId: number | null;
    tagName: string;
    attributes: ElementAttributes;
    textChunks?: string[];
    finalTextContent?: string;
    isVisibleBasedOnAttrs: boolean;
    role: string | null;
    isDataTableLikely: boolean;
    isCodeBlock: boolean;
    readability?: {
        contentScore: number;
    };
}
/** Metadata collected in Phase 1 */
export interface Metadata {
    title?: string;
    byline?: string;
    excerpt?: string;
    siteName?: string;
    publishedTime?: string;
    lang?: string;
    dir?: string;
    jsonLd?: any;
}
/** Formatting options used in Phase 4 */
export interface FormattingOptions {
    debug: boolean;
    allowedVideoRegex?: RegExp;
}
/** Phase 4 formatting options (for Handler) */
export interface Phase4HandlerOptions {
    baseURI: URL;
    keepClasses?: boolean;
    classesToPreserve?: string[];
    formattingOptions?: FormattingOptions;
}
/** Options for HtmlRewriterReadability constructor */
export interface ReadabilityOptions {
    debug?: boolean;
    maxElemsToParse?: number;
    nbTopCandidates?: number;
    charThreshold?: number;
    classesToPreserve?: string[];
    keepClasses?: boolean;
    allowedVideoRegex?: RegExp;
    linkDensityModifier?: number;
}
/** Type for function that returns the next unique element ID */
export type NextElementIdGetter = () => number;