shrink-dom/dist/types.d.ts

适用于网页分析、内容提取、AI训练数据准备和网页爬虫等场景，帮助开发者更高效地处理和优化DOM结构。

/**
 * DOM节点类型常量
 */
export declare enum NodeTypeEnum {
    ELEMENT_NODE = 1,
    TEXT_NODE = 3,
    DOCUMENT_NODE = 9
}
/**
 * 节点的JSON表示
 */
export interface JsonNode {
    type: 'element' | 'text' | 'template';
    path?: string;
    tag?: string;
    attrs?: Record<string, string>;
    children?: JsonNode[];
    text?: string;
    templateHash?: string;
    templateId?: string;
    params?: number[];
}
/**
 * 潜在模板
 */
export interface PossibleTemplate {
    hash: string;
    structure: JsonNode;
    occurrences: JsonNode[];
    depth: number;
    path?: string;
}
/**
 * 优化后的模板
 */
export interface OptimizedTemplate {
    structure: JsonNode;
    inlineValues: Record<string, string | null>;
    occurrences: JsonNode[];
    depth: number;
    hash: string;
    path?: string;
}
/**
 * 选择的模板
 */
export interface ChosenTemplate extends OptimizedTemplate {
    id: string;
    replacements: Map<JsonNode, number[]>;
}
export interface DOMShrinkerOptions {
    minTemplateDepth?: number;
    minTemplateOccurrences?: number;
    templateIdPrefix?: string;
    semanticAttributes?: string[];
    useHeuristicRules?: boolean;
    uiPatterns?: {
        forms?: boolean;
        navigation?: boolean;
        cards?: boolean;
        tables?: boolean;
        custom?: Record<string, string[]>;
    };
    semanticPreservationLevel?: 'low' | 'medium' | 'high';
    preserveDataAttributes?: boolean;
    preserveAriaAttributes?: boolean;
    preserveRoles?: boolean;
    criticalAttributes?: string[];
}
/**
 * ContentExtractor 的配置选项
 */
export interface ExtractorOptions {
    skipKeywords?: string[];
    preservedAttributes?: string[];
    formElementTags?: string[];
    emphasisTags?: string[];
    briefTextThreshold?: number;
}