sec-edgar-toolkit/dist/parsers/item-extractor.d.ts

Open source toolkit to facilitate working with the SEC EDGAR database

/**
 * SEC Filing Item Extractor
 *
 * Extracts individual items from SEC filings (10-K, 10-Q, 8-K, etc.)
 * based on standard item numbering and structure.
 */
export declare enum FormType {
    FORM_10K = "10-K",
    FORM_10Q = "10-Q",
    FORM_8K = "8-K",
    FORM_20F = "20-F",
    FORM_40F = "40-F"
}
export interface ItemDefinition {
    number: string;
    title: string;
    aliases?: string[];
    required?: boolean;
}
export interface ExtractedItem {
    itemNumber: string;
    title: string;
    content: string;
    startPosition: number;
    endPosition: number;
}
export declare class ItemExtractor {
    private static readonly FORM_10K_ITEMS;
    private static readonly FORM_10Q_ITEMS;
    private static readonly FORM_8K_ITEMS;
    private readonly formItems;
    constructor();
    /**
     * Extract all items from a filing
     * @param content The filing content (HTML or text)
     * @param formType The type of form (e.g., "10-K", "10-Q", "8-K")
     * @returns Dictionary mapping item numbers to their content
     */
    extractItems(content: string, formType: string | FormType): Record<string, string>;
    /**
     * Extract specific items from a filing
     * @param content The filing content
     * @param formType The type of form
     * @param itemNumbers List of item numbers to extract
     * @returns Dictionary with only the requested items
     */
    extractSpecificItems(content: string, formType: string | FormType, itemNumbers: string[]): Record<string, string>;
    /**
     * Get item definitions for a specific form type
     * @param formType The form type
     * @returns List of item definitions
     */
    getItemDefinitions(formType: string | FormType): ItemDefinition[];
    private parseFormType;
    private cleanContent;
    private extractTableOfContents;
    private extractItemsFromContent;
    private isInToc;
    private findItemEnd;
    private postProcessItems;
    private escapeRegex;
}
//# sourceMappingURL=item-extractor.d.ts.map