export declare const CHARSET_PRESETS: Record<string, string>;
/** A named charset bundling a Tesseract whitelist with expected OCR confusions. */
export interface Charset {
    /** Tesseract character whitelist passed directly to `tessedit_char_whitelist`. */
    chars: string;
    /** Expected glyph → OCR glyphs that are acceptable in its place. */
    swaps?: OcrSwaps;
}
/** Global OCR strategy set by init(). */
export interface OcrStrategy {
    /** Fallback charset when an element has no explicit charset and name-inference finds nothing. */
    defaultCharset?: string;
    /** Map from element-name substring (case-insensitive) → charset name. Checked before built-in heuristics. */
    infer?: Record<string, string>;
    /** Global swaps applied when no element-level swaps are configured. */
    swaps?: OcrSwaps;
    /** Global overflow applied when no element-level overflow is configured. */
    overflow?: OcrOverflow;
    /** Global read mode applied when no element-level read is configured. */
    read?: FieldRead;
}
/**
 * Called by `init()` to register user-defined charsets.
 * Merges into the existing registry — subsequent calls accumulate charsets rather than replacing them.
 * Call `setCharsetRegistry({})` (or `release()`) to clear.
 */
export declare function setCharsetRegistry(registry: Record<string, Charset>): void;
export declare function setOcrStrategy(strategy: OcrStrategy | undefined): void;
export declare function getOcrStrategy(): OcrStrategy | undefined;
/**
 * Resolve a charset using Strategies.Ocr: checks `infer` map first (substring match),
 * then falls back to `defaultCharset`. Returns undefined when no strategy is set.
 */
export declare function resolveOcrCharset(elementName: string): string | undefined;
/** Look up a registered charset by name, or return undefined if not found. */
export declare function lookupCharset(name: string): Charset | undefined;
/**
 * Resolve the bundled swaps for a charset name string or inline Charset object.
 * Returns undefined if the name is not registered or the charset has no swaps.
 */
export declare function resolveCharsetSwaps(charset: string | Charset | undefined): OcrSwaps | undefined;
export declare function charsetForField(name?: string, type?: string, preset?: string): string | undefined;
export declare function normalizeOcrText(text: string): string;
export declare function pickFromOptions(text: string, options?: string[]): string;
/** Expected glyph → OCR glyphs allowed in its place. `{ '@': ['Q', 'C'], '5': 'S' }` */
export type OcrSwaps = Record<string, string | readonly string[]>;
/** How a field is allowed to clip when the value does not fit the box. */
export type OcrOverflow = 'start' | 'end' | 'both';
/** How to read a field's value. `clipboard` is click / select-all / copy. */
export type FieldRead = 'ocr' | 'clipboard';
export declare function ocrTextMatches(actual: string, expected: string | RegExp, options?: {
    swaps?: OcrSwaps;
    exact?: boolean;
    overflow?: OcrOverflow;
    overflowSlop?: number;
}): boolean;
/**
 * OCR utility for extracting text from images using Tesseract.js
 */
export declare class OCRUtil {
    private worker;
    private language;
    initialize(language?: string): Promise<void>;
    private ocrParams;
    private ensureWorker;
    /**
     * Extract text from an image buffer
     */
    extractText(imageBuffer: Buffer, options?: {
        charset?: string;
        psm?: string;
    }): Promise<string>;
    /**
     * Extract text with confidence scores and bounding boxes
     */
    extractDetailedText(imageBuffer: Buffer): Promise<{
        text: string;
        confidence: number;
        words: any;
        lines: any;
    }>;
    /**
     * Full-page word boxes for authoring (labels next to controls).
     * Sparse text mode, no charset whitelist.
     */
    extractPageWords(imageBuffer: Buffer): Promise<Array<{
        text: string;
        confidence: number;
        x: number;
        y: number;
        width: number;
        height: number;
    }>>;
    terminate(): Promise<void>;
}
/**
 * Get or create a shared OCR utility instance
 */
export declare function getOCRUtil(language?: string): Promise<OCRUtil>;
/**
 * Clean up the shared OCR utility
 */
export declare function cleanupOCR(): Promise<void>;
//# sourceMappingURL=ocr.d.ts.map