doc-extract/dist/index.d.ts

A Node.js library for reading and extracting text from various document formats (PDF, DOCX, DOC, PPT, PPTX, TXT)

export interface DocumentContent {
    text: string;
    metadata?: {
        pages?: number;
        words?: number;
        characters?: number;
        fileSize?: number;
        fileName?: string;
    };
}
export interface PdfContent extends DocumentContent {
    metadata: DocumentContent['metadata'] & {
        pages: number;
        info?: any;
    };
}
export interface DocxContent extends DocumentContent {
    html?: string;
    messages?: any[];
}
export declare enum SupportedFormats {
    PDF = "pdf",
    DOCX = "docx",
    DOC = "doc",
    PPTX = "pptx",
    PPT = "ppt",
    TXT = "txt"
}
export declare class DocumentReaderError extends Error {
    readonly code?: string | undefined;
    constructor(message: string, code?: string | undefined);
}
export declare class DocumentReader {
    private readonly textractFromFile;
    private debug;
    constructor(options?: {
        debug?: boolean;
    });
    /**
     * Read any supported document format
     */
    readDocument(filePath: string): Promise<DocumentContent>;
    /**
     * Read multiple documents from file paths
     */
    readMultipleDocuments(filePaths: string[]): Promise<DocumentContent[]>;
    /**
     * Read PDF file
     */
    readPdf(filePath: string, fileSize?: number): Promise<PdfContent>;
    /**
     * Read DOCX file
     */
    readDocx(filePath: string, fileSize?: number): Promise<DocxContent>;
    /**
     * Read PPT/PPTX files using textract
     */
    readPowerPoint(filePath: string, fileSize?: number): Promise<DocumentContent>;
    /**
     * Read documents using textract (fallback for various formats)
     */
    private readWithTextract;
    /**
     * Read document from buffer
     */
    readDocumentFromBuffer(buffer: Buffer, fileName: string, mimeType?: string): Promise<DocumentContent>;
    /**
     * Read multiple documents from buffers
     */
    readMultipleFromBuffers(buffers: Array<{
        buffer: Buffer;
        fileName: string;
        mimeType?: string;
    }>): Promise<DocumentContent[]>;
    /**
     * Read PDF from buffer
     */
    private readPdfFromBuffer;
    /**
     * Read DOCX from buffer
     */
    private readDocxFromBuffer;
    /**
     * Read text from buffer
     */
    private readTextFromBuffer;
    /**
     * Read documents from buffer using textract
     */
    private readWithTextractFromBuffer;
    /**
     * Check if file format is supported
     */
    isFormatSupported(filePath: string): boolean;
    /**
     * Check if file format is supported by filename
     */
    isFormatSupportedByName(fileName: string): boolean;
    /**
     * Get supported formats
     */
    getSupportedFormats(): string[];
    /**
     * Validate file exists and is readable
     */
    validateFile(filePath: string): Promise<void>;
    /**
     * Utility methods
     */
    private getFileExtension;
    private getFileExtensionFromName;
    private getExtensionFromMimeType;
    private countWords;
    /**
     * Read text file
     */
    private readTextFile;
    private log;
}
export declare function readDocument(filePath: string): Promise<DocumentContent>;
export declare function readDocumentFromBuffer(buffer: Buffer, fileName: string, mimeType?: string): Promise<DocumentContent>;
export default DocumentReader;