@minded-ai/mindedjs 1.0.0-ec2-beta-1

package/dist/internalTools/documentExtraction/documentExtraction.d.ts

@@ -0,0 +1,129 @@
+import { ZodSchema } from 'zod';
+import { BaseLanguageModel } from '@langchain/core/language_models/base';
+import { DocumentExtractionOptions, DocumentProcessingResult, DocumentProcessorConfig } from './types';
+/**
+ * Extract data from documents using AI or return raw text.
+ *
+ * This function allows you to process various document types (PDFs, images, Word docs, etc.)
+ * in multiple ways:
+ * - With LLM + Schema: Extract structured data according to a Zod schema
+ * - With LLM + System Prompt: Extract unstructured data based on prompt instructions
+ * - Without LLM: Extract raw text content using LlamaParse
+ *
+ * @param options - Document extraction options
+ * @param options.llm - Optional language model for AI-powered extraction
+ * @param options.documentPath - Path to the document file
+ * @param options.documentContent - Document content as Buffer or string
+ * @param options.documentUrl - URL to fetch the document from
+ * @param options.schema - Optional Zod schema for structured data extraction
+ * @param options.systemPrompt - Optional prompt for guiding extraction
+ * @param options.config - Optional document processor configuration
+ *
+ * @returns Promise resolving to extracted data/text and metadata
+ *
+ * @example
+ * ```typescript
+ * import { extractFromDocument } from '@minded-ai/mindedjs';
+ * import { z } from 'zod';
+ *
+ * // Extract structured data with schema
+ * const result1 = await extractFromDocument({
+ *   llm: agent.llm,
+ *   documentPath: './invoice.pdf',
+ *   schema: z.object({
+ *     invoiceNumber: z.string(),
+ *     amount: z.number(),
+ *   })
+ * });
+ *
+ * // Extract unstructured data with prompt
+ * const result2 = await extractFromDocument({
+ *   llm: agent.llm,
+ *   documentPath: './contract.pdf',
+ *   systemPrompt: 'Extract all payment terms and conditions'
+ * });
+ *
+ * // Extract raw text without LLM
+ * const result3 = await extractFromDocument({
+ *   documentPath: './document.pdf'
+ * });
+ * ```
+ */
+export declare function extractFromDocument<T = string>(options: {
+    llm?: BaseLanguageModel;
+    documentPath?: string;
+    documentContent?: Buffer | string;
+    documentUrl?: string;
+    schema?: ZodSchema<T>;
+    systemPrompt?: string;
+    config?: DocumentProcessorConfig;
+}): Promise<{
+    data: T;
+    metadata: {
+        fileSize?: number;
+        fileType: string;
+        processingTime: number;
+        contentLength: number;
+    };
+}>;
+/**
+ * Generic document processor that can extract structured data from various document types
+ * including images, PDFs, Word documents, spreadsheets, and more.
+ */
+export declare class DocumentProcessor {
+    private config;
+    private llm;
+    private llamaCloudApiKey;
+    private sharpModule;
+    private sharpLoadAttempted;
+    constructor(config?: DocumentProcessorConfig, llm?: BaseLanguageModel);
+    /**
+     * Parse document using LlamaCloud REST API
+     */
+    private parseWithLlamaCloud;
+    /**
+     * Get MIME type for file extension
+     */
+    private getMimeType;
+    /**
+     * Extract data from a document - either structured data using AI or raw text
+     */
+    extractFromDocument<T = string>(options: DocumentExtractionOptions): Promise<DocumentProcessingResult<T>>;
+    /**
+     * Get document content from various sources
+     */
+    private getDocumentContent;
+    /**
+     * Process image documents by converting them to a standardized format
+     */
+    private processImageDocument;
+    /**
+     * Process text-based documents using LlamaParser or fallback methods
+     */
+    private processTextDocument;
+    /**
+     * Extract data using LLM - either structured with schema or unstructured with prompt
+     */
+    private extractStructuredData;
+    /**
+     * Check if file is an image type
+     */
+    private isImageFile;
+    /**
+     * Infer file type from buffer content
+     */
+    private inferFileTypeFromBuffer;
+    /**
+     * Infer file type from URL
+     */
+    private inferFileTypeFromUrl;
+    /**
+     * Get list of supported document types
+     */
+    static getSupportedDocumentTypes(): string[];
+    /**
+     * Check if a file type is supported
+     */
+    static isDocumentTypeSupported(fileType: string): boolean;
+}
+//# sourceMappingURL=documentExtraction.d.ts.map

package/dist/internalTools/documentExtraction/documentExtraction.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"documentExtraction.d.ts","sourceRoot":"","sources":["../../../src/internalTools/documentExtraction/documentExtraction.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAc,MAAM,KAAK,CAAC;AAC5C,OAAO,EAAE,iBAAiB,EAAE,MAAM,sCAAsC,CAAC;AAKzE,OAAO,EAAE,yBAAyB,EAAE,wBAAwB,EAAE,uBAAuB,EAA4B,MAAM,SAAS,CAAC;AAEjI;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+CG;AACH,wBAAsB,mBAAmB,CAAC,CAAC,GAAG,MAAM,EAAE,OAAO,EAAE;IAC7D,GAAG,CAAC,EAAE,iBAAiB,CAAC;IACxB,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,eAAe,CAAC,EAAE,MAAM,GAAG,MAAM,CAAC;IAClC,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,MAAM,CAAC,EAAE,SAAS,CAAC,CAAC,CAAC,CAAC;IACtB,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,MAAM,CAAC,EAAE,uBAAuB,CAAC;CAClC,GAAG,OAAO,CAAC;IACV,IAAI,EAAE,CAAC,CAAC;IACR,QAAQ,EAAE;QACR,QAAQ,CAAC,EAAE,MAAM,CAAC;QAClB,QAAQ,EAAE,MAAM,CAAC;QACjB,cAAc,EAAE,MAAM,CAAC;QACvB,aAAa,EAAE,MAAM,CAAC;KACvB,CAAC;CACH,CAAC,CAYD;AAED;;;GAGG;AACH,qBAAa,iBAAiB;IAC5B,OAAO,CAAC,MAAM,CAA0B;IACxC,OAAO,CAAC,GAAG,CAAkC;IAC7C,OAAO,CAAC,gBAAgB,CAAuB;IAC/C,OAAO,CAAC,WAAW,CAAa;IAChC,OAAO,CAAC,kBAAkB,CAAS;gBAEvB,MAAM,GAAE,uBAA4B,EAAE,GAAG,CAAC,EAAE,iBAAiB;IAYzE;;OAEG;YACW,mBAAmB;IAoNjC;;OAEG;IACH,OAAO,CAAC,WAAW;IA0BnB;;OAEG;IACG,mBAAmB,CAAC,CAAC,GAAG,MAAM,EAAE,OAAO,EAAE,yBAAyB,GAAG,OAAO,CAAC,wBAAwB,CAAC,CAAC,CAAC,CAAC;IAuE/G;;OAEG;YACW,kBAAkB;IAoFhC;;OAEG;YACW,oBAAoB;IAsFlC;;OAEG;YACW,mBAAmB;IAmDjC;;OAEG;YACW,qBAAqB;IAoFnC;;OAEG;IACH,OAAO,CAAC,WAAW;IAKnB;;OAEG;IACH,OAAO,CAAC,uBAAuB;IAsB/B;;OAEG;IACH,OAAO,CAAC,oBAAoB;IAU5B;;OAEG;IACH,MAAM,CAAC,yBAAyB,IAAI,MAAM,EAAE;IAI5C;;OAEG;IACH,MAAM,CAAC,uBAAuB,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO;CAG1D"}