@retab/node 1.0.73 → 1.0.75

package/dist/api/documents/client.d.ts

@@ -1,5 +1,6 @@
 import { CompositionClient, RequestOptions } from "../../client.js";
-import { DocumentExtractRequest, RetabParsedChatCompletion, ParseRequest, ParseResult, DocumentCreateMessageRequest, DocumentMessage, DocumentCreateInputRequest, RetabParsedChatCompletionChunk, EditRequest, EditResponse } from "../../types.js";
+import { DocumentExtractRequest, RetabParsedChatCompletion, ParseRequest, ParseResult, DocumentCreateMessageRequest, DocumentMessage, DocumentCreateInputRequest, RetabParsedChatCompletionChunk, EditRequest, EditResponse, SplitRequest } from "../../types.js";
+import type { SplitResponse } from "../../generated_types.js";
 export default class APIDocuments extends CompositionClient {
     constructor(client: CompositionClient);
     extract(params: DocumentExtractRequest, options?: RequestOptions): Promise<RetabParsedChatCompletion>;
@@ -14,13 +15,48 @@ export default class APIDocuments extends CompositionClient {
      * and filling. It accepts any PDF document and natural language instructions
      * describing the values to fill in.
      *
+     * Either `document` OR `template_id` must be provided, but not both.
+     * - When `document` is provided: OCR + LLM inference to detect and fill form fields
+     * - When `template_id` is provided: Uses pre-defined form fields from the template (PDF only)
+     *
      * @param params - EditRequest containing:
-     *   - document: MIMEData object with filename and url (data URI with base64 content)
-     *   - filling_instructions: Natural language instructions for filling
-     *   - model: LLM model for inference (default: "gemini-2.5-pro")
+     *   - filling_instructions: Natural language instructions for filling (required)
+     *   - document: MIMEData object, file path, Buffer, or Readable stream (optional, mutually exclusive with template_id)
+     *   - model: LLM model for inference (default: "retab-small")
+     *   - template_id: Template ID to use for filling with pre-defined form fields (optional, mutually exclusive with document)
      * @param options - Optional request options
-     * @returns EditResponse containing ocr_result, form_data (filled fields), and filled_document (MIMEData)
+     * @returns EditResponse containing form_data (filled fields) and filled_document (MIMEData)
      */
     edit(params: EditRequest, options?: RequestOptions): Promise<EditResponse>;
+    /**
+     * Split a document into sections based on provided categories.
+     *
+     * This method analyzes a multi-page document and classifies pages into
+     * user-defined categories, returning the page ranges for each section.
+     *
+     * @param params - SplitRequest containing:
+     *   - document: MIMEData object, file path, Buffer, or Readable stream
+     *   - categories: Array of categories with 'name' and 'description'
+     *   - model: LLM model for inference (e.g., "retab-small")
+     * @param options - Optional request options
+     * @returns SplitResponse containing splits array with name, start_page, and end_page for each section
+     *
+     * @example
+     * ```typescript
+     * const response = await retab.documents.split({
+     *     document: "invoice_batch.pdf",
+     *     model: "retab-small",
+     *     categories: [
+     *         { name: "invoice", description: "Invoice documents with billing information" },
+     *         { name: "receipt", description: "Receipt documents for payments" },
+     *         { name: "contract", description: "Legal contract documents" },
+     *     ]
+     * });
+     * for (const split of response.splits) {
+     *     console.log(`${split.name}: pages ${split.start_page}-${split.end_page}`);
+     * }
+     * ```
+     */
+    split(params: SplitRequest, options?: RequestOptions): Promise<SplitResponse>;
 }
 //# sourceMappingURL=client.d.ts.map

package/dist/api/documents/client.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/api/documents/client.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAC;AACpE,OAAO,EAA2B,sBAAsB,EAAE,yBAAyB,EAA8B,YAAY,EAAE,WAAW,EAA+B,4BAA4B,EAAE,eAAe,EAAmD,0BAA0B,EAA+B,8BAA8B,EAAmC,WAAW,EAAE,YAAY,EAA+B,MAAM,gBAAgB,CAAC;AAGld,MAAM,CAAC,OAAO,OAAO,YAAa,SAAQ,iBAAiB;gBAC3C,MAAM,EAAE,iBAAiB;IAG/B,OAAO,CAAC,MAAM,EAAE,sBAAsB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,yBAAyB,CAAC;IAUrG,cAAc,CAAC,MAAM,EAAE,sBAAsB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,cAAc,CAAC,8BAA8B,CAAC,CAAC;IAUjI,KAAK,CAAC,MAAM,EAAE,YAAY,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,WAAW,CAAC;IAS3E,eAAe,CAAC,MAAM,EAAE,4BAA4B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,eAAe,CAAC;IASzG,aAAa,CAAC,MAAM,EAAE,0BAA0B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,eAAe,CAAC;IAS3G;;;;;;;;;;;;;OAaG;IACG,IAAI,CAAC,MAAM,EAAE,WAAW,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,YAAY,CAAC;CASnF"}
+{"version":3,"file":"client.d.ts","sourceRoot":"","sources":["../../../src/api/documents/client.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,iBAAiB,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAC;AACpE,OAAO,EAA2B,sBAAsB,EAAE,yBAAyB,EAA8B,YAAY,EAAE,WAAW,EAA+B,4BAA4B,EAAE,eAAe,EAAmD,0BAA0B,EAA+B,8BAA8B,EAAmC,WAAW,EAAE,YAAY,EAA+B,YAAY,EAAiC,MAAM,gBAAgB,CAAC;AAC/f,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,0BAA0B,CAAC;AAG9D,MAAM,CAAC,OAAO,OAAO,YAAa,SAAQ,iBAAiB;gBAC3C,MAAM,EAAE,iBAAiB;IAG/B,OAAO,CAAC,MAAM,EAAE,sBAAsB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,yBAAyB,CAAC;IAUrG,cAAc,CAAC,MAAM,EAAE,sBAAsB,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,cAAc,CAAC,8BAA8B,CAAC,CAAC;IAUjI,KAAK,CAAC,MAAM,EAAE,YAAY,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,WAAW,CAAC;IAS3E,eAAe,CAAC,MAAM,EAAE,4BAA4B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,eAAe,CAAC;IASzG,aAAa,CAAC,MAAM,EAAE,0BAA0B,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,eAAe,CAAC;IAS3G;;;;;;;;;;;;;;;;;;OAkBG;IACG,IAAI,CAAC,MAAM,EAAE,WAAW,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,YAAY,CAAC;IAShF;;;;;;;;;;;;;;;;;;;;;;;;;;;;OA4BG;IACG,KAAK,CAAC,MAAM,EAAE,YAAY,EAAE,OAAO,CAAC,EAAE,cAAc,GAAG,OAAO,CAAC,aAAa,CAAC;CAStF"}

package/dist/api/documents/client.js

@@ -1,5 +1,5 @@
 import { CompositionClient } from "../../client.js";
-import { ZDocumentExtractRequest, ZRetabParsedChatCompletion, ZParseResult, ZParseRequest, ZDocumentMessage, ZDocumentCreateMessageRequest, ZDocumentCreateInputRequest, ZRetabParsedChatCompletionChunk, ZEditRequest, ZEditResponse } from "../../types.js";
+import { ZDocumentExtractRequest, ZRetabParsedChatCompletion, ZParseResult, ZParseRequest, ZDocumentMessage, ZDocumentCreateMessageRequest, ZDocumentCreateInputRequest, ZRetabParsedChatCompletionChunk, ZEditRequest, ZEditResponse, ZSplitRequest, ZSplitResponse } from "../../types.js";
 export default class APIDocuments extends CompositionClient {
     constructor(client) {
         super(client);
@@ -58,12 +58,17 @@ export default class APIDocuments extends CompositionClient {
      * and filling. It accepts any PDF document and natural language instructions
      * describing the values to fill in.
      *
+     * Either `document` OR `template_id` must be provided, but not both.
+     * - When `document` is provided: OCR + LLM inference to detect and fill form fields
+     * - When `template_id` is provided: Uses pre-defined form fields from the template (PDF only)
+     *
      * @param params - EditRequest containing:
-     *   - document: MIMEData object with filename and url (data URI with base64 content)
-     *   - filling_instructions: Natural language instructions for filling
-     *   - model: LLM model for inference (default: "gemini-2.5-pro")
+     *   - filling_instructions: Natural language instructions for filling (required)
+     *   - document: MIMEData object, file path, Buffer, or Readable stream (optional, mutually exclusive with template_id)
+     *   - model: LLM model for inference (default: "retab-small")
+     *   - template_id: Template ID to use for filling with pre-defined form fields (optional, mutually exclusive with document)
      * @param options - Optional request options
-     * @returns EditResponse containing ocr_result, form_data (filled fields), and filled_document (MIMEData)
+     * @returns EditResponse containing form_data (filled fields) and filled_document (MIMEData)
      */
     async edit(params, options) {
         return this._fetchJson(ZEditResponse, {
@@ -74,4 +79,42 @@ export default class APIDocuments extends CompositionClient {
             headers: options?.headers,
         });
     }
+    /**
+     * Split a document into sections based on provided categories.
+     *
+     * This method analyzes a multi-page document and classifies pages into
+     * user-defined categories, returning the page ranges for each section.
+     *
+     * @param params - SplitRequest containing:
+     *   - document: MIMEData object, file path, Buffer, or Readable stream
+     *   - categories: Array of categories with 'name' and 'description'
+     *   - model: LLM model for inference (e.g., "retab-small")
+     * @param options - Optional request options
+     * @returns SplitResponse containing splits array with name, start_page, and end_page for each section
+     *
+     * @example
+     * ```typescript
+     * const response = await retab.documents.split({
+     *     document: "invoice_batch.pdf",
+     *     model: "retab-small",
+     *     categories: [
+     *         { name: "invoice", description: "Invoice documents with billing information" },
+     *         { name: "receipt", description: "Receipt documents for payments" },
+     *         { name: "contract", description: "Legal contract documents" },
+     *     ]
+     * });
+     * for (const split of response.splits) {
+     *     console.log(`${split.name}: pages ${split.start_page}-${split.end_page}`);
+     * }
+     * ```
+     */
+    async split(params, options) {
+        return this._fetchJson(ZSplitResponse, {
+            url: "/v1/documents/split",
+            method: "POST",
+            body: { ...(await ZSplitRequest.parseAsync(params)), ...(options?.body || {}) },
+            params: options?.params,
+            headers: options?.headers,
+        });
+    }
 }