test ocr

Author: ianardee

src/v2/product/index.ts

@@ -1,5 +1,14 @@
 export { Classification, ClassificationResponse } from "./classification/index.js";
-export { Crop, CropResponse, CropItem } from "./crop/index.js";
+export * as classification from "./classification/index.js";
+
+export { Crop, CropResponse } from "./crop/index.js";
+export * as crop from "./crop/index.js";
+
 export { Extraction, ExtractionResponse } from "./extraction/index.js";
+export * as extraction from "./extraction/index.js";
+
 export { Ocr, OcrResponse } from "./ocr/index.js";
+export * as ocr from "./ocr/index.js";
+
 export { Split, SplitResponse } from "./split/index.js";
+export * as split from "./split/index.js";

src/v2/product/ocr/index.ts

@@ -2,3 +2,6 @@ export { Ocr } from "./ocr.js";
 export { OcrParameters } from "./ocrParameters.js";
 export { OcrResponse } from "./ocrResponse.js";
 export { OcrInference } from "./ocrInference.js";
+export { OcrResult } from "./ocrResult.js";
+export { OcrPage } from "./ocrPage.js";
+export { OcrWord } from "./ocrWord.js";

tests/v2/product/classification.spec.ts

@@ -1,7 +1,8 @@
 import { expect } from "chai";
 import path from "node:path";
-import { V2_PRODUCT_PATH } from "../../index.js";
 import { ClassificationResponse } from "@/v2/product/index.js";
+
+import { V2_PRODUCT_PATH } from "../../index.js";
 import { loadV2Response } from "./utils.js";
 
 

tests/v2/product/crop.spec.ts

@@ -1,14 +1,15 @@
 import { expect } from "chai";
 import path from "node:path";
+import { Polygon } from "@/geometry/index.js";
+import { crop } from "@/v2/product/index.js";
+
 import { V2_PRODUCT_PATH } from "../../index.js";
-import { CropResponse, CropItem } from "@/v2/product/index.js";
 import { loadV2Response } from "./utils.js";
-import { Polygon } from "@/geometry/index.js";
 
 describe("MindeeV2 - Crop Response", async () => {
   it("should load a single result", async () => {
     const response = await loadV2Response(
-      CropResponse,
+      crop.CropResponse,
       path.join(V2_PRODUCT_PATH, "crop", "crop_single.json")
     );
     // Validate inference metadata
@@ -21,14 +22,14 @@ describe("MindeeV2 - Crop Response", async () => {
     expect(response.inference.file.mimeType).to.equal("image/jpeg");
 
     // Validate crops
-    const crops = response.inference.result.crops;
+    const crops: crop.CropItem[] = response.inference.result.crops;
     expect(crops).to.be.an("array").that.has.lengthOf(1);
 
-    const crop: CropItem = crops[0];
-    expect(crop.objectType).to.equal("invoice");
-    expect(crop.location.page).to.equal(0);
+    const firstCrop = crops[0];
+    expect(firstCrop.objectType).to.equal("invoice");
+    expect(firstCrop.location.page).to.equal(0);
 
-    const polygon: Polygon = crop.location.polygon!;
+    const polygon: Polygon = firstCrop.location.polygon!;
     expect(polygon.length).to.equal(4);
     expect(polygon.length).to.equal(4);
     expect(polygon[0][0]).to.equal(0.15);
@@ -43,7 +44,7 @@ describe("MindeeV2 - Crop Response", async () => {
 
   it("should load multiple results", async () => {
     const response = await loadV2Response(
-      CropResponse,
+      crop.CropResponse,
       path.join(V2_PRODUCT_PATH, "crop", "crop_multiple.json")
     );
     // Validate inference metadata
@@ -55,11 +56,11 @@ describe("MindeeV2 - Crop Response", async () => {
     expect(response.inference.file.pageCount).to.equal(1);
     expect(response.inference.file.mimeType).to.equal("image/jpeg");
 
-    const crops: CropItem[] = response.inference.result.crops;
+    const crops: crop.CropItem[] = response.inference.result.crops;
     expect(crops).to.be.an("array").that.has.lengthOf(2);
 
     // Validate first crop item
-    const firstCrop: CropItem = crops[0];
+    const firstCrop: crop.CropItem = crops[0];
     expect(firstCrop.objectType).to.equal("invoice");
     expect(firstCrop.location.page).to.equal(0);
     const firstPolygon: Polygon = firstCrop.location.polygon!;
@@ -74,7 +75,7 @@ describe("MindeeV2 - Crop Response", async () => {
     expect(firstPolygon[3][1]).to.equal(0.979);
 
     // Validate second crop item
-    const secondCrop: CropItem = crops[1];
+    const secondCrop: crop.CropItem = crops[1];
     expect(secondCrop.objectType).to.equal("invoice");
     expect(secondCrop.location.page).to.equal(0);
     const secondPolygon: Polygon = secondCrop.location.polygon!;

tests/v2/product/extraction.spec.ts

@@ -9,8 +9,9 @@ import {
   SimpleField,
 } from "@/v2/parsing/inference/field/index.js";
 import { field } from "@/v2/parsing/index.js";
-import { V2_PRODUCT_PATH } from "../../index.js";
 import { ExtractionResponse } from "@/v2/product/index.js";
+
+import { V2_PRODUCT_PATH } from "../../index.js";
 import { loadV2Response } from "./utils.js";
 
 const findocPath = path.join(V2_PRODUCT_PATH, "extraction", "financial_document");

…sts/v2/client/inferenceParameter.spec.ts …s/v2/product/extractionParameter.spec.tstests/v2/client/inferenceParameter.spec.ts renamed to tests/v2/product/extractionParameter.spec.ts tests/v2/client/inferenceParameter.spec.ts renamed to tests/v2/product/extractionParameter.spec.ts

@@ -3,19 +3,19 @@ import path from "path";
 import { V2_PRODUCT_PATH } from "../../index.js";
 import { expect } from "chai";
 import { promises as fs } from "fs";
-import { ExtractionParameters, DataSchema } from "@/v2/product/extraction/index.js";
+import { extraction } from "@/v2/product/index.js";
 
 let expectedDataSchemaDict: StringDict;
 let expectedDataSchemaString: string;
-let expectedDataSchemaObject: DataSchema;
+let expectedDataSchemaObject: extraction.DataSchema;
 
-describe("MindeeV2 - Inference Parameter", () => {
+describe("MindeeV2 - Extraction Parameter", () => {
   const modelIdValue = "test-model-id";
 
   describe("Polling Options", () => {
     it("should provide sensible defaults", () => {
 
-      const paramsInstance = new ExtractionParameters({
+      const paramsInstance = new extraction.ExtractionParameters({
         modelId: modelIdValue,
       });
       expect(paramsInstance.modelId).to.equal(modelIdValue);
@@ -34,26 +34,26 @@ describe("MindeeV2 - Inference Parameter", () => {
       );
       expectedDataSchemaDict = JSON.parse(fileContents.toString());
       expectedDataSchemaString = JSON.stringify(expectedDataSchemaDict);
-      expectedDataSchemaObject = new DataSchema(expectedDataSchemaDict);
+      expectedDataSchemaObject = new extraction.DataSchema(expectedDataSchemaDict);
     });
 
     it("shouldn't replace when unset", () => {
-      const params = new ExtractionParameters({
+      const params = new extraction.ExtractionParameters({
         modelId: modelIdValue,
       });
       expect(params.dataSchema).to.be.undefined;
     });
 
     it("should equate no matter the type", () => {
-      const paramsDict = new ExtractionParameters({
+      const paramsDict = new extraction.ExtractionParameters({
         modelId: modelIdValue,
         dataSchema: expectedDataSchemaDict,
       });
-      const paramsString = new ExtractionParameters({
+      const paramsString = new extraction.ExtractionParameters({
         modelId: modelIdValue,
         dataSchema: expectedDataSchemaString,
       });
-      const paramsObject = new ExtractionParameters({
+      const paramsObject = new extraction.ExtractionParameters({
         modelId: modelIdValue,
         dataSchema: expectedDataSchemaObject,
       });

tests/v2/product/ocr.spec.ts

@@ -1,26 +1,60 @@
 import { expect } from "chai";
 import path from "node:path";
+import { ocr } from "@/v2/product/index.js";
+import { Polygon } from "@/geometry/index.js";
+
 import { V2_PRODUCT_PATH } from "../../index.js";
-import { OcrResponse } from "@/v2/product/index.js";
 import { loadV2Response } from "./utils.js";
 
 
 describe("MindeeV2 - OCR Response", async () => {
   it("should load a single result", async () => {
     const response = await loadV2Response(
-      OcrResponse,
+      ocr.OcrResponse,
       path.join(V2_PRODUCT_PATH, "ocr", "ocr_single.json")
     );
-    const pages = response.inference.result.pages;
+    // Validate inference metadata
+    expect(response.inference.id).to.equal("12345678-1234-1234-1234-123456789abc");
+    expect(response.inference.model.id).to.equal("test-model-id");
+
+    // Validate file metadata
+    expect(response.inference.file.name).to.equal("default_sample.jpg");
+    expect(response.inference.file.pageCount).to.equal(1);
+    expect(response.inference.file.mimeType).to.equal("image/jpeg");
+
+    // Validate pages
+    const pages: ocr.OcrPage[] = response.inference.result.pages;
     expect(pages).to.be.an("array").that.has.lengthOf(1);
+
+    // Validate first page
+    const firstPage: ocr.OcrPage = pages[0];
+    expect(firstPage.words).to.be.an("array");
+
+    // Check first word
+    const firstWord: ocr.OcrWord = firstPage.words[0];
+    expect(firstWord.content).to.equal("Shipper:");
+    const firstPolygon: Polygon = firstWord.polygon;
+    expect(firstPolygon.length).to.equal(4);
+
+    // Check another word (5th word: "INC.")
+    const fifthWord: ocr.OcrWord = firstPage.words[4];
+    expect(fifthWord.content).to.equal("INC.");
+    const fifthPolygon: Polygon = fifthWord.polygon;
+    expect(fifthPolygon.length).to.equal(4);
   });
 
   it("should load multiple results", async () => {
     const response = await loadV2Response(
-      OcrResponse,
+      ocr.OcrResponse,
       path.join(V2_PRODUCT_PATH, "ocr", "ocr_multiple.json")
     );
-    const pages = response.inference.result.pages;
+    const pages: ocr.OcrPage[] = response.inference.result.pages;
     expect(pages).to.be.an("array").that.has.lengthOf(3);
+
+    // Validate that each page has words and content
+    pages.forEach((page: ocr.OcrPage): void => {
+      expect(page.words).to.be.an("array");
+      expect(page.content).to.be.a("string");
+    });
   });
 });

tests/v2/product/split.spec.ts

@@ -1,26 +1,27 @@
 import { expect } from "chai";
 import path from "node:path";
+import { split } from "@/v2/product/index.js";
+
 import { V2_PRODUCT_PATH } from "../../index.js";
-import { SplitResponse } from "@/v2/product/index.js";
 import { loadV2Response } from "./utils.js";
 
 
 describe("MindeeV2 - Split Response", async () => {
   it("should load a single result", async () => {
     const response = await loadV2Response(
-      SplitResponse,
+      split.SplitResponse,
       path.join(V2_PRODUCT_PATH, "split", "split_single.json")
     );
-    const splits = response.inference.result.splits;
+    const splits: split.SplitRange[] = response.inference.result.splits;
     expect(splits).to.be.an("array").that.has.lengthOf(1);
   });
 
   it("should load multiple results", async () => {
     const response = await loadV2Response(
-      SplitResponse,
+      split.SplitResponse,
       path.join(V2_PRODUCT_PATH, "split", "split_multiple.json")
     );
-    const splits = response.inference.result.splits;
+    const splits: split.SplitRange[] = response.inference.result.splits;
     expect(splits).to.be.an("array").that.has.lengthOf(3);
   });
 });

tests/v2/product/splitParameter.spec.ts

@@ -0,0 +1,36 @@
+import { expect } from "chai";
+import { split } from "@/v2/product/index.js";
+
+describe("MindeeV2 - Split Parameter", () => {
+  const modelIdValue = "test-model-id";
+
+  describe("Polling Options", () => {
+    it("should provide sensible defaults", () => {
+
+      const paramsInstance = new split.SplitParameters({
+        modelId: modelIdValue,
+      });
+      expect(paramsInstance.modelId).to.equal(modelIdValue);
+      expect(paramsInstance.getValidatedPollingOptions()).to.deep.equal({
+        delaySec: 1.5,
+        initialDelaySec: 2,
+        maxRetries: 80
+      });
+    });
+  });
+
+  describe("Invalid Options", () => {
+    it("should not set invalid options", () => {
+
+      const paramsInstance = new split.SplitParameters({
+        modelId: modelIdValue,
+        // @ts-expect-error - rag is not a valid option
+        rag: true,
+      });
+      expect(paramsInstance.modelId).to.equal(modelIdValue);
+      // @ts-expect-error - rag is not a valid option
+      expect(paramsInstance.rag).to.be.undefined;
+    });
+  });
+
+});