🐛 make sure input sources are only initialized once (#392)

Author: ianardee

src/clientV2.ts

@@ -163,9 +163,7 @@ export class ClientV2 {
     if (inputSource === undefined) {
       throw new Error("The 'enqueue' function requires an input document.");
     }
-    if (!inputSource.isInitialized()) {
-      await inputSource.init();
-    }
+    await inputSource.init();
     return await this.mindeeApi.reqPostInferenceEnqueue(inputSource, params);
   }
 

src/input/sources/base64Input.ts

@@ -1,5 +1,6 @@
 import { LocalInputSource } from "./localInputSource";
 import { INPUT_TYPE_BASE64 } from "./inputSource";
+import { logger } from "../../logger";
 
 interface Base64InputProps {
   inputString: string;
@@ -19,6 +20,10 @@ export class Base64Input extends LocalInputSource {
   }
 
   async init() {
+    if (this.initialized) {
+      return;
+    }
+    logger.debug("Loading from base64");
     this.fileObject = Buffer.from(this.inputString, "base64");
     this.mimeType = await this.checkMimetype();
     // clear out the string

src/input/sources/bufferInput.ts

@@ -1,5 +1,6 @@
 import { LocalInputSource } from "./localInputSource";
 import { INPUT_TYPE_BUFFER } from "./inputSource";
+import { logger } from "../../logger";
 
 interface BufferInputProps {
   buffer: Buffer;
@@ -16,6 +17,10 @@ export class BufferInput extends LocalInputSource {
   }
 
   async init(): Promise<void> {
+    if (this.initialized) {
+      return;
+    }
+    logger.debug("Loading from buffer");
     this.mimeType = await this.checkMimetype();
     this.initialized = true;
   }

src/input/sources/bytesInput.ts

@@ -1,5 +1,6 @@
 import { INPUT_TYPE_BYTES } from "./inputSource";
 import { LocalInputSource } from "./localInputSource";
+import { logger } from "../../logger";
 
 interface BytesInputProps {
   inputBytes: Uint8Array;
@@ -19,6 +20,10 @@ export class BytesInput extends LocalInputSource {
   }
 
   async init() {
+    if (this.initialized) {
+      return;
+    }
+    logger.debug("Loading from bytes");
     this.fileObject = Buffer.from(this.inputBytes);
     this.mimeType = await this.checkMimetype();
     this.inputBytes = new Uint8Array(0);

src/input/sources/pathInput.ts

@@ -21,7 +21,10 @@ export class PathInput extends LocalInputSource {
   }
 
   async init() {
-    logger.debug(`Loading from: ${this.inputPath}`);
+    if (this.initialized) {
+      return;
+    }
+    logger.debug(`Loading from path: ${this.inputPath}`);
     this.fileObject = Buffer.from(await fs.readFile(this.inputPath));
     this.mimeType = await this.checkMimetype();
     this.initialized = true;

src/input/sources/streamInput.ts

@@ -1,6 +1,7 @@
 import { Readable } from "stream";
 import { LocalInputSource } from "./localInputSource";
 import { INPUT_TYPE_STREAM } from "./inputSource";
+import { logger } from "../../logger";
 
 interface StreamInputProps {
   inputStream: Readable;
@@ -20,6 +21,10 @@ export class StreamInput extends LocalInputSource {
   }
 
   async init() {
+    if (this.initialized) {
+      return;
+    }
+    logger.debug("Loading from stream");
     this.fileObject = await this.stream2buffer(this.inputStream);
     this.mimeType = await this.checkMimetype();
     this.initialized = true;

src/input/sources/urlInput.ts

@@ -6,6 +6,7 @@ import { writeFile } from "fs/promises";
 import { request as httpsRequest } from "https";
 import { IncomingMessage } from "http";
 import { BytesInput } from "./bytesInput";
+import { logger } from "../../logger";
 
 export class UrlInput extends InputSource {
   public readonly url: string;
@@ -16,6 +17,10 @@ export class UrlInput extends InputSource {
   }
 
   async init() {
+    if (this.initialized) {
+      return;
+    }
+    logger.debug(`source URL: ${this.url}`);
     if (!this.url.toLowerCase().startsWith("https")) {
       throw new Error("URL must be HTTPS");
     }

tests/v2/clientV2.integration.ts

@@ -1,9 +1,10 @@
 import { expect } from "chai";
 import path from "node:path";
 
-import { ClientV2, InferenceParameters, PathInput, UrlInput } from "../../src";
+import { ClientV2, InferenceParameters, PathInput, UrlInput, Base64Input } from "../../src";
 import { SimpleField } from "../../src/parsing/v2/field";
 import { MindeeHttpErrorV2 } from "../../src/errors/mindeeError";
+import * as fs from "node:fs";
 
 describe("MindeeClientV2 – integration tests (V2)", () => {
   let client: ClientV2;
@@ -22,6 +23,11 @@ describe("MindeeClientV2 – integration tests (V2)", () => {
     "financial_document",
     "default_sample.jpg",
   );
+  const sampleBase64Path = path.join(
+    dataDir,
+    "file_types",
+    "receipt.txt",
+  );
 
   beforeEach(async () => {
     const apiKey = process.env["MINDEE_V2_API_KEY"] ?? "";
@@ -30,7 +36,7 @@ describe("MindeeClientV2 – integration tests (V2)", () => {
     client = new ClientV2({ apiKey });
   });
 
-  it("Empty, multi-page PDF – enqueue & get inference must succeed", async () => {
+  it("Empty, multi-page PDF – PathInput - enqueueAndGetInference must succeed", async () => {
     const source = new PathInput({ inputPath: emptyPdfPath });
     const params: InferenceParameters = {
       modelId,
@@ -61,16 +67,16 @@ describe("MindeeClientV2 – integration tests (V2)", () => {
     expect(inference.activeOptions?.confidence).to.be.false;
   }).timeout(60000);
 
-  it("Filled, single-page image – enqueue & get inference must succeed", async () => {
+  it("Filled, single-page image – PathInput - enqueueAndGetInference must succeed", async () => {
     const source = new PathInput({ inputPath: sampleImagePath });
     const params: InferenceParameters = {
       modelId,
       rag: false,
       rawText: true,
-      polygon: false,
+      polygon: true,
       confidence: false,
       webhookIds: [],
-      alias: "ts_integration_filled_single"
+      alias: "ts_integration_binary_filled_single"
     };
 
     const response = await client.enqueueAndGetInference(source, params);
@@ -86,17 +92,49 @@ describe("MindeeClientV2 – integration tests (V2)", () => {
     expect(supplierField).to.be.instanceOf(SimpleField);
     expect(supplierField.value).to.equal("John Smith");
 
-
     expect(inference.result.rawText).to.exist;
     expect(inference.activeOptions).to.not.be.null;
     expect(inference.activeOptions?.rag).to.be.false;
     expect(inference.activeOptions?.rawText).to.be.true;
-    expect(inference.activeOptions?.polygon).to.be.false;
+    expect(inference.activeOptions?.polygon).to.be.true;
     expect(inference.activeOptions?.confidence).to.be.false;
 
     expect(inference.result.rawText?.pages).to.have.lengthOf(1);
   }).timeout(120000);
 
+  it("Filled, single-page image – Base64Input - enqueueAndGetInference must succeed", async () => {
+    const data = fs.readFileSync(sampleBase64Path, "utf8");
+    const source = new Base64Input({ inputString: data, filename: "receipt.jpg" });
+    const params: InferenceParameters = {
+      modelId,
+      rag: false,
+      rawText: false,
+      polygon: false,
+      confidence: false,
+      webhookIds: [],
+      alias: "ts_integration_base64_filled_single"
+    };
+
+    const response = await client.enqueueAndGetInference(source, params);
+
+    const inference = response.inference;
+    expect(inference.file?.name).to.equal("receipt.jpg");
+    expect(inference.model?.id).to.equal(modelId);
+
+    expect(inference.result).to.exist;
+    expect(inference.result.rawText).to.be.undefined;
+
+    const supplierField = inference.result.fields.get("supplier_name") as SimpleField;
+    expect(supplierField).to.be.instanceOf(SimpleField);
+    expect(supplierField.value).to.equal("Clachan");
+
+    expect(inference.activeOptions).to.not.be.null;
+    expect(inference.activeOptions?.rag).to.be.false;
+    expect(inference.activeOptions?.rawText).to.be.false;
+    expect(inference.activeOptions?.polygon).to.be.false;
+    expect(inference.activeOptions?.confidence).to.be.false;
+  }).timeout(120000);
+
   it("Invalid model ID – enqueue must raise 422", async () => {
     const source = new PathInput({ inputPath: emptyPdfPath });
     const badParams: InferenceParameters = { modelId: "INVALID MODEL ID" };