✨ add support for URL inputs in V2 + bump dependencies (#365)

Author: sebastianMindee

.github/workflows/_test-integrations.yml

@@ -47,4 +47,5 @@ jobs:
           WORKFLOW_ID: ${{ secrets.WORKFLOW_ID_SE_TESTS }}
           MINDEE_V2_API_KEY: ${{ secrets.MINDEE_V2_SE_TESTS_API_KEY }}
           MINDEE_V2_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
+          MINDEE_V2_SE_TESTS_BLANK_PDF_URL: ${{ secrets.MINDEE_V2_SE_TESTS_BLANK_PDF_URL }}
         run: npm run test-integration

package-lock.json

@@ -1,6 +1,6 @@
 import {
-  Base64Input, BufferInput, BytesInput,
-  LocalInputSource, PathInput, StreamInput, UrlInput,
+  Base64Input, BufferInput, BytesInput, InputSource,
+  PathInput, StreamInput, UrlInput,
 } from "./input";
 import { errorHandler } from "./errors/handler";
 import { LOG_LEVELS, logger } from "./logger";
@@ -124,7 +124,7 @@ export interface ClientOptions {
  * @category ClientV2
  */
 export class ClientV2 {
-  /** Key of the API. */
+  /** Mindee API handler. */
   protected mindeeApi: MindeeApiV2;
 
   /**
@@ -148,13 +148,13 @@ export class ClientV2 {
 
   /**
    * Send the document to an asynchronous endpoint and return its ID in the queue.
-   * @param inputSource file to parse.
+   * @param inputSource file or URL to parse.
    * @param params parameters relating to prediction options.
    * @category Asynchronous
    * @returns a `Promise` containing the job (queue) corresponding to a document.
    */
   async enqueueInference(
-    inputSource: LocalInputSource,
+    inputSource: InputSource,
     params: InferenceParameters
   ): Promise<JobResponse> {
     if (inputSource === undefined) {
@@ -240,7 +240,7 @@ export class ClientV2 {
    * @returns a `Promise` containing parsing results.
    */
   async enqueueAndGetInference(
-    inputDoc: LocalInputSource,
+    inputDoc: InputSource,
     params: InferenceParameters
   ): Promise<InferenceResponse> {
     const validatedAsyncParams = this.#setAsyncParams(params.pollingOptions);

src/clientV2.ts

@@ -4,7 +4,7 @@ import { InferenceResponse, JobResponse } from "../parsing/v2";
 import FormData from "form-data";
 import { RequestOptions } from "https";
 import { BaseEndpoint, EndpointResponse } from "./baseEndpoint";
-import { LocalInputSource } from "../input";
+import { InputSource, LocalInputSource, UrlInput } from "../input";
 import { MindeeApiV2Error, MindeeHttpErrorV2 } from "../errors/mindeeError";
 import { logger } from "../logger";
 
@@ -17,18 +17,18 @@ export class MindeeApiV2 {
 
   /**
    * Sends a file to the inference queue.
-   * @param inputDoc Local file loaded as an input.
+   * @param inputSource Local file loaded as an input.
    * @param params {InferenceParameters} parameters relating to the enqueueing options.
    * @category V2
    * @throws Error if the server's response contains one.
    * @returns a `Promise` containing a job response.
    */
-  async reqPostInferenceEnqueue(inputDoc: LocalInputSource, params: InferenceParameters): Promise<JobResponse> {
-    await inputDoc.init();
+  async reqPostInferenceEnqueue(inputSource: InputSource, params: InferenceParameters): Promise<JobResponse> {
+    await inputSource.init();
     if (params.modelId === undefined || params.modelId === null || params.modelId === "") {
       throw new Error("Model ID must be provided");
     }
-    const result: EndpointResponse = await this.#documentEnqueuePost(inputDoc, params);
+    const result: EndpointResponse = await this.#documentEnqueuePost(inputSource, params);
     if (result.data.error?.code !== undefined) {
       throw new MindeeHttpErrorV2(
         result.data.error.code,
@@ -86,10 +86,10 @@ export class MindeeApiV2 {
   /**
    * Sends a document to the inference queue.
    *
-   * @param inputDoc Local file loaded as an input.
+   * @param inputSource Local or remote file as an input.
    * @param params {InferenceParameters} parameters relating to the enqueueing options.
    */
-  #documentEnqueuePost(inputDoc: LocalInputSource, params: InferenceParameters): Promise<EndpointResponse> {
+  #documentEnqueuePost(inputSource: InputSource, params: InferenceParameters): Promise<EndpointResponse> {
     const form = new FormData();
 
     form.append("model_id", params.modelId);
@@ -99,9 +99,13 @@ export class MindeeApiV2 {
     if (params.webhookIds && params.webhookIds.length > 0) {
       form.append("webhook_ids", params.webhookIds.join(","));
     }
-    form.append("file", inputDoc.fileObject, {
-      filename: inputDoc.filename,
-    });
+    if (inputSource instanceof LocalInputSource) {
+      form.append("file", inputSource.fileObject, {
+        filename: inputSource.filename,
+      });
+    } else {
+      form.append("url", (inputSource as UrlInput).url);
+    }
     const path = "/v2/inferences/enqueue";
     const headers = { ...this.settings.baseHeaders, ...form.getHeaders() };
     const options: RequestOptions = {

src/http/mindeeApiV2.ts

@@ -8,7 +8,7 @@ import { IncomingMessage } from "http";
 import { BytesInput } from "./bytesInput";
 
 export class UrlInput extends InputSource {
-  private readonly url: string;
+  public readonly url: string;
 
   constructor({ url }: { url: string }) {
     super();

src/input/sources/urlInput.ts

@@ -1,19 +1,12 @@
 import { FieldConfidence } from "./fieldConfidence";
-import { FieldLocation } from "./fieldLocation";
 import { StringDict } from "../../common";
 
 export abstract class BaseField {
   protected _indentLevel: number;
-  public locations: Array<FieldLocation> | undefined;
   public confidence: FieldConfidence | undefined;
 
   protected constructor(rawResponse: StringDict, indentLevel = 0) {
     this._indentLevel = indentLevel;
-    if (rawResponse["locations"]) {
-      this.locations = rawResponse["locations"].map((location: StringDict | undefined) => {
-        return location ? new FieldLocation(location) : "";
-      });
-    }
     if (rawResponse["confidence"] !== undefined) {
       this.confidence = rawResponse["confidence"] as FieldConfidence;
     }

src/parsing/v2/field/baseField.ts

@@ -0,0 +1,17 @@
+import { BaseField } from "./baseField";
+import { FieldLocation } from "./fieldLocation";
+import { StringDict } from "../../common";
+
+
+export class DynamicField extends BaseField {
+  public locations: Array<FieldLocation> | undefined;
+
+  constructor(rawResponse: StringDict, indentLevel = 0) {
+    super(rawResponse, indentLevel);
+    if (rawResponse["locations"]) {
+      this.locations = rawResponse["locations"].map((location: StringDict | undefined) => {
+        return location ? new FieldLocation(location) : "";
+      });
+    }
+  }
+}