♻️ update raw text to new format (#381)

Author: ianardee

src/index.ts

@@ -10,8 +10,10 @@ export {
   Page,
 } from "./parsing/common";
 export {
+  InferenceFile,
   InferenceResponse,
   JobResponse,
+  RawText,
 } from "./parsing/v2";
 export {
   InputSource,

src/parsing/v2/inferenceResult.ts

@@ -1,12 +1,14 @@
 import { InferenceFields } from "./field/inferenceFields";
 import { InferenceResultOptions } from "./inferenceResultOptions";
 import { StringDict } from "../common";
+import { RawText } from "./rawText";
 
 export class InferenceResult {
   /**
    * Fields contained in the inference.
    */
   public fields: InferenceFields;
+  public rawText?: RawText;
 
   /**
    * Potential options retrieved alongside the inference.
@@ -15,8 +17,8 @@ export class InferenceResult {
 
   constructor(serverResponse: StringDict) {
     this.fields = new InferenceFields(serverResponse["fields"]);
-    if (serverResponse["options"]) {
-      this.options = new InferenceResultOptions(serverResponse["options"]);
+    if (serverResponse["raw_text"]) {
+      this.rawText = new RawText(serverResponse["raw_text"]);
     }
   }
 

src/parsing/v2/rawText.ts

@@ -1,21 +1,12 @@
 import { StringDict } from "../common";
-
+import { RawTextPage } from "./rawTextPage";
 
 export class RawText {
-  /**
-   * The page number the text was found on.
-   */
-  public page: number;
-  /**
-   * The text content found on the page.
-   */
-  public content: string;
+  pages: Array<RawTextPage>;
 
-  /**
-   * @param serverResponse JSON response from the server.
-   */
   constructor(serverResponse: StringDict) {
-    this.page = serverResponse["page"];
-    this.content = serverResponse["content"];
+    this.pages = serverResponse["pages"] ? serverResponse["pages"].map(
+      (rawTextPage: StringDict) => new RawTextPage(rawTextPage)
+    ) : [];
   }
 }

src/parsing/v2/rawTextPage.ts

@@ -0,0 +1,15 @@
+import { StringDict } from "../common";
+
+export class RawTextPage {
+  /**
+   * The text content found on the page.
+   */
+  public content: string;
+
+  /**
+   * @param serverResponse JSON response from the server.
+   */
+  constructor(serverResponse: StringDict) {
+    this.content = serverResponse["content"];
+  }
+}

tests/data

@@ -1,6 +1,6 @@
 import { expect } from "chai";
 import path from "node:path";
-import { LocalResponse, InferenceResponse } from "../../../src";
+import { LocalResponse, InferenceResponse, RawText } from "../../../src";
 import { FieldConfidence, ListField, ObjectField, SimpleField } from "../../../src/parsing/v2/field";
 import { promises as fs } from "node:fs";
 import { Polygon } from "../../../src/geometry";
@@ -236,22 +236,19 @@ describe("inference", async () => {
     });
   });
 
-  describe("options", async () => {
-    it("raw texts should be exposed", async () => {
+  describe("raw text", async () => {
+    it("raw text should be exposed", async () => {
       const response = await loadV2Inference(rawTextPath);
-      const opts = response.inference.result.options;
+      const rawText = response.inference.result.rawText;
 
-      expect(opts).to.not.be.undefined;
-      const rawTexts =
-        (opts as any).rawTexts ?? (opts as any).getRawTexts?.() ?? [];
+      expect(rawText).to.be.instanceOf(RawText);
 
-      expect(rawTexts).to.be.an("array").and.have.lengthOf(2);
+      const pages = rawText?.pages;
+      if (pages === undefined) throw new Error("pages is undefined");
 
-      const first = rawTexts[0];
-      expect(first.page).to.eq(0);
-      expect(first.content).to.eq(
-        "This is the raw text of the first page..."
-      );
+      expect(pages).to.be.an("array").and.have.lengthOf(2);
+      const first = pages[0];
+      expect(first.content).to.eq("This is the raw text of the first page...");
     });
   });