🐛 💥 harmonize Crop and Split extraction (#465)

ianardee · web-flow · commit ea7a78e53a82 · 2026-06-22T15:06:57.000+02:00
diff --git a/.github/workflows/_publish-code.yml b/.github/workflows/_publish-code.yml
@@ -11,7 +11,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
 
       - name: Install Dependencies
         run: sudo apt-get install pcregrep
diff --git a/.github/workflows/_publish-docs.yml b/.github/workflows/_publish-docs.yml
@@ -11,7 +11,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
 
       - name: Set up Node.js
         uses: actions/setup-node@v6
diff --git a/.github/workflows/_static-analysis.yml b/.github/workflows/_static-analysis.yml
@@ -10,7 +10,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
 
       - name: Set up Node.js
         uses: actions/setup-node@v6
diff --git a/.github/workflows/_test-integrations.yml b/.github/workflows/_test-integrations.yml
@@ -33,7 +33,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           submodules: recursive
 
@@ -84,7 +84,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           submodules: recursive
 
diff --git a/.github/workflows/_test-smoke.yml b/.github/workflows/_test-smoke.yml
@@ -27,7 +27,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           submodules: recursive
 
diff --git a/.github/workflows/_test-units.yml b/.github/workflows/_test-units.yml
@@ -18,7 +18,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           submodules: recursive
 
@@ -56,7 +56,7 @@ jobs:
 
     steps:
       - name: Check out Git repository
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
           submodules: recursive
 
diff --git a/src/image/extractedImages.ts b/src/image/extractedImages.ts
@@ -1,6 +1,6 @@
 import { ExtractedImage } from "@/image/index.js";
 
-export class CropFiles extends Array<ExtractedImage> {
+export class ExtractedImages extends Array<ExtractedImage> {
   constructor(...items: ExtractedImage[]) {
     super(...items);
   }
diff --git a/src/image/index.ts b/src/image/index.ts
@@ -1,3 +1,4 @@
 export { compressImage } from "./imageCompressor.js";
 export { ExtractedImage } from "./extractedImage.js";
+export { ExtractedImages } from "./extractedImages.js";
 export { extractFromPage } from "./imageExtractor.js";
diff --git a/src/pdf/extractedPdfs.ts b/src/pdf/extractedPdfs.ts
@@ -0,0 +1,7 @@
+import { ExtractedPdf } from "@/pdf/extractedPdf.js";
+
+export class ExtractedPdfs extends Array<ExtractedPdf> {
+  constructor(...items: ExtractedPdf[]) {
+    super(...items);
+  }
+}
diff --git a/src/pdf/index.ts b/src/pdf/index.ts
@@ -2,3 +2,5 @@ export { extractPages, countPages } from "./pdfOperation.js";
 export type { SplitPdf } from "./pdfOperation.js";
 export { compressPdf } from "./pdfCompressor.js";
 export { hasSourceText } from "./pdfUtils.js";
+export { ExtractedPdf } from "./extractedPdf.js";
+export { ExtractedPdfs } from "./extractedPdfs.js";
diff --git a/src/v2/fileOperations/crop.ts b/src/v2/fileOperations/crop.ts
@@ -3,18 +3,20 @@ import { CropItem } from "@/v2/product/crop/index.js";
 import { MindeeError } from "@/errors/index.js";
 import { extractImagesFromPolygon } from "@/image/imageExtractor.js";
 import { Polygon } from "@/geometry/index.js";
-import { CropFiles } from "@/v2/fileOperations/cropFiles.js";
-import { ExtractedImage } from "@/image/index.js";
+import { ExtractedImage, ExtractedImages } from "@/image/index.js";
 import { logger } from "@/logger.js";
 
 
 /**
  * Extracts a single specified crop from a given input source.
  * @param inputSource Local input source.
  * @param crop Crop to extract.
+ * @param quality JPEG quality of extracted image.
  */
-export async function extractSingleCrop(inputSource: LocalInputSource, crop: CropItem): Promise<ExtractedImage> {
-  return (await extractCrops(inputSource, [crop]))[0];
+export async function extractSingleCrop(
+  inputSource: LocalInputSource, crop: CropItem, quality?: number
+): Promise<ExtractedImage> {
+  return (await extractMultipleCrops(inputSource, [crop], quality))[0];
 }
 
 
@@ -25,11 +27,11 @@ export async function extractSingleCrop(inputSource: LocalInputSource, crop: Cro
  * @param quality JPEG quality of extracted images.
  * @return a list of extracted files, as a CropFiles object.
  */
-export async function extractCrops(
+export async function extractMultipleCrops(
   inputSource: LocalInputSource,
   crops: CropItem[],
   quality?: number ,
-): Promise<CropFiles> {
+): Promise<ExtractedImages> {
   if (crops.length === 0) {
     throw new MindeeError("No crop indexes provided.");
   }
@@ -44,5 +46,5 @@ export async function extractCrops(
     polygonsByPage.get(pageId)!.push(crop.location.polygon);
   }
   const extractedCrops = await extractImagesFromPolygon(inputSource, polygonsByPage, quality);
-  return new CropFiles(...extractedCrops);
+  return new ExtractedImages(...extractedCrops);
 }
diff --git a/src/v2/fileOperations/split.ts b/src/v2/fileOperations/split.ts
@@ -1,17 +1,17 @@
 import { LocalInputSource } from "@/input/index.js";
 import { MindeeError } from "@/errors/index.js";
 import { PdfExtractor } from "@/pdf/pdfExtractor.js";
-import { SplitFiles } from "@/v2/fileOperations/splitFiles.js";
 import { logger } from "@/logger.js";
 import { ExtractedPdf } from "@/pdf/extractedPdf.js";
+import { ExtractedPdfs } from "@/pdf/extractedPdfs.js";
 
 /**
  * Extracts a single specified split from a
  * @param inputSource
  * @param split
  */
 export async function extractSingleSplit(inputSource: LocalInputSource, split: number[]) {
-  return await extractSplits(inputSource, [split]);
+  return await extractMultipleSplits(inputSource, [split]);
 }
 
 /**
@@ -21,7 +21,7 @@ export async function extractSingleSplit(inputSource: LocalInputSource, split: n
  * @return a list of extracted files.
  * @throws MindeeError if no indexes are provided.
  */
-export async function extractSplits(inputSource: LocalInputSource, splits: number[][]): Promise<SplitFiles> {
+export async function extractMultipleSplits(inputSource: LocalInputSource, splits: number[][]): Promise<ExtractedPdfs> {
   const pageGroups = splits.filter(e => e.length > 0);
   if (pageGroups.length === 0) {
     throw new MindeeError("No valid split indexes provided.");
@@ -32,14 +32,14 @@ export async function extractSplits(inputSource: LocalInputSource, splits: numbe
   await pdfExtractor.init();
 
   if (splits.length === 0) {
-    return new SplitFiles();
+    return new ExtractedPdfs();
   }
   const pageCount = await pdfExtractor.getPageCount();
   if (splits.length === 1 && splits[0].at(-1) === pageCount-1) {
-    return new SplitFiles(new ExtractedPdf(inputSource.fileObject as Buffer, inputSource.filename, pageCount));
+    return new ExtractedPdfs(new ExtractedPdf(inputSource.fileObject as Buffer, inputSource.filename, pageCount));
   }
   const subDocuments = await pdfExtractor.extractSubDocuments(pageGroups);
-  return new SplitFiles(...subDocuments);
+  return new ExtractedPdfs(...subDocuments);
 }
 
 /**
diff --git a/src/v2/fileOperations/splitFiles.ts b/src/v2/fileOperations/splitFiles.ts
diff --git a/src/v2/product/crop/cropItem.ts b/src/v2/product/crop/cropItem.ts
@@ -1,7 +1,7 @@
 import { FieldLocation } from "@/v2/parsing/inference/field/index.js";
 import { StringDict } from "@/parsing/index.js";
 import { LocalInputSource } from "@/input/index.js";
-import { extractCrops } from "@/v2/fileOperations/crop.js";
+import { extractSingleCrop } from "@/v2/fileOperations/crop.js";
 import { ExtractedImage } from "@/image/index.js";
 import { ExtractionResponse } from "@/v2/product/index.js";
 
@@ -36,7 +36,7 @@ export class CropItem {
    * @param inputSource The input file to extract from.
    * @param quality Optional quality parameter for image extraction, default is undefined (full quality).
    */
-  async extractFromFile(inputSource: LocalInputSource, quality: number = 1): Promise<ExtractedImage>{
-    return (await extractCrops(inputSource, [this], quality))[0];
+  async extractFromInputSource(inputSource: LocalInputSource, quality: number = 1): Promise<ExtractedImage>{
+    return (await extractSingleCrop(inputSource, this, quality));
   }
 }
diff --git a/src/v2/product/crop/cropResponse.ts b/src/v2/product/crop/cropResponse.ts
@@ -1,7 +1,4 @@
-import { LocalInputSource } from "@/input/index.js";
 import { StringDict } from "@/parsing/stringDict.js";
-import { extractCrops } from "@/v2/fileOperations/crop.js";
-import { CropFiles } from "@/v2/fileOperations/cropFiles.js";
 import { BaseResponse } from "@/v2/parsing/index.js";
 import { CropInference } from "./cropInference.js";
 
@@ -18,13 +15,4 @@ export class CropResponse extends BaseResponse {
     super(serverResponse);
     this.inference = new CropInference(serverResponse["inference"]);
   }
-
-  /**
-   * Extracts all crops from an input.
-   * @param inputSource The input file to extract from.
-   * @param quality Optional quality parameter for image extraction, default is undefined (full quality).
-   */
-  async extractFromFile(inputSource: LocalInputSource, quality: number = 1): Promise<CropFiles> {
-    return await extractCrops(inputSource, this.inference.result.crops, quality);
-  }
 }
diff --git a/src/v2/product/crop/cropResult.ts b/src/v2/product/crop/cropResult.ts
@@ -1,5 +1,8 @@
 import { StringDict } from "@/parsing/stringDict.js";
 import { CropItem } from "@/v2/product/crop/cropItem.js";
+import { LocalInputSource } from "@/input/index.js";
+import { extractMultipleCrops } from "@/v2/fileOperations/crop.js";
+import { ExtractedImages } from "@/image/extractedImages.js";
 
 export class CropResult {
   /**
@@ -15,4 +18,13 @@ export class CropResult {
     const crops = this.crops.map(item => item.toString()).join("\n");
     return `Crops\n=====\n${crops}`;
   }
+
+  /**
+   * Extracts a single crop from an input.
+   * @param inputSource The input file to extract from.
+   * @param quality Optional quality parameter for image extraction, default is undefined (full quality).
+   */
+  async extractFromInputSource(inputSource: LocalInputSource, quality: number = 1): Promise<ExtractedImages>{
+    return (await extractMultipleCrops(inputSource, this.crops, quality));
+  }
 }
diff --git a/src/v2/product/split/splitRange.ts b/src/v2/product/split/splitRange.ts
@@ -1,7 +1,8 @@
 import { StringDict } from "@/parsing/index.js";
 import { LocalInputSource } from "@/input/index.js";
-import { expandRange, extractSplits } from "@/v2/fileOperations/split.js";
+import { expandRange, extractMultipleSplits } from "@/v2/fileOperations/split.js";
 import { ExtractionResponse } from "@/v2/product/index.js";
+import { ExtractedPdf } from "@/pdf/index.js";
 
 /**
  * Split inference result.
@@ -38,8 +39,8 @@ export class SplitRange {
    * Extracts a single split from the input file.
    * @param inputSource The input file to extract from.
    */
-  async extractFromFile(inputSource: LocalInputSource) {
+  async extractFromFile(inputSource: LocalInputSource): Promise<ExtractedPdf> {
     const pageRange = [expandRange(this.pageRange as [number, number])];
-    return (await extractSplits(inputSource, pageRange))[0];
+    return (await extractMultipleSplits(inputSource, pageRange))[0];
   }
 }
diff --git a/src/v2/product/split/splitResponse.ts b/src/v2/product/split/splitResponse.ts
@@ -1,9 +1,6 @@
 import { StringDict } from "@/parsing/stringDict.js";
 import { SplitInference } from "./splitInference.js";
 import { BaseResponse } from "@/v2/parsing/index.js";
-import { LocalInputSource } from "@/input/index.js";
-import { expandRange, extractSplits } from "@/v2/fileOperations/split.js";
-import { SplitFiles } from "@/v2/fileOperations/splitFiles.js";
 
 export class SplitResponse extends BaseResponse {
   /**
@@ -18,16 +15,4 @@ export class SplitResponse extends BaseResponse {
     super(serverResponse);
     this.inference = new SplitInference(serverResponse["inference"]);
   }
-
-  /**
-   * Extracts all splits from an input PDF.
-   * @param inputSource The input file to extract from.
-   */
-  async extractFromFile(inputSource: LocalInputSource): Promise<SplitFiles>{
-    const splits: number[][] = [];
-    for (const split of this.inference.result.splits) {
-      splits.push(expandRange(split.pageRange as [number, number]));
-    }
-    return await extractSplits(inputSource, splits);
-  }
 }
diff --git a/src/v2/product/split/splitResult.ts b/src/v2/product/split/splitResult.ts
@@ -1,5 +1,8 @@
 import { SplitRange } from "./splitRange.js";
 import { StringDict } from "@/parsing/index.js";
+import { LocalInputSource } from "@/input/index.js";
+import { ExtractedPdfs } from "@/pdf/index.js";
+import { extractMultipleSplits, expandRange } from "@/v2/fileOperations/split.js";
 
 /**
  * Split result info.
@@ -14,6 +17,18 @@ export class SplitResult {
     this.splits = rawResponse.splits.map((split: StringDict) => new SplitRange(split));
   }
 
+  /**
+   * Extracts all splits from an input PDF.
+   * @param inputSource The input file to extract from.
+   */
+  async extractFromInputSource(inputSource: LocalInputSource): Promise<ExtractedPdfs>{
+    const splits: number[][] = [];
+    for (const split of this.splits) {
+      splits.push(expandRange(split.pageRange as [number, number]));
+    }
+    return await extractMultipleSplits(inputSource, splits);
+  }
+
   toString(): string {
     let splits = "\n";
     if (this.splits.length > 0) {
diff --git a/tests/data b/tests/data
@@ -1 +1 @@
-Subproject commit 13093f3a48de212ef26889df71199c1a2a9d1478
+Subproject commit 2d7fcf8f591f6d7f40e39862965325e6a8a21874
diff --git a/tests/v2/fileOperations/crop.spec.ts b/tests/v2/fileOperations/crop.spec.ts
diff --git a/tests/v2/fileOperations/split.spec.ts b/tests/v2/fileOperations/split.spec.ts

Original file line number	Diff line number	Diff line change
`@@ -1,6 +1,6 @@`
`1`	`1`	`import { ExtractedImage } from "@/image/index.js";`
`2`	`2`
`3`		`-export class CropFiles extends Array<ExtractedImage> {`
	`3`	`+export class ExtractedImages extends Array<ExtractedImage> {`
`4`	`4`	`constructor(...items: ExtractedImage[]) {`
`5`	`5`	`super(...items);`
`6`	`6`	`}`