Merge pull request #40 from robertknight/add-clear-image

Add `clearImage` method to OCREngine/OCRClient

src/lib.cpp

@@ -160,7 +160,11 @@ class OCREngine {
     return {};
   }
 
-  void ClearImage() { tesseract_->Clear(); }
+  void ClearImage() {
+    tesseract_->Clear();
+    layout_analysis_done_ = false;
+    ocr_done_ = false;
+  }
 
   std::vector<TextRect> GetBoundingBoxes(TextUnit unit) {
     if (!layout_analysis_done_) {
@@ -304,12 +308,12 @@ EMSCRIPTEN_BINDINGS(ocrlib) {
   class_<OCREngine>("OCREngine")
       .constructor<>()
       .function("clearImage", &OCREngine::ClearImage)
-      .function("loadModel", &OCREngine::LoadModel)
-      .function("loadImage", &OCREngine::LoadImage)
       .function("getBoundingBoxes", &OCREngine::GetBoundingBoxes)
       .function("getOrientation", &OCREngine::GetOrientation)
+      .function("getText", &OCREngine::GetText)
       .function("getTextBoxes", &OCREngine::GetTextBoxes)
-      .function("getText", &OCREngine::GetText);
+      .function("loadImage", &OCREngine::LoadImage)
+      .function("loadModel", &OCREngine::LoadModel);
 
   value_object<OCRResult>("OCRResult").field("error", &OCRResult::error);
 

src/ocr-client.ts

@@ -149,6 +149,21 @@ export class OCRClient {
     return engine.loadImage(image);
   }
 
+  /**
+   * Clear the current image and text recognition results.
+   *
+   * This will clear the loaded image data internally, but keep the text
+   * recognition model loaded.
+   *
+   * At present there is no way to shrink WebAssembly memory, so this will not
+   * return the memory used by the image to the OS/browser. To release memory,
+   * the web worker needs to be shut down via {@link destroy}.
+   */
+  async clearImage(): Promise<void> {
+    const engine = await this._ocrEngine;
+    return engine.clearImage();
+  }
+
   /**
    * Perform layout analysis on the current image, if not already done, and
    * return bounding boxes for a given unit of text.

src/ocr-engine.ts

@@ -180,6 +180,21 @@ export class OCREngine {
     this._imageLoaded = true;
   }
 
+  /**
+   * Clear the current image and text recognition results.
+   *
+   * This will clear the loaded image data internally, but keep the text
+   * recognition model loaded.
+   *
+   * At present there is no way to shrink WebAssembly memory, so this will not
+   * return the memory used by the image to the OS/browser. To release memory,
+   * the `OCREngine` instance needs to be destroyed via {@link destroy}.
+   */
+  clearImage() {
+    this._engine.clearImage();
+    this._imageLoaded = false;
+  }
+
   /**
    * Perform layout analysis on the current image, if not already done, and
    * return bounding boxes for a given unit of text.

test/ocr-client-test.js

@@ -118,4 +118,22 @@ describe("OCRClient", () => {
     assert.equal(orient.rotation, 0);
     assert.equal(orient.confidence, 1.0);
   });
+
+  it("clears the image", async () => {
+    const imageData = await loadImage(resolve("./small-test-page.jpg"));
+    await ocr.loadImage(imageData);
+    await ocr.getBoundingBoxes("word");
+
+    await ocr.clearImage();
+
+    let error;
+    try {
+      await ocr.getBoundingBoxes("word");
+    } catch (e) {
+      error = e;
+    }
+
+    assert.instanceOf(error, Error);
+    assert.equal(error.message, "No image loaded");
+  });
 });

test/ocr-engine-test.js

@@ -308,4 +308,15 @@ describe("OCREngine", () => {
       assert.equal(estimatedOrient.confidence, 1);
     }
   });
+
+  it("clears the image", async () => {
+    ocr.loadImage(emptyImage(100, 100));
+    ocr.getBoundingBoxes("word");
+
+    ocr.clearImage();
+
+    assert.throws(() => {
+      ocr.getBoundingBoxes("word");
+    }, "No image loaded");
+  });
 });