QRCode Detection working.

functions.js

@@ -1,7 +1,9 @@
 import PDFLib from 'pdf-lib';
-import OpenCV from 'opencv-wasm';
 import PDFJS from "pdfjs-dist";
+
+delete global.crypto; // TODO: I hate to do this, but the new node version forces me to, if anyone finds a better solution, please tell me!
 import * as pdfcpuWraopper from "./public/wasm/pdfcpu-wrapper-node.js";
+import OpenCV from 'opencv-wasm';
 
 import { extractPages as dependantExtractPages } from "./public/functions/extractPages.js";
 import { impose as dependantImpose } from './public/functions/impose.js';
@@ -13,6 +15,7 @@ import { splitPDF as dependantSplitPDF } from './public/functions/splitPDF.js';
 import { editMetadata as dependantEditMetadata } from './public/functions/editMetadata.js';
 import { organizePages as dependantOrganizePages } from './public/functions/organizePages.js';
 import { removeBlankPages as dependantRemoveBlankPages} from './public/functions/removeBlankPages.js';
+import { splitOn as dependantSplitOn } from "./public/functions/splitOn.js";
 
 export async function extractPages(snapshot, pagesToExtractArray) {
     return dependantExtractPages(snapshot, pagesToExtractArray, PDFLib);
@@ -52,4 +55,8 @@ export async function organizePages(snapshot, operation, customOrderString) {
 
 export async function removeBlankPages(snapshot, whiteThreashold) {
     return dependantRemoveBlankPages(snapshot, whiteThreashold, PDFJS, OpenCV, PDFLib);
+}
+
+export async function splitOn(snapshot, type, whiteThreashold) {
+    return dependantSplitOn(snapshot, type, whiteThreashold, PDFJS, OpenCV, PDFLib);
 }

package.json

@@ -13,6 +13,7 @@
     "archiver": "^6.0.1",
     "express": "^4.18.2",
     "express-fileupload": "^1.4.1",
+    "jsqr": "^1.4.0",
     "opencv-wasm": "^4.3.0-10",
     "pdf-lib": "^1.17.1",
     "pdfjs-dist": "^2.0.943"

public/functions.js

@@ -13,6 +13,7 @@ import { splitPDF as dependantSplitPDF } from './functions/splitPDF.js';
 import { editMetadata as dependantEditMetadata} from "./functions/editMetadata.js";
 import { organizePages as dependantOrganizePages} from "./functions/organizePages.js";
 import { removeBlankPages as dependantRemoveBlankPages} from "./functions/removeBlankPages.js";
+import { splitOn as dependantSplitOn } from "./functions/splitOn.js";
 
 export async function extractPages(snapshot, pagesToExtractArray) {
     return dependantExtractPages(snapshot, pagesToExtractArray, PDFLib);
@@ -52,4 +53,8 @@ export async function organizePages(snapshot, operation, customOrderString) {
 
 export async function removeBlankPages(snapshot, whiteThreashold) {
     return dependantRemoveBlankPages(snapshot, whiteThreashold, pdfjsLib, OpenCV, PDFLib);
+}
+
+export async function splitOn(snapshot, type, whiteThreashold) {
+    return dependantSplitOn(snapshot, type, whiteThreashold, pdfjsLib, OpenCV, PDFLib);
 }

public/functions/shared/detectEmptyPages.js

@@ -1,3 +1,5 @@
+import { getImagesOnPage } from "./getImagesOnPage.js";
+
 export async function detectEmptyPages(snapshot, whiteThreashold, PDFJS, OpenCV) {
     const pdfDoc = await PDFJS.getDocument(snapshot).promise;
 
@@ -27,15 +29,10 @@ export async function detectEmptyPages(snapshot, whiteThreashold, PDFJS, OpenCV)
     }
 
     async function areImagesBlank(page, threshold) {
-        const ops = await page.getOperatorList();
-    
-        for (var j=0; j < ops.fnArray.length; j++) {
-            if (ops.fnArray[j] == PDFJS.OPS.paintJpegXObject || ops.fnArray[j] == PDFJS.OPS.paintImageXObject) {
-                const image = page.objs.get(ops.argsArray[j][0]);
-                if(image.data) {
-                    return isImageBlank(image, threshold);
-                }
-            }
+        const images = getImagesOnPage(page, PDFJS);
+        for (const image of images) {
+            if(!isImageBlank(image, threshold))
+                return false;
         }
         return true;
     }

public/functions/shared/getImagesOnPage.js

@@ -0,0 +1,11 @@
+export async function getImagesOnPage(page, PDFJS) {
+    const ops = await page.getOperatorList();
+    const images = [];
+    for (var j=0; j < ops.fnArray.length; j++) {
+        if (ops.fnArray[j] == PDFJS.OPS.paintJpegXObject || ops.fnArray[j] == PDFJS.OPS.paintImageXObject) {
+            const image = page.objs.get(ops.argsArray[j][0]);
+            images.push(image);
+        }
+    }
+    return images;
+}

public/functions/splitOn.js

@@ -1,4 +1,7 @@
-import { detectEmptyPages } from "./shared/detectEmptyPages";
+import { detectEmptyPages } from "./shared/detectEmptyPages.js";
+import { getImagesOnPage } from "./shared/getImagesOnPage.js";
+
+import jsQR from "jsQR";
 
 /**
  * @typedef {"BAR_CODE"|"QR_CODE"|"BLANK_PAGE"} SplitType
@@ -9,24 +12,22 @@ import { detectEmptyPages } from "./shared/detectEmptyPages";
  * @param {Uint16Array} snapshot
  * @param {SplitType} type
  * @param {} PDFJS
- * @param {} OpenCV
+ * @param {import('opencv-wasm')} OpenCV
  * @param {} PDFLib
- * @param {} QRCode
  * @returns 
  */
-export async function splitOn(snapshot, type, whiteThreashold, PDFJS, OpenCV, PDFLib, QRCode) {
+export async function splitOn(snapshot, type, whiteThreashold, PDFJS, OpenCV, PDFLib) {
     
     let splitAtPages = [];
 
     switch (type) {
         case "BAR_CODE":
             // TODO: Implement
-            throw new Error("This split-type has not been implemented yet")
+            throw new Error("This split-type has not been implemented yet");
             break;
 
         case "QR_CODE":
-            // TODO: Implement
-            throw new Error("This split-type has not been implemented yet")
+            splitAtPages = await getPagesWithQRCode(snapshot);
             break;
 
         case "BLANK_PAGE":
@@ -45,4 +46,49 @@ export async function splitOn(snapshot, type, whiteThreashold, PDFJS, OpenCV, PD
     // TODO: Remove detected Pages & Split
 
     return pdfDoc.save();
+
+    async function getPagesWithQRCode(snapshot) {
+        const pdfDoc = await PDFJS.getDocument(snapshot).promise;
+
+        const pagesWithQR = [];
+        for (let i = 1; i <= pdfDoc.numPages; i++) {
+            const page = await pdfDoc.getPage(i);
+            console.log("Checking page " + i);
+
+            const images = await getImagesOnPage(page, PDFJS);
+
+            for (const image of images) {
+                const data = await checkForQROnImage(image);
+                if(data == "https://github.com/Frooodle/Stirling-PDF") {
+                    pagesWithQR.push(i);
+                }
+            }
+        }
+        return pagesWithQR;
+    }
+
+    async function checkForQROnImage(image) {
+        console.log(image.data, image.width, image.height, image.width * image.height * 4);
+        
+        // TODO: There is an issue with the jsQR package, and the package seems to be stale, we could create a fork and fix the issue (The package expects rgba but sometimes we have rgb). In the meanwhile we just force rgba:
+        if(image.data.length == image.width * image.height * 3) {
+            const tmpArray = new Uint8ClampedArray(image.width * image.height * 4);
+
+            // Iterate through the original array and add an alpha channel
+            for (let i = 0, j = 0; i < image.data.length; i += 3, j += 4) {
+                tmpArray[j] = image.data[i];     // Red channel
+                tmpArray[j + 1] = image.data[i + 1]; // Green channel
+                tmpArray[j + 2] = image.data[i + 2]; // Blue channel
+                tmpArray[j + 3] = 255;               // Alpha channel (fully opaque)
+            }
+
+            image.data = tmpArray;
+        }
+
+        const code = jsQR(image.data, image.width, image.height);
+        if(code)
+            return code.data;
+        else
+            return null;
+    }
 };

public/traverseOperations.js

@@ -115,6 +115,12 @@ export async function * traverseOperations(operations, input, Functions) {
                     input.buffer = await Functions.removeBlankPages(input.buffer, operation.values["whiteThreashold"]);
                 });
                 break;
+            case "splitOn":
+                yield* oneToN(input, operation, async (input) => {
+                    input.fileName += "_split";
+                    input.buffer = await Functions.splitOn(input.buffer, operation.values["type"], operation.values["whiteThreashold"]);
+                });
+                break;
             default:
                 throw new Error(`${operation.type} not implemented yet.`);
                 break;