mindee 4.36.0 → 4.36.1

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # CHANGELOG
 
+## v4.36.1 - 2025-12-30
+### Fixes
+* :bug: extracted images should take rotation into account
+* :recycle: fix minor issues with param syntaxes
+
+
 ## v4.36.0 - 2025-12-19
 ### Changes
 * :sparkles: add support for dataschema param

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mindee",
-    "version": "4.36.0",
+    "version": "4.36.1",
     "description": "Mindee Client Library for Node.js",
     "main": "src/index.js",
     "bin": "bin/mindee.js",

package/src/geometry/polygonUtils.d.ts

@@ -44,3 +44,4 @@ export declare function getMinYCoordinate(polygon: Array<Point>): number;
 export declare function getMinXCoordinate(polygon: Array<Point>): number;
 export declare function compareOnY(polygon1: Array<Point>, polygon2: Array<Point>): number;
 export declare function compareOnX(polygon1: Array<Point>, polygon2: Array<Point>): number;
+export declare function adjustForRotation(polygon: Array<Point>, orientation: number): Array<Point>;

package/src/geometry/polygonUtils.js

@@ -13,6 +13,7 @@ exports.getMinYCoordinate = getMinYCoordinate;
 exports.getMinXCoordinate = getMinXCoordinate;
 exports.compareOnY = compareOnY;
 exports.compareOnX = compareOnX;
+exports.adjustForRotation = adjustForRotation;
 /**
  * Get the central point (centroid) given a list of points.
  */
@@ -124,3 +125,15 @@ function compareOnX(polygon1, polygon2) {
     }
     return sort < 0 ? -1 : 1;
 }
+function adjustForRotation(polygon, orientation) {
+    if (orientation === 90) {
+        return polygon.map(([x, y]) => [y, 1 - x]);
+    }
+    if (orientation === 180) {
+        return polygon.map(([x, y]) => [1 - x, 1 - y]);
+    }
+    if (orientation === 270) {
+        return polygon.map(([x, y]) => [1 - y, x]);
+    }
+    return polygon;
+}

package/src/imageOperations/common/imageExtractor.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.extractFromPage = extractFromPage;
 const pdf_lib_1 = require("@cantoo/pdf-lib");
 const geometry_1 = require("../../geometry");
+const polygonUtils_1 = require("../../geometry/polygonUtils");
 /**
  * Extracts elements from a page based off of a list of bounding boxes.
  *
@@ -15,7 +16,9 @@ async function extractFromPage(pdfPage, polygons) {
     // Manual upscale.
     // Fixes issues with the OCR.
     const qualityScale = 300 / 72;
-    for (const polygon of polygons) {
+    const orientation = pdfPage.getRotation().angle;
+    for (const origPolygon of polygons) {
+        const polygon = (0, polygonUtils_1.adjustForRotation)(origPolygon, orientation);
         const tempPdf = await pdf_lib_1.PDFDocument.create();
         const newWidth = width * ((0, geometry_1.getMinMaxX)(polygon).max - (0, geometry_1.getMinMaxX)(polygon).min);
         const newHeight = height * ((0, geometry_1.getMinMaxY)(polygon).max - (0, geometry_1.getMinMaxY)(polygon).min);
@@ -25,17 +28,59 @@ async function extractFromPage(pdfPage, polygons) {
             top: height - ((0, geometry_1.getMinMaxY)(polygon).min * height),
             bottom: height - ((0, geometry_1.getMinMaxY)(polygon).max * height),
         });
-        const samplePage = tempPdf.addPage([newWidth * qualityScale, newHeight * qualityScale]);
+        // Determine the final page dimensions based on orientation
+        let finalWidth;
+        let finalHeight;
+        if (orientation === 90 || orientation === 270) {
+            // For 90/270 rotations, swap width and height
+            finalWidth = newHeight * qualityScale;
+            finalHeight = newWidth * qualityScale;
+        }
+        else {
+            finalWidth = newWidth * qualityScale;
+            finalHeight = newHeight * qualityScale;
+        }
+        const samplePage = tempPdf.addPage([finalWidth, finalHeight]);
         samplePage.drawRectangle({
             x: 0,
             y: 0,
-            width: newWidth * qualityScale,
-            height: newHeight * qualityScale,
-        });
-        samplePage.drawPage(cropped, {
-            width: newWidth * qualityScale,
-            height: newHeight * qualityScale,
+            width: finalWidth,
+            height: finalHeight,
         });
+        // Draw the cropped page with rotation applied
+        if (orientation === 0) {
+            samplePage.drawPage(cropped, {
+                width: newWidth * qualityScale,
+                height: newHeight * qualityScale,
+            });
+        }
+        else if (orientation === 90) {
+            samplePage.drawPage(cropped, {
+                x: 0,
+                y: finalHeight,
+                width: newWidth * qualityScale,
+                height: newHeight * qualityScale,
+                rotate: (0, pdf_lib_1.degrees)(270),
+            });
+        }
+        else if (orientation === 180) {
+            samplePage.drawPage(cropped, {
+                x: finalWidth,
+                y: finalHeight,
+                width: newWidth * qualityScale,
+                height: newHeight * qualityScale,
+                rotate: (0, pdf_lib_1.degrees)(180),
+            });
+        }
+        else if (orientation === 270) {
+            samplePage.drawPage(cropped, {
+                x: finalWidth,
+                y: 0,
+                width: newWidth * qualityScale,
+                height: newHeight * qualityScale,
+                rotate: (0, pdf_lib_1.degrees)(90),
+            });
+        }
         extractedElements.push(await tempPdf.save());
     }
     return extractedElements;

package/src/imageOperations/multiReceiptsExtractor/multiReceiptsExtractor.js

@@ -65,6 +65,7 @@ async function extractReceipts(inputFile, inference) {
     const pdfDoc = await loadPdfDoc(inputFile);
     for (let pageId = 0; pageId < pdfDoc.getPageCount(); pageId++) {
         const [page] = await pdfDoc.copyPages(pdfDoc, [pageId]);
+        page.setRotation((0, pdf_lib_1.degrees)(inference.pages[pageId].orientation?.value ?? 0));
         const receiptPositions = inference.pages[pageId].prediction.receipts.map((receipt) => receipt.boundingBox);
         const extractedReceipts = await extractReceiptsFromPage(page, receiptPositions, pageId);
         images.push(...extractedReceipts);

package/src/input/dataSchema.js

@@ -71,8 +71,8 @@ class DataSchema {
         if (typeof dataSchema === "string") {
             this.replace = new DataSchemaReplace(JSON.parse(dataSchema)["replace"]);
         }
-        else if (dataSchema["replace"] instanceof DataSchemaReplace) {
-            this.replace = dataSchema["replace"];
+        else if (dataSchema instanceof DataSchema) {
+            this.replace = dataSchema.replace;
         }
         else {
             this.replace = new DataSchemaReplace(dataSchema["replace"]);