mindee 4.2.0 → 4.3.1

package/CHANGELOG.md

@@ -1,5 +1,15 @@
 # CHANGELOG
 
+## v4.3.1 - 2023-09-20
+### Fixes
+* :bug: take line height tolerance into account when evaluating fields
+
+
+## v4.3.0 - 2023-09-19
+### Changes
+* :sparkles: add line items reconstruction for API builder
+
+
 ## v4.2.0 - 2023-09-15
 ### Changes
 * :sparkles: add support for W9 V1

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mindee",
-    "version": "4.2.0",
+    "version": "4.3.1",
     "description": "Mindee Client Library for Node.js",
     "main": "src/index.js",
     "bin": "bin/mindee.js",

package/src/geometry/boundingBox.d.ts

@@ -1,5 +1,11 @@
 import { Point } from "./point";
 /** A simple bounding box defined by 4 coordinates: xMin, yMin, xMax, yMax */
-export type BBox = [number, number, number, number];
+export declare class BBox {
+    xMin: number;
+    yMin: number;
+    xMax: number;
+    yMax: number;
+    constructor(xMin: number, yMin: number, xMax: number, yMax: number);
+}
 /** A bounding box defined by 4 points. */
 export type BoundingBox = [Point, Point, Point, Point];

package/src/geometry/boundingBox.js

@@ -1,2 +1,13 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.BBox = void 0;
+/** A simple bounding box defined by 4 coordinates: xMin, yMin, xMax, yMax */
+class BBox {
+    constructor(xMin, yMin, xMax, yMax) {
+        this.xMin = xMin;
+        this.yMin = yMin;
+        this.xMax = xMax;
+        this.yMax = yMax;
+    }
+}
+exports.BBox = BBox;

package/src/geometry/boundingBoxUtils.d.ts

@@ -13,10 +13,10 @@ export declare function getBoundingBoxFromBBox(bbox: BBox): BoundingBox;
  */
 export declare function mergeBbox(bbox1: BBox, bbox2: BBox): BBox;
 /**
- * Given a Polygon, calculate a bounding box that encompasses all points.
+ * Given a Polygon, calculate a BBox that encompasses all points.
  */
 export declare function getBbox(polygon: Polygon): BBox;
 /**
- * Given polygons, calculate a bounding box that encompasses all points.
+ * Given polygons, calculate a BBox that encompasses all points.
  */
 export declare function getBBoxForPolygons(polygons: Polygon[]): BBox;

package/src/geometry/boundingBoxUtils.js

@@ -1,17 +1,13 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.getBBoxForPolygons = exports.getBbox = exports.mergeBbox = exports.getBoundingBoxFromBBox = exports.getBoundingBox = void 0;
+const boundingBox_1 = require("./boundingBox");
 /**
  * Given a Polygon, calculate a polygon that encompasses all points.
  */
 function getBoundingBox(polygon) {
     const bbox = getBbox(polygon);
-    return [
-        [bbox[0], bbox[1]],
-        [bbox[2], bbox[1]],
-        [bbox[2], bbox[3]],
-        [bbox[0], bbox[3]],
-    ];
+    return getBoundingBoxFromBBox(bbox);
 }
 exports.getBoundingBox = getBoundingBox;
 /**
@@ -19,10 +15,10 @@ exports.getBoundingBox = getBoundingBox;
  */
 function getBoundingBoxFromBBox(bbox) {
     return [
-        [bbox[0], bbox[1]],
-        [bbox[2], bbox[1]],
-        [bbox[2], bbox[3]],
-        [bbox[0], bbox[3]],
+        [bbox.xMin, bbox.yMin],
+        [bbox.xMax, bbox.yMin],
+        [bbox.xMax, bbox.yMax],
+        [bbox.xMin, bbox.yMax],
     ];
 }
 exports.getBoundingBoxFromBBox = getBoundingBoxFromBBox;
@@ -30,37 +26,24 @@ exports.getBoundingBoxFromBBox = getBoundingBoxFromBBox;
  * Given 2 bbox, merge them.
  */
 function mergeBbox(bbox1, bbox2) {
-    return [
-        Math.min(bbox1[0], bbox2[0]),
-        Math.min(bbox1[1], bbox2[1]),
-        Math.max(bbox1[2], bbox2[2]),
-        Math.max(bbox1[3], bbox2[3]),
-    ];
+    return new boundingBox_1.BBox(Math.min(bbox1.xMin, bbox2.xMin), Math.min(bbox1.yMin, bbox2.yMin), Math.max(bbox1.xMax, bbox2.xMax), Math.max(bbox1.yMax, bbox2.yMax));
 }
 exports.mergeBbox = mergeBbox;
 /**
- * Given a Polygon, calculate a bounding box that encompasses all points.
+ * Given a Polygon, calculate a BBox that encompasses all points.
  */
 function getBbox(polygon) {
     const allY = polygon.map((point) => point[1]);
     const allX = polygon.map((point) => point[0]);
-    const yMin = Math.min(...allY);
-    const yMax = Math.max(...allY);
-    const xMin = Math.min(...allX);
-    const xMax = Math.max(...allX);
-    return [xMin, yMin, xMax, yMax];
+    return new boundingBox_1.BBox(Math.min(...allX), Math.min(...allY), Math.max(...allX), Math.max(...allY));
 }
 exports.getBbox = getBbox;
 /**
- * Given polygons, calculate a bounding box that encompasses all points.
+ * Given polygons, calculate a BBox that encompasses all points.
  */
 function getBBoxForPolygons(polygons) {
     const allY = polygons.flatMap((polygon) => polygon.map((point) => point[1]));
     const allX = polygons.flatMap((polygon) => polygon.map((point) => point[0]));
-    const yMin = Math.min(...allY);
-    const yMax = Math.max(...allY);
-    const xMin = Math.min(...allX);
-    const xMax = Math.max(...allX);
-    return [xMin, yMin, xMax, yMax];
+    return new boundingBox_1.BBox(Math.min(...allX), Math.min(...allY), Math.max(...allX), Math.max(...allY));
 }
 exports.getBBoxForPolygons = getBBoxForPolygons;

package/src/geometry/index.js

@@ -1,6 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.mergeBbox = exports.getBoundingBoxFromBBox = exports.getBoundingBox = exports.getBBoxForPolygons = exports.getBbox = exports.getMinMaxY = exports.getMinMaxX = exports.relativeY = exports.relativeX = exports.isPointInPolygonY = exports.isPointInPolygonX = exports.getCentroid = exports.compareOnY = exports.compareOnX = exports.Polygon = void 0;
+exports.mergeBbox = exports.getBoundingBoxFromBBox = exports.getBoundingBox = exports.getBBoxForPolygons = exports.getBbox = exports.BBox = exports.getMinMaxY = exports.getMinMaxX = exports.relativeY = exports.relativeX = exports.isPointInPolygonY = exports.isPointInPolygonX = exports.getCentroid = exports.compareOnY = exports.compareOnX = exports.Polygon = void 0;
 var polygon_1 = require("./polygon");
 Object.defineProperty(exports, "Polygon", { enumerable: true, get: function () { return polygon_1.Polygon; } });
 var polygonUtils_1 = require("./polygonUtils");
@@ -13,6 +13,8 @@ Object.defineProperty(exports, "relativeX", { enumerable: true, get: function ()
 Object.defineProperty(exports, "relativeY", { enumerable: true, get: function () { return polygonUtils_1.relativeY; } });
 Object.defineProperty(exports, "getMinMaxX", { enumerable: true, get: function () { return polygonUtils_1.getMinMaxX; } });
 Object.defineProperty(exports, "getMinMaxY", { enumerable: true, get: function () { return polygonUtils_1.getMinMaxY; } });
+var boundingBox_1 = require("./boundingBox");
+Object.defineProperty(exports, "BBox", { enumerable: true, get: function () { return boundingBox_1.BBox; } });
 var boundingBoxUtils_1 = require("./boundingBoxUtils");
 Object.defineProperty(exports, "getBbox", { enumerable: true, get: function () { return boundingBoxUtils_1.getBbox; } });
 Object.defineProperty(exports, "getBBoxForPolygons", { enumerable: true, get: function () { return boundingBoxUtils_1.getBBoxForPolygons; } });

package/src/parsing/custom/index.d.ts

@@ -1,3 +1,3 @@
-export { Line, LineItems, getLineItems } from "./lineItems";
+export { CustomLine, CustomLines, getLineItems } from "./lineItems";
 export { ClassificationField } from "./classificationField";
 export { ListField, ListFieldValue } from "./listField";

package/src/parsing/custom/index.js

@@ -1,9 +1,9 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.ListFieldValue = exports.ListField = exports.ClassificationField = exports.getLineItems = exports.LineItems = exports.Line = void 0;
+exports.ListFieldValue = exports.ListField = exports.ClassificationField = exports.getLineItems = exports.CustomLines = exports.CustomLine = void 0;
 var lineItems_1 = require("./lineItems");
-Object.defineProperty(exports, "Line", { enumerable: true, get: function () { return lineItems_1.Line; } });
-Object.defineProperty(exports, "LineItems", { enumerable: true, get: function () { return lineItems_1.LineItems; } });
+Object.defineProperty(exports, "CustomLine", { enumerable: true, get: function () { return lineItems_1.CustomLine; } });
+Object.defineProperty(exports, "CustomLines", { enumerable: true, get: function () { return lineItems_1.CustomLines; } });
 Object.defineProperty(exports, "getLineItems", { enumerable: true, get: function () { return lineItems_1.getLineItems; } });
 var classificationField_1 = require("./classificationField");
 Object.defineProperty(exports, "ClassificationField", { enumerable: true, get: function () { return classificationField_1.ClassificationField; } });

package/src/parsing/custom/lineItems.d.ts

@@ -1,6 +1,6 @@
 import { BBox, Polygon } from "../../geometry";
 import { ListField, ListFieldValue } from "./listField";
-export declare class Line {
+export declare class CustomLine {
     /**
      * Number of the current line.
      * Starts at 1.
@@ -11,15 +11,10 @@ export declare class Line {
      */
     fields: Map<string, ListFieldValue>;
     /**
-     * The BBox of the current line.
+     * The BBox of the entire line, all fields included.
      */
     bbox: BBox;
-    /**
-     * The height tolerance used to build the line.
-     * It helps when the height of a line can vary unexpectedly.
-     */
-    heightTolerance: number;
-    constructor(rowNumber: number, heightTolerance: number);
+    constructor(rowNumber: number);
     /**
      * Extends the current bbox of the line with the bbox.
      */
@@ -28,14 +23,11 @@ export declare class Line {
      * Extends the current bbox of the line with the polygon.
      */
     extendWith(polygon: Polygon): void;
-    /**
-     * Check if the bbox fits the current line.
-     */
-    contains(bbox: BBox): boolean;
     updateField(name: string, fieldValue: ListFieldValue): void;
 }
-export declare class LineItems {
-    rows: Line[];
-    constructor(lines: Line[]);
+export declare class CustomLines extends Array<CustomLine> {
 }
-export declare function getLineItems(anchorNames: string[], heightLineTolerance: number, fieldNamesTargeted: string[], fields: Map<string, ListField>): LineItems;
+/**
+ * Get line items from fields.
+ */
+export declare function getLineItems(anchorNames: string[], fieldNames: string[], fields: Map<string, ListField>, heightLineTolerance: number): CustomLines;

package/src/parsing/custom/lineItems.js

@@ -1,16 +1,15 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.getLineItems = exports.LineItems = exports.Line = void 0;
+exports.getLineItems = exports.CustomLines = exports.CustomLine = void 0;
 const handler_1 = require("../../errors/handler");
 const errors_1 = require("../../errors");
 const geometry_1 = require("../../geometry");
 const listField_1 = require("./listField");
-class Line {
-    constructor(rowNumber, heightTolerance) {
+class CustomLine {
+    constructor(rowNumber) {
         this.rowNumber = rowNumber;
-        this.bbox = [1, 1, 0, 0];
+        this.bbox = new geometry_1.BBox(1, 1, 0, 0);
         this.fields = new Map();
-        this.heightTolerance = heightTolerance;
     }
     /**
      * Extends the current bbox of the line with the bbox.
@@ -24,15 +23,6 @@ class Line {
     extendWith(polygon) {
         this.bbox = (0, geometry_1.mergeBbox)(this.bbox, (0, geometry_1.getBbox)(polygon));
     }
-    /**
-     * Check if the bbox fits the current line.
-     */
-    contains(bbox) {
-        if (Math.abs(bbox[1] - this.bbox[1]) <= this.heightTolerance) {
-            return true;
-        }
-        return Math.abs(this.bbox[1] - bbox[1]) <= this.heightTolerance;
-    }
     updateField(name, fieldValue) {
         if (!this.fields.has(name)) {
             this.fields.set(name, fieldValue);
@@ -58,32 +48,34 @@ class Line {
         }
     }
 }
-exports.Line = Line;
-class LineItems {
-    constructor(lines) {
-        this.rows = [];
-        this.rows = lines;
-    }
+exports.CustomLine = CustomLine;
+class CustomLines extends Array {
 }
-exports.LineItems = LineItems;
-function getLineItems(anchorNames, heightLineTolerance, fieldNamesTargeted, fields) {
-    const fieldsToTransformIntoLines = new Map([...fields].filter(([k]) => fieldNamesTargeted.includes(k)));
+exports.CustomLines = CustomLines;
+/**
+ * Get line items from fields.
+ */
+function getLineItems(anchorNames, fieldNames, fields, heightLineTolerance) {
+    const fieldsToTransformIntoLines = new Map([...fields].filter(([k]) => fieldNames.includes(k)));
     const anchorName = findBestAnchor(anchorNames, fieldsToTransformIntoLines);
-    const lineItemsPrepared = prepare(anchorName, fieldsToTransformIntoLines, heightLineTolerance);
-    lineItemsPrepared.rows.forEach((currentLine) => {
+    const linesPrepared = prepare(anchorName, fieldsToTransformIntoLines, heightLineTolerance);
+    linesPrepared.forEach((currentLine) => {
         fieldsToTransformIntoLines.forEach((field, fieldName) => {
             field.values.forEach((listFieldValue) => {
-                const minYCurrentValue = (0, geometry_1.getMinMaxY)(listFieldValue.polygon).min;
-                if (minYCurrentValue < currentLine.bbox[3] &&
-                    minYCurrentValue >= currentLine.bbox[1]) {
+                const minMaxY = (0, geometry_1.getMinMaxY)(listFieldValue.polygon);
+                if (Math.abs(minMaxY.max - currentLine.bbox.yMax) <= heightLineTolerance &&
+                    Math.abs(minMaxY.min - currentLine.bbox.yMin) <= heightLineTolerance) {
                     currentLine.updateField(fieldName, listFieldValue);
                 }
             });
         });
     });
-    return lineItemsPrepared;
+    return linesPrepared;
 }
 exports.getLineItems = getLineItems;
+/**
+ * Loop through the possible anchor fields and find the one with the most values.
+ */
 function findBestAnchor(possibleAnchorNames, fields) {
     let anchorName = "";
     let anchorRows = 0;
@@ -99,31 +91,40 @@ function findBestAnchor(possibleAnchorNames, fields) {
     }
     return anchorName;
 }
+/**
+ * Check if the bbox fits inside the line.
+ */
+function isBboxInLine(line, bbox, heightTolerance) {
+    if (Math.abs(bbox.yMin - line.bbox.yMin) <= heightTolerance) {
+        return true;
+    }
+    return Math.abs(line.bbox.yMin - bbox.yMin) <= heightTolerance;
+}
 function prepare(anchorName, fields, heightLineTolerance) {
-    const lineItemsPrepared = [];
+    const linesPrepared = [];
     const anchorField = fields.get(anchorName);
     if (anchorField === undefined || anchorField.values.length === 0) {
         handler_1.errorHandler.throw(new errors_1.MindeeError("No lines have been detected."));
     }
     let currentLineNumber = 1;
-    let currentLine = new Line(currentLineNumber, heightLineTolerance);
+    let currentLine = new CustomLine(currentLineNumber);
     if (anchorField !== undefined) {
         let currentValue = anchorField.values[0];
         currentLine.extendWith(currentValue.polygon);
         for (let index = 1; index < anchorField.values.length; index++) {
             currentValue = anchorField.values[index];
             const currentFieldBbox = (0, geometry_1.getBbox)(currentValue.polygon);
-            if (!currentLine.contains(currentFieldBbox)) {
-                lineItemsPrepared.push(currentLine);
+            if (!isBboxInLine(currentLine, currentFieldBbox, heightLineTolerance)) {
+                linesPrepared.push(currentLine);
                 currentLineNumber++;
-                currentLine = new Line(currentLineNumber, heightLineTolerance);
+                currentLine = new CustomLine(currentLineNumber);
             }
             currentLine.extendWithBbox(currentFieldBbox);
         }
-        if (lineItemsPrepared.filter((line) => line.rowNumber === currentLineNumber)
+        if (linesPrepared.filter((line) => line.rowNumber === currentLineNumber)
             .length === 0) {
-            lineItemsPrepared.push(currentLine);
+            linesPrepared.push(currentLine);
         }
     }
-    return new LineItems(lineItemsPrepared);
+    return linesPrepared;
 }

package/src/product/custom/customV1Document.d.ts

@@ -1,5 +1,5 @@
 import { StringDict, Prediction } from "../../parsing/common";
-import { ClassificationField, ListField } from "../../parsing/custom";
+import { ClassificationField, ListField, CustomLines } from "../../parsing/custom";
 /**
  * Document data for Custom builds.
  */
@@ -18,6 +18,13 @@ export declare class CustomV1Document implements Prediction {
      * @param pageId page the field was found on.
      */
     protected setField(fieldName: string, fieldValue: any, pageId?: number): void;
+    /**
+     * Order column fields into line items.
+     * @param anchorNames list of possible anchor fields.
+     * @param fieldNames list of all column fields.
+     * @param heightTolerance height tolerance to apply to lines.
+     */
+    columnsToLineItems(anchorNames: string[], fieldNames: string[], heightTolerance?: number): CustomLines;
     /**
      * Default string representation.
      */

package/src/product/custom/customV1Document.js

@@ -40,6 +40,15 @@ class CustomV1Document {
             throw new Error(`Unknown API field type for field ${fieldName} : ${fieldValue}`);
         }
     }
+    /**
+     * Order column fields into line items.
+     * @param anchorNames list of possible anchor fields.
+     * @param fieldNames list of all column fields.
+     * @param heightTolerance height tolerance to apply to lines.
+     */
+    columnsToLineItems(anchorNames, fieldNames, heightTolerance = 0.01) {
+        return (0, custom_1.getLineItems)(anchorNames, fieldNames, this.fields, heightTolerance);
+    }
     /**
      * Default string representation.
      */

package/src/product/custom/customV1Page.d.ts

@@ -1,5 +1,5 @@
 import { StringDict, Prediction } from "../../parsing/common";
-import { ListField } from "../../parsing/custom";
+import { ListField, CustomLines } from "../../parsing/custom";
 /**
  * Page data for Custom builds.
  */
@@ -7,6 +7,13 @@ export declare class CustomV1Page implements Prediction {
     /** List of page-specific fields for a Custom build. Cannot include Classification fields. */
     fields: Map<string, ListField>;
     constructor(rawPrediction: StringDict, pageId?: number);
+    /**
+     * Order column fields into line items.
+     * @param anchorNames list of possible anchor fields.
+     * @param fieldNames list of all column fields.
+     * @param heightTolerance height tolerance to apply to lines.
+     */
+    columnsToLineItems(anchorNames: string[], fieldNames: string[], heightTolerance?: number): CustomLines;
     /**
      * Default string representation.
      */

package/src/product/custom/customV1Page.js

@@ -17,6 +17,15 @@ class CustomV1Page {
             }));
         });
     }
+    /**
+     * Order column fields into line items.
+     * @param anchorNames list of possible anchor fields.
+     * @param fieldNames list of all column fields.
+     * @param heightTolerance height tolerance to apply to lines.
+     */
+    columnsToLineItems(anchorNames, fieldNames, heightTolerance = 0.01) {
+        return (0, custom_1.getLineItems)(anchorNames, fieldNames, this.fields, heightTolerance);
+    }
     /**
      * Default string representation.
      */