esearch-ocr - npm Package Compare versions

Comparing version 5.1.1 to 5.1.2

package.json

 {
     "name": "esearch-ocr",
-    "version": "5.1.1",
+    "version": "5.1.2",
     "description": "paddleocr models run on onnx",
@@ -5,0 +5,0 @@ "main": "./dist/esearch-ocr.umd.cjs",

src/main.ts

@@ -50,2 +50,4 @@ // biome-ignore lint/suspicious/noImplicitAnyLet: 可自定义cv
 
+let onProgress = (type: "det" | "rec", total: number, count: number) => {};
+
 async function init(op: {
@@ -69,2 +71,4 @@ detPath?: string;
     cv?;
+
+    onProgress?: (type: "det" | "rec", total: number, count: number) => void;
 }) {
@@ -96,2 +100,3 @@ ort = op.ort;
     else if (typeof require !== "undefined") cv = require("opencv.js");
+    if (op.onProgress) onProgress = op.onProgress;
     return { ocr: x, det: Det, rec: Rec };
@@ -109,2 +114,3 @@ }
     const box = await Det(img);
+    onProgress("det", 1, 1);
 
@@ -114,3 +120,3 @@ const mainLine = await Rec(box);
     const newMainLine = afAfRec(mainLine);
-    console.log(mainLine, newMainLine);
+    log(mainLine, newMainLine);
     task.l("end");
@@ -151,5 +157,8 @@ return { src: mainLine, ...newMainLine };
     const recL = beforeRec(box);
+    let runCount = 0;
     const recPromises = recL.map(async (item) => {
         const { b, imgH, imgW } = item;
         const recResults = await runRec(b, imgH, imgW, rec);
+        runCount++;
+        onProgress("rec", recL.length, runCount);
         return afterRec(recResults, dic);
@@ -214,3 +223,3 @@ });
     const transposedData = toPaddleInput(image, [0.485, 0.456, 0.406], [0.229, 0.224, 0.225]);
-    console.log(image);
+    log(image);
     if (dev) {
@@ -288,3 +297,3 @@ const srcCanvas = data2canvas(image);
 
-    console.log(edgeRect);
+    log(edgeRect);
 
@@ -627,2 +636,54 @@ src.delete();
     // 短轴扩散，合并为段
+
+    const newL_ = structuredClone(l).sort((a, b) => a.box[0][1] - b.box[0][1]) as resultType[0][];
+    const newLZ: resultType[0][][] = [];
+    // 合并行
+    for (const j of newL_) {
+        const last = newLZ.at(-1)?.at(-1);
+        if (!last) {
+            newLZ.push([j]);
+            continue;
+        }
+        const thisCy = (j.box[2][1] + j.box[0][1]) / 2;
+        const lastCy = (last.box[2][1] + last.box[0][1]) / 2;
+        if (Math.abs(thisCy - lastCy) < 0.5 * (j.box[2][1] - j.box[0][1])) {
+            const lLast = newLZ.at(-1);
+            if (!lLast) {
+                newLZ.push([j]);
+            } else {
+                lLast.push(j);
+            }
+        } else {
+            newLZ.push([j]);
+        }
+    }
+
+    // 根据距离，合并或保持拆分
+    const newL: (resultType[0] | null)[] = [];
+    for (const l of newLZ) {
+        if (l.length === 1) {
+            newL.push(l.at(0) as resultType[0]);
+            continue;
+        }
+
+        const em = average(l.map((i) => i.box[2][1] - i.box[0][1]));
+        l.sort((a, b) => a.box[0][0] - b.box[0][0]);
+
+        let last = l.at(0) as resultType[0];
+
+        for (const this_ of l.slice(1)) {
+            const lastBoxRightX = last.box[1][0] ?? Number.NEGATIVE_INFINITY;
+            const thisLeftX = this_.box[0][0];
+            if (thisLeftX - lastBoxRightX > em) {
+                newL.push(last);
+                last = this_;
+            } else {
+                last.text += this_.text;
+                last.mean = (last.mean + this_.mean) / 2;
+                last.box = outerRect([last.box, this_.box]);
+            }
+        }
+        newL.push(last);
+    }
+
     // todo 分割线为边界
@@ -633,3 +694,2 @@ // 分栏
 
-    const newL = structuredClone(l).sort((a, b) => a.box[0][1] - b.box[0][1]) as (resultType[0] | null)[];
     const maxY = newL.reduce((a, b) => Math.max(a, b?.box[2][1] ?? 0), 0);
@@ -688,3 +748,3 @@ for (let i = 0; i <= maxY; i++) {
             if (!last) continue;
-            const jl = r(centerPoint(b.box), centerPoint(last.box));
+            const jl = centerPoint(b.box)[1] - centerPoint(last.box)[1];
             if (jl < _jl) {
@@ -777,2 +837,3 @@ nearest = Number(i);
         const c = v.src;
+        // gap feq
         const gs: Record<number, number> = {};
@@ -794,6 +855,6 @@ for (let i = 1; i < c.length; i++) {
             const maxN = Math.max(...Object.values(gs));
+            const minX = Math.min(...Object.keys(gs).map(Number));
             const gapsL = Object.keys(gs)
                 .sort()
-                .map(Number)
-                .filter((g) => g > 0)
+                .map((i) => Number(i) - minX)
                 .filter((g) => gs[g] !== maxN); // 去掉一个最大值
@@ -830,2 +891,15 @@ for (let i = 1; i < gapsL.length; i++) {
 
+    if (dev) {
+        const color: string[] = [];
+        for (let h = 10; h < 360; h += Math.floor(360 / p.length)) {
+            color.push(`hsl(${h}, 100%, 50%)`);
+        }
+
+        for (const i in p) {
+            for (const b of p[i].parragraphs) {
+                drawBox(b.parse.box, b.parse.text, color[i]);
+            }
+        }
+    }
+
     const pss = p.flatMap((v) => v.parragraphs.map((p) => p.parse));
@@ -868,3 +942,4 @@
     ctx.stroke();
+    ctx.strokeStyle = "black";
     ctx.strokeText(id, box[0][0], box[0][1]);
 }

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

320730

increased by1.4%

Lines of code

4855

increased by2.3%

No dependency changes