mirrors/CyberChef
1
Fork 0
mirror of https://github.com/gchq/CyberChef.git synced 2026-01-30 12:20:33 -08:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

feat(image+ocr): add OCR quick-action on Render Image and OCR whitelist option

Browse source 
- Render Image (browser): overlay a small ✎ button linking to an OCR recipe for quick access
- OCR: add optional "Character whitelist" arg and apply via tesseract parameter to improve accuracy on restricted alphabets

Authored-by: Izai Alejandro Zalles Merino <zallesrene@gmail.com>
This commit is contained in:
Izai Alejandro Zalles Merino 2025-10-13 10:02:48 -04:00
parent de2a83c0a0
commit d3c512e878
2 changed files with 18 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

12
src/core/operations/OpticalCharacterRecognition.mjs
View file

@ -15,6 +15,7 @@ import { isWorkerEnvironment } from "../Utils.mjs";
import { createWorker } from "tesseract.js";
const OEM_MODES = ["Tesseract only", "LSTM only", "Tesseract/LSTM Combined"];
const OCR_DEFAULT_WHITELIST = "";
/**
* Optical Character Recognition operation
@ -34,6 +35,11 @@ class OpticalCharacterRecognition extends Operation {
this.inputType = "ArrayBuffer";
this.outputType = "string";
this.args = [
{
name: "Character whitelist (optional)",
type: "string",
value: OCR_DEFAULT_WHITELIST
},
{
name: "Show confidence",
type: "boolean",
@ -61,7 +67,7 @@ class OpticalCharacterRecognition extends Operation {
* @returns {string}
*/
async run(input, args) {
const [showConfidence, oemChoice] = args;
const [whitelist, showConfidence, oemChoice] = args;
if (!isWorkerEnvironment()) throw new OperationError("This operation only works in a browser");
@ -86,6 +92,10 @@ class OpticalCharacterRecognition extends Operation {
}
}
});
self.sendStatusMessage("Configuring OCR parameters...");
if (whitelist && whitelist.length) {
await worker.setParameters({ /* eslint-disable camelcase */ tessedit_char_whitelist: whitelist /* eslint-enable camelcase */ });
}
self.sendStatusMessage("Finding text...");
const result = await worker.recognize(image);

9
src/core/operations/RenderImage.mjs
View file

@ -8,7 +8,7 @@ import { fromBase64, toBase64 } from "../lib/Base64.mjs";
import { fromHex } from "../lib/Hex.mjs";
import Operation from "../Operation.mjs";
import OperationError from "../errors/OperationError.mjs";
import Utils from "../Utils.mjs";
import Utils, { isWorkerEnvironment } from "../Utils.mjs";
import {isImage} from "../lib/FileType.mjs";
/**
@ -104,7 +104,12 @@ class RenderImage extends Operation {
// Add image data to URI
dataURI += "base64," + toBase64(data);
return "<img src='" + dataURI + "'>";
let html = "<img src='" + dataURI + "'>";
if (isWorkerEnvironment()) {
const ocrLink = "#recipe=Optical_Character_Recognition('Show confidence',true,'OCR Engine Mode','LSTM only')";
html = "<div style=\"position:relative; display:inline-block;\">" + html + "<a href=\"" + ocrLink + "\" title=\"Run OCR\" style=\"position:absolute; top:8px; right:8px; background:rgba(0,0,0,.6); color:#fff; padding:4px 6px; border-radius:4px; text-decoration:none; font-weight:bold;\">✎</a></div>";
}
return html;
}
}