panleicim/puppeteerjs
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

use jimp

Browse Source
This commit is contained in:
Lei PAN
2023-05-26 23:38:06 +02:00
parent b42d257ab9
commit 4de3bf606c
4 changed files with 128 additions and 105 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/workers/OCRChecker.js
+46 -26
View File
@@ -1,6 +1,7 @@
const {v4: uuidv4} = require("uuid");
const tesseract = require("node-tesseract-ocr");
const OCRResult = require("../models/OCRResult");
const Jimp = require('jimp');
function delay(delayInMs) {
return new Promise(resolve => {
@@ -30,9 +31,18 @@ const WELCOME_MESSAGE_FR = "Bienvenue dans Chrome"
const PAGE_OPTIMIZATION_CHROME_FR = "Vous pouvez changer d'avis a tout moment dans"
const PAGE_OPTIMIZATION_CHROME_FR_2 = "Vous pouvez modifier vos options a tout moment"
const PAGE_OPTIMIZATION_CHROME_FR_3 = "Vous pouvez effectuer des modifications"
const PAGE_OPTIMIZATION_CHROME_FR_4 = "de nouvelles fonctionnalités"
const PAGE_OPTIMIZATION_CHROME_FR_5 = "Avec la mesure des performance"
const ONLINE_APPOINTMENT = "Online Appointment"
const CONFIRM_RESEND_FORM_FR = "Confirmer le nouvel envoi"
const CLOSED_MESSAGE_FR = "Depuis plus de 130 ans"
const ABOUT_BLANK = " about:blank"
async function convertImageToWhiteBlack(image_path) {
const image = await Jimp.read(image_path);
image.grayscale().write(image_path + ".wb");
return image_path + ".wb";
}
class OCRChecker {
@@ -46,36 +56,46 @@ class OCRChecker {
return this.contact.passportNumber + "_" + uuid + ".png"
}
async get_result() {
let fileName = await this.take_screen_shot()
let result = await tesseract
.recognize(fileName, config)
console.log(result)
if (result.includes(MESSAGE_URL_VALIDATION_EN) || result.includes(MESSAGE_URL_VALIDATION_FR)) {
return OCRResult.SUCCESS
} else if (result.includes(CAPTCHA_ERROR_MESSAGE) || result.includes(CAPTCHA_ERROR_MESSAGE_FR)) {
return OCRResult.RECAPTCHA_ERROR
} else if (result.includes(BLOCKED_MSG_EN) || result.includes(BLOCKED_MSG_FR)) {
return OCRResult.BLOCKED
} else if (result.includes(ERR_CACHE_MISS)) {
return OCRResult.TO_REFRESH
} else if (result.includes(CHECKING_MSG_FR)) {
let screenShot = await convertImageToWhiteBlack(fileName);
try {
let result = await tesseract
.recognize(screenShot, config)
console.log(result)
if (result.includes(MESSAGE_URL_VALIDATION_EN) || result.includes(MESSAGE_URL_VALIDATION_FR)) {
return OCRResult.SUCCESS
} else if (result.includes(CAPTCHA_ERROR_MESSAGE) || result.includes(CAPTCHA_ERROR_MESSAGE_FR)) {
return OCRResult.RECAPTCHA_ERROR
} else if (result.includes(BLOCKED_MSG_EN) || result.includes(BLOCKED_MSG_FR)) {
return OCRResult.BLOCKED
} else if (result.includes(ERR_CACHE_MISS)) {
return OCRResult.TO_REFRESH
} else if (result.includes(CHECKING_MSG_FR)) {
return OCRResult.RECHECK
} else if (result.includes(SLIDING_CAPTCHA_FR)) {
return OCRResult.SLIDING_CAPTCHA
} else if (result.includes(MESSAGE_FILL_FIELD_FR)) {
return OCRResult.FILL_FIELD
} else if (result.includes(WELCOME_MESSAGE_FR)) {
return OCRResult.NEED_TO_CLICK_LATE_BTN
} else if (result.toLowerCase().includes(ONLINE_APPOINTMENT.toLowerCase())) {
return OCRResult.ONLINE_APPOINTMENT
} else if (result.includes(PAGE_OPTIMIZATION_CHROME_FR) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_2) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_3) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_4) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_5)) {
return OCRResult.PAGE_OPTIMIZATION
} else if (result.includes(CONFIRM_RESEND_FORM_FR)) {
return OCRResult.CONFIRM_RESEND_FORM
} else if (result.includes(CLOSED_MESSAGE_FR)) {
return OCRResult.CLOSED
} else if (result.includes(ABOUT_BLANK)) {
return OCRResult.TERMINAED
}
} catch (e) {
console.log(e)
return OCRResult.RECHECK
} else if (result.includes(SLIDING_CAPTCHA_FR)) {
return OCRResult.SLIDING_CAPTCHA
} else if (result.includes(MESSAGE_FILL_FIELD_FR)) {
return OCRResult.FILL_FIELD
} else if (result.includes(WELCOME_MESSAGE_FR)) {
return OCRResult.NEED_TO_CLICK_LATE_BTN
} else if (result.toLowerCase().includes(ONLINE_APPOINTMENT.toLowerCase())) {
return OCRResult.ONLINE_APPOINTMENT
} else if (result.includes(PAGE_OPTIMIZATION_CHROME_FR) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_2) || result.includes(PAGE_OPTIMIZATION_CHROME_FR_3)) {
return OCRResult.PAGE_OPTIMIZATION
} else if (result.includes(CONFIRM_RESEND_FORM_FR)) {
return OCRResult.CONFIRM_RESEND_FORM
} else if (result.includes(CLOSED_MESSAGE_FR)) {
return OCRResult.CLOSED
}
}
async take_screen_shot() {