app-manager-edge-worker 1.3.2 → 1.3.3

package/README.md

@@ -201,6 +201,14 @@ Log levels: `[INFO]` · `[SUCCESS]` · `[WARN]` · `[ERROR]` · `[TASK]` (job di
 
 ---
 
+### v1.3.3
+- **Captcha Session Preservation & Asynchronous Image Resynchronization**:
+  - **Deterministic Error Differentiation**: Disentangled captcha validation rejections from business errors (e.g., invalid GSTIN). When captcha validation fails, the worker and backend preserve the browser page and interactive session alive, reset the session TTL timer, and never abort prematurely.
+  - **Expanded Captcha Failure Signatures**: Added broad pattern matching (`letters?`, `characters?`, `image below`, `shown below`, `security/verification code`, `invalid_captcha`) and DOM attribute inspection (`input[name='cap'].ng-invalid`, `[data-ng-bind*='ERR_CAPTCHA']`) to immediately classify captcha errors even when portals omit the word "captcha".
+  - **Asynchronous Captcha Settle Logic**: Introduced post-rejection settle delays (`waitForFunction` monitoring `img.src` change and AJAX completion) before taking the replacement captcha screenshot, preventing stale/expired captcha image captures.
+  - **Session TTL Auto-Refresh**: Automatically resets the session expiration timer on both edge worker and central backend whenever a captcha is retried or refreshed, granting users the full interaction window.
+  - **Clean Session Termination on Business Errors**: Genuine business rejections (such as invalid GSTIN or record not found) immediately purge browser contexts and trigger clean state cleanup across worker, backend, and UI test playground.
+
 ### v1.3.2
 - **Sub-Second Inline Error Resolution & Strict Outcome Race Protocol**:
   - **Dynamic In-Page Error Detection**: Expanded `defaultBroadErrorSelector` to capture generic inline portal validation elements (`.err`, `span.err`, `p.err`, `div.err`, `[class*='err-']`, `[data-ng-bind*='ERR']`, `[data-ng-show*='err']`, `[data-ng-show*='w_gst']`, `[role='alert']`), eliminating 15-second hangs when portal rejections insert error lines without page reload.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "app-manager-edge-worker",
-  "version": "1.3.2",
+  "version": "1.3.3",
   "description": "Distributed Scraper Edge Worker for Apps Manager. Run on any local machine, VPS, or Termux Android device to serve browser and HTTP scraping jobs.",
   "type": "module",
   "bin": {

package/src/banner.js

@@ -1,7 +1,7 @@
 import { createRequire } from "module";
 
 const require = createRequire(import.meta.url);
-let workerVersion = "1.3.2";
+let workerVersion = "1.3.3";
 try {
     const pkg = require("../package.json");
     if (pkg?.version) workerVersion = pkg.version;

package/src/client.js

@@ -20,7 +20,7 @@ import fs from "fs";
 import { getWhatsAppHandler } from "./handlers/whatsappHandler.js";
 import { printBanner, logInfo, logSuccess, logWarn, logError, logJob } from "./banner.js";
 
-let workerPkgVersion = "1.3.2";
+let workerPkgVersion = "1.3.3";
 try {
     const pkg = JSON.parse(fs.readFileSync(new URL("../package.json", import.meta.url), "utf-8"));
     if (pkg?.version) workerPkgVersion = pkg.version;

package/src/handlers/interactiveHandler.js

@@ -529,6 +529,13 @@ export async function handleInteractiveSessionRefresh(payload) {
                 });
                 const pageScreenshotBase64 = `data:image/jpeg;base64,${Buffer.from(pageScreenshotBuffer).toString("base64")}`;
 
+                // Reset auto-purge TTL timer to grant user full window to solve refreshed captcha
+                if (session.timeoutTimer) clearTimeout(session.timeoutTimer);
+                const sessionTtlSeconds = cfg.sessionTimeoutSeconds || 180;
+                session.timeoutTimer = setTimeout(() => {
+                    purgeSession(sessionId).catch(() => { });
+                }, sessionTtlSeconds * 1000);
+
                 return {
                     sessionId,
                     imageBase64: `data:image/png;base64,${imageBase64}`,
@@ -587,7 +594,19 @@ export async function handleInteractiveSessionRefresh(payload) {
                     .filter(Boolean)
                     .join(", ");
 
-                const invalidPattern = cfg.errorDetection?.invalidCaptchaPattern || "captcha.*(incorrect|invalid|wrong|mismatch|enter text)";
+                const defaultCaptchaPattern =
+                    "captcha.*(incorrect|invalid|wrong|mismatch|enter|letters?|text|try again)|" +
+                    "(invalid|incorrect|wrong|mismatch).*captcha|" +
+                    "enter\\s*valid\\s*(letters?|characters?|code)|" +
+                    "letters?\\s*shown\\s*(below|in\\s*the\\s*image)|" +
+                    "characters?\\s*shown|" +
+                    "invalid_captcha|" +
+                    "verification\\s*code.*(invalid|wrong|incorrect)|" +
+                    "security\\s*code.*(invalid|wrong|incorrect)";
+
+                const invalidPattern = cfg.errorDetection?.invalidCaptchaPattern
+                    ? `${cfg.errorDetection.invalidCaptchaPattern}|${defaultCaptchaPattern}`
+                    : defaultCaptchaPattern;
 
                 // 1. Type captcha solution with robust candidate fallbacks & frame search
                 const solveTimeout = Math.max(cfg.inputTimeoutMs || 0, 15000);
@@ -880,9 +899,38 @@ export async function handleInteractiveSessionRefresh(payload) {
                         }
                     }
 
+                    // 5. DOM-level captcha error detection: check if captcha input or form has invalid captcha state
+                    let hasDomCaptchaError = false;
+                    try {
+                        const capInput = document.querySelector("input[name='cap'], input[id*='captcha' i], input[name*='captcha' i], #fo-captcha");
+                        if (capInput) {
+                            const cls = capInput.className || "";
+                            if (cls.includes("invalid_captcha") || cls.includes("invalid-captcha") || cls.includes("ng-invalid-invalid_captcha")) {
+                                hasDomCaptchaError = true;
+                            }
+                        }
+                        const form = document.querySelector("form");
+                        if (form && !hasDomCaptchaError) {
+                            const formCls = form.className || "";
+                            if (formCls.includes("invalid_captcha") || formCls.includes("invalid-captcha") || formCls.includes("ng-invalid-invalid_captcha")) {
+                                hasDomCaptchaError = true;
+                            }
+                        }
+                        if (!hasDomCaptchaError) {
+                            const capErrSpan = document.querySelector("[data-ng-bind*='ERR_CAPTCHA'], [ng-bind*='ERR_CAPTCHA'], [data-ng-if*='invalid_captcha'], [ng-if*='invalid_captcha']");
+                            if (capErrSpan) {
+                                const s = window.getComputedStyle(capErrSpan);
+                                if (s.display !== "none" && s.visibility !== "hidden") {
+                                    hasDomCaptchaError = true;
+                                }
+                            }
+                        }
+                    } catch (_) { }
+
                     return {
                         matchedRuleId: fallbackRuleId,
                         text: foundText,
+                        hasDomCaptchaError,
                     };
                 }, customRules, combinedErrorSelectors);
 
@@ -892,13 +940,32 @@ export async function handleInteractiveSessionRefresh(payload) {
                 errorText = (checkRulesAndText && checkRulesAndText.text) || "";
 
                 const regex = new RegExp(invalidPattern, "i");
-                const isCaptchaError = matchedRule
-                    ? matchedRule.action === "RETRY_CAPTCHA"
-                    : (Boolean(errorText) && regex.test(errorText));
+                const isCaptchaError =
+                    Boolean(checkRulesAndText?.hasDomCaptchaError) ||
+                    (matchedRule ? matchedRule.action === "RETRY_CAPTCHA" : false) ||
+                    (Boolean(errorText) && regex.test(errorText));
 
-                // Check if new captcha image was rendered on page
+                // If captcha error occurred, give target portal a brief window to load the refreshed captcha image
                 const imageSelector = cfg.imageSelector || "img[src*='captcha'], #imgCaptcha";
                 let newImageBase64 = null;
+                if (isCaptchaError) {
+                    try {
+                        const initialSrc = await page.$eval(imageSelector, (el) => el.src || el.getAttribute("src")).catch(() => null);
+                        await Promise.race([
+                            page.waitForFunction(
+                                (sel, old) => {
+                                    const el = document.querySelector(sel);
+                                    return el && (el.src || el.getAttribute("src")) !== old && el.complete;
+                                },
+                                { timeout: 1600 },
+                                imageSelector,
+                                initialSrc
+                            ).catch(() => { }),
+                            new Promise((r) => setTimeout(r, 650)),
+                        ]);
+                    } catch (_) { }
+                }
+
                 try {
                     const newCaptchaEl = await page.$(imageSelector);
                     if (newCaptchaEl) {
@@ -936,8 +1003,9 @@ export async function handleInteractiveSessionRefresh(payload) {
 
                 // Runtime non-inspection mode:
                 if (isCaptchaError) {
+                    // For captcha errors: NEVER abort session! Keep the browser session alive and reset TTL timer
                     if (session.timeoutTimer) clearTimeout(session.timeoutTimer);
-                    const retryTtlSeconds = cfg.sessionTimeoutSeconds || 90;
+                    const retryTtlSeconds = cfg.sessionTimeoutSeconds || 180;
                     session.timeoutTimer = setTimeout(() => {
                         purgeSession(sessionId).catch(() => { });
                     }, retryTtlSeconds * 1000);
@@ -945,14 +1013,16 @@ export async function handleInteractiveSessionRefresh(payload) {
                     return {
                         status: "INVALID_CAPTCHA",
                         success: false,
-                        message: (matchedRule && matchedRule.customMessage) || errorText || "The captcha entered was incorrect.",
+                        message: (matchedRule && matchedRule.customMessage) || errorText || "The captcha entered was incorrect. Please retry with the updated image.",
                         canRetry: true,
                         newImageBase64,
                         errorCode: (matchedRule && matchedRule.id) || "INVALID_CAPTCHA",
+                        matchedRule: matchedRule ? { id: matchedRule.id, name: matchedRule.name, action: matchedRule.action } : null,
                     };
                 }
 
                 if (errorText || matchedRule) {
+                    // For business errors (e.g. invalid GSTIN, not found): abort and purge session immediately
                     await purgeSession(sessionId);
                     return {
                         status: "BUSINESS_ERROR",
@@ -960,6 +1030,7 @@ export async function handleInteractiveSessionRefresh(payload) {
                         errorCode: (matchedRule && matchedRule.id) || "BUSINESS_ERROR",
                         message: (matchedRule && matchedRule.customMessage) || errorText || "Target portal returned an error.",
                         canRetry: false,
+                        matchedRule: matchedRule ? { id: matchedRule.id, name: matchedRule.name, action: matchedRule.action } : null,
                     };
                 }