vodorod/parcer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'develop' into internationalization2

Browse Source
This commit is contained in:
Amit Chauhan
2024-12-10 23:26:21 +05:30
committed by GitHub
parent 2872e1b51e ebb6e3af00
commit 100fdb9f0d
15 changed files with 555 additions and 146 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
maxun-core/package.json
View File

@@ -1,6 +1,6 @@
{
"name": "maxun-core",
"version": "0.0.4",
"version": "0.0.5",
"description": "Core package for Maxun, responsible for data extraction",
"main": "build/index.js",
"typings": "build/index.d.ts",

10
maxun-core/src/browserSide/scraper.js
View File

@@ -283,13 +283,13 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
} else if (attribute === 'innerHTML') {
record[label] = fieldElement.innerHTML.trim();
} else if (attribute === 'src') {
// Handle relative 'src' URLs
const src = fieldElement.getAttribute('src');
record[label] = src ? new URL(src, baseUrl).href : null;
// Handle relative 'src' URLs
const src = fieldElement.getAttribute('src');
record[label] = src ? new URL(src, window.location.origin).href : null;
} else if (attribute === 'href') {
// Handle relative 'href' URLs
const href = fieldElement.getAttribute('href');
record[label] = href ? new URL(href, baseUrl).href : null;
record[label] = href ? new URL(href, window.location.origin).href : null;
} else {
record[label] = fieldElement.getAttribute(attribute);
}
@@ -346,5 +346,5 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
return results;
};
})(window);

212
maxun-core/src/interpret.ts
View File

@@ -121,6 +121,53 @@ export default class Interpreter extends EventEmitter {
}
}
// private getSelectors(workflow: Workflow, actionId: number): string[] {
// const selectors: string[] = [];
// // Validate actionId
// if (actionId <= 0) {
// console.log("No previous selectors to collect.");
// return selectors; // Empty array as there are no previous steps
// }
// // Iterate from the start up to (but not including) actionId
// for (let index = 0; index < actionId; index++) {
// const currentSelectors = workflow[index]?.where?.selectors;
// console.log(`Selectors at step ${index}:`, currentSelectors);
// if (currentSelectors && currentSelectors.length > 0) {
// currentSelectors.forEach((selector) => {
// if (!selectors.includes(selector)) {
// selectors.push(selector); // Avoid duplicates
// }
// });
// }
// }
// console.log("Collected Selectors:", selectors);
// return selectors;
// }
private getSelectors(workflow: Workflow): string[] {
const selectorsSet = new Set<string>();
if (workflow.length === 0) {
return [];
}
for (let index = workflow.length - 1; index >= 0; index--) {
const currentSelectors = workflow[index]?.where?.selectors;
if (currentSelectors && currentSelectors.length > 0) {
currentSelectors.forEach((selector) => selectorsSet.add(selector));
return Array.from(selectorsSet);
}
}
return [];
}
/**
* Returns the context object from given Page and the current workflow.\
* \
@@ -130,52 +177,63 @@ export default class Interpreter extends EventEmitter {
* @param workflow Current **initialized** workflow (array of where-what pairs).
* @returns {PageState} State of the current page.
*/
private async getState(page: Page, workflow: Workflow): Promise<PageState> {
private async getState(page: Page, workflowCopy: Workflow, selectors: string[]): Promise<PageState> {
/**
* All the selectors present in the current Workflow
*/
const selectors = Preprocessor.extractSelectors(workflow);
// const selectors = Preprocessor.extractSelectors(workflow);
// console.log("Current selectors:", selectors);
/**
* Determines whether the element targetted by the selector is [actionable](https://playwright.dev/docs/actionability).
* @param selector Selector to be queried
* @returns True if the targetted element is actionable, false otherwise.
*/
const actionable = async (selector: string): Promise<boolean> => {
try {
const proms = [
page.isEnabled(selector, { timeout: 500 }),
page.isVisible(selector, { timeout: 500 }),
];
// const actionable = async (selector: string): Promise<boolean> => {
// try {
// const proms = [
// page.isEnabled(selector, { timeout: 5000 }),
// page.isVisible(selector, { timeout: 5000 }),
// ];
return await Promise.all(proms).then((bools) => bools.every((x) => x));
} catch (e) {
// log(<Error>e, Level.ERROR);
return false;
}
};
// return await Promise.all(proms).then((bools) => bools.every((x) => x));
// } catch (e) {
// // log(<Error>e, Level.ERROR);
// return false;
// }
// };
/**
* Object of selectors present in the current page.
*/
const presentSelectors: SelectorArray = await Promise.all(
selectors.map(async (selector) => {
if (await actionable(selector)) {
return [selector];
}
return [];
}),
).then((x) => x.flat());
// const presentSelectors: SelectorArray = await Promise.all(
// selectors.map(async (selector) => {
// if (await actionable(selector)) {
// return [selector];
// }
// return [];
// }),
// ).then((x) => x.flat());
const action = workflowCopy[workflowCopy.length - 1];
// console.log("Next action:", action)
let url: any = page.url();
if (action && action.where.url !== url && action.where.url !== "about:blank") {
url = action.where.url;
}
return {
url: page.url(),
url,
cookies: (await page.context().cookies([page.url()]))
.reduce((p, cookie) => (
{
...p,
[cookie.name]: cookie.value,
}), {}),
selectors: presentSelectors,
selectors,
};
}
@@ -365,6 +423,7 @@ export default class Interpreter extends EventEmitter {
console.log("MERGED results:", mergedResult);
await this.options.serializableCallback(mergedResult);
// await this.options.serializableCallback(scrapeResult);
},
scrapeList: async (config: { listSelector: string, fields: any, limit?: number, pagination: any }) => {
@@ -410,6 +469,16 @@ export default class Interpreter extends EventEmitter {
}),
};
const executeAction = async (invokee: any, methodName: string, args: any) => {
console.log("Executing action:", methodName, args);
if (!args || Array.isArray(args)) {
await (<any>invokee[methodName])(...(args ?? []));
} else {
await (<any>invokee[methodName])(args);
}
};
for (const step of steps) {
this.log(`Launching ${String(step.action)}`, Level.LOG);
@@ -427,10 +496,20 @@ export default class Interpreter extends EventEmitter {
invokee = invokee[level];
}
if (!step.args || Array.isArray(step.args)) {
await (<any>invokee[methodName])(...(step.args ?? []));
if (methodName === 'waitForLoadState') {
try {
await executeAction(invokee, methodName, step.args);
} catch (error) {
await executeAction(invokee, methodName, 'domcontentloaded');
}
} else if (methodName === 'click') {
try {
await executeAction(invokee, methodName, step.args);
} catch (error) {
await executeAction(invokee, methodName, [step.args[0], { force: true }]);
}
} else {
await (<any>invokee[methodName])(step.args);
await executeAction(invokee, methodName, step.args);
}
}
@@ -475,6 +554,8 @@ export default class Interpreter extends EventEmitter {
case 'clickNext':
const pageResults = await page.evaluate((cfg) => window.scrapeList(cfg), config);
// console.log("Page results:", pageResults);
// Filter out already scraped items
const newResults = pageResults.filter(item => {
const uniqueKey = JSON.stringify(item);
@@ -482,9 +563,9 @@ export default class Interpreter extends EventEmitter {
scrapedItems.add(uniqueKey); // Mark as scraped
return true;
});
allResults = allResults.concat(newResults);
if (config.limit && allResults.length >= config.limit) {
return allResults.slice(0, config.limit);
}
@@ -494,7 +575,7 @@ export default class Interpreter extends EventEmitter {
return allResults; // No more pages to scrape
}
await Promise.all([
nextButton.click(),
nextButton.dispatchEvent('click'),
page.waitForNavigation({ waitUntil: 'networkidle' })
]);
@@ -510,7 +591,7 @@ export default class Interpreter extends EventEmitter {
return allResults;
}
// Click the 'Load More' button to load additional items
await loadMoreButton.click();
await loadMoreButton.dispatchEvent('click');
await page.waitForTimeout(2000); // Wait for new items to load
// After clicking 'Load More', scroll down to load more items
await page.evaluate(() => window.scrollTo(0, document.body.scrollHeight));
@@ -546,11 +627,31 @@ export default class Interpreter extends EventEmitter {
return allResults;
}
private getMatchingActionId(workflow: Workflow, pageState: PageState, usedActions: string[]) {
for (let actionId = workflow.length - 1; actionId >= 0; actionId--) {
const step = workflow[actionId];
const isApplicable = this.applicable(step.where, pageState, usedActions);
console.log("-------------------------------------------------------------");
console.log(`Where:`, step.where);
console.log(`Page state:`, pageState);
console.log(`Match result: ${isApplicable}`);
console.log("-------------------------------------------------------------");
if (isApplicable) {
return actionId;
}
}
}
private async runLoop(p: Page, workflow: Workflow) {
const workflowCopy: Workflow = JSON.parse(JSON.stringify(workflow));
// apply ad-blocker to the current page
await this.applyAdBlocker(p);
const usedActions: string[] = [];
let selectors: string[] = [];
let lastAction = null;
let actionId = -1
let repeatCount = 0;
/**
@@ -559,7 +660,7 @@ export default class Interpreter extends EventEmitter {
* e.g. via `enqueueLinks`.
*/
p.on('popup', (popup) => {
this.concurrency.addJob(() => this.runLoop(popup, workflow));
this.concurrency.addJob(() => this.runLoop(popup, workflowCopy));
});
/* eslint no-constant-condition: ["warn", { "checkLoops": false }] */
@@ -578,8 +679,11 @@ export default class Interpreter extends EventEmitter {
}
let pageState = {};
let getStateTest = "Hello";
try {
pageState = await this.getState(p, workflow);
pageState = await this.getState(p, workflowCopy, selectors);
selectors = [];
console.log("Empty selectors:", selectors)
} catch (e: any) {
this.log('The browser has been closed.');
return;
@@ -589,32 +693,52 @@ export default class Interpreter extends EventEmitter {
this.log(`Current state is: \n${JSON.stringify(pageState, null, 2)}`, Level.WARN);
}
const actionId = workflow.findIndex((step) => {
const isApplicable = this.applicable(step.where, pageState, usedActions);
console.log(`Where:`, step.where);
console.log(`Page state:`, pageState);
console.log(`Match result: ${isApplicable}`);
return isApplicable;
});
// const actionId = workflow.findIndex((step) => {
// const isApplicable = this.applicable(step.where, pageState, usedActions);
// console.log("-------------------------------------------------------------");
// console.log(`Where:`, step.where);
// console.log(`Page state:`, pageState);
// console.log(`Match result: ${isApplicable}`);
// console.log("-------------------------------------------------------------");
// return isApplicable;
// });
const action = workflow[actionId];
actionId = this.getMatchingActionId(workflowCopy, pageState, usedActions);
const action = workflowCopy[actionId];
console.log("MATCHED ACTION:", action);
console.log("MATCHED ACTION ID:", actionId);
this.log(`Matched ${JSON.stringify(action?.where)}`, Level.LOG);
if (action) { // action is matched
if (this.options.debugChannel?.activeId) {
this.options.debugChannel.activeId(actionId);
}
repeatCount = action === lastAction ? repeatCount + 1 : 0;
if (this.options.maxRepeats && repeatCount >= this.options.maxRepeats) {
console.log("REPEAT COUNT", repeatCount);
if (this.options.maxRepeats && repeatCount > this.options.maxRepeats) {
return;
}
lastAction = action;
try {
console.log("Carrying out:", action.what);
await this.carryOutSteps(p, action.what);
usedActions.push(action.id ?? 'undefined');
workflowCopy.splice(actionId, 1);
console.log(`Action with ID ${action.id} removed from the workflow copy.`);
// const newSelectors = this.getPreviousSelectors(workflow, actionId);
const newSelectors = this.getSelectors(workflowCopy);
newSelectors.forEach(selector => {
if (!selectors.includes(selector)) {
selectors.push(selector);
}
});
} catch (e) {
this.log(<Error>e, Level.ERROR);
}

8
package.json
View File

@@ -1,6 +1,6 @@
{
"name": "maxun",
"version": "0.0.3",
"version": "0.0.4",
"author": "Maxun",
"license": "AGPL-3.0-or-later",
"dependencies": {
@@ -36,16 +36,20 @@
"fortawesome": "^0.0.1-security",
"google-auth-library": "^9.14.1",
"googleapis": "^144.0.0",
"i18next": "^24.0.2",
"i18next-browser-languagedetector": "^8.0.0",
"i18next-http-backend": "^3.0.1",
"idcac-playwright": "^0.1.3",
"ioredis": "^5.4.1",
"joi": "^17.6.0",
"jsonwebtoken": "^9.0.2",
"jwt-decode": "^4.0.0",
"loglevel": "^1.8.0",
"loglevel-plugin-remote": "^0.6.8",
"maxun-core": "0.0.4",
"maxun-core": "^0.0.5",
"minio": "^8.0.1",
"moment-timezone": "^0.5.45",
"node-cron": "^3.0.3",

18
server/src/api/record.ts
View File

@@ -15,6 +15,8 @@ import { io, Socket } from "socket.io-client";
import { BinaryOutputService } from "../storage/mino";
import { AuthenticatedRequest } from "../routes/record"
import {capture} from "../utils/analytics";
import { Page } from "playwright";
import { WorkflowFile } from "maxun-core";
chromium.use(stealthPlugin());
const formatRecording = (recordingData: any) => {
@@ -533,6 +535,17 @@ function resetRecordingState(browserId: string, id: string) {
id = '';
}
function AddGeneratedFlags(workflow: WorkflowFile) {
const copy = JSON.parse(JSON.stringify(workflow));
for (let i = 0; i < workflow.workflow.length; i++) {
copy.workflow[i].what.unshift({
action: 'flag',
args: ['generated'],
});
}
return copy;
};
async function executeRun(id: string) {
try {
const run = await Run.findOne({ where: { runId: id } });
@@ -560,13 +573,14 @@ async function executeRun(id: string) {
throw new Error('Could not access browser');
}
const currentPage = await browser.getCurrentPage();
let currentPage = await browser.getCurrentPage();
if (!currentPage) {
throw new Error('Could not create a new page');
}
const workflow = AddGeneratedFlags(recording.recording);
const interpretationInfo = await browser.interpreter.InterpretRecording(
recording.recording, currentPage, plainRun.interpreterSettings
workflow, currentPage, (newPage: Page) => currentPage = newPage, plainRun.interpreterSettings
);
const binaryOutputService = new BinaryOutputService('maxun-run-screenshots');

83
server/src/browser-management/classes/RemoteBrowser.ts
View File

@@ -15,6 +15,7 @@ import { InterpreterSettings, RemoteBrowserOptions } from "../../types";
import { WorkflowGenerator } from "../../workflow-management/classes/Generator";
import { WorkflowInterpreter } from "../../workflow-management/classes/Interpreter";
import { getDecryptedProxyConfig } from '../../routes/proxy';
import { getInjectableScript } from 'idcac-playwright';
chromium.use(stealthPlugin());
@@ -65,6 +66,8 @@ export class RemoteBrowser {
maxRepeats: 1,
};
private lastEmittedUrl: string | null = null;
/**
* {@link WorkflowGenerator} instance specific to the remote browser.
*/
@@ -87,6 +90,64 @@ export class RemoteBrowser {
this.generator = new WorkflowGenerator(socket);
}
/**
* Normalizes URLs to prevent navigation loops while maintaining consistent format
*/
private normalizeUrl(url: string): string {
try {
const parsedUrl = new URL(url);
// Remove trailing slashes except for root path
parsedUrl.pathname = parsedUrl.pathname.replace(/\/+$/, '') || '/';
// Ensure consistent protocol handling
parsedUrl.protocol = parsedUrl.protocol.toLowerCase();
return parsedUrl.toString();
} catch {
return url;
}
}
/**
* Determines if a URL change is significant enough to emit
*/
private shouldEmitUrlChange(newUrl: string): boolean {
if (!this.lastEmittedUrl) {
return true;
}
const normalizedNew = this.normalizeUrl(newUrl);
const normalizedLast = this.normalizeUrl(this.lastEmittedUrl);
return normalizedNew !== normalizedLast;
}
private async setupPageEventListeners(page: Page) {
page.on('framenavigated', async (frame) => {
if (frame === page.mainFrame()) {
const currentUrl = page.url();
if (this.shouldEmitUrlChange(currentUrl)) {
this.lastEmittedUrl = currentUrl;
this.socket.emit('urlChanged', currentUrl);
}
}
});
// Handle page load events with retry mechanism
page.on('load', async () => {
const injectScript = async (): Promise<boolean> => {
try {
await page.waitForLoadState('networkidle', { timeout: 5000 });
await page.evaluate(getInjectableScript());
return true;
} catch (error: any) {
logger.log('warn', `Script injection attempt failed: ${error.message}`);
return false;
}
};
const success = await injectScript();
console.log("Script injection result:", success);
});
}
/**
* An asynchronous constructor for asynchronously initialized properties.
* Must be called right after creating an instance of RemoteBrowser class.
@@ -166,11 +227,7 @@ export class RemoteBrowser {
this.context = await this.browser.newContext(contextOptions);
this.currentPage = await this.context.newPage();
this.currentPage.on('framenavigated', (frame) => {
if (frame === this.currentPage?.mainFrame()) {
this.socket.emit('urlChanged', this.currentPage.url());
}
});
await this.setupPageEventListeners(this.currentPage);
// await this.currentPage.setExtraHTTPHeaders({
// 'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3'
@@ -370,11 +427,7 @@ export class RemoteBrowser {
await this.stopScreencast();
this.currentPage = page;
this.currentPage.on('framenavigated', (frame) => {
if (frame === this.currentPage?.mainFrame()) {
this.socket.emit('urlChanged', this.currentPage.url());
}
});
await this.setupPageEventListeners(this.currentPage);
//await this.currentPage.setViewportSize({ height: 400, width: 900 })
this.client = await this.currentPage.context().newCDPSession(this.currentPage);
@@ -402,14 +455,8 @@ export class RemoteBrowser {
await this.currentPage?.close();
this.currentPage = newPage;
if (this.currentPage) {
this.currentPage.on('framenavigated', (frame) => {
if (frame === this.currentPage?.mainFrame()) {
this.socket.emit('urlChanged', this.currentPage.url());
}
});
// this.currentPage.on('load', (page) => {
// this.socket.emit('urlChanged', page.url());
// })
await this.setupPageEventListeners(this.currentPage);
this.client = await this.currentPage.context().newCDPSession(this.currentPage);
await this.subscribeToScreencast();
} else {

18
server/src/routes/storage.ts
View File

@@ -18,6 +18,8 @@ import { AuthenticatedRequest } from './record';
import { computeNextRun } from '../utils/schedule';
import { capture } from "../utils/analytics";
import { tryCatch } from 'bullmq';
import { WorkflowFile } from 'maxun-core';
import { Page } from 'playwright';
chromium.use(stealthPlugin());
export const router = Router();
@@ -422,6 +424,17 @@ router.get('/runs/run/:id', requireSignIn, async (req, res) => {
}
});
function AddGeneratedFlags(workflow: WorkflowFile) {
const copy = JSON.parse(JSON.stringify(workflow));
for (let i = 0; i < workflow.workflow.length; i++) {
copy.workflow[i].what.unshift({
action: 'flag',
args: ['generated'],
});
}
return copy;
};
/**
* PUT endpoint for finishing a run and saving it to the storage.
*/
@@ -443,10 +456,11 @@ router.post('/runs/run/:id', requireSignIn, async (req: AuthenticatedRequest, re
// interpret the run in active browser
const browser = browserPool.getRemoteBrowser(plainRun.browserId);
const currentPage = browser?.getCurrentPage();
let currentPage = browser?.getCurrentPage();
if (browser && currentPage) {
const workflow = AddGeneratedFlags(recording.recording);
const interpretationInfo = await browser.interpreter.InterpretRecording(
recording.recording, currentPage, plainRun.interpreterSettings);
workflow, currentPage, (newPage: Page) => currentPage = newPage, plainRun.interpreterSettings);
const binaryOutputService = new BinaryOutputService('maxun-run-screenshots');
const uploadedBinaryOutput = await binaryOutputService.uploadAndStoreBinaryOutput(run, interpretationInfo.binaryOutput);
await destroyRemoteBrowser(plainRun.browserId);

26
server/src/workflow-management/classes/Interpreter.ts
View File

@@ -244,7 +244,12 @@ export class WorkflowInterpreter {
* @param page The page instance used to interact with the browser.
* @param settings The settings to use for the interpretation.
*/
public InterpretRecording = async (workflow: WorkflowFile, page: Page, settings: InterpreterSettings) => {
public InterpretRecording = async (
workflow: WorkflowFile,
page: Page,
updatePageOnPause: (page: Page) => void,
settings: InterpreterSettings
) => {
const params = settings.params ? settings.params : null;
delete settings.params;
@@ -262,7 +267,7 @@ export class WorkflowInterpreter {
this.socket.emit('debugMessage', msg)
},
},
serializableCallback: (data: string) => {
serializableCallback: (data: any) => {
this.serializableData.push(data);
this.socket.emit('serializableCallback', data);
},
@@ -275,6 +280,23 @@ export class WorkflowInterpreter {
const interpreter = new Interpreter(decryptedWorkflow, options);
this.interpreter = interpreter;
interpreter.on('flag', async (page, resume) => {
if (this.activeId !== null && this.breakpoints[this.activeId]) {
logger.log('debug', `breakpoint hit id: ${this.activeId}`);
this.socket.emit('breakpointHit');
this.interpretationIsPaused = true;
}
if (this.interpretationIsPaused) {
this.interpretationResume = resume;
logger.log('debug', `Paused inside of flag: ${page.url()}`);
updatePageOnPause(page);
this.socket.emit('log', '----- The interpretation has been paused -----', false);
} else {
resume();
}
});
const status = await interpreter.run(page, params);
const lastArray = this.serializableData.length > 1

19
server/src/workflow-management/scheduler/index.ts
View File

@@ -11,6 +11,8 @@ import Run from "../../models/Run";
import { getDecryptedProxyConfig } from "../../routes/proxy";
import { BinaryOutputService } from "../../storage/mino";
import { capture } from "../../utils/analytics";
import { WorkflowFile } from "maxun-core";
import { Page } from "playwright";
chromium.use(stealthPlugin());
async function createWorkflowAndStoreMetadata(id: string, userId: string) {
@@ -79,6 +81,17 @@ async function createWorkflowAndStoreMetadata(id: string, userId: string) {
}
}
function AddGeneratedFlags(workflow: WorkflowFile) {
const copy = JSON.parse(JSON.stringify(workflow));
for (let i = 0; i < workflow.workflow.length; i++) {
copy.workflow[i].what.unshift({
action: 'flag',
args: ['generated'],
});
}
return copy;
};
async function executeRun(id: string) {
try {
const run = await Run.findOne({ where: { runId: id } });
@@ -106,13 +119,15 @@ async function executeRun(id: string) {
throw new Error('Could not access browser');
}
const currentPage = await browser.getCurrentPage();
let currentPage = await browser.getCurrentPage();
if (!currentPage) {
throw new Error('Could not create a new page');
}
const workflow = AddGeneratedFlags(recording.recording);
const interpretationInfo = await browser.interpreter.InterpretRecording(
recording.recording, currentPage, plainRun.interpreterSettings);
workflow, currentPage, (newPage: Page) => currentPage = newPage, plainRun.interpreterSettings
);
const binaryOutputService = new BinaryOutputService('maxun-run-screenshots');
const uploadedBinaryOutput = await binaryOutputService.uploadAndStoreBinaryOutput(run, interpretationInfo.binaryOutput);

210
server/src/workflow-management/selector.ts
View File

@@ -20,49 +20,6 @@ type Workflow = WorkflowFile["workflow"];
* @category WorkflowManagement-Selectors
* @returns {Promise<Rectangle|undefined|null>}
*/
export const getRect = async (page: Page, coordinates: Coordinates) => {
try {
const rect = await page.evaluate(
async ({ x, y }) => {
const el = document.elementFromPoint(x, y) as HTMLElement;
if (el) {
const { parentElement } = el;
// Match the logic in recorder.ts for link clicks
const element = parentElement?.tagName === 'A' ? parentElement : el;
const rectangle = element?.getBoundingClientRect();
// @ts-ignore
if (rectangle) {
return {
x: rectangle.x,
y: rectangle.y,
width: rectangle.width,
height: rectangle.height,
top: rectangle.top,
right: rectangle.right,
bottom: rectangle.bottom,
left: rectangle.left,
};
}
}
},
{ x: coordinates.x, y: coordinates.y },
);
return rect;
} catch (error) {
const { message, stack } = error as Error;
logger.log('error', `Error while retrieving selector: ${message}`);
logger.log('error', `Stack: ${stack}`);
}
}
/**
* Checks the basic info about an element and returns a {@link BaseActionInfo} object.
* If the element is not found, returns undefined.
* @param page The page instance.
* @param coordinates Coordinates of an element.
* @category WorkflowManagement-Selectors
* @returns {Promise<BaseActionInfo|undefined>}
*/
export const getElementInformation = async (
page: Page,
coordinates: Coordinates
@@ -70,10 +27,50 @@ export const getElementInformation = async (
try {
const elementInfo = await page.evaluate(
async ({ x, y }) => {
const el = document.elementFromPoint(x, y) as HTMLElement;
if (el) {
const { parentElement } = el;
const element = parentElement?.tagName === 'A' ? parentElement : el;
const originalEl = document.elementFromPoint(x, y) as HTMLElement;
if (originalEl) {
let element = originalEl;
// if (originalEl.tagName === 'A') {
// element = originalEl;
// } else if (originalEl.parentElement?.tagName === 'A') {
// element = originalEl.parentElement;
// } else {
// Generic parent finding logic based on visual containment
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
// Check if parent visually contains the child
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
// Additional checks for more comprehensive containment
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
// }
} }
let info: {
tagName: string;
@@ -98,7 +95,7 @@ export const getElementInformation = async (
);
}
// Gather specific information based on the tag
// Existing tag-specific logic
if (element?.tagName === 'A') {
info.url = (element as HTMLAnchorElement).href;
info.innerText = element.innerText ?? '';
@@ -112,7 +109,6 @@ export const getElementInformation = async (
info.innerHTML = element.innerHTML;
info.outerHTML = element.outerHTML;
return info;
}
return null;
@@ -127,6 +123,80 @@ export const getElementInformation = async (
}
};
export const getRect = async (page: Page, coordinates: Coordinates) => {
try {
const rect = await page.evaluate(
async ({ x, y }) => {
const originalEl = document.elementFromPoint(x, y) as HTMLElement;
if (originalEl) {
let element = originalEl;
// if (originalEl.tagName === 'A') {
// element = originalEl;
// } else if (originalEl.parentElement?.tagName === 'A') {
// element = originalEl.parentElement;
// } else {
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
// }
}}
//element = element?.parentElement?.tagName === 'A' ? element?.parentElement : element;
const rectangle = element?.getBoundingClientRect();
if (rectangle) {
return {
x: rectangle.x,
y: rectangle.y,
width: rectangle.width,
height: rectangle.height,
top: rectangle.top,
right: rectangle.right,
bottom: rectangle.bottom,
left: rectangle.left,
};
}
}
},
{ x: coordinates.x, y: coordinates.y },
);
return rect;
} catch (error) {
const { message, stack } = error as Error;
logger.log('error', `Error while retrieving selector: ${message}`);
logger.log('error', `Stack: ${stack}`);
}
}
/**
* Returns the best and unique css {@link Selectors} for the element on the page.
@@ -742,7 +812,6 @@ interface SelectorResult {
export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates): Promise<SelectorResult> => {
try {
const selectors = await page.evaluate(({ x, y }: { x: number, y: number }) => {
function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase();
@@ -774,8 +843,44 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
return path.join(' > ');
}
const element = document.elementFromPoint(x, y) as HTMLElement | null;
if (!element) return null;
const originalEl = document.elementFromPoint(x, y) as HTMLElement;
if (!originalEl) return null;
let element = originalEl;
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
}
}
const generalSelector = getSelectorPath(element);
return {
@@ -790,7 +895,6 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
}
};
export const getChildSelectors = async (page: Page, parentSelector: string): Promise<string[]> => {
try {
const childSelectors = await page.evaluate((parentSelector: string) => {

5
src/components/molecules/IntegrationSettings.tsx
View File

@@ -15,11 +15,13 @@ import { useGlobalInfoStore } from "../../context/globalInfo";
import { getStoredRecording } from "../../api/storage";
import { apiUrl } from "../../apiConfig.js";
import Cookies from 'js-cookie';
interface IntegrationProps {
isOpen: boolean;
handleStart: (data: IntegrationSettings) => void;
handleClose: () => void;
}
export interface IntegrationSettings {
spreadsheetId: string;
spreadsheetName: string;
@@ -75,8 +77,7 @@ export const IntegrationSettingsModal = ({
);
notify(
"error",
`Error fetching spreadsheet files: ${
error.response?.data?.message || error.message
`Error fetching spreadsheet files: ${error.response?.data?.message || error.message
}`
);
}

84
src/components/molecules/NavBar.tsx
View File

@@ -1,17 +1,32 @@
import React, { useState, useContext } from "react";
import axios from "axios";
import { useTranslation } from "react-i18next"; // Import useTranslation hook
import React, { useState, useContext, useEffect } from 'react';
import axios from 'axios';
import styled from "styled-components";
import { stopRecording } from "../../api/recording";
import { useGlobalInfoStore } from "../../context/globalInfo";
import { IconButton, Menu, MenuItem, Typography, Chip } from "@mui/material";
import { AccountCircle, Logout, Clear, Language } from "@mui/icons-material";
import { useNavigate } from "react-router-dom";
import { AuthContext } from "../../context/auth";
import { SaveRecording } from "../molecules/SaveRecording";
import DiscordIcon from "../atoms/DiscordIcon";
import { apiUrl } from "../../apiConfig";
import { IconButton, Menu, MenuItem, Typography, Chip, Button, Modal, Tabs, Tab, Box, Snackbar } from "@mui/material";
import { AccountCircle, Logout, Clear, YouTube, X, Update, Close,Language } from "@mui/icons-material";
import { useNavigate } from 'react-router-dom';
import { AuthContext } from '../../context/auth';
import { SaveRecording } from '../molecules/SaveRecording';
import DiscordIcon from '../atoms/DiscordIcon';
import { apiUrl } from '../../apiConfig';
import MaxunLogo from "../../assets/maxunlogo.png";
import { useTranslation } from "react-i18next"; // Import useTranslation hook
import packageJson from "../../../package.json"
interface NavBarProps {
recordingName: string;
@@ -29,8 +44,43 @@ export const NavBar: React.FC<NavBarProps> = ({
const { t, i18n } = useTranslation(); // Get translation function and i18n methods
const [anchorEl, setAnchorEl] = useState<null | HTMLElement>(null);
const [langAnchorEl, setLangAnchorEl] = useState<null | HTMLElement>(null);
const currentVersion = packageJson.version;
const [open, setOpen] = useState(false);
const [latestVersion, setLatestVersion] = useState<string | null>(null);
const [tab, setTab] = useState(0);
const [isUpdateAvailable, setIsUpdateAvailable] = useState(false);
const fetchLatestVersion = async (): Promise<string | null> => {
try {
const response = await fetch("https://api.github.com/repos/getmaxun/maxun/releases/latest");
const data = await response.json();
const version = data.tag_name.replace(/^v/, ""); // Remove 'v' prefix
return version;
} catch (error) {
console.error("Failed to fetch latest version:", error);
return null; // Handle errors gracefully
}
};
const handleUpdateOpen = () => {
setOpen(true);
fetchLatestVersion();
};
const handleUpdateClose = () => {
setOpen(false);
setTab(0); // Reset tab to the first tab
};
const handleUpdateTabChange = (event: React.SyntheticEvent, newValue: number) => {
setTab(newValue);
};
const handleMenuOpen = (event: React.MouseEvent<HTMLElement>) => {
setAnchorEl(event.currentTarget);
};
@@ -66,7 +116,19 @@ export const NavBar: React.FC<NavBarProps> = ({
localStorage.setItem("language", lang); // Persist language to localStorage
};
useEffect(() => {
const checkForUpdates = async () => {
const latestVersion = await fetchLatestVersion();
setLatestVersion(latestVersion); // Set the latest version state
if (latestVersion && latestVersion !== currentVersion) {
setIsUpdateAvailable(true); // Show a notification or highlight the "Upgrade" button
}
};
checkForUpdates();
}, []);
return (
<NavBarWrapper>
<div
@@ -332,6 +394,8 @@ export const NavBar: React.FC<NavBarProps> = ({
</NavBarWrapper>
);
};

2
src/components/molecules/Pair.tsx
View File

@@ -1,5 +1,5 @@
import React, { FC, useState } from 'react';
import { Stack, Button, IconButton, Tooltip, Chip, Badge } from "@mui/material";
import { Stack, Button, IconButton, Tooltip, Badge } from "@mui/material";
import { AddPair, deletePair, UpdatePair } from "../../api/workflow";
import { WorkflowFile } from "maxun-core";
import { ClearButton } from "../atoms/buttons/ClearButton";

2
src/components/molecules/RobotDuplicate.tsx
View File

@@ -1,6 +1,6 @@
import React, { useState, useEffect } from 'react';
import { GenericModal } from "../atoms/GenericModal";
import { TextField, Typography, Box, Button, Chip } from "@mui/material";
import { TextField, Typography, Box, Button } from "@mui/material";
import { modalStyle } from "./AddWhereCondModal";
import { useGlobalInfoStore } from '../../context/globalInfo';
import { duplicateRecording, getStoredRecording } from '../../api/storage';

2
src/context/browserActions.tsx
View File

@@ -53,6 +53,7 @@ export const ActionProvider = ({ children }: { children: ReactNode }) => {
const startPaginationMode = () => {
setPaginationMode(true);
setCaptureStage('pagination');
socket?.emit('setGetList', { getList: false });
};
const stopPaginationMode = () => setPaginationMode(false);
@@ -75,7 +76,6 @@ export const ActionProvider = ({ children }: { children: ReactNode }) => {
const stopGetList = () => {
setGetList(false);
socket?.emit('setGetList', { getList: false });
setPaginationType('');
setLimitType('');
setCustomLimit('');