Freelance projects

Freelance projects

Program (Python + Playwright or Node.js + Puppeteer) for Windows

Javascript and Typescript, Python — incorrectly specified categories?

34 USD

Client's review of cooperation with Oleksandr Leonov

Program (Python + Playwright or Node.js + Puppeteer) for Windows

34 USD

Oleksandr — a true genius in his field!

I am impressed with his work! Oleksandr solved a problem that more than ten other specialists could not. He did it quickly, perfectly, and without unnecessary questions. Everything was done with maximum precision and attention to detail.

Always in touch, clearly understands the task, and most importantly — he not only offers solutions but also proposed options that are much better than I could have imagined! It’s simply incredible! I was amazed by his ability to find solutions that exceeded all my expectations.

Oleksandr doesn’t just do the work — he creates effective strategies that improve not only the technical side of the project but also the end-user experience. His attentiveness to every detail and ability to quickly adapt to new conditions are truly impressive.

Working with Oleksandr is a real pleasure! His professionalism, friendliness, and willingness to help at any time create an atmosphere of mutual trust. If you need a talented, reliable, and responsible specialist, I confidently recommend Oleksandr!

Thanks again for the impeccable work and extremely productive collaboration! I will definitely contact you again!

5.0

Igor D.

Reciprocal Feedback

1 August 2025

Freelancer's review of cooperation with Igor D.

Program (Python + Playwright or Node.js + Puppeteer) for Windows

34 USD

Working was comfortable and productive. The technical task was clear, and unclear moments were easily clarified during the process. The client is always in touch, responds quickly, and provides feedback. The impression of cooperation is positive. I recommend.

5.0

Oleksandr Leonov

Reciprocal Feedback

1 August 2025

Project translated automatically. Log in or register, to view the original

Required program (Python / Node.js) for Windows - Runs manually and:

Visits the website: https://threeplus.three.co.uk
Uses my already logged-in Chrome profile (without authorization) or username - password.
In each account:
- Goes to the section with codes
- Copies 1 numeric code
- Saves it in CSV
Processes the first 20 accounts from the CSV file, then automatically terminates.

EVERYTHING!

📂 The same — in other words:
I have a CSV file with 30–50... accounts.
(I paid and legally registered on the site — completed SMS verification).
The program should process the first 20 accounts sequentially upon launch and turn off.

❗ Requirements:

❌ Do not use Selenium, WebDriver, ZennoPoster, headless browsers, etc.
❌ Do not use proxies / VPNs
✅ Use real Chrome profiles
✅ Mimic human behavior (small delays, genuine clicks)
✅ Fully local solution (no hosting or cloud)

💸 Expect affordable implementation
(including — because the site changes several times a month — I will regularly contact you with payment for minor adjustments)

💬 Please specify in your application:

Which language and library you plan to use (Python/Playwright, Node.js/Puppeteer, etc.)
How your program will work

🎯 Goal

Create a desktop program or script (in Python or Node.js) that automatically opens real Chrome profiles, logs into 20 Three+ accounts, extracts the code (number) from the page after login, and adds it to the same CSV table.

⚙️ Functionality

The program is launched manually on Windows (without background services or auto-start).
Processes the first 20 accounts per launch (number set in config).
Works through real Chrome with local profiles (user-data-dir + profile-directory).
Does not use headless modes, Selenium, or HTTP requests.
Simulates human actions: opening browser, navigation, clicks, pauses.
Skips accounts from which the code has already been obtained — determined via data in the same CSV.

🔄 Workflow Algorithm

Opens Chrome with the appropriate profile, already logged into Three+.
Goes to the Cineworld section, clicks the Get Reward button.
Getting the code:
- a) The code appears in a popup — reads and saves it.
- b) In rare cases, a link “Go to your codes” appears — then the program navigates there and retrieves the code.
  In any case, after clicking “Get Reward” the code always goes to the My Codes section, so the program should read it from there.
Saves the code in the corresponding column in the same CSV file next to the login.
Closes the browser, but does not log out.
Processes the next accounts (up to 20), skipping already processed.
Ends the operation.
Next time the program is launched, it continues from the next unprocessed entries.

⚙️ Configuration (settings.json or .ini)

Path to Chrome (chrome.exe)
List of profiles (names or paths)
Rewards URL (changes weekly)
Delays between actions: 2–3 minutes (random)
Path to CSV file

🗺️ Site features

Rewards URL changes every Monday (should be configurable in settings)
Site built on Flutter Web (canvas), HSTS active, no Captcha

🛠️ Technologies

Python + Playwright or Node.js + Puppeteer
Use of real Chrome profiles
Pauses: random + time.sleep()
Configuration via JSON or INI

📝 Waiting for output

.exe file for Windows
Full source code with comments
README file with instructions
Example configuration file

🚨 Important

🧍‍♂️ All actions must imitate manual user work.

The program should operate as “human-like” as possible (with pauses, genuine User-Agent, and without sudden simultaneous requests), to avoid blocks or SMS verification requests.
✅ Manual site operation is stable — I can easily get codes from 40-60 accounts in 10 minutes.

The program may not perform login — instead, it uses already logged-in profiles.
All behavior must be as “human” as possible: open browser, delays, real clicks, avoid simultaneous requests.
IP remains unchanged, proxies are not used.

🔹 Python + Playwright or Node.js + Puppeteer — are truly the right choices because:

They allow launching real Chrome with profiles, without headless mode and unnecessary emulation.
They work reliably even with Flutter Web sites (like Three+).

🔹 No authorization — login is done via existing profiles, so risk of SMS confirmation is minimal (with proper user simulation).

🔹 Code collection via page.evaluate() — effective and natural, especially when navigating through My Codes.

🔹 Delays, clicks, user behavior — crucial: mimic manual work (each profile opens separately, acts with pauses) — key to avoiding blocks.

🔹 CSV logic — correct: reading, checking, updating — all within one file, no extra databases.

🔹 No headless, Selenium, or proxies — exactly what is needed for safe automation in such a sensitive environment.

✅ If everything is implemented as described — the program will fully meet the requirements and work most reliably as manual interaction.

Update #1 from 27 July 2025

1. General Requirements

The program must be launched manually, work locally on Windows.
Use a real Chrome browser, launch it with existing user profiles.
Ensure maximum imitation of human behavior (delays, clicks, navigation).
The use of headless modes, Selenium/WebDriver, API requests, proxies, multi-threaded simultaneous launches is prohibited.
Record results into a CSV file in the appropriate columns.
The program processes up to 20 accounts per run.

2. Input Parameters (file `settings.json`)

json{
  "chrome_path": "C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
  "profiles": [
    {
      "user_data_dir": "C:\\Users\\User\\AppData\\Local\\Google\\Chrome\\User Data",
      "profile_dir": "Profile 1"
    },
    ...
  ],
  "rewards_url": "https://threeplus.three.co.uk/rewards",
  "csv_path": "C:\\path\\to\\accounts.csv",
  "delay_range_seconds": [90, 240]
}

chrome_path – full path to chrome.exe.
profiles – array of objects with two parameters: user_data_dir (Chrome root directory) and profile_dir (profile folder).
rewards_url – current URL of the rewards page.
csv_path – full path to the CSV file with accounts.
delay_range_seconds – array of two numbers specifying the minimum and maximum random delay between actions (in seconds).

3. CSV File Format

The file contains at least the following columns:
- account — unique account identifier (email, login, or other).
- code — column for recording the received code (initially empty).

4. Program Logic

Loading Configuration
- The program reads settings.json.
- Checks for Chrome at the path chrome_path.
- Loads the list of profiles.
Loading CSV
- Reads the CSV file.
- Identifies accounts with an empty code field (unprocessed).
Processing Accounts
- For each unprocessed account (up to 20 per run), performs:
  a. Launches Chrome via subprocess with parameters:
```
ini--user-data-dir<user_data_dir>  
--profile-directory<profile_dir>  
--remote-debugging-port=9222
```
  b. Connects to the browser via CDP (Playwright/Puppeteer).
  c. Navigates to rewards_url.
  d. Waits for the page to fully load.
  e. Clicks the Get Reward button.
  f. If a modal window with a code appears — reads the code (textually or via OCR if needed).
  g. If there is a Go to My Codes link — navigates there and reads the code.
  h. Records the code in the corresponding code column of the CSV file.
  i. Closes the browser without exiting the profile (browser must be fully closed).
  j. Waits for a random delay within delay_range_seconds before proceeding to the next account.
Termination
- After processing up to 20 accounts, saves the updated CSV.
- Ends the work.

5. Important Restrictions

Action	Reason for Restriction
Using Selenium/WebDriver	Easily detectable and leads to SMS confirmations
Launching Chrome in headless mode	Detected via `navigator.webdriver` and other signals
Making HTTP or API requests to three.co.uk	The site uses Flutter Web and canvas rendering, regular requests do not work
Running more than 1 browser simultaneously	Leads to account blocking as bots
Delays less than 5 seconds	Do not mimic real human behavior, raise suspicion
Changing IP or using proxies	Violates IP-profile consistency, causes re-authorization

6. Recommended Technologies

Component	Recommendations
Language	Python 3.10+ or Node.js v18+
Tool	Playwright (Python) or Puppeteer (Node.js)
Automation	`launchPersistentContext` (Playwright) for a real profile
Action Simulation	Clicking buttons, scrolling, random delays
Data Storage	pandas (Python) or csv-parser (Node.js) for working with CSV

7. Additional Requirements

Logging actions for diagnostics (saving to a separate log file).
Error handling: proper termination on failures, ability to restart from the point of stop.
Concise and clear command-line interface (e.g., informing about the current account, processing status).

Update #2 from 28 July 2025

This is the final updated version of the technical task text, already considering all corrections, including a single method for obtaining the code via "My Codes", without OCR and modal windows:

🎯 Goal

Create a desktop application (in Python or Node.js) that automatically opens real Chrome profiles, logs into 20 Three+ accounts, clicks Get Reward, navigates to My Codes, extracts the code in the form of TH3DC\d{9,} and adds it to the corresponding row in CSV.

⚙️ Functionality

Runs manually on Windows (without auto-start or services).
Works through real Chrome with local profiles (user-data-dir + profile-directory).
Does not use headless mode, Selenium, HTTP requests, or proxies.
Simulates human actions: launching the browser, clicking, navigating, pauses.
Handles up to 20 accounts per run (specified in config).
Skips accounts that already have a code in CSV.

🔄 Working Algorithm

Step 1: Load Configuration

Reads settings.json (Chrome path, profile list, CSV path, delays).
Reads CSV, identifies accounts without a code.

Step 2: Process Each Profile

Launches Chrome with specified profile:

iniCopyEdit--user-data-dir=<user_data_dir>
--profile-directory=<profile_dir>
--remote-debugging-port=9222

Connects via CDP (Playwright or Puppeteer).
Opens rewards_url.
Waits for the page to fully load.
Clicks the Get Reward button (actual click, not .click()): a real click (using Playwright/Puppeteer with mouse.move + mouse.down + mouse.up, without .click()).
Waits 3–5 seconds.

Step 3: Read Code from My Codes

Navigate to https://threeplus.three.co.uk/my-rewards
(or click Go to my codes if it appeared after Get Reward).
Searches for the last code in the format TH3DC\d{9,} in the DOM (via innerText / textContent).
Saves to CSV in the format:
```
CopyEditProfile_01,TH3DC103897998490
```
Closes the browser (but does not log out).
Waits for a random delay (20–40 seconds).
Repeats for up to 20 accounts.

🧠 Exact Method of Reading the Code (Clear Algorithm)

OCR, screenshots, and modal windows are not used.
The code is always read from the My Codes page — after clicking "Get Reward"
Code appearance: TH3DC103897998490 (pattern: TH3DC\d{9,}).
Reading: via DOM (innerText, textContent, evaluate()).

⚙️ Configuration (settings.json)

jsonCopyEdit{
  "chrome_path": "C:\\Program Files\\Google\\Chrome\\Application\\chrome.exe",
  "profiles": [
    {
      "user_data_dir": "C:\\Users\\User\\AppData\\Local\\Google\\Chrome\\User Data",
      "profile_dir": "Profile 1"
    }
    // ...
  ],
  "rewards_url": "https://threeplus.three.co.uk/rewards",
  "csv_path": "C:\\path\\to\\accounts.csv",
  "delay_range_seconds": [90, 240]
}

📄 CSV Format

csvCopyEditaccount,code
profile_01,
profile_02,

account — profile identifier
code — field where the obtained code will be written

🚫 Prohibited

Action	Reason
Selenium/WebDriver	Easily detected, causes SMS
Headless mode	Detected by the site
HTTP/API requests	Flutter Web, Canvas rendering
Multithreaded processing	Raises suspicion/blocking
Proxy or IP change	Triggers re-authorization
Delays < 5 seconds	Unusual behavior

📦 Expected Output Files

.exe for Windows (1 file).
Full source code with comments.
README.md with instructions.
Example settings.json.

🛠️ Recommended Technologies

Component	Tool
Language	Python 3.10+ or Node.js 18+
Automation	Playwright or Puppeteer
Working with CSV	pandas (Python) or csv-parser
Action simulation	Clicks, pauses, scrolls, real profiles

Proposals 4 Withdrawn 1 Rejected 2 Discussions 4

Please select

Nik Ivashkovsky

3 0

Budget: 150 USD Deadline: 5 days

Hello! I am ready to develop a high-quality bot for automatic parsing. The technical specifications are very clear, so there are no questions. I will use Python + Playwright to develop the script. I have been familiar with these technologies for over 3 years and have developed very diverse solutions throughout my career. For example, using asynchronous Playwright, I set up product parsing by query from Aliexpress. I look forward to working with you!

Oleksandr Leonov

Winning proposal

7 0

Budget: 34 USD Deadline: 3 days

Hello! I have experience creating similar solutions on Node.js + Puppeteer with support for real Chrome profiles. I will implement according to all requirements.

Maksim Neheiev

7 0

Projects 9
Rating -
Rating 418

Budget: 50 USD Deadline: 7 days

Good day! I have been coding in Python for a long time, specifically with the Playwright library. I already have a bunch of parsers implemented and would be happy to help you with this task.
The idea is interesting. I have actually done similar projects with "live" interaction before. Feel free to message me privately; I would like to receive the necessary input data, clarify some questions, and then start working.

Andrii Melnik

6 0

Budget: 35 USD Deadline: 2 days

Hello, I will do it on Node.js + Puppeteer. I have made similar programs for spam sites and account farming. I know what to pay attention to so that the services do not detect the script

Oleg Polyakov

52 0

Budget: 400 USD Deadline: 16 days

Good day. I write programs for automating various processes in Python. I also mainly use Playwright. I am 99 percent sure I can implement the project. I just need to conduct tests before starting - and for the tests, one of the profiles will be needed.