Skip to content

Building a Standalone Executable — urdu.exe بنانا

This guide explains how to compile the Urdu Programming Language into a standalone urdu.exe executable using Nuitka. The resulting executable runs on any Windows machine without requiring Python to be installed.

اردو: یہ رہنما بتاتا ہے کہ Nuitka سے اردو پروگرامنگ زبان کو ایک خودکفیل urdu.exe قابل عمل میں کیسے مرتب کریں۔ نتیجتاً قابل عمل Python نصب کیے بغیر کسی بھی Windows مشین پر چلتی ہے۔

Overview — جائزہ

The build system uses Nuitka to compile the Python interpreter and all bundled runtime libraries into a self-contained Windows executable. The process:

  1. Nuitka analyses the urdu package and all included pip packages.
  2. It compiles everything to C, then links a native Windows binary.
  3. The output is a dist/urdu.dist/ folder containing urdu.exe and all required DLLs.

اردو: تعمیراتی نظام Nuitka استعمال کرتا ہے جو Python تفسیر کار اور تمام بنڈل رن ٹائم لائبریریاں ایک خودکفیل Windows قابل عمل میں مرتب کرتا ہے۔

Output size: approximately 600 MB for the full standalone build (including web frameworks, databases, ML libraries, and cryptography). A fast build (core + web only) is roughly 250–350 MB.

Build time: - Full build: 20–60 minutes on a modern machine. - Fast build (core + web): 10–20 minutes. - Adding TensorFlow (--with-tf): adds ~30 minutes and ~1.1 GB to the output.

اردو: آؤٹ پٹ کا سائز: مکمل خودکفیل تعمیر کے لیے تقریباً 600 MB۔ تیز تعمیر (صرف مرکز + ویب) تقریباً 250-350 MB ہے۔

Requirements — ضروریات

System — نظام

اردو: نظامی ضروریات: Windows 10/11 (64 بٹ)، کم از کم 4 GB RAM، 5 GB خالی ڈسک جگہ، اور Visual Studio Build Tools۔

  • Windows 10 or Windows 11 (64-bit)
  • At least 4 GB RAM (8 GB recommended)
  • At least 5 GB free disk space
  • A C compiler: install Visual Studio Build Tools (free) or full Visual Studio.
  • Download: https://visualstudio.microsoft.com/visual-cpp-build-tools/
  • Select workload: "Desktop development with C++"

Python environment — Python ماحول

Important: Use a clean virtual environment — not Anaconda base. Anaconda includes hundreds of extra packages (scipy, matplotlib, IPython, Jupyter, etc.) that Nuitka will include even if you do not use them, inflating build time from ~1 hour to 6+ hours and doubling the output size.

اردو: اہم: صاف ورچوئل ماحول استعمال کریں — Anaconda بیس نہیں۔ Anaconda میں سینکڑوں اضافی پیکیج ہوتے ہیں جو Nuitka شامل کر لیتا ہے اور تعمیراتی وقت 1 گھنٹے سے 6+ گھنٹے تک بڑھ جاتا ہے۔

The project ships a pre-configured build_env/ virtual environment in the repository for convenience. If it does not exist or you prefer to create your own, follow the steps below.

Step-by-Step Build Instructions — قدم بہ قدم تعمیری ہدایات

Step 1 — Create a clean virtual environment — مرحلہ ۱ — صاف ورچوئل ماحول بنائیں

Open a terminal in the project root folder:

python -m venv build_env

Step 2 — Activate the virtual environment — مرحلہ ۲ — ورچوئل ماحول فعال کریں

Windows (PowerShell): 

build_env\Scripts\Activate.ps1

Windows (Command Prompt): 

build_env\Scripts\activate.bat

Your prompt should now show (build_env) at the beginning.

اردو: آپ کی پرامپٹ اب شروع میں (build_env) دکھائے گی۔

Step 3 — Install required packages — مرحلہ ۳ — درکار پیکیج نصب کریں

Install Nuitka and all the runtime packages you want to bundle:

اردو: Nuitka اور تمام رن ٹائم پیکیج نصب کریں جو آپ بنڈل کرنا چاہتے ہیں۔

pip install nuitka

# Core
pip install -e .

# Web frameworks
pip install fastapi flask django uvicorn websockets python-socketio python-multipart
pip install httpx aiohttp requests pydantic

# Cryptography
pip install cryptography bcrypt python-jose[cryptography]

# Databases
pip install sqlalchemy pymysql psycopg2-binary aiosqlite pymongo

# Machine learning (skip if not needed — saves ~30 min build time)
pip install numpy pandas scikit-learn scipy

# Scraping and files
pip install beautifulsoup4 lxml openpyxl

# Logging
pip install colorlog

You do not need to install all packages. The build script (build.py) automatically detects which packages are installed and includes only those. Packages that are not installed are listed as "skipped" in the build summary and will not be available in the compiled executable.

اردو: تمام پیکیج نصب کرنا ضروری نہیں۔ تعمیری اسکرپٹ خودکار طور پر نصب شدہ پیکیج پہچانتا اور صرف انہی کو شامل کرتا ہے۔

Step 4 — Run the build script — مرحلہ ۴ — تعمیری اسکرپٹ چلائیں

python build.py

The build script will: 1. Check which packages are installed. 2. Print a summary of what will be included and what will be skipped. 3. Invoke Nuitka with the correct flags. 4. Report success and the output path when complete.

اردو: تعمیری اسکرپٹ نصب شدہ پیکیج جانچے گا، خلاصہ پیش کرے گا، Nuitka چلائے گا، اور مکمل ہونے پر نتیجہ رپورٹ کرے گا۔

Build Modes — تعمیری طریقے

Default — Full standalone build — ڈیفالٹ — مکمل خودکفیل تعمیر

python build.py

Includes: core interpreter, web frameworks, databases, ML (excluding TensorFlow), cryptography, scraping, file I/O, threading, date/time, logging.

Output: dist\urdu.dist\urdu.exe

--fast — Core + Web only — تیز — صرف مرکز + ویب

python build.py --fast

Skips database and ML packages. Faster to build, smaller output. Ideal for web application deployments.

Output: dist\urdu.dist\urdu.exe (~250 MB)

--onefile — Single executable file — یک فائل قابل عمل

python build.py --onefile

Packages everything into a single urdu.exe file (using a self-extracting archive). Startup is slower because the archive is unpacked into a temp folder on first run, but distribution is a single file.

Output: dist\urdu.exe (single file)

اردو: سب کچھ ایک urdu.exe فائل میں بند کرتا ہے۔ پہلی بار چلانے پر سست ہوتا ہے مگر تقسیم ایک فائل سے ہوتی ہے۔

--with-tf — Include TensorFlow — TensorFlow شامل کریں

python build.py --with-tf

Also bundles TensorFlow, Keras, and TensorBoard. Adds approximately 1.1 GB and 30 minutes to the build.

Requires: pip install tensorflow before running the build.

--check — Dry run (print Nuitka command only) — جانچ — صرف Nuitka حکم پرنٹ کریں

python build.py --check
python build.py --fast --check
python build.py --onefile --with-tf --check

Prints the full Nuitka command that would be executed, without actually running it. Useful for inspecting or customising the command.

اردو: چلائے بغیر Nuitka حکم پرنٹ کرتا ہے — معائنے یا تبدیلی کے لیے مفید۔

Combining flags — جھنڈے ملائیں

python build.py --fast --onefile
python build.py --with-tf --onefile

Output Structure — آؤٹ پٹ کی ساخت

After a successful standalone build:

اردو: کامیاب خودکفیل تعمیر کے بعد آؤٹ پٹ کی ساخت۔

dist\
  urdu.dist\
    urdu.exe              ← main executable (run this)
    python312.dll         ← embedded CPython runtime
    _asyncio.pyd          ┐
    _ctypes.pyd           │ compiled Python extension modules
    _decimal.pyd          ┘
    fastapi\              ┐
    flask\                │ bundled pip packages (as compiled modules)
    django\               ┘
    ...                   ← ~600 MB total

After a --onefile build:

dist\
  urdu.exe                ← single self-extracting executable

Using the Built Executable — تیار شدہ قابل عمل استعمال کرنا

From the dist\urdu.dist\ folder:

urdu.exe run myprogram.urdu
urdu.exe repl
urdu.exe version
urdu.exe مدد
urdu.exe compile myfile.urdu

Or add dist\urdu.dist\ to your system PATH so you can type urdu from anywhere.

اردو: dist\urdu.dist\ فولڈر کو PATH میں شامل کریں تاکہ urdu کہیں سے بھی ٹائپ کر سکیں۔

Distribution — تقسیم

To distribute the compiled executable:

اردو: تیار شدہ قابل عمل تقسیم کرنے کے لیے۔

  1. Copy the entire dist\urdu.dist\ folder to the target machine.
  2. No Python installation is required on the target machine.
  3. The target machine must be Windows 64-bit (same architecture as the build machine).
  4. The Visual C++ runtime (vcruntime140.dll) is included automatically by Nuitka.

For GitHub releases, zip the urdu.dist folder and attach it as a release asset.

build.py Options Reference — build.py اختیارات کا حوالہ

Flag Description Default
(none) Full standalone build with all detected packages —
--fast Skip database and ML packages False
--onefile Single self-extracting .exe instead of folder False
--with-tf Include TensorFlow (+1.1 GB, +~30 min) False
--check Dry run — print the Nuitka command, do not build False

Nuitka flags used internally — اندرونی طور پر استعمال شدہ Nuitka جھنڈے

Nuitka flag Purpose
--standalone Self-contained folder with all DLLs
--onefile Alternate: single file with embedded archive
--output-dir=dist Place output in dist/
--output-filename=urdu.exe Name the executable urdu.exe
--windows-console-mode=attach Attach to existing console (no new window)
--lto=no Disable link-time optimisation (faster build)
--jobs=4 Parallel C compilation with 4 threads
--enable-plugins=no-qt Do not bundle Qt bindings
--include-package=<pkg> Explicitly include each detected package
--include-distribution-metadata=<dist> Bundle .dist-info for packages that call importlib.metadata at startup (e.g. opentelemetry-api, uvicorn, anyio, flask). Nuitka 2.3.9 does not support the * wildcard — list each package name explicitly
--nofollow-import-to=django.core.management Exclude Django's management command infrastructure (not needed at runtime; importing it in the compiled exe causes a crash)
--python-flag=-m Run as python -m urdu

Troubleshooting — مسائل کا حل

اردو: عام مسائل اور ان کے حل۔

Problem Solution
nuitka: command not found Activate the virtual environment first
Build fails after 1–2 minutes Ensure Visual Studio Build Tools are installed with the C++ workload
Build takes 6+ hours You are building from Anaconda base — create a clean venv instead
urdu.exe crashes on another machine Ensure the target machine is 64-bit Windows; check that the full urdu.dist folder was copied
Missing library at runtime The package was not installed when build.py ran — install it, then rebuild
TensorFlow not found Run pip install tensorflow inside the build venv first
Low disk space error Nuitka needs ~3 GB working space during the build
FATAL: could not find distribution '*' Nuitka 2.3.9 does not support --include-distribution-metadata=*. Use explicit package names (already fixed in build.py)
cannot locate package '' associated with metadata at runtime The compiled exe is missing dist-info for a package that calls importlib_metadata.entry_points() (typically opentelemetry-api). Rebuild with the explicit --include-distribution-metadata flags in build.py
Django app crashes with ModuleNotFoundError: django.core.management django.core.management is excluded from the compiled exe by --nofollow-import-to. Remove django.contrib.auth and django.contrib.contenttypes from INSTALLED_APPS — they transitively import that module. The ڈجانگو wrapper no longer includes them by default

Quick Reference — فوری حوالہ

اردو: تعمیر کی فوری یاد دہانی۔

# One-time setup
python -m venv build_env
build_env\Scripts\Activate.ps1
pip install -e . nuitka fastapi flask django uvicorn cryptography ...

# Build
python build.py              # full build  → dist\urdu.dist\urdu.exe
python build.py --fast       # quick build → dist\urdu.dist\urdu.exe
python build.py --onefile    # single file → dist\urdu.exe
python build.py --check      # dry run, no compilation

# Run the result
dist\urdu.dist\urdu.exe run hello.urdu

Previous: Quick Start | Next: Contributing →