Python Translation AI - Translate Python scripts or web pages on the fly to any human language
Project description
pytlai
Python Translation AI — Translate Python scripts or web pages on the fly to any human language.
Features
- HTML Translation — Extract text nodes, skip code/scripts, reconstruct with
lang/dirattributes - Python Translation — Translate docstrings and comments while preserving code structure
- Context-Aware — Captures surrounding context to disambiguate words like "Run", "Post", "Save"
- Smart Caching — Memory, Redis, or file-based caching to avoid re-translating identical content
- Offline Mode — Export to JSON/YAML/PO files, ship with your project, no AI at runtime
- 40+ Languages — Full support for major languages including RTL (Arabic, Hebrew, etc.)
- Batch Processing — Deduplicate and batch API calls for efficiency
Installation
pip install pytlai
With optional dependencies:
pip install pytlai[redis] # Redis cache support
pip install pytlai[yaml] # YAML file support
pip install pytlai[all] # All optional dependencies
Quick Start
Translate HTML
from pytlai import Pytlai
from pytlai.providers import OpenAIProvider
translator = Pytlai(
target_lang="es_ES",
provider=OpenAIProvider(),
)
result = translator.process("<h1>Hello World</h1><p>Welcome to our site.</p>")
print(result.content)
# <html lang="es-ES" dir="ltr"><h1>Hola Mundo</h1><p>Bienvenido a nuestro sitio.</p></html>
Translate Python Script
from pytlai import Pytlai
from pytlai.providers import OpenAIProvider
translator = Pytlai(
target_lang="ja_JP",
provider=OpenAIProvider(),
)
code = '''
def greet(name):
"""Return a greeting message."""
# Build the greeting
return f"Hello, {name}!"
'''
result = translator.process(code, content_type="python")
print(result.content)
# def greet(name):
# """挨拶メッセージを返します。"""
# # 挨拶を作成する
# return f"Hello, {name}!"
Translate Files
# Auto-detects content type from extension
result = translator.process_file("app.py")
result = translator.process_file("index.html")
Command Line Interface
# Translate a file
pytlai translate index.html -l es_ES -o index_es.html
# Translate Python script
pytlai translate app.py -l fr_FR -o app_fr.py -t python
# Extract strings for translation
pytlai extract app.py -o strings.json
# Export to gettext PO format
pytlai export translations.json -o locale/es.po -f po
Offline Workflow
For projects that want to ship translations without runtime AI dependencies:
# 1. During development: translate and export
from pytlai import Pytlai
from pytlai.providers import OpenAIProvider
from pytlai.export import TranslationExporter
translator = Pytlai(target_lang="es_ES", provider=OpenAIProvider())
result = translator.process(content)
# Export translations
exporter = TranslationExporter()
exporter.export_from_cache(translator._cache, "locale/es_ES.json", target_lang="es_ES")
# 2. At runtime: use cached translations (no AI needed)
from pytlai import Pytlai
from pytlai.cache import FileCache
translator = Pytlai(
target_lang="es_ES",
cache=FileCache("locale/es_ES.json"),
provider=None, # No AI provider needed
)
result = translator.process(content)
Configuration
from pytlai import Pytlai, TranslationConfig, PythonOptions
from pytlai.providers import OpenAIProvider
from pytlai.cache import RedisCache
config = TranslationConfig(
target_lang="fr_FR",
source_lang="en",
excluded_terms=["API", "SDK", "pytlai"], # Never translate these
context="Technical documentation for a Python library",
python_options=PythonOptions(
translate_docstrings=True,
translate_comments=True,
translate_strings=False,
),
)
translator = Pytlai(
config=config,
provider=OpenAIProvider(model="gpt-4o"),
cache=RedisCache(url="redis://localhost:6379"),
)
Context-Aware Translation
Single words can translate differently depending on context. pytlai automatically captures surrounding information to help the AI choose the correct translation:
"Run" in <button class="execute-btn"> → "Ejecutar" (execute)
"Run" in a sports article → "Correr" (physical running)
"Post" in a blog interface → "Publicar" (publish)
"Post" in mail context → "Correo" (postal mail)
"Save" in a file dialog → "Guardar" (save file)
"Save" in a banking app → "Ahorrar" (save money)
Context captured automatically:
| Content | Context Information |
|---|---|
| HTML | Parent tag, CSS classes, sibling text, ancestor path |
| Python docstring | Function/class name, parameters, return type |
| Python comment | Surrounding code lines |
This happens automatically — no configuration needed.
Supported Languages
Tier 1 (High Quality): English, German, Spanish, French, Italian, Japanese, Portuguese, Chinese (Simplified/Traditional)
Tier 2 (Good Quality): Arabic, Bengali, Czech, Danish, Greek, Finnish, Hebrew, Hindi, Hungarian, Indonesian, Korean, Dutch, Norwegian, Polish, Romanian, Russian, Swedish, Thai, Turkish, Ukrainian, Vietnamese
Tier 3 (Functional): Bulgarian, Catalan, Persian, Croatian, Lithuanian, Latvian, Malay, Slovak, Slovenian, Serbian, Swahili, Filipino, Urdu
Environment Variables
| Variable | Description | Default |
|---|---|---|
OPENAI_API_KEY |
OpenAI API key | - |
OPENAI_MODEL |
Model to use | gpt-4o-mini |
OPENAI_BASE_URL |
Custom API base URL | - |
REDIS_URL |
Redis connection string | redis://localhost:6379 |
API Reference
Pytlai
The main translator class.
Pytlai(
target_lang: str, # Target language code (e.g., "es_ES")
provider: AIProvider = None, # AI provider (OpenAIProvider, etc.)
cache: TranslationCache = None, # Cache backend (InMemoryCache, RedisCache, FileCache)
config: TranslationConfig = None, # Full configuration object
source_lang: str = "en", # Source language code
excluded_terms: list[str] = None, # Terms to never translate
context: str = None, # Context for AI translations
)
Methods:
process(content, content_type=None)→ProcessedContent— Translate contentprocess_file(path, content_type=None)→ProcessedContent— Translate a filetranslate_text(text)→str— Translate a single string
ProcessedContent
Result of translation.
ProcessedContent(
content: str, # Translated content
translated_count: int, # Number of newly translated items
cached_count: int, # Number of cache hits
total_nodes: int, # Total translatable nodes found
)
TextNode
Internal representation of a translatable text (for advanced usage).
TextNode(
id: str, # Unique identifier
text: str, # Original text content
hash: str, # SHA-256 hash for caching
node_type: str, # "html_text", "docstring", "comment", etc.
context: str, # Surrounding context for disambiguation
metadata: dict, # Additional info (line number, parent tag, etc.)
)
The context field is automatically populated with surrounding information:
- HTML:
"in <button class='primary'> | with: Cancel | inside: form > div" - Python:
"docstring for function 'save_file' | parameters: path, data"
Contributing
Contributions are welcome! Please read our contributing guidelines and submit pull requests.
License
MIT License - see LICENSE for details.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file pytlai-0.1.1.tar.gz.
File metadata
- Download URL: pytlai-0.1.1.tar.gz
- Upload date:
- Size: 46.4 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
7bc3477feab2a8a47bc92b917965bcaae4beec0dbc34e98f2f05203016650458
|
|
| MD5 |
d7bfaf46efd7f53b446599cd69a11fb7
|
|
| BLAKE2b-256 |
a6497a985ca75c9ce2406e3d6a22e26f03deecdef3e7af56f99e4e5028ae3023
|
File details
Details for the file pytlai-0.1.1-py3-none-any.whl.
File metadata
- Download URL: pytlai-0.1.1-py3-none-any.whl
- Upload date:
- Size: 43.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.1.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
f1664441976964bc74b34a3524efa6a1ea6f024bbc60ba2a5ad9d0a345b8e3ce
|
|
| MD5 |
d27b6784d46cc91dbb8bd6ac970b222b
|
|
| BLAKE2b-256 |
35693313f1204ce5e425761156ef169a68a1de8202d320c6744d7f80a1c13a91
|
