# Sawalni Python SDK
This is the official Python SDK for the Sawalni API, providing easy access to language-related services such as embedding generation, language identification, and translation. Sawalni API is developed by [Omneity Labs](https://sawalni.com/developers), and provides unique multilingual models and NLP capabilities, including pioneering Moroccan Darija support.
## Installation
Install the package using pip:
```bash
pip install sawalni
```
## Quick Start
To use the Sawalni SDK, you'll need an API key. You can set it as an environment variable or pass it directly to the client:
```py
from sawalni import Sawalni
client = Sawalni(api_key='your_api_key_here')
# or specify the key via SAWALNI_API_KEY in the environment
```
## Features
The SDK supports both synchronous and asynchronous operations for the following services:
1. **Chat**: Generate chat completions using our `Sawalni` multilingual models, supporting Moroccan Darija, English, French, Arabic and many other languages.
2. **Search**: Perform internet searches in multiple languages with a single query.
3. **Generate Embeddings**: Create embeddings for text in multiple languages using `Madmon`.
4. **Identify Language**: Detect the language of a given text with `Gherbal`, supporting up to 33 languages.
5. **Translate Text**: Translate text between 13 supported languages with `Tarjamli`.
6. **Transliterate Text**: Convert Moroccan Arabic script to Moroccan Latin script with `Daktilo`.
The Sawalni SDK includes an OpenAI compatible client, which can be accessed via the `chat` and `embeddings` properties, or direct use via the OpenAI client as detailed below.
### Chat
```py
# Available models: sawalni-micro, sawalni-mini, sawalni-small
chat = client.chat.completions.create(messages=[{"role": "user", "content": "Hello, how are you?"}], model="sawalni-small")
# Stream
stream = client.chat.completions.create(messages=[{"role": "user", "content": "Hello, how are you?"}], model="sawalni-small", stream=True)
for chunk in stream:
print(chunk.choices[0].delta.content)
```
### Search
```py
search = client.search("Hello, how are you?")
```
### Generate Embeddings
```py
embeddings = client.embed("Hello, world!")
```
### Identify Language
```py
language = client.identify("Bonjour le monde")
```
### Translate Text
```py
# You can specify a source language or let the model detect it automatically
translation = client.translate("Hello", source="auto", target="ary_Latn")
```
### Transliterate Text
```py
transliteration = client.transliterate("اهلا بيك", model="daktilo-mini", to="latn", temperature=0.1)
# {"text": "ahlane bik"}
```
## Asynchronous Usage
For asynchronous operations, use the SawalniAsync client:
```py
from sawalni import SawalniAsync
async_client = SawalniAsync(api_key='your_api_key_here')
embeddings = await async_client.embed("Hello, world!")
```
## OpenAI compatible client
The SDK also includes an OpenAI compatible client, which can be accessed via the `chat` and `embeddings` properties:
```py
chat = client.chat
embeddings = client.embeddings
```
You can also use the OpenAI client directly with the base URL set to `https://api.sawalni.com/v1` and the API key set to your Sawalni API key.
```py
import openai
client = openai.OpenAI(base_url="https://api.sawalni.com/v1", api_key="your_api_key_here")
```
Only the `chat` and `embeddings` properties are supported with this approach.
## Documentation
For detailed information about available models, parameters, languages and and response formats, please refer to the complete API documentation at https://api.sawalni.com.
## Support
If you encounter any issues or have questions, please contact api@sawalni.com.
Raw data
{
"_id": null,
"home_page": null,
"name": "sawalni",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.7",
"maintainer_email": null,
"keywords": "nlp language processing embedding translation transliteration identification api sdk moroccan darija arabic multilingual low-resource languages",
"author": "Omar Kamali",
"author_email": "api@sawalni.com",
"download_url": "https://files.pythonhosted.org/packages/65/d4/128478eb83e9bbc045ed7456b8147948beddfa403adf08de3d01f3572332/sawalni-0.2.11.tar.gz",
"platform": null,
"description": "# Sawalni Python SDK\n\nThis is the official Python SDK for the Sawalni API, providing easy access to language-related services such as embedding generation, language identification, and translation. Sawalni API is developed by [Omneity Labs](https://sawalni.com/developers), and provides unique multilingual models and NLP capabilities, including pioneering Moroccan Darija support.\n\n## Installation\n\nInstall the package using pip:\n\n```bash\npip install sawalni\n```\n\n## Quick Start\n\nTo use the Sawalni SDK, you'll need an API key. You can set it as an environment variable or pass it directly to the client:\n\n```py\nfrom sawalni import Sawalni\n\nclient = Sawalni(api_key='your_api_key_here') \n# or specify the key via SAWALNI_API_KEY in the environment\n```\n\n## Features\n\nThe SDK supports both synchronous and asynchronous operations for the following services:\n\n1. **Chat**: Generate chat completions using our `Sawalni` multilingual models, supporting Moroccan Darija, English, French, Arabic and many other languages.\n2. **Search**: Perform internet searches in multiple languages with a single query.\n3. **Generate Embeddings**: Create embeddings for text in multiple languages using `Madmon`.\n4. **Identify Language**: Detect the language of a given text with `Gherbal`, supporting up to 33 languages.\n5. **Translate Text**: Translate text between 13 supported languages with `Tarjamli`.\n6. **Transliterate Text**: Convert Moroccan Arabic script to Moroccan Latin script with `Daktilo`.\n\nThe Sawalni SDK includes an OpenAI compatible client, which can be accessed via the `chat` and `embeddings` properties, or direct use via the OpenAI client as detailed below.\n\n### Chat\n\n```py\n# Available models: sawalni-micro, sawalni-mini, sawalni-small\nchat = client.chat.completions.create(messages=[{\"role\": \"user\", \"content\": \"Hello, how are you?\"}], model=\"sawalni-small\")\n\n# Stream\nstream = client.chat.completions.create(messages=[{\"role\": \"user\", \"content\": \"Hello, how are you?\"}], model=\"sawalni-small\", stream=True)\nfor chunk in stream:\n print(chunk.choices[0].delta.content)\n```\n\n### Search\n\n```py\nsearch = client.search(\"Hello, how are you?\")\n```\n\n### Generate Embeddings\n\n```py\nembeddings = client.embed(\"Hello, world!\")\n```\n\n### Identify Language\n\n```py\nlanguage = client.identify(\"Bonjour le monde\")\n```\n\n### Translate Text\n\n```py\n# You can specify a source language or let the model detect it automatically\ntranslation = client.translate(\"Hello\", source=\"auto\", target=\"ary_Latn\")\n```\n\n### Transliterate Text\n\n```py\ntransliteration = client.transliterate(\"\u0627\u0647\u0644\u0627 \u0628\u064a\u0643\", model=\"daktilo-mini\", to=\"latn\", temperature=0.1)\n\n# {\"text\": \"ahlane bik\"}\n```\n\n## Asynchronous Usage\n\nFor asynchronous operations, use the SawalniAsync client:\n\n```py\nfrom sawalni import SawalniAsync\n\nasync_client = SawalniAsync(api_key='your_api_key_here')\nembeddings = await async_client.embed(\"Hello, world!\")\n```\n\n## OpenAI compatible client\n\nThe SDK also includes an OpenAI compatible client, which can be accessed via the `chat` and `embeddings` properties:\n\n```py\nchat = client.chat\nembeddings = client.embeddings\n```\n\nYou can also use the OpenAI client directly with the base URL set to `https://api.sawalni.com/v1` and the API key set to your Sawalni API key.\n\n```py\nimport openai\nclient = openai.OpenAI(base_url=\"https://api.sawalni.com/v1\", api_key=\"your_api_key_here\")\n```\n\nOnly the `chat` and `embeddings` properties are supported with this approach.\n\n## Documentation\n\nFor detailed information about available models, parameters, languages and and response formats, please refer to the complete API documentation at https://api.sawalni.com.\n\n## Support\n\nIf you encounter any issues or have questions, please contact api@sawalni.com.\n",
"bugtrack_url": null,
"license": null,
"summary": "Official Python SDK for the Sawalni API",
"version": "0.2.11",
"project_urls": null,
"split_keywords": [
"nlp",
"language",
"processing",
"embedding",
"translation",
"transliteration",
"identification",
"api",
"sdk",
"moroccan",
"darija",
"arabic",
"multilingual",
"low-resource",
"languages"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "bd8d6c10f20ffe856ab1a20517b5b823f7421d66ea8652c3a409977ffcc6ba78",
"md5": "6ea6a02897c6913267d4de0a647051d5",
"sha256": "9703376f6d3eb15fdb5c6823b249c8d7a0532e67408ec423e341c7e65fe9290c"
},
"downloads": -1,
"filename": "sawalni-0.2.11-py3-none-any.whl",
"has_sig": false,
"md5_digest": "6ea6a02897c6913267d4de0a647051d5",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.7",
"size": 5860,
"upload_time": "2025-01-04T00:43:06",
"upload_time_iso_8601": "2025-01-04T00:43:06.803586Z",
"url": "https://files.pythonhosted.org/packages/bd/8d/6c10f20ffe856ab1a20517b5b823f7421d66ea8652c3a409977ffcc6ba78/sawalni-0.2.11-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "65d4128478eb83e9bbc045ed7456b8147948beddfa403adf08de3d01f3572332",
"md5": "2cc008d3c5e12c41a07f9d14ce03ae69",
"sha256": "87daca6df15ac35475245029a79a9d466f823efcce42312f630a29291ef6383f"
},
"downloads": -1,
"filename": "sawalni-0.2.11.tar.gz",
"has_sig": false,
"md5_digest": "2cc008d3c5e12c41a07f9d14ce03ae69",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.7",
"size": 6057,
"upload_time": "2025-01-04T00:43:07",
"upload_time_iso_8601": "2025-01-04T00:43:07.834171Z",
"url": "https://files.pythonhosted.org/packages/65/d4/128478eb83e9bbc045ed7456b8147948beddfa403adf08de3d01f3572332/sawalni-0.2.11.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-01-04 00:43:07",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "sawalni"
}
JSON