Metadata-Version: 2.4
Name: octen
Version: 0.2.0
Summary: Official Python SDK for Octen API - Web Search, Text Embeddings, and LLM Chat
Author-email: Octen Team <support@octen.ai>
License: MIT
Keywords: octen,search,embedding,chat,llm,api,sdk,web-search,text-embedding
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.8
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Classifier: Topic :: Internet :: WWW/HTTP
Requires-Python: >=3.8
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: httpx[http2]>=0.25.0
Requires-Dist: pydantic>=2.0.0
Provides-Extra: dev
Requires-Dist: pytest>=7.0.0; extra == "dev"
Requires-Dist: pytest-asyncio>=0.21.0; extra == "dev"
Requires-Dist: pytest-cov>=4.0.0; extra == "dev"
Requires-Dist: pytest-mock>=3.10.0; extra == "dev"
Requires-Dist: black>=23.0.0; extra == "dev"
Requires-Dist: ruff>=0.1.0; extra == "dev"
Requires-Dist: mypy>=1.0.0; extra == "dev"
Requires-Dist: build>=1.0.0; extra == "dev"
Requires-Dist: twine>=4.0.0; extra == "dev"
Provides-Extra: async
Requires-Dist: httpx[http2]>=0.25.0; extra == "async"
Provides-Extra: all
Requires-Dist: octen[async,dev]; extra == "all"
Dynamic: license-file

# Octen Python SDK

[![PyPI version](https://badge.fury.io/py/octen.svg)](https://badge.fury.io/py/octen)
[![Python Support](https://img.shields.io/pypi/pyversions/octen.svg)](https://pypi.org/project/octen/)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

Official Python SDK for the [Octen API](https://octen.ai) — web search, text embeddings, and multi-model LLM chat in one package.

## ✨ Features

- 🔍 **Web Search** — search and retrieve ranked web results with filtering, highlighting, and full content
- 💬 **Multi-model Chat** — access 10+ LLMs (GPT, Claude, Gemini, Kimi, MiniMax) through a single unified API
- 🧮 **Text Embeddings** — convert text into high-quality vector representations
- ⚡ **Streaming (SSE)** — real-time token streaming with typed event objects
- 🔄 **Auto Retry** — exponential backoff for transient errors
- 🛡️ **Type Safe** — full Pydantic models with IDE auto-completion
- 🔀 **Async Support** — native `asyncio` client for concurrent workloads
- 📦 **HTTP/2** — connection pooling and keep-alive out of the box

## 📦 Installation

```bash
pip install octen
```

Requires Python 3.8 or higher.

### Development Version

```bash
pip install octen[dev]
```

### Async Support

```bash
pip install octen[async]
```

## 🚀 Quick Start

### Search

```python
from octen import Octen

with Octen(api_key="your-api-key") as client:
    response = client.search.search(query="Python programming", count=5)

    for result in response.results:
        print(f"Title: {result['title']}")
        print(f"URL: {result['url']}")
        print(f"Highlight: {result.get('highlight', '')}")
```

### Chat

```python
from octen import Octen, ChatMessage

with Octen(api_key="your-api-key") as client:
    response = client.chat.create(
        model="openai/gpt-5.4",
        messages=[ChatMessage(role="user", content="Hello!")],
        web_search="on"
    )
    print(response.text)
```

### Embeddings

```python
from octen import Octen

with Octen(api_key="your-api-key") as client:
    embedding = client.embedding.create(
        input=["Hello, world!"],
        model="octen-embedding-4b"
    )
    vector = embedding.get_first_embedding()
    print(f"Vector dimension: {len(vector)}")
```

## 🔍 Search API

### Advanced Search

```python
from octen import Octen, HighlightOptions, FullContentOptions

with Octen(api_key="your-api-key") as client:
    response = client.search.search(
        query="machine learning best practices",
        count=10,
        search_type="semantic",  # Semantic search
        include_domains=["github.com", "arxiv.org"],  # Search only these domains
        start_time="2024-01-01T00:00:00Z",  # Time filtering
        highlight=HighlightOptions(
            enable=True,
            max_tokens=500
        ),
        full_content=FullContentOptions(
            enable=True,
            max_tokens=2000
        ),
        timeout=60.0  # Custom timeout
    )

    print(f"Found {len(response.results)} results")
    print(f"Actual search type: {response.search_type}")
    print(f"Token usage: {response.usage}")
```

## 💬 Chat API

### Non-streaming

```python
from octen import Octen, ChatMessage, WebSearchOptions

with Octen(api_key="your-api-key") as client:
    response = client.chat.create(
        model="openai/gpt-5.4",
        messages=[
            ChatMessage(role="system", content="You are a helpful assistant."),
            ChatMessage(role="user", content="What happened in tech today?"),
        ],
        web_search="on",
        web_search_options=WebSearchOptions(safesearch="off", count=5),
        max_tokens=500,
        temperature=0.7
    )

    print(response.text)
    print(f"Tokens used: {response.usage.total_tokens}")

    # Access search results
    if response.search_results:
        for group in response.search_results:
            for item in group.results:
                print(f"  - {item.title}: {item.url}")
```

### Streaming

```python
from octen import Octen, ChatMessage

with Octen(api_key="your-api-key") as client:
    for event in client.chat.create(
        model="openai/gpt-5.4",
        messages=[ChatMessage(role="user", content="Tell me a story")],
        stream=True,
        web_search="on"
    ):
        if event.type == "search_done":
            print(f"[{len(event.search_results or [])} search groups]")

        elif event.type == "content" and event.choices:
            print(event.choices[0].delta.content or "", end="", flush=True)

        elif event.type == "finish":
            print()  # newline

        elif event.type == "usage" and event.usage:
            print(f"[total tokens: {event.usage.total_tokens}]")
```

### Tool Calling

```python
from octen import Octen
from octen.models import ChatMessage, Tool, ToolFunction

weather_tool = Tool(
    function=ToolFunction(
        name="get_weather",
        description="Get current weather for a city",
        parameters={
            "type": "object",
            "properties": {
                "city": {"type": "string", "description": "City name"},
            },
            "required": ["city"],
        }
    )
)

with Octen(api_key="your-api-key") as client:
    response = client.chat.create(
        model="openai/gpt-5.4",
        messages=[ChatMessage(role="user", content="What's the weather in London?")],
        tools=[weather_tool],
        tool_choice="auto"
    )
    if response.choices[0].finish_reason == "tool_calls":
        tc = response.choices[0].message.tool_calls[0]
        print(f"Tool: {tc.function.name}, Args: {tc.function.arguments}")
```

### JSON Output Mode

```python
from octen import Octen, ChatMessage
from octen.models import ResponseFormat

with Octen(api_key="your-api-key") as client:
    response = client.chat.create(
        model="google/gemini-3-flash-preview",
        messages=[ChatMessage(role="user", content="Return a JSON list of 3 programming languages")],
        response_format=ResponseFormat(type="json_object"),
        web_search="off"
    )
    print(response.text)
```

### Web Search with Full Page Content

```python
from octen import Octen, ChatMessage, WebSearchOptions
from octen.models.chat import ChatFullContentOptions

with Octen(api_key="your-api-key") as client:
    response = client.chat.create(
        model="openai/gpt-5.4",
        messages=[ChatMessage(role="user", content="Latest Python 3.13 features?")],
        web_search="on",
        web_search_options=WebSearchOptions(
            safesearch="off",
            full_content=ChatFullContentOptions(enable=True, max_tokens=1000)
        )
    )
    print(f"Full content tokens: {response.usage.full_content_tokens}")
```

### 🤖 Supported Chat Models

For the full and up-to-date list of supported models, visit the [Octen official website](https://docs.octen.ai/).

## 🧮 Embeddings API

### Batch Embeddings

```python
from octen import Octen

with Octen(api_key="your-api-key") as client:
    # Process multiple texts
    texts = [
        "Artificial intelligence is transforming the world",
        "Applications of deep learning",
        "Natural language processing technology"
    ]

    response = client.embedding.create(
        input=texts,
        model="octen-embedding-8b",
        input_type="document"
    )

    vectors = response.get_embeddings()
    print(f"Generated {len(vectors)} vectors")

    # Or use convenience methods
    query_vector = client.embedding.embed_query("search query")
    doc_vectors = client.embedding.embed_documents(["document 1", "document 2"])
```

### Custom Configuration

```python
from octen import Octen

client = Octen(
    api_key="your-api-key",
    base_url="https://api.octen.ai",  # Custom API endpoint
    timeout=10.0,  # Global default timeout (seconds)
    max_retries=3,  # Maximum retry attempts
    http2=True  # Enable HTTP/2
)

try:
    # This request uses global timeout (10 seconds)
    response1 = client.search.search("query 1")

    # This request overrides timeout to 30 seconds
    response2 = client.search.search("complex query", timeout=30.0)
finally:
    client.close()  # Release connection pool resources
```

## 📚 API Documentation

### Search API

#### `client.search.search()`

Perform a web search query.

**Parameters:**

- `query` (str, required): Search query string, max 500 characters
- `count` (int, optional): Number of results to return, range 1-100, default 5
- `search_type` (str, optional): Search type, options:
  - `"auto"` - Automatically select (default)
  - `"keyword"` - Keyword search
  - `"semantic"` - Semantic search
- `include_domains` (List[str], optional): Include only results from these domains
- `exclude_domains` (List[str], optional): Exclude results from these domains
- `include_text` (List[str], optional): Results must contain these texts
- `exclude_text` (List[str], optional): Results must exclude these texts
- `time_basis` (str, optional): Time basis, options: `"auto"`, `"published"`, `"crawled"`
- `start_time` (str, optional): Start time in ISO 8601 format
- `end_time` (str, optional): End time in ISO 8601 format
- `highlight` (HighlightOptions, optional): Highlight options configuration
- `format` (str, optional): Content format, options: `"text"`, `"markdown"`
- `safesearch` (str, optional): Safe search, options: `"off"`, `"strict"` (default)
- `full_content` (FullContentOptions, optional): Full content options configuration
- `timeout` (float, optional): Request timeout in seconds

**Returns:** `SearchResponse` object

**Response Properties:**

- `results` - List of search results
- `query` - The actual query used
- `search_type` - The actual search type used
- `usage` - Token usage information
- `latency` - Latency information

### Chat API

#### `client.chat.create()`

Create a chat completion (non-streaming or streaming).

**Parameters:**

- `messages` (List[ChatMessage | dict], required): Conversation history. Each item can be a `ChatMessage` object or a plain dict `{"role": ..., "content": ...}`
- `model` (str, required): Model ID (e.g. `"openai/gpt-5.4"`). See [Supported Chat Models](#-supported-chat-models) for the full list
- `stream` (bool, optional): If `True`, return a `Stream` iterator of `StreamEvent` objects. Default `False`
- `web_search` (str, optional): `"on"` to augment with live web search, `"off"` to disable
- `web_search_options` (WebSearchOptions, optional): Fine-grained search configuration
  - `safesearch` (str): `"off"` or `"strict"` (default `"off"`)
  - `count` (int): Number of search results, range 1-100
  - `country` (str): Country code for localised results (e.g. `"CN"`)
  - `include_domains` / `exclude_domains` (List[str]): Domain filtering
  - `include_text` / `exclude_text` (List[str]): Text filtering
  - `time_basis` (str): `"auto"`, `"published"`, or `"crawled"`
  - `start_time` / `end_time` (str): ISO 8601 time range
  - `format` (str): `"text"` or `"markdown"`
  - `full_content` (ChatFullContentOptions): Full page content options
  - `highlight` (ChatHighlightOptions): Highlight snippet options
- `max_tokens` (int, optional): Maximum number of output tokens
- `max_completion_tokens` (int, optional): Alternative max-token parameter
- `temperature` (float, optional): Sampling temperature `[0, 2]`
- `top_p` (float, optional): Nucleus sampling probability `(0, 1]`
- `frequency_penalty` (float, optional): Frequency penalty `[-2, 2]`
- `presence_penalty` (float, optional): Presence penalty `[-2, 2]`
- `response_format` (ResponseFormat, optional): Output format — `ResponseFormat(type="text")`, `ResponseFormat(type="json_object")`, or `ResponseFormat(type="json_schema", json_schema=...)`
- `stop` (List[str], optional): Up to 4 stop sequences
- `seed` (int, optional): Integer seed for deterministic sampling
- `reasoning_effort` (str, optional): Chain-of-thought effort: `"low"`, `"medium"`, or `"high"`
- `logprobs` (bool, optional): Whether to return log probabilities
- `top_logprobs` (int, optional): Number of most-likely tokens `[0, 20]`. Requires `logprobs=True`
- `logit_bias` (Dict[str, float], optional): Token ID to bias value mapping
- `tools` (List[Tool | dict], optional): Tool/function definitions available to the model
- `tool_choice` (str | dict, optional): `"none"`, `"auto"`, `"required"`, or a dict specifying a particular tool
- `user` (str, optional): Opaque end-user identifier
- `timeout` (float, optional): Per-request timeout in seconds (default 60s for chat)

**Returns:**

- `ChatCompletion` when `stream=False`
- `Stream` (iterable of `StreamEvent`) when `stream=True`

**`ChatCompletion` Properties:**

- `id` - Unique completion ID
- `model` - Model used for generation
- `choices` - List of `Choice` objects
- `text` - Convenience accessor for the first choice's content
- `usage` - `Usage` object (prompt_tokens, completion_tokens, total_tokens, num_search_queries, reasoning_tokens)
- `search_results` - List of `ChatSearchResult` (when `web_search="on"`)
- `citations` - Citation string referencing search results
- `warning` - Optional warning message

**`StreamEvent` Properties:**

- `type` - Event type: `"search_done"`, `"content"`, `"finish"`, `"usage"`, `"error"`
- `choices` - List of `StreamChoice` (with `delta.content` for incremental text)
- `search_results` - Web search results (on `search_done` event)
- `usage` - Token usage (on `usage` event)
- `citations` - Citation string (on `search_done` event)
- `error` - `StreamError` with `message` and `code` (on `error` event)

### Embedding API

#### `client.embedding.create()`

Create text embedding vectors.

**Parameters:**

- `input` (str | List[str], required): Input text or list of texts
- `model` (str, optional): Model name, options:
  - `"octen-embedding-0.6b"` - Lightweight model
  - `"octen-embedding-4b"` - Balanced performance
  - `"octen-embedding-8b"` - Highest quality
- `dimension` (int, optional): Vector dimension
- `input_type` (str, optional): Input type, options: `"query"` or `"document"`
- `truncation` (bool, optional): Whether to truncate long inputs, default True
- `timeout` (float, optional): Request timeout in seconds

**Returns:** `EmbeddingResponse` object

**Response Methods:**

- `get_embeddings()` - Get all vectors
- `get_first_embedding()` - Get first vector (for single input)

**Convenience Methods:**

- `embed_query(text)` - Embed a single query text
- `embed_documents(texts)` - Batch embed document texts

## 🔧 Async Support

```python
import asyncio
from octen import AsyncOcten, ChatMessage

async def main():
    async with AsyncOcten(api_key="your-api-key") as client:
        # Concurrent chat requests
        task1 = client.chat.create(
            model="openai/gpt-5.4",
            messages=[ChatMessage(role="user", content="Explain deep learning")],
            web_search="off"
        )
        task2 = client.chat.create(
            model="anthropic/claude-sonnet-4.6",
            messages=[ChatMessage(role="user", content="Explain reinforcement learning")],
            web_search="off"
        )
        r1, r2 = await asyncio.gather(task1, task2)
        print(r1.text)
        print(r2.text)

        # Async streaming
        stream = await client.chat.create(
            model="openai/gpt-5.4",
            messages=[ChatMessage(role="user", content="Hello!")],
            stream=True
        )
        async for event in stream:
            if event.type == "content" and event.choices:
                print(event.choices[0].delta.content or "", end="", flush=True)

        # Search and embeddings also work async
        results = await client.search.search(query="AI")
        embedding = await client.embedding.create(input=["Hello"], model="octen-embedding-4b")

asyncio.run(main())
```

## ⚠️ Error Handling

```python
from octen import (
    Octen,
    ChatMessage,
    OctenAPIError,
    OctenTimeoutError,
    OctenConnectionError,
    OctenRateLimitError,
    OctenAuthenticationError,
    OctenStreamError,
)

with Octen(api_key="your-api-key") as client:
    try:
        response = client.chat.create(
            model="openai/gpt-5.4",
            messages=[ChatMessage(role="user", content="Hello")]
        )
    except OctenAuthenticationError:
        print("Invalid or missing API key")
    except OctenRateLimitError as e:
        print(f"Rate limited — retry after {e.retry_after}s")
    except OctenStreamError as e:
        print(f"Stream error: {e.message} (code {e.code})")
    except OctenTimeoutError as e:
        print(f"Request timed out after {e.timeout}s")
    except OctenAPIError as e:
        print(f"API error {e.status_code}: {e.message}")
```

## 🧪 Development

### Install Development Dependencies

```bash
# Install development version from source
pip install -e ".[dev]"
```

### Run Tests

```bash
pytest tests/
```

### Code Formatting

```bash
black octen/
ruff check octen/ --fix
```

### Type Checking

```bash
mypy octen/
```

## 📝 License

MIT License - See [LICENSE](LICENSE) file for details

## 🔗 Links

- [Official Website](https://octen.ai)
- [API Documentation](https://docs.octen.ai)

## 📧 Support

For questions or help, please:

- Check the [Documentation](https://docs.octen.ai)
- Email us at support@octen.ai
