fix: resolve windows tempfile locking, tool schema generation, and add client improvements

CONTRIBUTING.md

@@ -0,0 +1,117 @@
+# Contributing to Ollama Python
+
+Thank you for your interest in contributing to the official Ollama Python library! We welcome bug reports, bug fixes, feature proposals, documentation improvements, and example additions.
+
+---
+
+## Prerequisites
+
+- **Python 3.8+** (Python 3.10+ recommended for development)
+- **[uv](https://docs.astral.sh/uv/)** or **[hatch](https://hatch.pypa.io/)**
+- A running instance of **[Ollama](https://ollama.com/download)** (optional for running unit/mock tests, required for real model execution)
+
+---
+
+## Development Setup
+
+1. **Fork and clone the repository:**
+
+   ```bash
+   git clone https://github.com/<your-username>/ollama-python.git
+   cd ollama-python
+   ```
+
+2. **Set up virtual environment and install dependencies:**
+
+   Using `uv`:
+   ```bash
+   uv sync --all-extras --all-groups
+   ```
+
+   Using `hatch`:
+   ```bash
+   hatch env create
+   ```
+
+---
+
+## Code Style and Quality
+
+We use [Ruff](https://docs.astral.sh/ruff/) to format and lint our codebase.
+
+- **Indentation**: 2 spaces (`indent-width = 2`, `indent-style = 'space'`)
+- **Quotes**: Single quotes (`quote-style = 'single'`)
+- **Line length**: 320 characters max
+
+### Formatting & Linting Commands
+
+Check formatting and linting:
+```bash
+uv run ruff format --check .
+uv run ruff check .
+```
+
+Automatically fix format and lint issues:
+```bash
+uv run ruff format .
+uv run ruff check --fix .
+```
+
+Or using `hatch`:
+```bash
+hatch fmt --check
+```
+
+---
+
+## Running Tests
+
+All unit tests use mock HTTP servers with `pytest-httpserver` and doctests across module docstrings.
+
+Run the test suite with `uv`:
+```bash
+uv run pytest
+```
+
+Run tests across environments with `hatch`:
+```bash
+hatch test
+```
+
+Run a specific test file:
+```bash
+uv run pytest tests/test_client.py
+uv run pytest tests/test_utils.py
+uv run pytest tests/test_type_serialization.py
+```
+
+---
+
+## Updating Lockfile and Requirements
+
+If dependencies in `pyproject.toml` are modified, make sure the lockfile and exported requirements remain in sync:
+
+```bash
+uv lock
+uv export > requirements.txt
+```
+
+Verify that requirements match:
+```bash
+git diff --exit-code requirements.txt
+```
+
+---
+
+## Pull Request Guidelines
+
+1. **Branch naming**: Use descriptive branch names (e.g. `fix/named-tempfile-windows`, `feat/client-api-key-arg`).
+2. **Commit messages**: Follow concise conventional commit style:
+   - `fix: ...` for bug fixes
+   - `feat: ...` for new features
+   - `docs: ...` for documentation updates
+   - `test: ...` for test additions
+   - `refactor: ...` for code refactoring
+3. **Tests**: Add unit tests for any new functionality or bug fixes in `tests/`.
+4. **Documentation & Examples**: Update `README.md` and `examples/README.md` if changing public APIs or adding new features.
+5. **No regressions**: Ensure formatting (`ruff format`), linting (`ruff check`), and all tests pass before submitting.

README.md

@@ -113,7 +113,8 @@ curl https://ollama.com/api/tags
 import os
 from ollama import Client
 
-client = Client(host='https://ollama.com', headers={'Authorization': 'Bearer ' + os.environ.get('OLLAMA_API_KEY')})
+# You can pass api_key directly or rely on the OLLAMA_API_KEY environment variable
+client = Client(host='https://ollama.com', api_key=os.environ.get('OLLAMA_API_KEY'))
 
 messages = [
   {
@@ -129,12 +130,16 @@ for part in client.chat('gpt-oss:120b', messages=messages, stream=True):
 ## Custom client
 A custom client can be created by instantiating `Client` or `AsyncClient` from `ollama`.
 
-All extra keyword arguments are passed into the [`httpx.Client`](https://www.python-httpx.org/api/#client).
+Clients accept `host`, `api_key`, `headers`, `timeout`, and any keyword arguments supported by [`httpx.Client`](https://www.python-httpx.org/api/#client).
 
 ```python
 from ollama import Client
 
-client = Client(host='http://localhost:11434', headers={'x-some-header': 'some-value'})
+client = Client(
+  host='http://localhost:11434',
+  api_key='your_api_key',
+  headers={'x-some-header': 'some-value'},
+)
 response = client.chat(
   model='gemma4',
   messages=[
@@ -146,6 +151,13 @@ response = client.chat(
 )
 ```
 
+Clients can also be used as context managers to ensure immediate resource and socket cleanup:
+
+```python
+with Client() as client:
+  response = client.chat(model='gemma4', messages=[{'role': 'user', 'content': 'Why is the sky blue?'}])
+```
+
 ## Async client
 
 The `AsyncClient` class is used to make asynchronous requests. It can be configured with the same fields as the `Client` class.

examples/README.md

@@ -18,12 +18,14 @@ See [ollama/docs/api.md](https://github.com/ollama/ollama/blob/main/docs/api.md)
 - [async-chat.py](async-chat.py)
 - [chat-stream.py](chat-stream.py) - Streamed outputs
 - [chat-with-history.py](chat-with-history.py) - Chat with model and maintain history of the conversation
+- [chat-logprobs.py](chat-logprobs.py) - Return log probabilities for chat completions
 
 ### Generate - Generate text with a model
 
 - [generate.py](generate.py)
 - [async-generate.py](async-generate.py)
 - [generate-stream.py](generate-stream.py) - Streamed outputs
+- [generate-logprobs.py](generate-logprobs.py) - Return log probabilities for generation
 - [fill-in-middle.py](fill-in-middle.py) - Given a prefix and suffix, fill in the middle
 
 ### Tools/Function Calling - Call a function with a model
@@ -46,6 +48,7 @@ export OLLAMA_API_KEY="your_api_key_here"
 ```
 
 - [web-search.py](web-search.py)
+- [web-fetch.py](web-fetch.py) - Fetch and extract content from a web page
 - [web-search-gpt-oss.py](web-search-gpt-oss.py) - Using browser research tools with gpt-oss
 
 #### MCP server
@@ -131,3 +134,7 @@ Requirement: `pip install tqdm`
 ### System One - Score typed questions
 
 - [systemone.py](systemone.py) - Choice, boolean probability, and rubric score questions
+
+### Client Lifecycle & Context Managers - Manage connections and resources
+
+- [context-manager.py](context-manager.py) - Use sync and async clients with context managers for automatic resource cleanup

examples/context-manager.py

@@ -0,0 +1,44 @@
+# /// script
+# requires-python = ">=3.8"
+# dependencies = [
+#     "ollama",
+# ]
+# ///
+"""
+Example demonstrating how to use Client and AsyncClient as context managers
+to ensure proper cleanup and release of underlying HTTP connections.
+"""
+
+import asyncio
+
+from ollama import AsyncClient, Client
+
+
+def sync_example():
+  print('=== Synchronous Client Context Manager ===')
+  # Entering the context manager initializes the client;
+  # exiting cleanly closes all httpx connection pools.
+  with Client() as client:
+    models = client.list()
+    print(f'Discovered {len(models.models)} installed models:')
+    for model in models.models[:3]:
+      print(f' - {model.model} (modified: {model.modified_at})')
+
+
+async def async_example():
+  print('\n=== Asynchronous Client Context Manager ===')
+  async with AsyncClient() as client:
+    response = await client.generate(
+      model='gemma4',
+      prompt='Why is the sky blue in one sentence?',
+    )
+    print(f'Response: {response.response}')
+
+
+if __name__ == '__main__':
+  sync_example()
+  # Run async example if an async event loop is available
+  try:
+    asyncio.run(async_example())
+  except Exception as e:
+    print(f'Async example skipped or encountered error: {e}')

examples/web-fetch.py

@@ -0,0 +1,37 @@
+# /// script
+# requires-python = ">=3.11"
+# dependencies = [
+#     "rich",
+#     "ollama",
+# ]
+# ///
+import os
+import sys
+
+from rich import print
+
+from ollama import Client
+
+api_key = os.getenv('OLLAMA_API_KEY')
+if not api_key:
+  print('[bold red]Error:[/bold red] OLLAMA_API_KEY environment variable is required.')
+  print('Get your API key at https://ollama.com/settings/keys')
+  sys.exit(1)
+
+# You can pass api_key directly or rely on OLLAMA_API_KEY
+client = Client(api_key=api_key)
+
+url = 'https://ollama.com'
+print(f'Fetching content from: [bold cyan]{url}[/bold cyan]\n')
+
+# Fetch using the client
+response = client.web_fetch(url)
+
+print(f'Title: [bold green]{response.title}[/bold green]')
+if response.links:
+  print(f'Found {len(response.links)} links on the page:')
+  for link in response.links[:5]:
+    print(f'  - {link}')
+
+print('\nContent Preview:')
+print(response.content[:500] if response.content else 'No content returned')

ollama/__init__.py

@@ -63,6 +63,22 @@
   'Tool',
   'WebFetchResponse',
   'WebSearchResponse',
+  'chat',
+  'copy',
+  'create',
+  'create_blob',
+  'delete',
+  'embed',
+  'embeddings',
+  'generate',
+  'list',
+  'ps',
+  'pull',
+  'push',
+  'show',
+  'systemone',
+  'web_fetch',
+  'web_search',
 ]
 
 _client = Client()
@@ -74,6 +90,7 @@
 pull = _client.pull
 push = _client.push
 create = _client.create
+create_blob = _client.create_blob
 delete = _client.delete
 list = _client.list
 copy = _client.copy