Merge branch 'main' into feature/sambanovacloud_integration

docs/examples/bulk_classification.md

@@ -528,6 +528,4 @@ async def get_tags(text: List[str], tags: List[Tag]) -> List[Tag]:
 tag_results = asyncio.run(get_tags(text, tags))
 for tag in tag_results:
     print(tag)
-    #> id=0 name='personal'
     #> id=1 name='phone'
-```

docs/integrations/anthropic.md

@@ -15,6 +15,8 @@ pip install "instructor[anthropic]"
 
 Once we've done so, getting started is as simple as using our `from_anthropic` method to patch the client up.
 
+### Basic Usage
+
 ```python
 from pydantic import BaseModel
 from typing import List
@@ -71,6 +73,19 @@ print(user_response.model_dump_json(indent=2))
 """
 ```
 
+### Beta API Features
+
+Anthropic provides beta features through their beta API endpoint. To use these features, enable the beta parameter when creating the client:
+
+```python
+client = instructor.from_anthropic(
+    anthropic.Anthropic(),
+    beta=True  # Enable beta API features
+)
+```
+
+This allows you to access beta features like PDF support and other experimental capabilities. For more information about available beta features, refer to [Anthropic's documentation](https://docs.anthropic.com/claude/docs/beta-features).
+
 ## Streaming Support
 
 Instructor has two main ways that you can use to stream responses out

instructor/client_anthropic.py

@@ -13,6 +13,7 @@ def from_anthropic(
     ),
     mode: instructor.Mode = instructor.Mode.ANTHROPIC_TOOLS,
     enable_prompt_caching: bool = False,
+    beta: bool = False,
     **kwargs: Any,
 ) -> instructor.Instructor: ...
 
@@ -26,6 +27,7 @@ def from_anthropic(
     ),
     mode: instructor.Mode = instructor.Mode.ANTHROPIC_TOOLS,
     enable_prompt_caching: bool = False,
+    beta: bool = False,
     **kwargs: Any,
 ) -> instructor.AsyncInstructor: ...
 
@@ -41,8 +43,25 @@ def from_anthropic(
     ),
     mode: instructor.Mode = instructor.Mode.ANTHROPIC_TOOLS,
     enable_prompt_caching: bool = False,
+    beta: bool = False,
     **kwargs: Any,
 ) -> instructor.Instructor | instructor.AsyncInstructor:
+    """Create an Instructor instance from an Anthropic client.
+
+    Args:
+        client: An instance of Anthropic client (sync or async)
+        mode: The mode to use for the client (ANTHROPIC_JSON or ANTHROPIC_TOOLS)
+        enable_prompt_caching: Whether to enable prompt caching (requires Anthropic or AsyncAnthropic client)
+        beta: Whether to use beta API features (uses client.beta.messages.create)
+        **kwargs: Additional keyword arguments to pass to the Instructor constructor
+
+    Returns:
+        An Instructor instance (sync or async depending on the client type)
+
+    Raises:
+        TypeError: If enable_prompt_caching is True and client is not Anthropic or AsyncAnthropic
+        AssertionError: If mode is not ANTHROPIC_JSON or ANTHROPIC_TOOLS
+    """
     assert (
         mode
         in {
@@ -70,6 +89,8 @@ def from_anthropic(
             raise TypeError(
                 "Client must be an instance of {anthropic.Anthropic, anthropic.AsyncAnthropic} to enable prompt caching"
             )
+    elif beta:
+        create = client.beta.messages.create
     else:
         create = client.messages.create
 

pyproject.toml

@@ -18,7 +18,7 @@ rich = "^13.7.0"
 aiohttp = "^3.9.1"
 tenacity = ">=9.0.0,<10.0.0"
 pydantic-core = "^2.18.0"
-jiter = ">=0.6.1,<0.7"
+jiter = ">=0.6.1,<0.9"
 jinja2 = "^3.1.4"
 requests = "^2.32.3"
 
@@ -30,9 +30,9 @@ pandas = { version = "^2.2.0", optional = true }
 tabulate = { version = "^0.9.0", optional = true }
 pydantic_extra_types = { version = "^2.6.0", optional = true }
 litellm = { version = "^1.35.31", optional = true }
-anthropic = { version = ">=0.36.2,<0.38.0", optional = true }
+anthropic = { version = ">=0.36.2,<0.41.0", optional = true }
 xmltodict = { version = ">=0.13,<0.15", optional = true }
-groq = { version = ">=0.4.2,<0.12.0", optional = true }
+groq = { version = ">=0.4.2,<0.14.0", optional = true }
 cohere = { version = "^5.1.8", optional = true }
 mistralai = { version = "^1.0.3", optional = true }
 google-generativeai = { version = "^0.8.2", optional = true }
@@ -81,16 +81,16 @@ black = "^24.10.0"
 [tool.poetry.group.docs.dependencies]
 mkdocs = "^1.4.3"
 mkdocs-material = { extras = ["imaging"], version = "^9.5.9" }
-mkdocstrings = "^0.26.1"
+mkdocstrings = ">=0.26.1,<0.28.0"
 mkdocstrings-python = "^1.11.1"
-pytest-examples = "^0.0.13"
+pytest-examples = ">=0.0.13,<0.0.16"
 mkdocs-jupyter = ">=0.24.6,<0.26.0"
 mkdocs-rss-plugin = "^1.12.0"
 mkdocs-minify-plugin = "^0.8.0"
 mkdocs-redirects = "^1.2.1"
 
 [tool.poetry.group.anthropic.dependencies]
-anthropic = ">=0.36.2,<0.38.0"
+anthropic = ">=0.36.2,<0.41.0"
 
 [tool.poetry.group.test-docs.dependencies]
 fastapi = ">=0.109.2,<0.116.0"
@@ -100,17 +100,17 @@ pandas = "^2.2.0"
 tabulate = "^0.9.0"
 pydantic_extra_types = "^2.6.0"
 litellm = "^1.35.31"
-anthropic = ">=0.36.2,<0.38.0"
+anthropic = ">=0.36.2,<0.41.0"
 xmltodict = ">=0.13,<0.15"
-groq = ">=0.4.2,<0.12.0"
+groq = ">=0.4.2,<0.14.0"
 phonenumbers = "^8.13.33"
 cohere = "^5.1.8"
 mistralai = "^1.0.3"
 cerebras_cloud_sdk = "^1.5.0"
 fireworks-ai = "^0.15.4"
 graphviz = "^0.20.3"
 sqlmodel = "^0.0.22"
-trafilatura = "^1.12.2"
+trafilatura = ">=1.12.2,<3.0.0"
 pydub = "^0.25.1"
 datasets = "^3.0.1"
 writer-sdk = "^1.2.0"

requirements-examples.txt

@@ -3,9 +3,9 @@ pydantic
 docstring-parser
 rich
 aiohttp
-ruff==0.7.1
+ruff==0.8.1
 pre-commit==4.0.1
-pyright==1.1.386
+pyright==1.1.390
 typer
 cohere
 datasets

requirements.txt

@@ -3,8 +3,8 @@ pydantic
 docstring-parser
 rich
 aiohttp
-ruff==0.7.1
+ruff==0.8.1
 pre-commit==4.0.1
-pyright==1.1.386
+pyright==1.1.390
 typer
 cohere