feat: testing for pdf usage (#420)

## Description
<!-- What does this PR do? -->
Integration testing for pdf handling. Turns out that only a subset of
providers support pdf recognition.

We also need to modify the type definition of `annotations` to allow for
more permissive content.

I opted to not put this into the provider compatibility matrix doc page,
since it seemed like a niche feature and would risk cluttering an
already busy page.

## PR Type

<!-- Delete the types that don't apply --!>

🆕 New Feature

## Relevant issues

<!-- e.g. "Fixes #123" -->

## Checklist
- [x] I have added unit tests that prove my fix/feature works
- [x] New and existing tests pass locally
- [x] Documentation was updated where necessary
- [x] I have read and followed the [contribution
guidelines](https://github.com/mozilla-ai/any-llm/blob/main/CONTRIBUTING.md)```

Author: njbrake

src/any_llm/provider.py

@@ -114,6 +114,9 @@ class Provider(ABC):
     SUPPORTS_COMPLETION_IMAGE: bool
     """Image Support for Completion API"""
 
+    SUPPORTS_COMPLETION_PDF: bool
+    """PDF Support for Completion API"""
+
     SUPPORTS_EMBEDDING: bool
     """OpenAI Embedding API"""
 
@@ -205,6 +208,7 @@ def get_provider_metadata(cls) -> ProviderMetadata:
             reasoning=cls.SUPPORTS_COMPLETION_REASONING,
             completion=cls.SUPPORTS_COMPLETION,
             image=cls.SUPPORTS_COMPLETION_IMAGE,
+            pdf=cls.SUPPORTS_COMPLETION_PDF,
             embedding=cls.SUPPORTS_EMBEDDING,
             responses=cls.SUPPORTS_RESPONSES,
             list_models=cls.SUPPORTS_LIST_MODELS,

src/any_llm/providers/anthropic/anthropic.py

@@ -40,6 +40,7 @@ class AnthropicProvider(Provider):
     SUPPORTS_RESPONSES = False
     SUPPORTS_COMPLETION_REASONING = True
     SUPPORTS_COMPLETION_IMAGE = False  # Needs https://github.com/mozilla-ai/any-llm/issues/416
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/azure/azure.py

@@ -37,6 +37,7 @@ class AzureProvider(Provider):
     PROVIDER_DOCUMENTATION_URL = "https://azure.microsoft.com/en-us/products/ai-services/openai-service"
     SUPPORTS_COMPLETION_STREAMING = True
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = True
     SUPPORTS_COMPLETION_REASONING = False
     SUPPORTS_COMPLETION = True

src/any_llm/providers/bedrock/bedrock.py

@@ -41,6 +41,7 @@ class BedrockProvider(Provider):
     SUPPORTS_RESPONSES = False
     SUPPORTS_COMPLETION_REASONING = False
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = True
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/cerebras/cerebras.py

@@ -34,6 +34,7 @@ class CerebrasProvider(Provider):
     SUPPORTS_RESPONSES = False
     SUPPORTS_COMPLETION_REASONING = False
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/cohere/cohere.py

@@ -34,6 +34,7 @@ class CohereProvider(Provider):
     SUPPORTS_RESPONSES = False
     SUPPORTS_COMPLETION_REASONING = False
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/deepseek/deepseek.py

@@ -13,6 +13,7 @@ class DeepseekProvider(BaseOpenAIProvider):
     PROVIDER_DOCUMENTATION_URL = "https://platform.deepseek.com/"
 
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False  # DeepSeek doesn't host an embedding model
 
     async def acompletion(

src/any_llm/providers/fireworks/fireworks.py

@@ -18,6 +18,7 @@ class FireworksProvider(BaseOpenAIProvider):
     SUPPORTS_COMPLETION = True
     SUPPORTS_RESPONSES = True
     SUPPORTS_COMPLETION_REASONING = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/gemini/base.py

@@ -52,6 +52,7 @@ class GoogleProvider(Provider):
     SUPPORTS_RESPONSES = False
     SUPPORTS_COMPLETION_REASONING = True
     SUPPORTS_COMPLETION_IMAGE = False  # TODO: Add image support https://github.com/mozilla-ai/any-llm/issues/415
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = True
     SUPPORTS_LIST_MODELS = True
 

src/any_llm/providers/groq/groq.py

@@ -48,6 +48,7 @@ class GroqProvider(Provider):
     SUPPORTS_RESPONSES = True
     SUPPORTS_COMPLETION_REASONING = True
     SUPPORTS_COMPLETION_IMAGE = False
+    SUPPORTS_COMPLETION_PDF = False
     SUPPORTS_EMBEDDING = False
     SUPPORTS_LIST_MODELS = True