fix(factory.py): flexible handling of image file type in b64 encode

Fixes https://github.com/BerriAI/litellm/issues/4589

litellm/llms/prompt_templates/factory.py

@@ -704,22 +704,19 @@ def convert_to_anthropic_image_obj(openai_image_url: str) -> GenericImageParsing
     try:
         if openai_image_url.startswith("http"):
             openai_image_url = convert_url_to_base64(url=openai_image_url)
-        # Extract the base64 image data
+        # Extract the media type and base64 data
-        base64_data = openai_image_url.split("data:image/")[1].split(";base64,")[1]
+        media_type, base64_data = openai_image_url.split("data:")[1].split(";base64,")
-
-        # Infer image format from the URL
-        image_format = openai_image_url.split("data:image/")[1].split(";base64,")[0]
 
         return GenericImageParsingChunk(
             type="base64",
-            media_type=f"image/{image_format}",
+            media_type=media_type,
             data=base64_data,
         )
     except Exception as e:
         if "Error: Unable to fetch image from URL" in str(e):
             raise e
         raise Exception(
-            """Image url not in expected format. Example Expected input - "image_url": "data:image/jpeg;base64,{base64_image}". Supported formats - ['image/jpeg', 'image/png', 'image/gif', 'image/webp'] """
+            """Image url not in expected format. Example Expected input - "image_url": "data:image/jpeg;base64,{base64_image}". Supported formats - ['image/jpeg', 'image/png', 'image/gif', 'image/webp']."""
         )
 
 

litellm/tests/test_completion.py

@@ -848,6 +848,45 @@ def test_completion_function_plus_image(model):
     print(response)
 
 
+@pytest.mark.parametrize(
+    "model", ["gemini/gemini-1.5-pro"]  # "claude-3-sonnet-20240229",
+)
+def test_completion_function_plus_pdf(model):
+    litellm.set_verbose = True
+    try:
+        import base64
+
+        import requests
+
+        # URL of the file
+        url = "https://storage.googleapis.com/cloud-samples-data/generative-ai/pdf/2403.05530.pdf"
+
+        # Download the file
+        response = requests.get(url)
+        file_data = response.content
+
+        encoded_file = base64.b64encode(file_data).decode("utf-8")
+
+        image_content = [
+            {"type": "text", "text": "What's this file about?"},
+            {
+                "type": "image_url",
+                "image_url": {"url": f"data:application/pdf;base64,{encoded_file}"},
+            },
+        ]
+        image_message = {"role": "user", "content": image_content}
+
+        response = completion(
+            model=model,
+            messages=[image_message],
+            stream=False,
+        )
+
+        print(response)
+    except litellm.InternalServerError:
+        pass
+
+
 @pytest.mark.parametrize(
     "provider",
     ["azure", "azure_ai"],

litellm/tests/test_prompt_factory.py

@@ -15,6 +15,7 @@ from litellm.llms.prompt_templates.factory import (
     anthropic_messages_pt,
     anthropic_pt,
     claude_2_1_pt,
+    convert_to_anthropic_image_obj,
     convert_url_to_base64,
     llama_2_chat_pt,
     prompt_factory,
@@ -163,3 +164,16 @@ def test_convert_url_to_img():
     )
 
     assert "image/jpeg" in response_url
+
+
+@pytest.mark.parametrize(
+    "url, expected_media_type",
+    [
+        ("data:image/jpeg;base64,1234", "image/jpeg"),
+        ("data:application/pdf;base64,1234", "application/pdf"),
+    ],
+)
+def test_base64_image_input(url, expected_media_type):
+    response = convert_to_anthropic_image_obj(openai_image_url=url)
+
+    assert response["media_type"] == expected_media_type