add comments

docs/resources/llama-stack-spec.html

@@ -2446,10 +2446,12 @@
                         "type": "object",
                         "properties": {
                             "url": {
-                                "$ref": "#/components/schemas/URL"
+                                "$ref": "#/components/schemas/URL",
+                                "description": "A URL of the image or data URL in the format of data:image/{type};base64,{data}. Note that URL could have length limits."
                             },
                             "data": {
-                                "type": "string"
+                                "type": "string",
+                                "description": "base64 encoded image data as string"
                             }
                         },
                         "additionalProperties": false,

docs/resources/llama-stack-spec.yaml

@@ -1473,8 +1473,12 @@ components:
           properties:
             url:
               $ref: '#/components/schemas/URL'
+              description: >-
+                A URL of the image or data URL in the format of data:image/{type};base64,{data}.
+                Note that URL could have length limits.
             data:
               type: string
+              description: base64 encoded image data as string
           additionalProperties: false
           description: >-
             Image as a base64 encoded string or an URL

llama_stack/apis/common/content_types.py

@@ -19,8 +19,16 @@ class URL(BaseModel):
 
 
 class _URLOrData(BaseModel):
+    """
+    A URL or a base64 encoded string
+
+    :param url: A URL of the image or data URL in the format of data:image/{type};base64,{data}. Note that URL could have length limits.
+    :param data: base64 encoded image data as string
+    """
+
     url: Optional[URL] = None
     # data is a base64 encoded string
+    # TODO: annotate with contentEncoding="base64" in OpenAPI schema
     data: Optional[str] = None
 
     @model_validator(mode="before")

tests/client-sdk/inference/test_inference.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 import base64
-import os
+import pathlib
 
 import pytest
 from pydantic import BaseModel
@@ -48,14 +48,31 @@ def get_weather_tool_definition():
     }
 
 
+# @pytest.fixture
+# def base64_image_url():
+#     image_path = os.path.join(os.path.dirname(__file__), "dog.png")
+#     with open(image_path, "rb") as image_file:
+#         # Convert the image to base64
+#         base64_string = base64.b64encode(image_file.read()).decode("utf-8")
+#         base64_url = f"data:image/png;base64,{base64_string}"
+#         return base64_url
+
+
 @pytest.fixture
-def base64_image_url():
-    image_path = os.path.join(os.path.dirname(__file__), "dog.png")
-    with open(image_path, "rb") as image_file:
-        # Convert the image to base64
-        base64_string = base64.b64encode(image_file.read()).decode("utf-8")
-        base64_url = f"data:image/png;base64,{base64_string}"
-        return base64_url
+def image_path():
+    return pathlib.Path(__file__).parent / "dog.png"
+
+
+@pytest.fixture
+def base64_image_data(image_path):
+    # Convert the image to base64
+    return base64.b64encode(image_path.read_bytes()).decode("utf-8")
+
+
+@pytest.fixture
+def base64_image_url(base64_image_data, image_path):
+    # suffix includes the ., so we remove it
+    return f"data:image/{image_path.suffix[1:]};base64,{base64_image_data}"
 
 
 def test_text_completion_non_streaming(llama_stack_client, text_model_id):
@@ -353,25 +370,30 @@ def test_image_chat_completion_streaming(llama_stack_client, vision_model_id):
     assert any(expected in streamed_content for expected in {"dog", "puppy", "pup"})
 
 
-def test_image_chat_completion_base64_url(
-    llama_stack_client, vision_model_id, base64_image_url
+@pytest.mark.parametrize("type_", ["url", "data"])
+def test_image_chat_completion_base64(
+    llama_stack_client, vision_model_id, base64_image_data, base64_image_url, type_
 ):
-    message = {
-        "role": "user",
-        "content": [
-            {
-                "type": "image",
-                "image": {
-                    "url": {
-                        "uri": base64_image_url,
-                    },
+    image_spec = {
+        "url": {
+            "type": "image",
+            "image": {
+                "url": {
+                    "uri": base64_image_url,
                 },
             },
-            {
-                "type": "text",
-                "text": "Describe what is in this image.",
+        },
+        "data": {
+            "type": "image",
+            "image": {
+                "data": base64_image_data,
             },
-        ],
+        },
+    }[type_]
+
+    message = {
+        "role": "user",
+        "content": [image_spec],
     }
     response = llama_stack_client.inference.chat_completion(
         model_id=vision_model_id,