livekit-plugins-anthropic 0.2.6__py3-none-any.whl → 0.2.8__py3-none-any.whl

livekit/plugins/anthropic/llm.py

@@ -24,8 +24,8 @@ from typing import (
     Awaitable,
     List,
     Literal,
-    Tuple,
     Union,
+    cast,
     get_args,
     get_origin,
 )
@@ -40,6 +40,10 @@ from livekit.agents import (
     utils,
 )
 from livekit.agents.llm import ToolChoice
+from livekit.agents.llm.function_context import (
+    _create_ai_function_info,
+    _is_optional_type,
+)
 from livekit.agents.types import DEFAULT_API_CONNECT_OPTIONS, APIConnectOptions
 
 import anthropic
@@ -402,8 +406,10 @@ def _build_anthropic_message(
 
         return a_msg
     elif msg.role == "tool":
+        if isinstance(msg.content, dict):
+            msg.content = json.dumps(msg.content)
         if not isinstance(msg.content, str):
-            logger.warning("tool message content is not a string")
+            logger.warning("tool message content is not a string or dict")
             return None
         if not msg.tool_call_id:
             return None
@@ -425,11 +431,36 @@ def _build_anthropic_message(
 def _build_anthropic_image_content(
     image: llm.ChatImage, cache_key: Any
 ) -> anthropic.types.ImageBlockParam:
-    if isinstance(image.image, str):  # image url
-        logger.warning(
-            "image url not supported by anthropic, skipping image '%s'", image.image
-        )
-    elif isinstance(image.image, rtc.VideoFrame):  # VideoFrame
+    if isinstance(image.image, str):  # image is a URL
+        if not image.image.startswith("data:"):
+            raise ValueError("LiveKit Anthropic Plugin: Image URLs must be data URLs")
+
+        try:
+            header, b64_data = image.image.split(",", 1)
+            media_type = header.split(";")[0].split(":")[1]
+
+            supported_types = {"image/jpeg", "image/png", "image/webp", "image/gif"}
+            if media_type not in supported_types:
+                raise ValueError(
+                    f"LiveKit Anthropic Plugin: Unsupported media type {media_type}. Must be jpeg, png, webp, or gif"
+                )
+
+            return {
+                "type": "image",
+                "source": {
+                    "type": "base64",
+                    "data": b64_data,
+                    "media_type": cast(
+                        Literal["image/jpeg", "image/png", "image/gif", "image/webp"],
+                        media_type,
+                    ),
+                },
+            }
+        except (ValueError, IndexError) as e:
+            raise ValueError(
+                f"LiveKit Anthropic Plugin: Invalid image data URL {str(e)}"
+            )
+    elif isinstance(image.image, rtc.VideoFrame):  # image is a VideoFrame
         if cache_key not in image._cache:
             # inside our internal implementation, we allow to put extra metadata to
             # each ChatImage (avoid to reencode each time we do a chatcompletion request)
@@ -438,7 +469,7 @@ def _build_anthropic_image_content(
                 opts.resize_options = utils.images.ResizeOptions(
                     width=image.inference_width,
                     height=image.inference_height,
-                    strategy="center_aspect_fit",
+                    strategy="scale_aspect_fit",
                 )
 
             encoded_data = utils.images.encode(image.image, opts)
@@ -453,65 +484,8 @@ def _build_anthropic_image_content(
             },
         }
 
-    raise ValueError(f"unknown image type {type(image.image)}")
-
-
-def _create_ai_function_info(
-    fnc_ctx: llm.function_context.FunctionContext,
-    tool_call_id: str,
-    fnc_name: str,
-    raw_arguments: str,  # JSON string
-) -> llm.function_context.FunctionCallInfo:
-    if fnc_name not in fnc_ctx.ai_functions:
-        raise ValueError(f"AI function {fnc_name} not found")
-
-    parsed_arguments: dict[str, Any] = {}
-    try:
-        if raw_arguments:  # ignore empty string
-            parsed_arguments = json.loads(raw_arguments)
-    except json.JSONDecodeError:
-        raise ValueError(
-            f"AI function {fnc_name} received invalid JSON arguments - {raw_arguments}"
-        )
-
-    fnc_info = fnc_ctx.ai_functions[fnc_name]
-
-    # Ensure all necessary arguments are present and of the correct type.
-    sanitized_arguments: dict[str, Any] = {}
-    for arg_info in fnc_info.arguments.values():
-        if arg_info.name not in parsed_arguments:
-            if arg_info.default is inspect.Parameter.empty:
-                raise ValueError(
-                    f"AI function {fnc_name} missing required argument {arg_info.name}"
-                )
-            continue
-
-        arg_value = parsed_arguments[arg_info.name]
-        if get_origin(arg_info.type) is not None:
-            if not isinstance(arg_value, list):
-                raise ValueError(
-                    f"AI function {fnc_name} argument {arg_info.name} should be a list"
-                )
-
-            inner_type = get_args(arg_info.type)[0]
-            sanitized_value = [
-                _sanitize_primitive(
-                    value=v, expected_type=inner_type, choices=arg_info.choices
-                )
-                for v in arg_value
-            ]
-        else:
-            sanitized_value = _sanitize_primitive(
-                value=arg_value, expected_type=arg_info.type, choices=arg_info.choices
-            )
-
-        sanitized_arguments[arg_info.name] = sanitized_value
-
-    return llm.function_context.FunctionCallInfo(
-        tool_call_id=tool_call_id,
-        raw_arguments=raw_arguments,
-        function_info=fnc_info,
-        arguments=sanitized_arguments,
+    raise ValueError(
+        "LiveKit Anthropic Plugin: ChatImage must be an rtc.VideoFrame or a data URL"
     )
 
 
@@ -538,8 +512,10 @@ def _build_function_description(
         if arg_info.description:
             p["description"] = arg_info.description
 
-        if get_origin(arg_info.type) is list:
-            inner_type = get_args(arg_info.type)[0]
+        is_optional, inner_th = _is_optional_type(arg_info.type)
+
+        if get_origin(inner_th) is list:
+            inner_type = get_args(inner_th)[0]
             p["type"] = "array"
             p["items"] = {}
             p["items"]["type"] = type2str(inner_type)
@@ -547,7 +523,7 @@ def _build_function_description(
             if arg_info.choices:
                 p["items"]["enum"] = arg_info.choices
         else:
-            p["type"] = type2str(arg_info.type)
+            p["type"] = type2str(inner_th)
             if arg_info.choices:
                 p["enum"] = arg_info.choices
 
@@ -563,31 +539,3 @@ def _build_function_description(
         "description": fnc_info.description,
         "input_schema": input_schema,
     }
-
-
-def _sanitize_primitive(
-    *, value: Any, expected_type: type, choices: Tuple[Any] | None
-) -> Any:
-    if expected_type is str:
-        if not isinstance(value, str):
-            raise ValueError(f"expected str, got {type(value)}")
-    elif expected_type in (int, float):
-        if not isinstance(value, (int, float)):
-            raise ValueError(f"expected number, got {type(value)}")
-
-        if expected_type is int:
-            if value % 1 != 0:
-                raise ValueError("expected int, got float")
-
-            value = int(value)
-        elif expected_type is float:
-            value = float(value)
-
-    elif expected_type is bool:
-        if not isinstance(value, bool):
-            raise ValueError(f"expected bool, got {type(value)}")
-
-    if choices and value not in choices:
-        raise ValueError(f"invalid value {value}, not in {choices}")
-
-    return value

livekit/plugins/anthropic/version.py

@@ -12,4 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-__version__ = "0.2.6"
+__version__ = "0.2.8"

{livekit_plugins_anthropic-0.2.6.dist-info → livekit_plugins_anthropic-0.2.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: livekit-plugins-anthropic
-Version: 0.2.6
+Version: 0.2.8
 Summary: Agent Framework plugin for services from Anthropic
 Home-page: https://github.com/livekit/agents
 License: Apache-2.0
@@ -19,7 +19,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3 :: Only
 Requires-Python: >=3.9.0
 Description-Content-Type: text/markdown
-Requires-Dist: livekit-agents>=0.11
+Requires-Dist: livekit-agents>=0.12.3
 Requires-Dist: anthropic>=0.34
 
 # LiveKit Plugins Anthropic

livekit_plugins_anthropic-0.2.8.dist-info/RECORD

@@ -0,0 +1,10 @@
+livekit/plugins/anthropic/__init__.py,sha256=1WCyNEaR6qBsX54qJQM0SeY-QHIucww16PLXcSnMqRo,1175
+livekit/plugins/anthropic/llm.py,sha256=e65Z_YchNHCXN2F1kKb8lczfSY1_Ak8Y_94nT12pZGI,18975
+livekit/plugins/anthropic/log.py,sha256=fG1pYSY88AnT738gZrmzF9FO4l4BdGENj3VKHMQB3Yo,72
+livekit/plugins/anthropic/models.py,sha256=wyTr2nl6SL4ylN6s4mHJcqtmgV2mjJysZo89FknWdhI,213
+livekit/plugins/anthropic/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+livekit/plugins/anthropic/version.py,sha256=711Prlpzg5p2xCBjcE2dctFtzW_seKcOBI5-dvNUVK4,600
+livekit_plugins_anthropic-0.2.8.dist-info/METADATA,sha256=xG06Y2Xf9RjmmBqJ314_6ZmFlEM0ZCo00K82wJHEHHo,1265
+livekit_plugins_anthropic-0.2.8.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+livekit_plugins_anthropic-0.2.8.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
+livekit_plugins_anthropic-0.2.8.dist-info/RECORD,,

livekit_plugins_anthropic-0.2.6.dist-info/RECORD

@@ -1,10 +0,0 @@
-livekit/plugins/anthropic/__init__.py,sha256=1WCyNEaR6qBsX54qJQM0SeY-QHIucww16PLXcSnMqRo,1175
-livekit/plugins/anthropic/llm.py,sha256=Qz8POfM4A4Z-MuwcNoD9uJt8lpNi0SvatJvNtqls5p4,20650
-livekit/plugins/anthropic/log.py,sha256=fG1pYSY88AnT738gZrmzF9FO4l4BdGENj3VKHMQB3Yo,72
-livekit/plugins/anthropic/models.py,sha256=wyTr2nl6SL4ylN6s4mHJcqtmgV2mjJysZo89FknWdhI,213
-livekit/plugins/anthropic/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-livekit/plugins/anthropic/version.py,sha256=qiU8NmY5thT_I47ndTlA5TMw4pNi5-r9qqSsLmz-9ps,600
-livekit_plugins_anthropic-0.2.6.dist-info/METADATA,sha256=7v72icRCffOPETvp3GbhrJXKRO2yZG7O1eU3OpCeGnM,1263
-livekit_plugins_anthropic-0.2.6.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-livekit_plugins_anthropic-0.2.6.dist-info/top_level.txt,sha256=OoDok3xUmXbZRvOrfvvXB-Juu4DX79dlq188E19YHoo,8
-livekit_plugins_anthropic-0.2.6.dist-info/RECORD,,