livekit-plugins-aws 0.1.1__tar.gz → 1.0.0.dev5__tar.gz

livekit_plugins_aws-1.0.0.dev5/.gitignore

@@ -0,0 +1,168 @@
+**/.vscode
+**/.DS_Store
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+.idea/
+
+node_modules
+
+credentials.json
+pyrightconfig.json

{livekit_plugins_aws-0.1.1 → livekit_plugins_aws-1.0.0.dev5}/PKG-INFO

@@ -1,38 +1,28 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: livekit-plugins-aws
-Version: 0.1.1
+Version: 1.0.0.dev5
 Summary: LiveKit Agents Plugin for services from AWS
-Home-page: https://github.com/livekit/agents
-License: Apache-2.0
 Project-URL: Documentation, https://docs.livekit.io
 Project-URL: Website, https://livekit.io/
 Project-URL: Source, https://github.com/livekit/agents
-Keywords: webrtc,realtime,audio,video,livekit,aws
+Author-email: LiveKit <support@livekit.io>
+License-Expression: Apache-2.0
+Keywords: audio,aws,livekit,realtime,video,webrtc
 Classifier: Intended Audience :: Developers
 Classifier: License :: OSI Approved :: Apache Software License
-Classifier: Topic :: Multimedia :: Sound/Audio
-Classifier: Topic :: Multimedia :: Video
-Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3 :: Only
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
-Classifier: Programming Language :: Python :: 3 :: Only
+Classifier: Topic :: Multimedia :: Sound/Audio
+Classifier: Topic :: Multimedia :: Video
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: >=3.9.0
-Description-Content-Type: text/markdown
-Requires-Dist: livekit-agents[codecs]<1.0.0,>=0.12.16
 Requires-Dist: aiobotocore==2.19.0
-Requires-Dist: boto3==1.36.3
 Requires-Dist: amazon-transcribe>=0.6.2
-Dynamic: classifier
-Dynamic: description
-Dynamic: description-content-type
-Dynamic: home-page
-Dynamic: keywords
-Dynamic: license
-Dynamic: project-url
-Dynamic: requires-dist
-Dynamic: requires-python
-Dynamic: summary
+Requires-Dist: boto3==1.36.3
+Requires-Dist: livekit-agents>=1.0.0.dev5
+Description-Content-Type: text/markdown
 
 # LiveKit Plugins AWS
 
@@ -50,4 +40,4 @@ pip install livekit-plugins-aws
 
 ## Pre-requisites
 
-You'll need to specify an AWS Access Key and a Deployment Region. They can be set as environment variables: `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY` and `AWS_DEFAULT_REGION`, respectively.
+You'll need to specify an AWS Access Key and a Deployment Region. They can be set as environment variables: `AWS_ACCESS_KEY_ID`, `AWS_SECRET_ACCESS_KEY` and `AWS_DEFAULT_REGION`, respectively.

livekit_plugins_aws-1.0.0.dev5/livekit/plugins/aws/llm.py

@@ -0,0 +1,270 @@
+# Copyright 2023 LiveKit, Inc.
+#
+
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from __future__ import annotations
+
+import asyncio
+import os
+from dataclasses import dataclass
+from typing import Any, Literal
+
+import boto3
+
+from livekit.agents import APIConnectionError, APIStatusError, llm
+from livekit.agents.llm import ChatContext, FunctionTool, FunctionToolCall, ToolChoice
+from livekit.agents.types import (
+    DEFAULT_API_CONNECT_OPTIONS,
+    NOT_GIVEN,
+    APIConnectOptions,
+    NotGivenOr,
+)
+from livekit.agents.utils import is_given
+
+from .log import logger
+from .utils import get_aws_credentials, to_chat_ctx, to_fnc_ctx
+
+TEXT_MODEL = Literal["anthropic.claude-3-5-sonnet-20241022-v2:0"]
+DEFAULT_REGION = "us-east-1"
+
+
+@dataclass
+class _LLMOptions:
+    model: str | TEXT_MODEL
+    temperature: NotGivenOr[float]
+    tool_choice: NotGivenOr[ToolChoice | Literal["auto", "required", "none"]]
+    max_output_tokens: NotGivenOr[int]
+    top_p: NotGivenOr[float]
+    additional_request_fields: NotGivenOr[dict[str, Any]]
+
+
+class LLM(llm.LLM):
+    def __init__(
+        self,
+        *,
+        model: NotGivenOr[str | TEXT_MODEL] = NOT_GIVEN,
+        api_key: NotGivenOr[str] = NOT_GIVEN,
+        api_secret: NotGivenOr[str] = NOT_GIVEN,
+        region: NotGivenOr[str] = NOT_GIVEN,
+        temperature: NotGivenOr[float] = NOT_GIVEN,
+        max_output_tokens: NotGivenOr[int] = NOT_GIVEN,
+        top_p: NotGivenOr[float] = NOT_GIVEN,
+        tool_choice: NotGivenOr[ToolChoice | Literal["auto", "required", "none"]] = NOT_GIVEN,
+        additional_request_fields: NotGivenOr[dict[str, Any]] = NOT_GIVEN,
+    ) -> None:
+        """
+        Create a new instance of AWS Bedrock LLM.
+
+        ``api_key``  and ``api_secret`` must be set to your AWS Access key id and secret access key, either using the argument or by setting the
+        ``AWS_ACCESS_KEY_ID`` and ``AWS_SECRET_ACCESS_KEY`` environmental variables.
+
+        See https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/bedrock-runtime/client/converse_stream.html for more details on the AWS Bedrock Runtime API.
+
+        Args:
+            model (TEXT_MODEL, optional): model or inference profile arn to use(https://docs.aws.amazon.com/bedrock/latest/userguide/inference-profiles-use.html). Defaults to 'anthropic.claude-3-5-sonnet-20240620-v1:0'.
+            api_key(str, optional): AWS access key id.
+            api_secret(str, optional): AWS secret access key
+            region (str, optional): The region to use for AWS API requests. Defaults value is "us-east-1".
+            temperature (float, optional): Sampling temperature for response generation. Defaults to 0.8.
+            max_output_tokens (int, optional): Maximum number of tokens to generate in the output. Defaults to None.
+            top_p (float, optional): The nucleus sampling probability for response generation. Defaults to None.
+            tool_choice (ToolChoice or Literal["auto", "required", "none"], optional): Specifies whether to use tools during response generation. Defaults to "auto".
+            additional_request_fields (dict[str, Any], optional): Additional request fields to send to the AWS Bedrock Converse API. Defaults to None.
+        """
+        super().__init__()
+        self._api_key, self._api_secret, self._region = get_aws_credentials(
+            api_key, api_secret, region
+        )
+
+        model = model or os.environ.get("BEDROCK_INFERENCE_PROFILE_ARN")
+        if not is_given(model):
+            raise ValueError(
+                "model or inference profile arn must be set using the argument or by setting the BEDROCK_INFERENCE_PROFILE_ARN environment variable."
+            )
+        self._opts = _LLMOptions(
+            model=model,
+            temperature=temperature,
+            tool_choice=tool_choice,
+            max_output_tokens=max_output_tokens,
+            top_p=top_p,
+            additional_request_fields=additional_request_fields,
+        )
+
+    def chat(
+        self,
+        *,
+        chat_ctx: ChatContext,
+        tools: list[FunctionTool] | None = None,
+        conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
+        temperature: NotGivenOr[float] = NOT_GIVEN,
+        tool_choice: NotGivenOr[ToolChoice | Literal["auto", "required", "none"]] = NOT_GIVEN,
+    ) -> LLMStream:
+        opts = {}
+
+        if is_given(self._opts.model):
+            opts["modelId"] = self._opts.model
+
+        def _get_tool_config() -> dict[str, Any] | None:
+            nonlocal tool_choice
+
+            if not tools:
+                return None
+
+            tool_config: dict[str, Any] = {"tools": to_fnc_ctx(tools)}
+            tool_choice = tool_choice if is_given(tool_choice) else self._opts.tool_choice
+            if is_given(tool_choice):
+                if isinstance(tool_choice, ToolChoice):
+                    tool_config["toolChoice"] = {"tool": {"name": tool_choice.name}}
+                elif tool_choice == "required":
+                    tool_config["toolChoice"] = {"any": {}}
+                elif tool_choice == "auto":
+                    tool_config["toolChoice"] = {"auto": {}}
+                else:
+                    return None
+
+            return tool_config
+
+        tool_config = _get_tool_config()
+        if tool_config:
+            opts["toolConfig"] = tool_config
+        messages, system_message = to_chat_ctx(chat_ctx, id(self))
+        opts["messages"] = messages
+        if system_message:
+            opts["system"] = [system_message]
+
+        inference_config = {}
+        if is_given(self._opts.max_output_tokens):
+            inference_config["maxTokens"] = self._opts.max_output_tokens
+        temperature = temperature if is_given(temperature) else self._opts.temperature
+        if is_given(temperature):
+            inference_config["temperature"] = temperature
+        if is_given(self._opts.top_p):
+            inference_config["topP"] = self._opts.top_p
+
+        opts["inferenceConfig"] = inference_config
+        if is_given(self._opts.additional_request_fields):
+            opts["additionalModelRequestFields"] = self._opts.additional_request_fields
+
+        return LLMStream(
+            self,
+            aws_access_key_id=self._api_key,
+            aws_secret_access_key=self._api_secret,
+            region_name=self._region,
+            chat_ctx=chat_ctx,
+            tools=tools,
+            conn_options=conn_options,
+            extra_kwargs=opts,
+        )
+
+
+class LLMStream(llm.LLMStream):
+    def __init__(
+        self,
+        llm: LLM,
+        *,
+        aws_access_key_id: str,
+        aws_secret_access_key: str,
+        region_name: str,
+        chat_ctx: ChatContext,
+        conn_options: APIConnectOptions,
+        tools: list[FunctionTool] | None,
+        extra_kwargs: dict[str, Any],
+    ) -> None:
+        super().__init__(llm, chat_ctx=chat_ctx, tools=tools, conn_options=conn_options)
+        self._client = boto3.client(
+            "bedrock-runtime",
+            region_name=region_name,
+            aws_access_key_id=aws_access_key_id,
+            aws_secret_access_key=aws_secret_access_key,
+        )
+        self._llm: LLM = llm
+        self._opts = extra_kwargs
+
+        self._tool_call_id: str | None = None
+        self._fnc_name: str | None = None
+        self._fnc_raw_arguments: str | None = None
+        self._text: str = ""
+
+    async def _run(self) -> None:
+        retryable = True
+        try:
+            response = self._client.converse_stream(**self._opts)  # type: ignore
+            request_id = response["ResponseMetadata"]["RequestId"]
+            if response["ResponseMetadata"]["HTTPStatusCode"] != 200:
+                raise APIStatusError(
+                    f"aws bedrock llm: error generating content: {response}",
+                    retryable=False,
+                    request_id=request_id,
+                )
+
+            for chunk in response["stream"]:
+                chat_chunk = self._parse_chunk(request_id, chunk)
+                if chat_chunk is not None:
+                    retryable = False
+                    self._event_ch.send_nowait(chat_chunk)
+
+                # Let other coroutines run
+                await asyncio.sleep(0)
+
+        except Exception as e:
+            raise APIConnectionError(
+                f"aws bedrock llm: error generating content: {e}",
+                retryable=retryable,
+            ) from e
+
+    def _parse_chunk(self, request_id: str, chunk: dict) -> llm.ChatChunk | None:
+        if "contentBlockStart" in chunk:
+            tool_use = chunk["contentBlockStart"]["start"]["toolUse"]
+            self._tool_call_id = tool_use["toolUseId"]
+            self._fnc_name = tool_use["name"]
+            self._fnc_raw_arguments = ""
+        elif "contentBlockDelta" in chunk:
+            delta = chunk["contentBlockDelta"]["delta"]
+            if "toolUse" in delta:
+                self._fnc_raw_arguments += delta["toolUse"]["input"]
+            elif "text" in delta:
+                self._text += delta["text"]
+        elif "contentBlockStop" in chunk:
+            if self._text:
+                chat_chunk = llm.ChatChunk(
+                    id=request_id,
+                    delta=llm.ChoiceDelta(content=self._text, role="assistant"),
+                )
+                self._text = ""
+                return chat_chunk
+            elif self._tool_call_id:
+                if self._tool_call_id is None:
+                    logger.warning("aws bedrock llm: no tool call id in the response")
+                    return None
+                if self._fnc_name is None:
+                    logger.warning("aws bedrock llm: no function name in the response")
+                    return None
+                if self._fnc_raw_arguments is None:
+                    logger.warning("aws bedrock llm: no function arguments in the response")
+                    return None
+                chat_chunk = llm.ChatChunk(
+                    id=request_id,
+                    delta=llm.ChoiceDelta(
+                        role="assistant",
+                        tool_calls=[
+                            FunctionToolCall(
+                                arguments=self._fnc_raw_arguments,
+                                name=self._fnc_name,
+                                call_id=self._tool_call_id,
+                            ),
+                        ],
+                    ),
+                )
+                self._tool_call_id = self._fnc_name = self._fnc_raw_arguments = None
+                return chat_chunk
+        return None

{livekit_plugins_aws-0.1.1 → livekit_plugins_aws-1.0.0.dev5}/livekit/plugins/aws/stt.py

@@ -14,20 +14,15 @@ from __future__ import annotations
 
 import asyncio
 from dataclasses import dataclass
-from typing import Optional
 
 from amazon_transcribe.client import TranscribeStreamingClient
 from amazon_transcribe.model import Result, TranscriptEvent
+
 from livekit import rtc
-from livekit.agents import (
-    DEFAULT_API_CONNECT_OPTIONS,
-    APIConnectOptions,
-    stt,
-    utils,
-)
-
-from ._utils import _get_aws_credentials
+from livekit.agents import DEFAULT_API_CONNECT_OPTIONS, APIConnectOptions, stt, utils
+
 from .log import logger
+from .utils import get_aws_credentials
 
 
 @dataclass
@@ -36,16 +31,16 @@ class STTOptions:
     sample_rate: int
     language: str
     encoding: str
-    vocabulary_name: Optional[str]
-    session_id: Optional[str]
-    vocab_filter_method: Optional[str]
-    vocab_filter_name: Optional[str]
-    show_speaker_label: Optional[bool]
-    enable_channel_identification: Optional[bool]
-    number_of_channels: Optional[int]
-    enable_partial_results_stabilization: Optional[bool]
-    partial_results_stability: Optional[str]
-    language_model_name: Optional[str]
+    vocabulary_name: str | None
+    session_id: str | None
+    vocab_filter_method: str | None
+    vocab_filter_name: str | None
+    show_speaker_label: bool | None
+    enable_channel_identification: bool | None
+    number_of_channels: int | None
+    enable_partial_results_stabilization: bool | None
+    partial_results_stability: str | None
+    language_model_name: str | None
 
 
 class STT(stt.STT):
@@ -58,26 +53,24 @@ class STT(stt.STT):
         sample_rate: int = 48000,
         language: str = "en-US",
         encoding: str = "pcm",
-        vocabulary_name: Optional[str] = None,
-        session_id: Optional[str] = None,
-        vocab_filter_method: Optional[str] = None,
-        vocab_filter_name: Optional[str] = None,
-        show_speaker_label: Optional[bool] = None,
-        enable_channel_identification: Optional[bool] = None,
-        number_of_channels: Optional[int] = None,
-        enable_partial_results_stabilization: Optional[bool] = None,
-        partial_results_stability: Optional[str] = None,
-        language_model_name: Optional[str] = None,
+        vocabulary_name: str | None = None,
+        session_id: str | None = None,
+        vocab_filter_method: str | None = None,
+        vocab_filter_name: str | None = None,
+        show_speaker_label: bool | None = None,
+        enable_channel_identification: bool | None = None,
+        number_of_channels: int | None = None,
+        enable_partial_results_stabilization: bool | None = None,
+        partial_results_stability: str | None = None,
+        language_model_name: str | None = None,
     ):
-        super().__init__(
-            capabilities=stt.STTCapabilities(streaming=True, interim_results=True)
-        )
+        super().__init__(capabilities=stt.STTCapabilities(streaming=True, interim_results=True))
 
-        self._api_key, self._api_secret = _get_aws_credentials(
+        self._api_key, self._api_secret, self._speech_region = get_aws_credentials(
             api_key, api_secret, speech_region
         )
         self._config = STTOptions(
-            speech_region=speech_region,
+            speech_region=self._speech_region,
             language=language,
             sample_rate=sample_rate,
             encoding=encoding,
@@ -100,16 +93,14 @@ class STT(stt.STT):
         language: str | None,
         conn_options: APIConnectOptions,
     ) -> stt.SpeechEvent:
-        raise NotImplementedError(
-            "Amazon Transcribe does not support single frame recognition"
-        )
+        raise NotImplementedError("Amazon Transcribe does not support single frame recognition")
 
     def stream(
         self,
         *,
         language: str | None = None,
         conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
-    ) -> "SpeechStream":
+    ) -> SpeechStream:
         return SpeechStream(
             stt=self,
             conn_options=conn_options,
@@ -124,9 +115,7 @@ class SpeechStream(stt.SpeechStream):
         opts: STTOptions,
         conn_options: APIConnectOptions = DEFAULT_API_CONNECT_OPTIONS,
     ) -> None:
-        super().__init__(
-            stt=stt, conn_options=conn_options, sample_rate=opts.sample_rate
-        )
+        super().__init__(stt=stt, conn_options=conn_options, sample_rate=opts.sample_rate)
         self._opts = opts
         self._client = TranscribeStreamingClient(region=self._opts.speech_region)
 
@@ -151,9 +140,7 @@ class SpeechStream(stt.SpeechStream):
         async def input_generator():
             async for frame in self._input_ch:
                 if isinstance(frame, rtc.AudioFrame):
-                    await stream.input_stream.send_audio_event(
-                        audio_chunk=frame.data.tobytes()
-                    )
+                    await stream.input_stream.send_audio_event(audio_chunk=frame.data.tobytes())
             await stream.input_stream.end_stream()
 
         @utils.log_exceptions(logger=logger)
@@ -184,9 +171,7 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(
                             type=stt.SpeechEventType.INTERIM_TRANSCRIPT,
-                            alternatives=[
-                                _streaming_recognize_response_to_speech_data(resp)
-                            ],
+                            alternatives=[_streaming_recognize_response_to_speech_data(resp)],
                         )
                     )
 
@@ -194,16 +179,12 @@ class SpeechStream(stt.SpeechStream):
                     self._event_ch.send_nowait(
                         stt.SpeechEvent(
                             type=stt.SpeechEventType.FINAL_TRANSCRIPT,
-                            alternatives=[
-                                _streaming_recognize_response_to_speech_data(resp)
-                            ],
+                            alternatives=[_streaming_recognize_response_to_speech_data(resp)],
                         )
                     )
 
             if not resp.is_partial:
-                self._event_ch.send_nowait(
-                    stt.SpeechEvent(type=stt.SpeechEventType.END_OF_SPEECH)
-                )
+                self._event_ch.send_nowait(stt.SpeechEvent(type=stt.SpeechEventType.END_OF_SPEECH))
 
 
 def _streaming_recognize_response_to_speech_data(resp: Result) -> stt.SpeechData: