datapizza-ai-clients-google 0.0.1__tar.gz

datapizza_ai_clients_google-0.0.1/.gitignore

@@ -0,0 +1,207 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[codz]
+*$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
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+uv.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
+#poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#   pdm recommends including project-wide configuration in pdm.toml, but excluding .pdm-python.
+#   https://pdm-project.org/en/latest/usage/project/#working-with-version-control
+#pdm.lock
+#pdm.toml
+.pdm-python
+.pdm-build/
+
+# pixi
+#   Similar to Pipfile.lock, it is generally recommended to include pixi.lock in version control.
+#pixi.lock
+#   Pixi creates a virtual environment in the .pixi directory, just like venv module creates one
+#   in the .venv directory. It is recommended not to include this directory in version control.
+.pixi
+
+# 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
+.envrc
+.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/
+
+# Abstra
+# Abstra is an AI-powered process automation framework.
+# Ignore directories containing user credentials, local state, and settings.
+# Learn more at https://abstra.io/docs
+.abstra/
+
+# Visual Studio Code
+#  Visual Studio Code specific template is maintained in a separate VisualStudioCode.gitignore 
+#  that can be found at https://github.com/github/gitignore/blob/main/Global/VisualStudioCode.gitignore
+#  and can be added to the global gitignore or merged into this file. However, if you prefer, 
+#  you could uncomment the following to ignore the entire vscode folder
+# .vscode/
+
+# Ruff stuff:
+.ruff_cache/
+
+# PyPI configuration file
+.pypirc
+
+# Cursor
+#  Cursor is an AI-powered code editor. `.cursorignore` specifies files/directories to
+#  exclude from AI features like autocomplete and code analysis. Recommended for sensitive data
+#  refer to https://docs.cursor.com/context/ignore-files
+.cursorignore
+.cursorindexingignore
+
+# Marimo
+marimo/_static/
+marimo/_lsp/
+__marimo__/

datapizza_ai_clients_google-0.0.1/PKG-INFO

@@ -0,0 +1,12 @@
+Metadata-Version: 2.4
+Name: datapizza-ai-clients-google
+Version: 0.0.1
+Summary: Google (Gemini) client for the datapizza-ai framework
+License: MIT
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development :: Libraries :: Application Frameworks
+Requires-Python: <3.13.0,>=3.10.0
+Requires-Dist: datapizza-ai-core>=0.0.1
+Requires-Dist: google-genai<2.0.0,>=1.3.0

datapizza_ai_clients_google-0.0.1/datapizza/clients/google/__init__.py

@@ -0,0 +1,3 @@
+from .google_client import GoogleClient
+
+__all__ = ["GoogleClient"]

datapizza_ai_clients_google-0.0.1/datapizza/clients/google/google_client.py

@@ -0,0 +1,659 @@
+from collections.abc import AsyncIterator, Iterator
+from typing import Literal
+
+from datapizza.core.cache import Cache
+from datapizza.core.clients import Client, ClientResponse
+from datapizza.memory import Memory
+from datapizza.tools import Tool
+from datapizza.tools.tool_converter import ToolConverter
+from datapizza.type import (
+    FunctionCallBlock,
+    Model,
+    StructuredBlock,
+    TextBlock,
+    ThoughtBlock,
+)
+
+from google import genai
+from google.genai import types
+from google.oauth2 import service_account
+
+from .memory_adapter import GoogleMemoryAdapter
+
+
+class GoogleClient(Client):
+    """A client for interacting with Google's Generative AI APIs.
+
+    This class provides methods for invoking the Google GenAI API to generate responses
+    based on given input data. It extends the Client class.
+    """
+
+    def __init__(
+        self,
+        api_key: str | None = None,
+        model: str = "gemini-2.0-flash",
+        system_prompt: str = "",
+        temperature: float | None = None,
+        cache: Cache | None = None,
+        project_id: str | None = None,
+        location: str | None = None,
+        credentials_path: str | None = None,
+        use_vertexai: bool = False,
+    ):
+        """
+        Args:
+            api_key: The API key for the Google API.
+            model: The model to use for the Google API.
+            system_prompt: The system prompt to use for the Google API.
+            temperature: The temperature to use for the Google API.
+            cache: The cache to use for the Google API.
+            project_id: The project ID for the Google API.
+            location: The location for the Google API.
+            credentials_path: The path to the credentials for the Google API.
+            use_vertexai: Whether to use Vertex AI for the Google API.
+        """
+        if temperature and not 0 <= temperature <= 2:
+            raise ValueError("Temperature must be between 0 and 2")
+
+        super().__init__(
+            model_name=model,
+            system_prompt=system_prompt,
+            temperature=temperature,
+            cache=cache,
+        )
+        self.memory_adapter = GoogleMemoryAdapter()
+
+        try:
+            if use_vertexai:
+                if not credentials_path:
+                    raise ValueError("credentials_path must be provided")
+                if not project_id:
+                    raise ValueError("project_id must be provided")
+                if not location:
+                    raise ValueError("location must be provided")
+
+                credentials = service_account.Credentials.from_service_account_file(
+                    credentials_path,
+                    scopes=["https://www.googleapis.com/auth/cloud-platform"],
+                )
+                self.client = genai.Client(
+                    vertexai=True,
+                    project=project_id,
+                    location=location,
+                    credentials=credentials,
+                )
+            else:
+                if not api_key:
+                    raise ValueError("api_key must be provided")
+
+                self.client = genai.Client(api_key=api_key)
+
+        except Exception as e:
+            raise RuntimeError(
+                f"Failed to initialize Google GenAI client: {e!s}"
+            ) from None
+
+    def _convert_tool(self, tool: Tool) -> dict:
+        """Convert tools to Google function format"""
+        return ToolConverter.to_google_format(tool)
+
+    def _prepare_tools(self, tools: list[Tool] | None) -> list[types.Tool] | None:
+        if not tools:
+            return None
+
+        google_tools = []
+        function_declarations = []
+        has_google_search = False
+
+        for tool in tools:
+            # Check if tool has google search capability
+            if hasattr(tool, "name") and "google_search" in tool.name.lower():
+                has_google_search = True
+            elif isinstance(tool, Tool):
+                function_declarations.append(self._convert_tool(tool))
+            elif isinstance(tool, dict):
+                google_tools.append(tool)
+            else:
+                raise ValueError(f"Unknown tool type: {type(tool)}")
+
+        if function_declarations:
+            google_tools.append(types.Tool(function_declarations=function_declarations))
+
+        if has_google_search:
+            google_tools.append(types.Tool(google_search=types.GoogleSearch()))
+
+        return google_tools if google_tools else None
+
+    def _convert_tool_choice(
+        self, tool_choice: Literal["auto", "required", "none"] | list[str]
+    ) -> types.ToolConfig:
+        adjusted_tool_choice: types.ToolConfig
+        if isinstance(tool_choice, list):
+            adjusted_tool_choice = types.ToolConfig(
+                function_calling_config=types.FunctionCallingConfig(
+                    mode="ANY",  # type: ignore
+                    allowed_function_names=tool_choice,
+                )
+            )
+        elif tool_choice == "required":
+            adjusted_tool_choice = types.ToolConfig(
+                function_calling_config=types.FunctionCallingConfig(mode="ANY")  # type: ignore
+            )
+        elif tool_choice == "none":
+            adjusted_tool_choice = types.ToolConfig(
+                function_calling_config=types.FunctionCallingConfig(mode="NONE")  # type: ignore
+            )
+        elif tool_choice == "auto":
+            adjusted_tool_choice = types.ToolConfig(
+                function_calling_config=types.FunctionCallingConfig(mode="AUTO")  # type: ignore
+            )
+        return adjusted_tool_choice
+
+    def _invoke(
+        self,
+        *,
+        input: str,
+        tools: list[Tool] | None,
+        memory: Memory | None,
+        tool_choice: Literal["auto", "required", "none"] | list[str],
+        temperature: float | None,
+        max_tokens: int,
+        system_prompt: str | None,
+        **kwargs,
+    ) -> ClientResponse:
+        """Implementation of the abstract _invoke method"""
+        if tools is None:
+            tools = []
+        contents = self._memory_to_contents(None, input, memory)
+
+        tool_map = {tool.name: tool for tool in tools if isinstance(tool, Tool)}
+
+        prepared_tools = self._prepare_tools(tools)
+        config = types.GenerateContentConfig(
+            temperature=temperature or self.temperature,
+            system_instruction=system_prompt or self.system_prompt,
+            max_output_tokens=max_tokens or None,
+            tools=prepared_tools,  # type: ignore
+            tool_config=self._convert_tool_choice(tool_choice)
+            if tools and any(isinstance(tool, Tool) for tool in tools)
+            else None,
+            **kwargs,
+        )
+
+        response = self.client.models.generate_content(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=config,  # type: ignore
+        )
+        return self._response_to_client_response(response, tool_map)
+
+    async def _a_invoke(
+        self,
+        *,
+        input: str,
+        tools: list[Tool] | None,
+        memory: Memory | None,
+        tool_choice: Literal["auto", "required", "none"] | list[str],
+        temperature: float | None,
+        max_tokens: int,
+        system_prompt: str | None,
+        **kwargs,
+    ) -> ClientResponse:
+        """Implementation of the abstract _invoke method"""
+        if tools is None:
+            tools = []
+        contents = self._memory_to_contents(None, input, memory)
+
+        tool_map = {tool.name: tool for tool in tools if isinstance(tool, Tool)}
+
+        prepared_tools = self._prepare_tools(tools)
+        config = types.GenerateContentConfig(
+            temperature=temperature or self.temperature,
+            system_instruction=system_prompt or self.system_prompt,
+            max_output_tokens=max_tokens or None,
+            tools=prepared_tools,  # type: ignore
+            tool_config=self._convert_tool_choice(tool_choice)
+            if tools and any(isinstance(tool, Tool) for tool in tools)
+            else None,
+            **kwargs,
+        )
+
+        response = await self.client.aio.models.generate_content(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=config,  # type: ignore
+        )
+        return self._response_to_client_response(response, tool_map)
+
+    def _stream_invoke(
+        self,
+        input: str,
+        tools: list[Tool] | None,
+        memory: Memory | None,
+        tool_choice: Literal["auto", "required", "none"] | list[str],
+        temperature: float | None,
+        max_tokens: int,
+        system_prompt: str | None,
+        **kwargs,
+    ) -> Iterator[ClientResponse]:
+        """Implementation of the abstract _stream_invoke method"""
+        if tools is None:
+            tools = []
+        contents = self._memory_to_contents(None, input, memory)
+
+        prepared_tools = self._prepare_tools(tools)
+        config = types.GenerateContentConfig(
+            temperature=temperature or self.temperature,
+            system_instruction=system_prompt or self.system_prompt,
+            max_output_tokens=max_tokens or None,
+            tools=prepared_tools,  # type: ignore
+            tool_config=self._convert_tool_choice(tool_choice)
+            if tools and any(isinstance(tool, Tool) for tool in tools)
+            else None,
+            **kwargs,
+        )
+
+        message_text = ""
+        thought_block = ThoughtBlock(content="")
+
+        for chunk in self.client.models.generate_content_stream(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=config,
+        ):
+            if not chunk.candidates:
+                raise ValueError("No candidates in response")
+
+            finish_reason = chunk.candidates[0].finish_reason
+            stop_reason = (
+                finish_reason.value.lower()
+                if finish_reason is not None
+                else finish_reason
+            )
+
+            if not chunk.candidates[0].content:
+                raise ValueError("No content in response")
+
+            if not chunk.candidates[0].content.parts:
+                yield ClientResponse(
+                    content=[],
+                    delta=chunk.text or "",
+                    stop_reason=stop_reason,
+                    prompt_tokens_used=(
+                        chunk.usage_metadata.prompt_token_count
+                        if chunk.usage_metadata
+                        and chunk.usage_metadata.prompt_token_count
+                        else 0
+                    ),
+                    completion_tokens_used=(
+                        chunk.usage_metadata.candidates_token_count
+                        if chunk.usage_metadata
+                        and chunk.usage_metadata.candidates_token_count
+                        else 0
+                    ),
+                    cached_tokens_used=(
+                        chunk.usage_metadata.cached_content_token_count
+                        if chunk.usage_metadata
+                        and chunk.usage_metadata.cached_content_token_count
+                        else 0
+                    ),
+                )
+                continue
+
+            for part in chunk.candidates[0].content.parts:
+                if not part.text:
+                    continue
+                elif hasattr(part, "thought") and part.thought:
+                    thought_block.content += part.text
+                else:  # If it's not a thought, it's a message
+                    if part.text:
+                        message_text += str(chunk.text or "")
+
+                        yield ClientResponse(
+                            content=[],
+                            delta=chunk.text or "",
+                            stop_reason=stop_reason,
+                            prompt_tokens_used=(
+                                chunk.usage_metadata.prompt_token_count
+                                if chunk.usage_metadata
+                                and chunk.usage_metadata.prompt_token_count
+                                else 0
+                            ),
+                            completion_tokens_used=(
+                                chunk.usage_metadata.candidates_token_count
+                                if chunk.usage_metadata
+                                and chunk.usage_metadata.candidates_token_count
+                                else 0
+                            ),
+                            cached_tokens_used=(
+                                chunk.usage_metadata.cached_content_token_count
+                                if chunk.usage_metadata
+                                and chunk.usage_metadata.cached_content_token_count
+                                else 0
+                            ),
+                        )
+
+    async def _a_stream_invoke(
+        self,
+        input: str,
+        tools: list[Tool] | None = None,
+        memory: Memory | None = None,
+        tool_choice: Literal["auto", "required", "none"] | list[str] = "auto",
+        temperature: float | None = None,
+        max_tokens: int | None = None,
+        system_prompt: str | None = None,
+        **kwargs,
+    ) -> AsyncIterator[ClientResponse]:
+        """Implementation of the abstract _a_stream_invoke method for Google"""
+        if tools is None:
+            tools = []
+        contents = self._memory_to_contents(None, input, memory)
+
+        prepared_tools = self._prepare_tools(tools)
+        config = types.GenerateContentConfig(
+            temperature=temperature or self.temperature,
+            system_instruction=system_prompt or self.system_prompt,
+            max_output_tokens=max_tokens or None,
+            tools=prepared_tools,  # type: ignore
+            tool_config=self._convert_tool_choice(tool_choice)
+            if tools and any(isinstance(tool, Tool) for tool in tools)
+            else None,
+            **kwargs,
+        )
+
+        message_text = ""
+        thought_block = ThoughtBlock(content="")
+        async for chunk in await self.client.aio.models.generate_content_stream(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=config,
+        ):  # type: ignore
+            finish_reason = chunk.candidates[0].finish_reason
+            stop_reason = (
+                finish_reason.value.lower()
+                if finish_reason is not None
+                else finish_reason
+            )
+
+            # Handle the case where the response has no parts
+            if not chunk.candidates[0].content.parts:
+                yield ClientResponse(
+                    content=[],
+                    delta=chunk.text or "",
+                    stop_reason=stop_reason,
+                    prompt_tokens_used=chunk.usage_metadata.prompt_token_count
+                    if chunk.usage_metadata
+                    else 0,
+                    completion_tokens_used=chunk.usage_metadata.candidates_token_count
+                    if chunk.usage_metadata
+                    else 0,
+                    cached_tokens_used=chunk.usage_metadata.cached_content_token_count
+                    if chunk.usage_metadata
+                    else 0,
+                )
+                continue
+
+            for part in chunk.candidates[0].content.parts:
+                if not part.text:
+                    continue
+                elif hasattr(part, "thought") and part.thought:
+                    thought_block.content += part.text
+                else:  # If it's not a thought, it's a message
+                    if part.text:
+                        message_text += chunk.text or ""
+                        yield ClientResponse(
+                            content=[],
+                            delta=chunk.text or "",
+                            stop_reason=stop_reason,
+                            prompt_tokens_used=chunk.usage_metadata.prompt_token_count
+                            if chunk.usage_metadata
+                            and chunk.usage_metadata.prompt_token_count
+                            else 0,
+                            completion_tokens_used=chunk.usage_metadata.candidates_token_count
+                            if chunk.usage_metadata
+                            and chunk.usage_metadata.candidates_token_count
+                            else 0,
+                            cached_tokens_used=chunk.usage_metadata.cached_content_token_count
+                            if chunk.usage_metadata
+                            and chunk.usage_metadata.cached_content_token_count
+                            else 0,
+                        )
+
+    def _structured_response(
+        self,
+        input: str,
+        output_cls: type[Model],
+        memory: Memory | None,
+        temperature: float | None,
+        max_tokens: int,
+        system_prompt: str | None,
+        tools: list[Tool] | None,
+        tool_choice: Literal["auto", "required", "none"] | list[str] = "auto",
+        **kwargs,
+    ) -> ClientResponse:
+        """Implementation of the abstract _structured_response method"""
+        contents = self._memory_to_contents(self.system_prompt, input, memory)
+
+        prepared_tools = self._prepare_tools(tools)
+        response = self.client.models.generate_content(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=types.GenerateContentConfig(
+                system_instruction=system_prompt,
+                temperature=temperature,
+                max_output_tokens=max_tokens,
+                response_mime_type="application/json",
+                tools=prepared_tools,  # type: ignore
+                tool_config=self._convert_tool_choice(tool_choice)
+                if tools and any(isinstance(tool, Tool) for tool in tools)
+                else None,
+                response_schema=(
+                    output_cls.model_json_schema()
+                    if hasattr(output_cls, "model_json_schema")
+                    else output_cls
+                ),
+            ),
+        )
+        if not response or not response.candidates:
+            raise ValueError("No response from Google GenAI")
+
+        structured_data = output_cls.model_validate_json(str(response.text))
+        return ClientResponse(
+            content=[StructuredBlock(content=structured_data)],
+            stop_reason=response.candidates[0].finish_reason.value.lower()
+            if response.candidates[0].finish_reason
+            else None,
+            prompt_tokens_used=(
+                response.usage_metadata.prompt_token_count
+                if response.usage_metadata
+                and response.usage_metadata.prompt_token_count
+                else 0
+            ),
+            completion_tokens_used=(
+                response.usage_metadata.candidates_token_count
+                if response.usage_metadata
+                and response.usage_metadata.candidates_token_count
+                else 0
+            ),
+            cached_tokens_used=(
+                response.usage_metadata.cached_content_token_count
+                if response.usage_metadata
+                and response.usage_metadata.cached_content_token_count
+                else 0
+            ),
+        )
+
+    async def _a_structured_response(
+        self,
+        input: str,
+        output_cls: type[Model],
+        memory: Memory | None,
+        temperature: float | None,
+        max_tokens: int,
+        system_prompt: str | None,
+        tools: list[Tool] | None,
+        tool_choice: Literal["auto", "required", "none"] | list[str] = "auto",
+        **kwargs,
+    ) -> ClientResponse:
+        """Implementation of the abstract _structured_response method"""
+        contents = self._memory_to_contents(self.system_prompt, input, memory)
+        prepared_tools = self._prepare_tools(tools)
+        response = await self.client.aio.models.generate_content(
+            model=self.model_name,
+            contents=contents,  # type: ignore
+            config=types.GenerateContentConfig(
+                system_instruction=system_prompt,
+                temperature=temperature,
+                max_output_tokens=max_tokens,
+                response_mime_type="application/json",
+                tools=prepared_tools,  # type: ignore
+                tool_config=self._convert_tool_choice(tool_choice)
+                if tools and any(isinstance(tool, Tool) for tool in tools)
+                else None,
+                response_schema=(
+                    output_cls.model_json_schema()
+                    if hasattr(output_cls, "model_json_schema")
+                    else output_cls
+                ),
+            ),
+        )
+
+        if not response or not response.candidates:
+            raise ValueError("No response from Google GenAI")
+
+        structured_data = output_cls.model_validate_json(str(response.text))
+        return ClientResponse(
+            content=[StructuredBlock(content=structured_data)],
+            stop_reason=response.candidates[0].finish_reason.value.lower()
+            if response.candidates[0].finish_reason
+            else None,
+            prompt_tokens_used=(
+                response.usage_metadata.prompt_token_count
+                if response.usage_metadata
+                and response.usage_metadata.prompt_token_count
+                else 0
+            ),
+            completion_tokens_used=(
+                response.usage_metadata.candidates_token_count
+                if response.usage_metadata
+                and response.usage_metadata.candidates_token_count
+                else 0
+            ),
+            cached_tokens_used=(
+                response.usage_metadata.cached_content_token_count
+                if response.usage_metadata
+                and response.usage_metadata.cached_content_token_count
+                else 0
+            ),
+        )
+
+    def _embed(
+        self,
+        text: str | list[str],
+        model_name: str | None,
+        task_type: str = "RETRIEVAL_DOCUMENT",
+        output_dimensionality: int = 768,
+        title: str | None = None,
+        **kwargs,
+    ) -> list[float] | list[list[float] | None]:
+        """Embed a text using the model"""
+        response = self.client.models.embed_content(
+            model=model_name or self.model_name,
+            contents=text,  # type: ignore
+            config=types.EmbedContentConfig(
+                task_type=task_type,
+                output_dimensionality=output_dimensionality,
+                title=title,
+                **kwargs,
+            ),
+        )
+        # Extract the embedding values from the response
+        if not response.embeddings:
+            return []
+
+        embeddings = [embedding.values for embedding in response.embeddings]
+
+        if isinstance(text, str) and embeddings[0]:
+            return embeddings[0]
+
+        return embeddings
+
+    async def _a_embed(
+        self,
+        text: str | list[str],
+        model_name: str | None,
+        task_type: str = "RETRIEVAL_DOCUMENT",
+        output_dimensionality: int = 768,
+        title: str | None = None,
+        **kwargs,
+    ) -> list[float] | list[list[float] | None]:
+        """Embed a text using the model"""
+        response = await self.client.aio.models.embed_content(
+            model=model_name or self.model_name,
+            contents=text,  # type: ignore
+            config=types.EmbedContentConfig(
+                task_type=task_type,
+                output_dimensionality=output_dimensionality,
+                title=title,
+                **kwargs,
+            ),
+        )
+        # Extract the embedding values from the response
+        if not response.embeddings:
+            return []
+        embeddings = [embedding.values for embedding in response.embeddings]
+
+        if isinstance(text, str) and embeddings[0]:
+            return embeddings[0]
+
+        return embeddings
+
+    def _response_to_client_response(
+        self, response, tool_map: dict[str, Tool] | None = None
+    ) -> ClientResponse:
+        blocks = []
+        # Handle function calls if present
+        if hasattr(response, "function_calls") and response.function_calls:
+            for fc in response.function_calls:
+                if not tool_map:
+                    raise ValueError("Tool map is required")
+
+                tool = tool_map.get(fc.name, None)
+                if not tool:
+                    raise ValueError(f"Tool {fc.name} not found in tool map")
+
+                blocks.append(
+                    FunctionCallBlock(
+                        name=fc.name,
+                        arguments=fc.args,
+                        id=f"fc_{id(fc)}",
+                        tool=tool,
+                    )
+                )
+        else:
+            if hasattr(response, "text") and response.text:
+                blocks.append(TextBlock(content=response.text))
+
+        if hasattr(response, "candidates") and response.candidates:
+            for part in response.candidates[0].content.parts:
+                if not part.text:
+                    continue
+                if hasattr(part, "thought") and part.thought:
+                    blocks.append(ThoughtBlock(content=part.text))
+
+        usage_metadata = getattr(response, "usage_metadata", None)
+        return ClientResponse(
+            content=blocks,
+            stop_reason=(response.candidates[0].finish_reason.value.lower())
+            if hasattr(response, "candidates") and response.candidates
+            else None,
+            prompt_tokens_used=usage_metadata.prompt_token_count
+            if usage_metadata
+            else 0,
+            completion_tokens_used=usage_metadata.candidates_token_count
+            if usage_metadata
+            else 0,
+            cached_tokens_used=usage_metadata.cached_content_token_count
+            if usage_metadata
+            else 0,
+        )

datapizza_ai_clients_google-0.0.1/datapizza/clients/google/memory_adapter.py

@@ -0,0 +1,140 @@
+import base64
+
+from datapizza.memory.memory import Turn
+from datapizza.memory.memory_adapter import MemoryAdapter
+from datapizza.type import (
+    ROLE,
+    FunctionCallBlock,
+    FunctionCallResultBlock,
+    MediaBlock,
+    StructuredBlock,
+    TextBlock,
+)
+
+from google.genai import types
+
+
+class GoogleMemoryAdapter(MemoryAdapter):
+    def _turn_to_message(self, turn: Turn) -> dict:
+        content = []
+        for block in turn:
+            block_dict = {}
+
+            match block:
+                case TextBlock():
+                    block_dict = {"text": block.content}
+                case FunctionCallBlock():
+                    block_dict = {
+                        "function_call": {"name": block.name, "args": block.arguments}
+                    }
+                case FunctionCallResultBlock():
+                    block_dict = types.Part.from_function_response(
+                        name=block.tool.name,
+                        response={"result": block.result},
+                    )
+                case StructuredBlock():
+                    block_dict = {"text": str(block.content)}
+                case MediaBlock():
+                    match block.media.media_type:
+                        case "image":
+                            block_dict = self._process_image_block(block)
+                        case "pdf":
+                            block_dict = self._process_pdf_block(block)
+
+                        case "audio":
+                            block_dict = self._process_audio_block(block)
+
+                        case _:
+                            raise NotImplementedError(
+                                f"Unsupported media type: {block.media.media_type}"
+                            )
+
+            content.append(block_dict)
+
+        return {
+            "role": turn.role.google_role,
+            "parts": (content),
+        }
+
+    def _process_audio_block(self, block: MediaBlock) -> types.Part:
+        match block.media.source_type:
+            case "raw":
+                return types.Part.from_bytes(
+                    data=block.media.source,
+                    mime_type="audio/mp3",
+                )
+
+            case "path":
+                with open(block.media.source, "rb") as f:
+                    audio_bytes = f.read()
+
+                return types.Part.from_bytes(
+                    data=audio_bytes,
+                    mime_type="audio/mp3",
+                )
+
+            case _:
+                raise NotImplementedError(
+                    f"Unsupported media source type: {block.media.source_type} for audio, source type supported: raw, path"
+                )
+
+    def _process_pdf_block(self, block: MediaBlock) -> types.Part | dict:
+        match block.media.source_type:
+            case "raw":
+                return types.Part.from_bytes(
+                    data=block.media.source,
+                    mime_type="application/pdf",
+                )
+            case "base64":
+                return {
+                    "inline_data": {
+                        "mime_type": "application/pdf",
+                        "data": block.media.source,
+                    }
+                }
+            case "path":
+                with open(block.media.source, "rb") as f:
+                    pdf_bytes = f.read()
+
+                return {
+                    "inline_data": {
+                        "mime_type": "application/pdf",
+                        "data": pdf_bytes,
+                    }
+                }
+
+            case _:
+                raise NotImplementedError(
+                    f"Unsupported media source type: {block.media.source_type} only supported: raw, base64, path"
+                )
+
+    def _process_image_block(self, block: MediaBlock) -> dict:
+        match block.media.source_type:
+            case "url":
+                return types.Part.from_uri(
+                    file_uri=block.media.source,
+                    mime_type=f"image/{block.media.extension}",
+                )  # type: ignore
+            case "base64":
+                return {
+                    "inline_data": {
+                        "mime_type": f"image/{block.media.extension}",
+                        "data": block.media.source,
+                    }
+                }
+            case "path":
+                with open(block.media.source, "rb") as image_file:
+                    base64_image = base64.b64encode(image_file.read()).decode("utf-8")
+                return {
+                    "inline_data": {
+                        "mime_type": f"image/{block.media.extension}",
+                        "data": base64_image,
+                    }
+                }
+            case _:
+                raise NotImplementedError(
+                    f"Unsupported media source type: {block.media.source_type} for image, only url, base64, path are supported"
+                )
+
+    def _text_to_message(self, text: str, role: ROLE) -> dict:
+        return {"role": role.google_role, "parts": [{"text": text}]}

datapizza_ai_clients_google-0.0.1/pyproject.toml

@@ -0,0 +1,57 @@
+# Build system configuration
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+# Project metadata
+[project]
+name = "datapizza-ai-clients-google"
+version = "0.0.1"
+description = "Google (Gemini) client for the datapizza-ai framework"
+readme = "README.md"
+license = {text = "MIT"}
+
+requires-python = ">=3.10.0,<3.13.0"
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+    "Topic :: Software Development :: Libraries :: Application Frameworks",
+]
+dependencies = [
+    "datapizza-ai-core>=0.0.1",
+    "google-genai>=1.3.0,<2.0.0",
+]
+
+# Development dependencies
+[dependency-groups]
+dev = [
+    "deptry>=0.23.0",
+    "pytest",
+    "ruff>=0.11.5",
+]
+
+# Hatch build configuration
+[tool.hatch.build.targets.sdist]
+include = ["datapizza"]
+exclude = ["**/BUILD"]
+
+[tool.hatch.build.targets.wheel]
+include = ["datapizza"]
+exclude = ["**/BUILD"]
+
+# Ruff configuration
+[tool.ruff]
+line-length = 88
+
+[tool.ruff.lint]
+select = [
+    "W",   # pycodestyle warnings
+    "F",   # pyflakes
+    "B",   # flake8-bugbear
+    "I",   # isort
+    "UP",  # pyupgrade
+    "SIM", # flake8-simplify
+    "RUF", # Ruff-specific rules
+    "C4",  # flake8-comprehensions
+]