satif-ai 0.2.8__tar.gz → 0.2.10__tar.gz

{satif_ai-0.2.8 → satif_ai-0.2.10}/PKG-INFO

@@ -1,17 +1,19 @@
 Metadata-Version: 2.3
 Name: satif-ai
-Version: 0.2.8
+Version: 0.2.10
 Summary: AI Agents for Satif
 License: MIT
-Author: Bryan Djafer
-Author-email: bryan.djafer@syncpulse.fr
-Requires-Python: >=3.10,<4.0
+Author: Syncpulse
+Maintainer: Bryan Djafer
+Maintainer-email: bryan.djafer@syncpulse.fr
+Requires-Python: >=3.10,<3.14
 Classifier: License :: OSI Approved :: MIT License
 Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Provides-Extra: xlsx
 Requires-Dist: openai-agents (>=0.0.9,<0.0.10)
 Requires-Dist: satif-sdk (>=0.1.0,<1.0.0)
 Requires-Dist: sdif-mcp (>=0.1.0,<1.0.0)

{satif_ai-0.2.8 → satif_ai-0.2.10}/pyproject.toml

@@ -1,14 +1,17 @@
 [project]
 name = "satif-ai"
-version = "0.2.8"
+version = "0.2.10"
 description = "AI Agents for Satif"
 authors = [
+    {name = "Syncpulse"}
+]
+maintainers = [
     {name = "Bryan Djafer", email = "bryan.djafer@syncpulse.fr"}
 ]
 license = "MIT"
 readme = "README.md"
 
-requires-python = ">=3.10,<4.0"
+requires-python = ">=3.10,<3.14"
 
 [tool.poetry.dependencies]
 openai-agents = ">=0.0.9,<0.0.10"
@@ -19,6 +22,9 @@ sdif-mcp = ">=0.1.0,<1.0.0"
 requires = ["poetry-core>=2.0.0,<3.0.0"]
 build-backend = "poetry.core.masonry.api"
 
+[tool.poetry.extras]
+xlsx = ["xlsx-to-sdif"]
+
 [project.scripts]
 satif-ai = "satif.cli:main"
 
@@ -28,6 +34,7 @@ satif-core = {path = "../core", develop = true}
 satif-sdk = {path = "../sdk", develop = true}
 sdif-mcp = {path = "../mcp", develop = true}
 sdif-db = {path = "../sdif", develop = true}
+xlsx-to-sdif = {path = "../xlsx-to-sdif", develop = true}
 ipykernel = "^6.29.5"
 
 

satif_ai-0.2.10/satif_ai/__init__.py

@@ -0,0 +1,19 @@
+from .adapters.tidy import TidyAdapter
+from .standardize import astandardize
+from .standardizers.ai import AIStandardizer
+from .standardizers.ai_csv import AICSVStandardizer
+from .transform import atransform
+from .transformation_builders.syncpulse import SyncpulseTransformationBuilder
+from .utils import OpenAICompatibleMCP, extract_zip_archive_async, merge_sdif_files
+
+__all__ = [
+    "astandardize",
+    "atransform",
+    "TidyAdapter",
+    "AICSVStandardizer",
+    "AIStandardizer",
+    "SyncpulseTransformationBuilder",
+    "OpenAICompatibleMCP",
+    "extract_zip_archive_async",
+    "merge_sdif_files",
+]

{satif_ai-0.2.8 → satif_ai-0.2.10}/satif_ai/adapters/tidy.py

@@ -6,23 +6,19 @@ import shutil
 import sqlite3
 import tempfile
 from pathlib import Path
-from typing import Optional
+from typing import Optional, Union
 
-# MCP and Agent imports
 from agents import Agent, Runner, function_tool
-from agents.mcp.server import MCPServerStdio
+from agents.mcp.server import MCPServer
 from mcp import ClientSession
-
-# SATIF imports
 from satif_core.adapters.base import Adapter
-from satif_core.types import Datasource
+from satif_core.types import Datasource, SDIFPath
 from satif_sdk import SDIFDatabase
 from satif_sdk.adapters.code import AdapterError, CodeAdapter
 
 logger = logging.getLogger(__name__)
 
 
-# --- Tidy Transformation Prompt ---
 TIDY_TRANSFORMATION_PROMPT = """
 You are an expert Data Tidying Agent for SDIF databases.
 Your task is to write Python code to transform tables within a given SDIF database into a 'tidy' format, modifying the database *in place*.
@@ -130,12 +126,11 @@ def adapt_sdif(db: SDIFDatabase) -> None:
 - Ensure pandas and other necessary libraries (like `typing`, `AdapterError`) are imported within the code string if you use them.
 """
 
-# --- Global context for tools ---
-# These will be set within the TidyAdapter instance when adapt is called
+
 TOOL_CONTEXT = {
     "copied_input_sdif_path": None,
     "temp_dir": None,
-    "current_output_sdif_path": None,  # Path generated by the tool
+    "current_output_sdif_path": None,
 }
 
 
@@ -167,13 +162,10 @@ async def execute_tidy_adaptation(code: str) -> str:
     )
 
     try:
-        # 1. Instantiate CodeAdapter with the provided code
-        #    It will operate on a *copy* specified by copied_input_path
-        #    and write to a *new* file (_adapted suffix by default).
         adapter = CodeAdapter(
             function=code,
-            function_name="adapt_sdif",  # As specified in prompt
-            output_suffix="_adapted_tool_run",  # Give tool runs a distinct suffix
+            function_name="adapt_sdif",
+            output_suffix="_adapted_tool_run",
         )
         # Run the adaptation. It copies `copied_input_path` and modifies the copy.
         # The returned path is the newly created, adapted file.
@@ -232,16 +224,16 @@ class TidyAdapter(Adapter):
 
     def __init__(
         self,
-        mcp_server: MCPServerStdio,  # Use the server instance
-        mcp_session: ClientSession,  # Use the client session
-        llm_model: str = "o4-mini",  # Specify the LLM model
+        mcp_server: MCPServer,
+        mcp_session: ClientSession,
+        llm_model: str = "o4-mini",
         max_iterations: int = 5,
     ):
         """
         Initialize the TidyAdapter.
 
         Args:
-            mcp_server: An instance of MCPServerStdio for agent communication.
+            mcp_server: An instance of MCPServer for agent communication.
             mcp_session: An instance of ClientSession for resource/prompt fetching.
             llm_model: Name of the language model to use for the agent.
             max_iterations: Maximum number of attempts the agent gets to refine the code.
@@ -339,12 +331,12 @@ class TidyAdapter(Adapter):
                 return code_text.strip()
             return None  # Indicate no valid code found
 
-    async def adapt(self, sdif_database: SDIFDatabase) -> Datasource:
+    async def adapt(self, sdif: Union[SDIFPath, SDIFDatabase]) -> Datasource:
         """
         Transforms the data in the input SDIF to be tidy using an AI agent.
 
         Args:
-            sdif_database: The input SDIF database instance. Connection will be closed.
+            sdif: The input SDIF database instance. Connection will be closed.
 
         Returns:
             Path to the new SDIF file containing the tidied data.
@@ -354,13 +346,16 @@ class TidyAdapter(Adapter):
             RuntimeError: If the agent fails to produce valid tidy code.
             Exception: For unexpected errors during the process.
         """
-        input_path = Path(sdif_database.path)
+        if isinstance(sdif, SDIFDatabase):
+            input_path = Path(sdif.path)
+        else:
+            input_path = Path(sdif)
         if not input_path.exists():
             raise FileNotFoundError(f"Input SDIF file not found: {input_path}")
 
         # Ensure the input DB connection is closed before copying
         try:
-            sdif_database.close()
+            sdif.close()
         except Exception:
             pass
 
@@ -372,17 +367,14 @@ class TidyAdapter(Adapter):
                 input_schema_dict = db.get_schema()
                 input_sample_dict = db.get_sample_analysis()
 
-            # Get SDIFDatabase method signatures
             sdif_methods_str = self._get_sdif_methods()
 
-            # Prepare context for the prompt
             initial_context = {
                 "input_schema": json.dumps(input_schema_dict, indent=2),
                 "input_sample": json.dumps(input_sample_dict, indent=2),
                 "sdif_database_methods": sdif_methods_str,
             }
 
-            # Instantiate the Agent
             agent = Agent(
                 name="Tidy SDIF Adapter Agent",
                 mcp_servers=[self.mcp_server],
@@ -390,8 +382,6 @@ class TidyAdapter(Adapter):
                 model=self.llm_model,
             )
 
-            # Run the agent using the Runner
-            # Pass the prompt and initial context
             logger.info(f"Running Tidy Agent with model {self.llm_model}...")
             result = await Runner.run(
                 agent,
@@ -409,7 +399,6 @@ class TidyAdapter(Adapter):
                 f"Agent finished. Final output message:\n{result.final_output[:500]}..."
             )
 
-            # Parse the final code from the agent's response
             final_code = self.parse_code(result.final_output)
 
             if not final_code:
@@ -421,20 +410,16 @@ class TidyAdapter(Adapter):
             logger.info(
                 "Successfully parsed final adaptation code from agent response."
             )
-            # print(f"--- Final Code ---\n{final_code}\n------------------") # Debugging
 
-            # Execute the *final* code using CodeAdapter directly to create the definitive output
             logger.info("Executing final adaptation code...")
             final_adapter = CodeAdapter(
                 function=final_code,
                 function_name="adapt_sdif",
-                output_suffix="_tidy_final",  # Use a distinct suffix for the final output
+                output_suffix="_tidy_final",
             )
-            # Adapt the *original* copied input path
+
             final_adapted_path = final_adapter.adapt(copied_input_path)
 
-            # Move the final successful output SDIF to a persistent location
-            # Example: place it next to the original input file
             persistent_output_path = (
                 input_path.parent / final_adapted_path.name
             ).resolve()
@@ -444,9 +429,7 @@ class TidyAdapter(Adapter):
                 )
                 persistent_output_path.unlink()
 
-            shutil.move(
-                str(final_adapted_path), persistent_output_path
-            )  # Move needs strings sometimes
+            shutil.move(str(final_adapted_path), persistent_output_path)
             logger.info(
                 f"Successfully generated final tidy SDIF: {persistent_output_path}"
             )
@@ -455,8 +438,6 @@ class TidyAdapter(Adapter):
 
         except Exception as e:
             logger.exception(f"Error during TidyAdapter adapt process: {e}")
-            # Re-raise or handle as appropriate
             raise
         finally:
-            # Always clean up temporary files
             self._cleanup_temp_env()

satif_ai-0.2.10/satif_ai/standardize.py

@@ -0,0 +1,112 @@
+from pathlib import Path
+from typing import Any, Dict, Optional, Union
+
+from satif_core.standardizers.base import AsyncStandardizer
+from satif_core.types import Datasource, FilePath, SDIFPath, StandardizationResult
+
+from satif_ai.adapters.tidy import TidyAdapter
+from satif_ai.standardizers.ai import AIStandardizer
+
+
+async def astandardize(
+    datasource: Datasource,
+    output_path: SDIFPath,
+    *,
+    overwrite: bool = False,
+    sdif_schema: Optional[Union[FilePath, Dict[str, Any]]] = None,
+    tidy_adapter: Union[bool, TidyAdapter] = False,
+    config: Optional[Dict[str, Any]] = None,
+    standardizer: Optional[AsyncStandardizer] = None,
+    mcp_server: Optional[Any] = None,
+    mcp_session: Optional[Any] = None,
+    llm_model: Optional[str] = None,
+) -> StandardizationResult:
+    """
+    Asynchronously standardizes a datasource into a single, canonical SDIF SQLite file.
+
+    This function serves as the primary entry point for the SATIF standardization layer.
+    It orchestrates the conversion of various input file formats (e.g., CSV, Excel, PDF)
+    from the provided datasource into a unified SDIF (Standard Data Interchange Format)
+    SQLite file. The process may involve AI-driven parsing, schema adaptation, and
+    data tidying, depending on the configuration and the capabilities of the
+    underlying standardizer.
+
+    Args:
+        datasource: The source of the data to be standardized. This can be a
+                    single file path (str or Path), a list of file paths, or other
+                    datasource types supported by the chosen standardizer.
+        output_path: The path (str or Path) where the output SDIF SQLite database file
+                     will be created (e.g., "./output/my_data.sdif").
+        overwrite: If True, an existing SDIF file at `output_path` will be
+                   overwritten. Defaults to False.
+        sdif_schema: Optional. Path to an SDIF schema definition file (e.g., a JSON file)
+                     or a dictionary representing the schema. If provided, the
+                     standardization process (specifically if using the default
+                     `AIStandardizer`) may attempt to adapt the data to this
+                     target schema.
+        tidy_adapter: Optional. If True, a default `TidyAdapter` may be used.
+                      Alternatively, a specific `TidyAdapter` instance can be provided
+                      to perform data tidying processes (e.g., cleaning, normalization,
+                      restructuring tables). If False (default), no explicit tidying
+                      step is initiated by this top-level function, though underlying
+                      standardizers might perform their own internal tidying.
+                      The specifics depend on the standardizer's capabilities.
+        config: Optional. A dictionary for advanced or standardizer-specific
+                configurations. This config is passed directly to the
+                `standardize` method of the chosen standardizer.
+        standardizer: Optional. An instance of an `AsyncStandardizer` subclass.
+                      If provided, this instance will be used for standardization.
+                      This allows for using pre-configured or custom standardizers.
+                      If None, a default `AIStandardizer` is instantiated using
+                      `mcp_server`, `mcp_session`, `llm_model`, `sdif_schema`,
+                      and `tidy_adapter`.
+        mcp_server: Optional. The MCP (Model Coordination Platform) server instance.
+                    Used if `standardizer` is None for the default `AIStandardizer`.
+        mcp_session: Optional. The MCP session or transport object.
+                     Used if `standardizer` is None for the default `AIStandardizer`.
+        llm_model: Optional. The language model to be used by the default `AIStandardizer`
+                   if no `standardizer` instance is provided (e.g., "gpt-4o").
+                   Each standardizer may have its own default model.
+
+    Returns:
+        A `StandardizationResult` object containing:
+        - `output_path`: The absolute `Path` to the created or updated SDIF database file.
+        - `file_configs`: An optional dictionary detailing configurations used for
+                          each processed input file, if applicable and returned by
+                          the standardizer.
+
+    Raises:
+        FileNotFoundError: If the `datasource` (or parts of it) does not exist.
+        FileExistsError: If `output_path` exists and `overwrite` is False.
+        ValueError: If input arguments are invalid (e.g., unsupported datasource type).
+        RuntimeError: For general errors during the standardization process.
+                      Specific exceptions may also be raised by the underlying
+                      standardizer implementation.
+    """
+    if standardizer is None:
+        standardizer = AIStandardizer(
+            mcp_server=mcp_server,
+            mcp_session=mcp_session,
+            llm_model=llm_model,
+            sdif_schema=sdif_schema,
+            tidy_adapter=tidy_adapter
+            if isinstance(tidy_adapter, TidyAdapter)
+            else (TidyAdapter() if tidy_adapter else None),
+        )
+
+    result = await standardizer.standardize(
+        datasource=datasource,
+        output_path=output_path,
+        overwrite=overwrite,
+        config=config,
+    )
+
+    output_sdif_path = (
+        Path(result.output_path)
+        if isinstance(result.output_path, str)
+        else result.output_path
+    )
+
+    return StandardizationResult(
+        output_path=output_sdif_path, file_configs=result.file_configs
+    )