kumoai 2.13.0.dev202511231731__cp312-cp312-macosx_11_0_arm64.whl → 2.13.0.dev202512031731__cp312-cp312-macosx_11_0_arm64.whl

kumoai/experimental/rfm/{local_graph.py → graph.py}

@@ -2,8 +2,10 @@ import contextlib
 import io
 import warnings
 from collections import defaultdict
+from dataclasses import dataclass, field
 from importlib.util import find_spec
-from typing import TYPE_CHECKING, Dict, List, Optional, Union
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, Dict, List, Optional, Sequence, Union
 
 import pandas as pd
 from kumoapi.graph import ColumnKey, ColumnKeyGroup, GraphDefinition
@@ -12,19 +14,28 @@ from kumoapi.typing import Stype
 from typing_extensions import Self
 
 from kumoai import in_notebook
-from kumoai.experimental.rfm import LocalTable
+from kumoai.experimental.rfm import Table
 from kumoai.graph import Edge
+from kumoai.mixin import CastMixin
 
 if TYPE_CHECKING:
     import graphviz
+    from adbc_driver_sqlite.dbapi import AdbcSqliteConnection
+    from snowflake.connector import SnowflakeConnection
 
 
-class LocalGraph:
-    r"""A graph of :class:`LocalTable` objects, akin to relationships between
+@dataclass
+class SqliteConnectionConfig(CastMixin):
+    uri: Union[str, Path]
+    kwargs: Dict[str, Any] = field(default_factory=dict)
+
+
+class Graph:
+    r"""A graph of :class:`Table` objects, akin to relationships between
     tables in a relational database.
 
     Creating a graph is the final step of data definition; after a
-    :class:`LocalGraph` is created, you can use it to initialize the
+    :class:`Graph` is created, you can use it to initialize the
     Kumo Relational Foundation Model (:class:`KumoRFM`).
 
     .. code-block:: python
@@ -44,7 +55,7 @@ class LocalGraph:
         >>> table3 = rfm.LocalTable(name="table3", data=df3)
 
         >>> # Create a graph from a dictionary of tables:
-        >>> graph = rfm.LocalGraph({
+        >>> graph = rfm.Graph({
         ...     "table1": table1,
         ...     "table2": table2,
         ...     "table3": table3,
@@ -75,33 +86,47 @@ class LocalGraph:
 
     def __init__(
         self,
-        tables: List[LocalTable],
-        edges: Optional[List[Edge]] = None,
+        tables: Sequence[Table],
+        edges: Optional[Sequence[Edge]] = None,
     ) -> None:
 
-        self._tables: Dict[str, LocalTable] = {}
+        self._tables: Dict[str, Table] = {}
         self._edges: List[Edge] = []
 
         for table in tables:
             self.add_table(table)
 
+        for table in tables:
+            for fkey in table._source_foreign_key_dict.values():
+                if fkey.name not in table or fkey.dst_table not in self:
+                    continue
+                if self[fkey.dst_table].primary_key is None:
+                    self[fkey.dst_table].primary_key = fkey.primary_key
+                elif self[fkey.dst_table]._primary_key != fkey.primary_key:
+                    raise ValueError(f"Found duplicate primary key definition "
+                                     f"'{self[fkey.dst_table]._primary_key}' "
+                                     f"and '{fkey.primary_key}' in table "
+                                     f"'{fkey.dst_table}'.")
+                self.link(table.name, fkey.name, fkey.dst_table)
+
         for edge in (edges or []):
             _edge = Edge._cast(edge)
             assert _edge is not None
-            self.link(*_edge)
+            if _edge not in self._edges:
+                self.link(*_edge)
 
     @classmethod
     def from_data(
         cls,
         df_dict: Dict[str, pd.DataFrame],
-        edges: Optional[List[Edge]] = None,
+        edges: Optional[Sequence[Edge]] = None,
         infer_metadata: bool = True,
         verbose: bool = True,
     ) -> Self:
-        r"""Creates a :class:`LocalGraph` from a dictionary of
+        r"""Creates a :class:`Graph` from a dictionary of
         :class:`pandas.DataFrame` objects.
 
-        Automatically infers table metadata and links.
+        Automatically infers table metadata and links by default.
 
         .. code-block:: python
 
@@ -115,55 +140,186 @@ class LocalGraph:
             >>> df3 = pd.DataFrame(...)
 
             >>> # Create a graph from a dictionary of data frames:
-            >>> graph = rfm.LocalGraph.from_data({
+            >>> graph = rfm.Graph.from_data({
             ...     "table1": df1,
             ...     "table2": df2,
             ...     "table3": df3,
             ... })
 
-            >>> # Inspect table metadata:
-            >>> for table in graph.tables.values():
-            ...     table.print_metadata()
-
-            >>> # Visualize graph (if graphviz is installed):
-            >>> graph.visualize()
-
         Args:
             df_dict: A dictionary of data frames, where the keys are the names
                 of the tables and the values hold table data.
+            edges: An optional list of :class:`~kumoai.graph.Edge` objects to
+                add to the graph. If not provided, edges will be automatically
+                inferred from the data in case ``infer_metadata=True``.
             infer_metadata: Whether to infer metadata for all tables in the
                 graph.
+            verbose: Whether to print verbose output.
+        """
+        from kumoai.experimental.rfm.backend.local import LocalTable
+        tables = [LocalTable(df, name) for name, df in df_dict.items()]
+
+        graph = cls(tables, edges=edges or [])
+
+        if infer_metadata:
+            graph.infer_metadata(False)
+
+            if edges is None:
+                graph.infer_links(False)
+
+        if verbose:
+            graph.print_metadata()
+            graph.print_links()
+
+        return graph
+
+    @classmethod
+    def from_sqlite(
+        cls,
+        connection: Union[
+            'AdbcSqliteConnection',
+            SqliteConnectionConfig,
+            str,
+            Path,
+            Dict[str, Any],
+        ],
+        table_names: Optional[Sequence[str]] = None,
+        edges: Optional[Sequence[Edge]] = None,
+        infer_metadata: bool = True,
+        verbose: bool = True,
+    ) -> Self:
+        r"""Creates a :class:`Graph` from a :class:`sqlite` database.
+
+        Automatically infers table metadata and links by default.
+
+        .. code-block:: python
+
+            >>> # doctest: +SKIP
+            >>> import kumoai.experimental.rfm as rfm
+
+            >>> # Create a graph from a SQLite database:
+            >>> graph = rfm.Graph.from_sqlite('data.db')
+
+        Args:
+            connection: An open connection from
+                :meth:`~kumoai.experimental.rfm.backend.sqlite.connect` or the
+                path to the database file.
+            table_names: Set of table names to include. If ``None``, will add
+                all tables present in the database.
             edges: An optional list of :class:`~kumoai.graph.Edge` objects to
                 add to the graph. If not provided, edges will be automatically
-                inferred from the data.
+                inferred from the data in case ``infer_metadata=True``.
+            infer_metadata: Whether to infer metadata for all tables in the
+                graph.
             verbose: Whether to print verbose output.
+        """
+        from kumoai.experimental.rfm.backend.sqlite import (
+            Connection,
+            SQLiteTable,
+            connect,
+        )
+
+        if not isinstance(connection, Connection):
+            connection = SqliteConnectionConfig._cast(connection)
+            assert isinstance(connection, SqliteConnectionConfig)
+            connection = connect(connection.uri, **connection.kwargs)
+        assert isinstance(connection, Connection)
+
+        if table_names is None:
+            with connection.cursor() as cursor:
+                cursor.execute("SELECT name FROM sqlite_master "
+                               "WHERE type='table'")
+                table_names = [row[0] for row in cursor.fetchall()]
+
+        tables = [SQLiteTable(connection, name) for name in table_names]
 
-        Note:
-            This method will automatically infer metadata and links for the
-            graph.
+        graph = cls(tables, edges=edges or [])
+
+        if infer_metadata:
+            graph.infer_metadata(False)
+
+            if edges is None:
+                graph.infer_links(False)
+
+        if verbose:
+            graph.print_metadata()
+            graph.print_links()
+
+        return graph
+
+    @classmethod
+    def from_snowflake(
+        cls,
+        connection: Union['SnowflakeConnection', Dict[str, Any], None] = None,
+        table_names: Optional[Sequence[str]] = None,
+        edges: Optional[Sequence[Edge]] = None,
+        infer_metadata: bool = True,
+        verbose: bool = True,
+    ) -> Self:
+        r"""Creates a :class:`Graph` from a :class:`snowflake` database and
+        schema.
+
+        Automatically infers table metadata and links by default.
+
+        .. code-block:: python
 
-        Example:
             >>> # doctest: +SKIP
             >>> import kumoai.experimental.rfm as rfm
-            >>> df1 = pd.DataFrame(...)
-            >>> df2 = pd.DataFrame(...)
-            >>> df3 = pd.DataFrame(...)
-            >>> graph = rfm.LocalGraph.from_data(data={
-            ...     "table1": df1,
-            ...     "table2": df2,
-            ...     "table3": df3,
-            ... })
-            >>> graph.validate()
+
+            >>> # Create a graph directly in a Snowflake notebook:
+            >>> graph = rfm.Graph.from_snowflake()
+
+        Args:
+            connection: An open connection from
+                :meth:`~kumoai.experimental.rfm.backend.snow.connect` or the
+                :class:`snowflake` connector keyword arguments to open a new
+                connection. If ``None``, will re-use an active session in case
+                it exists, or create a new connection from credentials stored
+                in environment variables.
+            table_names: Set of table names to include. If ``None``, will add
+                all tables present in the database.
+            edges: An optional list of :class:`~kumoai.graph.Edge` objects to
+                add to the graph. If not provided, edges will be automatically
+                inferred from the data in case ``infer_metadata=True``.
+            infer_metadata: Whether to infer metadata for all tables in the
+                graph.
+            verbose: Whether to print verbose output.
         """
-        tables = [LocalTable(df, name) for name, df in df_dict.items()]
+        from kumoai.experimental.rfm.backend.snow import (
+            Connection,
+            SnowTable,
+            connect,
+        )
+
+        if not isinstance(connection, Connection):
+            connection = connect(**(connection or {}))
+        assert isinstance(connection, Connection)
+
+        if table_names is None:
+            with connection.cursor() as cursor:
+                cursor.execute("SELECT CURRENT_DATABASE(), CURRENT_SCHEMA()")
+                database, schema = cursor.fetchone()
+                query = f"""
+                SELECT TABLE_NAME
+                FROM {database}.INFORMATION_SCHEMA.TABLES
+                WHERE TABLE_SCHEMA = '{schema}'
+                """
+                cursor.execute(query)
+                table_names = [row[0] for row in cursor.fetchall()]
+
+        tables = [SnowTable(connection, name) for name in table_names]
 
         graph = cls(tables, edges=edges or [])
 
         if infer_metadata:
-            graph.infer_metadata(verbose)
+            graph.infer_metadata(False)
 
             if edges is None:
-                graph.infer_links(verbose)
+                graph.infer_links(False)
+
+        if verbose:
+            graph.print_metadata()
+            graph.print_links()
 
         return graph
 
@@ -175,7 +331,7 @@ class LocalGraph:
         """
         return name in self.tables
 
-    def table(self, name: str) -> LocalTable:
+    def table(self, name: str) -> Table:
         r"""Returns the table with name ``name`` in the graph.
 
         Raises:
@@ -186,11 +342,11 @@ class LocalGraph:
         return self.tables[name]
 
     @property
-    def tables(self) -> Dict[str, LocalTable]:
+    def tables(self) -> Dict[str, Table]:
         r"""Returns the dictionary of table objects."""
         return self._tables
 
-    def add_table(self, table: LocalTable) -> Self:
+    def add_table(self, table: Table) -> Self:
         r"""Adds a table to the graph.
 
         Args:
@@ -199,11 +355,21 @@ class LocalGraph:
         Raises:
             KeyError: If a table with the same name already exists in the
                 graph.
+            ValueError: If the table belongs to a different backend than the
+                rest of the tables in the graph.
         """
         if table.name in self._tables:
             raise KeyError(f"Cannot add table with name '{table.name}' to "
                            f"this graph; table names must be globally unique.")
 
+        if len(self._tables) > 0:
+            cls = next(iter(self._tables.values())).__class__
+            if table.__class__ != cls:
+                raise ValueError(f"Cannot register a "
+                                 f"'{table.__class__.__name__}' to this "
+                                 f"graph since other tables are of type "
+                                 f"'{cls.__name__}'.")
+
         self._tables[table.name] = table
 
         return self
@@ -241,7 +407,7 @@ class LocalGraph:
         Example:
             >>> # doctest: +SKIP
             >>> import kumoai.experimental.rfm as rfm
-            >>> graph = rfm.LocalGraph(tables=...).infer_metadata()
+            >>> graph = rfm.Graph(tables=...).infer_metadata()
             >>> graph.metadata  # doctest: +SKIP
                 name   primary_key  time_column  end_time_column
             0   users      user_id            -                -
@@ -263,7 +429,7 @@ class LocalGraph:
         })
 
     def print_metadata(self) -> None:
-        r"""Prints the :meth:`~LocalGraph.metadata` of the graph."""
+        r"""Prints the :meth:`~Graph.metadata` of the graph."""
         if in_notebook():
             from IPython.display import Markdown, display
             display(Markdown('### 🗂️ Graph Metadata'))
@@ -287,7 +453,7 @@ class LocalGraph:
 
         Note:
             For more information, please see
-            :meth:`kumoai.experimental.rfm.LocalTable.infer_metadata`.
+            :meth:`kumoai.experimental.rfm.Table.infer_metadata`.
         """
         for table in self.tables.values():
             table.infer_metadata(verbose=False)
@@ -305,7 +471,7 @@ class LocalGraph:
         return self._edges
 
     def print_links(self) -> None:
-        r"""Prints the :meth:`~LocalGraph.edges` of the graph."""
+        r"""Prints the :meth:`~Graph.edges` of the graph."""
         edges = [(edge.dst_table, self[edge.dst_table]._primary_key,
                   edge.src_table, edge.fkey) for edge in self.edges]
         edges = sorted(edges)
@@ -333,9 +499,9 @@ class LocalGraph:
 
     def link(
         self,
-        src_table: Union[str, LocalTable],
+        src_table: Union[str, Table],
         fkey: str,
-        dst_table: Union[str, LocalTable],
+        dst_table: Union[str, Table],
     ) -> Self:
         r"""Links two tables (``src_table`` and ``dst_table``) from the foreign
         key ``fkey`` in the source table to the primary key in the destination
@@ -358,11 +524,11 @@ class LocalGraph:
                 table does not exist in the graph, if the source key does not
                 exist in the source table.
         """
-        if isinstance(src_table, LocalTable):
+        if isinstance(src_table, Table):
             src_table = src_table.name
         assert isinstance(src_table, str)
 
-        if isinstance(dst_table, LocalTable):
+        if isinstance(dst_table, Table):
             dst_table = dst_table.name
         assert isinstance(dst_table, str)
 
@@ -396,9 +562,9 @@ class LocalGraph:
 
     def unlink(
         self,
-        src_table: Union[str, LocalTable],
+        src_table: Union[str, Table],
         fkey: str,
-        dst_table: Union[str, LocalTable],
+        dst_table: Union[str, Table],
     ) -> Self:
         r"""Removes an :class:`~kumoai.graph.Edge` from the graph.
 
@@ -410,11 +576,11 @@ class LocalGraph:
         Raises:
             ValueError: if the edge is not present in the graph.
         """
-        if isinstance(src_table, LocalTable):
+        if isinstance(src_table, Table):
             src_table = src_table.name
         assert isinstance(src_table, str)
 
-        if isinstance(dst_table, LocalTable):
+        if isinstance(dst_table, Table):
             dst_table = dst_table.name
         assert isinstance(dst_table, str)
 
@@ -428,17 +594,13 @@ class LocalGraph:
         return self
 
     def infer_links(self, verbose: bool = True) -> Self:
-        r"""Infers links for the tables and adds them as edges to the graph.
+        r"""Infers missing links for the tables and adds them as edges to the
+        graph.
 
         Args:
             verbose: Whether to print verbose output.
-
-        Note:
-            This function expects graph edges to be undefined upfront.
         """
-        if len(self.edges) > 0:
-            warnings.warn("Cannot infer links if graph edges already exist")
-            return self
+        known_edges = {(edge.src_table, edge.fkey) for edge in self.edges}
 
         # A list of primary key candidates (+score) for every column:
         candidate_dict: dict[
@@ -463,6 +625,9 @@ class LocalGraph:
                 src_table_name = src_table.name.lower()
 
                 for src_key in src_table.columns:
+                    if (src_table.name, src_key.name) in known_edges:
+                        continue
+
                     if src_key == src_table.primary_key:
                         continue  # Cannot link to primary key.
 
@@ -528,7 +693,9 @@ class LocalGraph:
                         score += 1.0
 
                     # Cardinality ratio:
-                    if len(src_table._data) > len(dst_table._data):
+                    if (src_table._num_rows is not None
+                            and dst_table._num_rows is not None
+                            and src_table._num_rows > dst_table._num_rows):
                         score += 1.0
 
                     if score < 5.0:
@@ -790,7 +957,7 @@ class LocalGraph:
     def __contains__(self, name: str) -> bool:
         return self.has_table(name)
 
-    def __getitem__(self, name: str) -> LocalTable:
+    def __getitem__(self, name: str) -> Table:
         return self.table(name)
 
     def __delitem__(self, name: str) -> None:

kumoai/experimental/rfm/infer/__init__.py

@@ -1,9 +1,15 @@
+from .dtype import infer_dtype
+from .pkey import infer_primary_key
+from .time_col import infer_time_column
 from .id import contains_id
 from .timestamp import contains_timestamp
 from .categorical import contains_categorical
 from .multicategorical import contains_multicategorical
 
 __all__ = [
+    'infer_dtype',
+    'infer_primary_key',
+    'infer_time_column',
     'contains_id',
     'contains_timestamp',
     'contains_categorical',

kumoai/experimental/rfm/infer/dtype.py

@@ -0,0 +1,90 @@
+from typing import Any, Dict
+
+import numpy as np
+import pandas as pd
+import pyarrow as pa
+from kumoapi.typing import Dtype
+
+PANDAS_TO_DTYPE: Dict[Any, Dtype] = {
+    np.dtype('bool'): Dtype.bool,
+    pd.BooleanDtype(): Dtype.bool,
+    pa.bool_(): Dtype.bool,
+    np.dtype('byte'): Dtype.int,
+    pd.UInt8Dtype(): Dtype.int,
+    np.dtype('int16'): Dtype.int,
+    pd.Int16Dtype(): Dtype.int,
+    np.dtype('int32'): Dtype.int,
+    pd.Int32Dtype(): Dtype.int,
+    np.dtype('int64'): Dtype.int,
+    pd.Int64Dtype(): Dtype.int,
+    np.dtype('float32'): Dtype.float,
+    pd.Float32Dtype(): Dtype.float,
+    np.dtype('float64'): Dtype.float,
+    pd.Float64Dtype(): Dtype.float,
+    np.dtype('object'): Dtype.string,
+    pd.StringDtype(storage='python'): Dtype.string,
+    pd.StringDtype(storage='pyarrow'): Dtype.string,
+    pa.string(): Dtype.string,
+    pa.binary(): Dtype.binary,
+    np.dtype('datetime64[ns]'): Dtype.date,
+    np.dtype('timedelta64[ns]'): Dtype.timedelta,
+    pa.list_(pa.float32()): Dtype.floatlist,
+    pa.list_(pa.int64()): Dtype.intlist,
+    pa.list_(pa.string()): Dtype.stringlist,
+}
+
+
+def infer_dtype(ser: pd.Series) -> Dtype:
+    """Extracts the :class:`Dtype` from a :class:`pandas.Series`.
+
+    Args:
+        ser: A :class:`pandas.Series` to analyze.
+
+    Returns:
+        The data type.
+    """
+    if pd.api.types.is_datetime64_any_dtype(ser.dtype):
+        return Dtype.date
+
+    if isinstance(ser.dtype, pd.CategoricalDtype):
+        return Dtype.string
+
+    if pd.api.types.is_object_dtype(ser.dtype):
+        index = ser.iloc[:1000].first_valid_index()
+        if index is not None and pd.api.types.is_list_like(ser[index]):
+            pos = ser.index.get_loc(index)
+            assert isinstance(pos, int)
+            ser = ser.iloc[pos:pos + 1000].dropna()
+
+            if not ser.map(pd.api.types.is_list_like).all():
+                raise ValueError("Data contains a mix of list-like and "
+                                 "non-list-like values")
+
+            # Remove all empty Python lists without known data type:
+            ser = ser[ser.map(lambda x: not isinstance(x, list) or len(x) > 0)]
+
+            # Infer unique data types in this series:
+            dtypes = ser.apply(lambda x: PANDAS_TO_DTYPE.get(
+                np.array(x).dtype, Dtype.string)).unique().tolist()
+
+            invalid_dtypes = set(dtypes) - {
+                Dtype.string,
+                Dtype.int,
+                Dtype.float,
+            }
+            if len(invalid_dtypes) > 0:
+                raise ValueError(f"Data contains unsupported list data types: "
+                                 f"{list(invalid_dtypes)}")
+
+            if Dtype.string in dtypes:
+                return Dtype.stringlist
+
+            if dtypes == [Dtype.int]:
+                return Dtype.intlist
+
+            return Dtype.floatlist
+
+    if ser.dtype not in PANDAS_TO_DTYPE:
+        raise ValueError(f"Unsupported data type '{ser.dtype}'")
+
+    return PANDAS_TO_DTYPE[ser.dtype]