Add a session keyword argument to to_sql and to_df

documentation/usage.md

@@ -425,6 +425,7 @@ The `to_sql` API takes in PyDough code and transforms it into SQL query text wit
 - `metadata`: the PyDough knowledge graph to use for the conversion (if omitted, `pydough.active_session.metadata` is used instead).
 - `config`: the PyDough configuration settings to use for the conversion (if omitted, `pydough.active_session.config` is used instead).
 - `database`: the database context to use for the conversion (if omitted, `pydough.active_session.database` is used instead). The database context matters because it controls which SQL dialect is used for the translation.
+- `session`: a PyDough session object which, if provided, is used instead of `pydough.active_session` or the `metadata` / `config` / `database` arguments. Note: this argument cannot be used alongside those arguments.
 
 Below is an example of using `pydough.to_sql` and the output (the SQL output may be outdated if PyDough's SQL conversion process has been updated):
 
@@ -436,34 +437,22 @@ pydough.to_sql(result, columns=["name", "n_custs"])
 ```
 
 ```sql
-SELECT name, COALESCE(agg_0, 0) AS n_custs
-FROM (
-    SELECT name, agg_0
-    FROM (
-        SELECT name, key
-        FROM (
-            SELECT _table_alias_0.name AS name, _table_alias_0.key AS key, _table_alias_1.name AS name_3
-            FROM (
-                SELECT n_name AS name, n_nationkey AS key, n_regionkey AS region_key FROM main.NATION
-            ) AS _table_alias_0
-            LEFT JOIN (
-                SELECT r_name AS name, r_regionkey AS key
-                FROM main.REGION
-            ) AS _table_alias_1
-            ON region_key = _table_alias_1.key
-        )
-        WHERE name_3 = 'EUROPE'
-    )
-    LEFT JOIN (
-        SELECT nation_key, COUNT(*) AS agg_0
-        FROM (
-            SELECT c_nationkey AS nation_key
-            FROM main.CUSTOMER
-        )
-        GROUP BY nation_key
-    )
-    ON key = nation_key
+WITH _s3 AS (
+  SELECT
+    c_nationkey,
+    COUNT(*) AS n_rows
+  FROM tpch.customer
+  GROUP BY
+    1
 )
+SELECT
+  nation.n_name AS name,
+  _s3.n_rows AS n_custs
+FROM tpch.nation AS nation
+JOIN tpch.region AS region
+  ON nation.n_regionkey = region.r_regionkey AND region.r_name = 'EUROPE'
+JOIN _s3 AS _s3
+  ON _s3.c_nationkey = nation.n_nationkey
 ```
 
 See the [demo notebooks](../demos/README.md) for more instances of how to use the `to_sql` API.
@@ -478,6 +467,7 @@ The `to_df` API does all the same steps as the [`to_sql` API](#pydoughto_sql), b
 - `metadata`: the PyDough knowledge graph to use for the conversion (if omitted, `pydough.active_session.metadata` is used instead).
 - `config`: the PyDough configuration settings to use for the conversion (if omitted, `pydough.active_session.config` is used instead).
 - `database`: the database context to use for the conversion (if omitted, `pydough.active_session.database` is used instead). The database context matters because it controls which SQL dialect is used for the translation.
+- `session`: a PyDough session object which, if provided, is used instead of `pydough.active_session` or the `metadata` / `config` / `database` arguments. Note: this argument cannot be used alongside those arguments.
 - `display_sql`: displays the sql before executing in a logger.
 
 Below is an example of using `pydough.to_df` and the output, attached to a sqlite database containing data for the TPC-H schema:
@@ -616,41 +606,35 @@ The value of `sql` is the following SQL query text as a Python string:
 ```sql
 WITH _s7 AS (
   SELECT
-    ROUND(
-      COALESCE(
-        SUM(
-          lineitem.l_extendedprice * (
-            1 - lineitem.l_discount
-          ) * (
-            1 - lineitem.l_tax
-          ) - lineitem.l_quantity * partsupp.ps_supplycost
-        ),
-        0
-      ),
-      2
-    ) AS revenue_year,
-    partsupp.ps_suppkey
+    partsupp.ps_suppkey,
+    SUM(
+      lineitem.l_extendedprice * (
+        1 - lineitem.l_discount
+      ) * (
+        1 - lineitem.l_tax
+      ) - lineitem.l_quantity * partsupp.ps_supplycost
+    ) AS sum_rev
   FROM main.partsupp AS partsupp
   JOIN main.part AS part
     ON part.p_name LIKE 'coral%' AND part.p_partkey = partsupp.ps_partkey
   JOIN main.lineitem AS lineitem
-    ON CAST(STRFTIME('%Y', lineitem.l_shipdate) AS INTEGER) = 1996
+    ON EXTRACT(YEAR FROM CAST(lineitem.l_shipdate AS DATETIME)) = 1996
     AND lineitem.l_partkey = partsupp.ps_partkey
     AND lineitem.l_shipmode = 'TRUCK'
     AND lineitem.l_suppkey = partsupp.ps_suppkey
   GROUP BY
-    partsupp.ps_suppkey
+    1
 )
 SELECT
   supplier.s_name AS name,
-  _s7.revenue_year
+  ROUND(COALESCE(_s7.sum_rev, 0), 2) AS revenue_year
 FROM main.supplier AS supplier
 JOIN main.nation AS nation
   ON nation.n_name = 'JAPAN' AND nation.n_nationkey = supplier.s_nationkey
 JOIN _s7 AS _s7
   ON _s7.ps_suppkey = supplier.s_suppkey
 ORDER BY
-  revenue_year DESC
+  2 DESC
 LIMIT 5
 ```
 
@@ -688,27 +672,27 @@ The value of `sql` is the following SQL query text as a Python string:
 ```sql
 WITH _s1 AS (
   SELECT
-    COALESCE(SUM(o_totalprice), 0) AS total,
+    o_custkey,
     COUNT(*) AS n_rows,
-    o_custkey
+    SUM(o_totalprice) AS sum_o_totalprice
   FROM main.orders
   WHERE
-    o_orderdate < '1997-01-01'
-    AND o_orderdate >= '1996-01-01'
+    o_orderdate < CAST('1997-01-01' AS DATE)
+    AND o_orderdate >= CAST('1996-01-01' AS DATE)
     AND o_orderpriority = '1-URGENT'
     AND o_totalprice > 100000
   GROUP BY
-    o_custkey
+    1
 )
 SELECT
   customer.c_name AS name,
   _s1.n_rows AS n_orders,
-  _s1.total
+  _s1.sum_o_totalprice AS total
 FROM main.customer AS customer
 JOIN _s1 AS _s1
   ON _s1.o_custkey = customer.c_custkey
 ORDER BY
-  total DESC
+  3 DESC
 ```
 
 <!-- TOC --><a name="exploration-apis"></a>
@@ -776,7 +760,7 @@ The `explain` API is a more generic explanation interface that can be called on
 - A specific property within a specific collection within a metadata graph object (can be accessed as `graph["collection_name"]["property_name"]`)
 - The PyDough code for a collection that could have `to_sql` or `to_df` called on it.
 
-The `explain` API also has an optional `verbose` argument (default=False) that enables displaying additional information.
+The `explain` API also has an optional `verbose` argument (default=False) that enables displaying additional information. It also has an optional `session` argument to specify what configs etc. to use when explaining certain terms (if not provided, uses `pydough.active_session`).
 
 Below are examples of each of these behaviors, using a knowledge graph for the TPCH schema.
 
@@ -994,7 +978,8 @@ The `explain` API is limited in that it can only be called on complete PyDough c
 
 To handle cases where you need to learn about a term within a collection, you can use the `explain_term` API. The first argument to `explain_term` is PyDough code for a collection, which can have `explain` called on it, and the second is PyDough code for a term that can be evaluated within the context of that collection (e.g. a scalar term of the collection, or one of its sub-collections).
 
-The `explain_term` API also has a `verbose` keyword argument (default False) to specify whether to include a more detailed explanation, as opposed to a more compact summary.
+The `explain_term` API also has a `verbose` keyword argument (default False) to specify whether to include a more detailed explanation, as opposed to a more compact summary. The `explain_term` API also has an optional `verbose` argument (default=False) that enables displaying additional information. It also has an optional `session` argument to specify what configs etc. to use when explaining certain terms (if not provided, uses `pydough.active_session`).
+
 
 Below are examples of using `explain_term`, using a knowledge graph for the TPCH schema. For each of these examples, `european_countries` is the "context" collection, which could have `to_sql` or `to_df` called on it, and `term` is the term being explained with regards to `european_countries`.
 

pydough/conversion/agg_split.py

@@ -7,7 +7,7 @@
 
 
 import pydough.pydough_operators as pydop
-from pydough.configs import PyDoughConfigs
+from pydough.configs import PyDoughSession
 from pydough.relational import (
     Aggregate,
     CallExpression,
@@ -51,15 +51,15 @@
 """
 
 
-def decompose_aggregations(node: Aggregate, config: PyDoughConfigs) -> RelationalNode:
+def decompose_aggregations(node: Aggregate, session: PyDoughSession) -> RelationalNode:
     """
     Splits up an aggregate node into an aggregate followed by a projection when
     the aggregate contains 1+ calls to functions that can be split into 1+
     calls to partial aggregates, e.g. how AVG(X) = SUM(X)/COUNT(X).
 
     Args:
         `node`: the aggregate node to be decomposed.
-        `config`: the current configuration settings.
+        `session`: the PyDough session used during the transformation.
 
     Returns:
         The projection node on top of the new aggregate, overall containing the
@@ -110,7 +110,7 @@ def decompose_aggregations(node: Aggregate, config: PyDoughConfigs) -> Relationa
             )
             # If the config specifies that the default value for AVG should be
             # zero, wrap the division in a DEFAULT_TO call.
-            if config.avg_default_zero:
+            if session.config.avg_default_zero:
                 avg_call = CallExpression(
                     pydop.DEFAULT_TO,
                     agg.data_type,
@@ -277,7 +277,7 @@ def transpose_aggregate_join(
 
 
 def attempt_join_aggregate_transpose(
-    node: Aggregate, join: Join, config: PyDoughConfigs
+    node: Aggregate, join: Join, session: PyDoughSession
 ) -> tuple[RelationalNode, bool]:
     """
     Determine whether the aggregate join transpose operation can occur, and if
@@ -396,7 +396,7 @@ def attempt_join_aggregate_transpose(
     for col in node.aggregations.values():
         if col.op in decomposable_aggfuncs:
             return split_partial_aggregates(
-                decompose_aggregations(node, config), config
+                decompose_aggregations(node, session), session
             ), False
 
     # Keep a dictionary for the projection columns that will be used to post-process
@@ -464,7 +464,7 @@ def attempt_join_aggregate_transpose(
 
 
 def split_partial_aggregates(
-    node: RelationalNode, config: PyDoughConfigs
+    node: RelationalNode, session: PyDoughSession
 ) -> RelationalNode:
     """
     Splits partial aggregates above joins into two aggregates, one above the
@@ -473,19 +473,21 @@ def split_partial_aggregates(
 
     Args:
         `node`: the root node of the relational plan to be transformed.
-        `config`: the current configuration settings.
+        `session`: the PyDough session used during the transformation.
 
     Returns:
         The transformed node. The transformation is also done-in-place.
     """
     # If the aggregate+join pattern is detected, attempt to do the transpose.
     handle_inputs: bool = True
     if isinstance(node, Aggregate) and isinstance(node.input, Join):
-        node, handle_inputs = attempt_join_aggregate_transpose(node, node.input, config)
+        node, handle_inputs = attempt_join_aggregate_transpose(
+            node, node.input, session
+        )
 
     # If needed, recursively invoke the procedure on all inputs to the node.
     if handle_inputs:
         node = node.copy(
-            inputs=[split_partial_aggregates(input, config) for input in node.inputs]
+            inputs=[split_partial_aggregates(input, session) for input in node.inputs]
         )
     return node

pydough/conversion/filter_pushdown.py

@@ -6,7 +6,7 @@
 
 
 import pydough.pydough_operators as pydop
-from pydough.configs import PyDoughConfigs
+from pydough.configs import PyDoughSession
 from pydough.relational import (
     Aggregate,
     CallExpression,
@@ -66,7 +66,7 @@ class FilterPushdownShuttle(RelationalShuttle):
     cannot be pushed further.
     """
 
-    def __init__(self, configs: PyDoughConfigs):
+    def __init__(self, session: PyDoughSession):
         # The set of filters that are currently being pushed down. When
         # visit_xxx is called, it is presumed that the set of conditions in
         # self.filters are the conditions that can be pushed down as far as the
@@ -76,7 +76,7 @@ def __init__(self, configs: PyDoughConfigs):
         # simplification logic to aid in advanced filter predicate inference,
         # such as determining that a left join is redundant because if the RHS
         # column is null then the filter will always be false.
-        self.simplifier: SimplificationShuttle = SimplificationShuttle(configs)
+        self.simplifier: SimplificationShuttle = SimplificationShuttle(session)
 
     def reset(self):
         self.filters = set()
@@ -300,7 +300,7 @@ def visit_empty_singleton(self, empty_singleton: EmptySingleton) -> RelationalNo
         return self.flush_remaining_filters(empty_singleton, self.filters, set())
 
 
-def push_filters(node: RelationalNode, configs: PyDoughConfigs) -> RelationalNode:
+def push_filters(node: RelationalNode, session: PyDoughSession) -> RelationalNode:
     """
     Transpose filter conditions down as far as possible.
 
@@ -314,5 +314,5 @@ def push_filters(node: RelationalNode, configs: PyDoughConfigs) -> RelationalNod
         the node or into one of its inputs, or possibly both if there are
         multiple filters.
     """
-    pusher: FilterPushdownShuttle = FilterPushdownShuttle(configs)
+    pusher: FilterPushdownShuttle = FilterPushdownShuttle(session)
     return node.accept_shuttle(pusher)

pydough/conversion/hybrid_translator.py

@@ -7,7 +7,7 @@
 from collections.abc import Iterable
 
 import pydough.pydough_operators as pydop
-from pydough.configs import PyDoughConfigs
+from pydough.configs import PyDoughSession
 from pydough.database_connectors import DatabaseDialect
 from pydough.errors import PyDoughSQLException
 from pydough.metadata import (
@@ -80,8 +80,8 @@ class HybridTranslator:
     Class used to translate PyDough QDAG nodes into the HybridTree structure.
     """
 
-    def __init__(self, configs: PyDoughConfigs, dialect: DatabaseDialect):
-        self.configs = configs
+    def __init__(self, session: PyDoughSession):
+        self.session = session
         # An index used for creating fake column names for aliases
         self.alias_counter: int = 0
         # A stack where each element is a hybrid tree being derived
@@ -91,7 +91,7 @@ def __init__(self, configs: PyDoughConfigs, dialect: DatabaseDialect):
         # If True, rewrites MEDIAN calls into an average of the 1-2 median rows
         # or rewrites QUANTILE calls to select the first qualifying row,
         # both derived from window functions, otherwise leaves as-is.
-        self.rewrite_median_quantile: bool = dialect not in {
+        self.rewrite_median_quantile: bool = session.database.dialect not in {
             DatabaseDialect.ANSI,
             DatabaseDialect.SNOWFLAKE,
         }
@@ -481,8 +481,8 @@ def postprocess_agg_output(
         # COUNT/NDISTINCT for left joins since the semantics of those functions
         # never allow returning NULL.
         if (
-            (agg_call.operator == pydop.SUM and self.configs.sum_default_zero)
-            or (agg_call.operator == pydop.AVG and self.configs.avg_default_zero)
+            (agg_call.operator == pydop.SUM and self.session.config.sum_default_zero)
+            or (agg_call.operator == pydop.AVG and self.session.config.avg_default_zero)
             or (
                 agg_call.operator in (pydop.COUNT, pydop.NDISTINCT)
                 and joins_can_nullify