pixeltable 0.3.6__py3-none-any.whl → 0.3.8__py3-none-any.whl

pixeltable/metadata/converters/convert_25.py

@@ -13,7 +13,7 @@ def _(engine: sql.engine.Engine) -> None:
 
 def __substitute_md(k: Optional[str], v: Any) -> Optional[tuple[Optional[str], Any]]:
     if k == 'path' and (
-        v in ['pixeltable.functions.huggingface.clip_text', 'pixeltable.functions.huggingface.clip_image']
+        v in {'pixeltable.functions.huggingface.clip_text', 'pixeltable.functions.huggingface.clip_image'}
     ):
         return 'path', 'pixeltable.functions.huggingface.clip'
     return None

pixeltable/metadata/converters/convert_27.py

@@ -1,12 +1,10 @@
 import logging
-from typing import Any, Optional
 from uuid import UUID
 
 import sqlalchemy as sql
 
 from pixeltable.metadata import register_converter
 from pixeltable.metadata.converters.util import convert_table_md
-from pixeltable.metadata.schema import Table
 
 _logger = logging.getLogger('pixeltable')
 

pixeltable/metadata/converters/convert_28.py

@@ -1,5 +1,3 @@
-import logging
-
 import sqlalchemy as sql
 
 from pixeltable.metadata import register_converter

pixeltable/metadata/converters/convert_29.py

@@ -63,13 +63,12 @@ def __substitute_md(k: Optional[str], v: Any) -> Optional[tuple[Optional[str], A
             # is an edge case that won't migrate properly.
             parameters: list[dict] = v['fn']['signature']['parameters']
             for i, param in enumerate(parameters):
-                if param['kind'] == 'VAR_POSITIONAL':
-                    if new_args_len > i:
-                        # For peculiar historical reasons, variable kwargs might show up in args. Thus variable
-                        # positional args is not necessarily the last element of args; it might be the second-to-last.
-                        assert new_args_len <= i + 2, new_args
-                        rolled_args = new_args[i]
-                        new_args = new_args[:i] + new_args[i + 1 :]
+                if param['kind'] == 'VAR_POSITIONAL' and new_args_len > i:
+                    # For peculiar historical reasons, variable kwargs might show up in args. Thus variable
+                    # positional args is not necessarily the last element of args; it might be the second-to-last.
+                    assert new_args_len <= i + 2, new_args
+                    rolled_args = new_args[i]
+                    new_args = new_args[:i] + new_args[i + 1 :]
                 if param['kind'] == 'VAR_KEYWORD':
                     # As noted above, variable kwargs might show up either in args or in kwargs. If it's in args, it
                     # is necessarily the last element.
@@ -81,7 +80,7 @@ def __substitute_md(k: Optional[str], v: Any) -> Optional[tuple[Optional[str], A
                         rolled_kwargs = kwargs.pop(param['name'])
 
         if rolled_args is not None:
-            assert rolled_args['_classname'] in ('InlineArray', 'InlineList')
+            assert rolled_args['_classname'] in {'InlineArray', 'InlineList'}
             new_args.extend(rolled_args['components'])
         if rolled_kwargs is not None:
             assert rolled_kwargs['_classname'] == 'InlineDict'

pixeltable/metadata/converters/util.py

@@ -34,12 +34,12 @@ def convert_table_md(
     """
     with engine.begin() as conn:
         for row in conn.execute(sql.select(Table)):
-            id = row[0]
+            tbl_id = row[0]
             table_md = row[2]
             assert isinstance(table_md, dict)
             updated_table_md = copy.deepcopy(table_md)
             if table_md_updater is not None:
-                table_md_updater(updated_table_md, id)
+                table_md_updater(updated_table_md, tbl_id)
             if column_md_updater is not None:
                 __update_column_md(updated_table_md, column_md_updater)
             if external_store_md_updater is not None:
@@ -47,19 +47,19 @@ def convert_table_md(
             if substitution_fn is not None:
                 updated_table_md = __substitute_md_rec(updated_table_md, substitution_fn)
             if updated_table_md != table_md:
-                __logger.info(f'Updating schema for table: {id}')
-                conn.execute(sql.update(Table).where(Table.id == id).values(md=updated_table_md))
+                __logger.info(f'Updating schema for table: {tbl_id}')
+                conn.execute(sql.update(Table).where(Table.id == tbl_id).values(md=updated_table_md))
 
         for row in conn.execute(sql.select(Function)):
-            id = row[0]
+            fn_id = row[0]
             function_md = row[2]
             assert isinstance(function_md, dict)
             updated_function_md = copy.deepcopy(function_md)
             if substitution_fn is not None:
                 updated_function_md = __substitute_md_rec(updated_function_md, substitution_fn)
             if updated_function_md != function_md:
-                __logger.info(f'Updating function: {id}')
-                conn.execute(sql.update(Function).where(Function.id == id).values(md=updated_function_md))
+                __logger.info(f'Updating function: {fn_id}')
+                conn.execute(sql.update(Function).where(Function.id == fn_id).values(md=updated_function_md))
 
 
 def __update_column_md(table_md: dict, column_md_updater: Callable[[dict], None]) -> None:

pixeltable/metadata/schema.py

@@ -4,16 +4,14 @@ import uuid
 from typing import Any, Optional, TypeVar, Union, get_type_hints
 
 import sqlalchemy as sql
-import sqlalchemy.orm as orm
-from sqlalchemy import BigInteger, ForeignKey, Integer, LargeBinary
+from sqlalchemy import BigInteger, ForeignKey, Integer, LargeBinary, orm
 from sqlalchemy.dialects.postgresql import JSONB, UUID
-from sqlalchemy.orm import declarative_base
 from sqlalchemy.orm.decl_api import DeclarativeMeta
 
 # Base has to be marked explicitly as a type, in order to be used elsewhere as a type hint. But in addition to being
 # a type, it's also a `DeclarativeMeta`. The following pattern enables us to expose both `Base` and `Base.metadata`
 # outside of the module in a typesafe way.
-Base: type = declarative_base()
+Base: type = orm.declarative_base()
 assert isinstance(Base, DeclarativeMeta)
 base_metadata = Base.metadata
 
@@ -23,7 +21,7 @@ T = TypeVar('T')
 def md_from_dict(data_class_type: type[T], data: Any) -> T:
     """Re-instantiate a dataclass instance that contains nested dataclasses from a dict."""
     if dataclasses.is_dataclass(data_class_type):
-        fieldtypes = {f: t for f, t in get_type_hints(data_class_type).items()}
+        fieldtypes = get_type_hints(data_class_type)
         return data_class_type(**{f: md_from_dict(fieldtypes[f], data[f]) for f in data})  # type: ignore[return-value]
 
     origin = typing.get_origin(data_class_type)
@@ -43,7 +41,7 @@ def md_from_dict(data_class_type: type[T], data: Any) -> T:
         elif origin is tuple:
             return tuple(md_from_dict(arg_type, elem) for arg_type, elem in zip(type_args, data))  # type: ignore[return-value]
         else:
-            assert False
+            raise AssertionError(origin)
     else:
         return data
 
@@ -85,7 +83,7 @@ class Dir(Base):
         UUID(as_uuid=True), primary_key=True, default=uuid.uuid4, nullable=False
     )
     parent_id: orm.Mapped[uuid.UUID] = orm.mapped_column(UUID(as_uuid=True), ForeignKey('dirs.id'), nullable=True)
-    md = sql.Column(JSONB, nullable=False)
+    md: orm.Mapped[dict[str, Any]] = orm.mapped_column(JSONB, nullable=False)  # DirMd
 
 
 @dataclasses.dataclass
@@ -131,13 +129,17 @@ class IndexMd:
     init_args: dict[str, Any]
 
 
+# a stored table version path is a list of (table id as str, effective table version)
+TableVersionPath = list[tuple[str, Optional[int]]]
+
+
 @dataclasses.dataclass
 class ViewMd:
     is_snapshot: bool
     include_base_columns: bool
 
     # (table id, version); for mutable views, all versions are None
-    base_versions: list[tuple[str, Optional[int]]]
+    base_versions: TableVersionPath
 
     # filter predicate applied to the base table; view-only
     predicate: Optional[dict[str, Any]]
@@ -192,7 +194,7 @@ class Table(Base):
 
     id: orm.Mapped[uuid.UUID] = orm.mapped_column(UUID(as_uuid=True), primary_key=True, nullable=False)
     dir_id: orm.Mapped[uuid.UUID] = orm.mapped_column(UUID(as_uuid=True), ForeignKey('dirs.id'), nullable=False)
-    md = sql.Column(JSONB, nullable=False)  # TableMd
+    md: orm.Mapped[dict[str, Any]] = orm.mapped_column(JSONB, nullable=False)  # TableMd
 
 
 @dataclasses.dataclass
@@ -205,9 +207,11 @@ class TableVersionMd:
 
 class TableVersion(Base):
     __tablename__ = 'tableversions'
-    tbl_id = sql.Column(UUID(as_uuid=True), ForeignKey('tables.id'), primary_key=True, nullable=False)
-    version = sql.Column(BigInteger, primary_key=True, nullable=False)
-    md = sql.Column(JSONB, nullable=False)  # TableVersionMd
+    tbl_id: orm.Mapped[uuid.UUID] = orm.mapped_column(
+        UUID(as_uuid=True), ForeignKey('tables.id'), primary_key=True, nullable=False
+    )
+    version: orm.Mapped[int] = orm.mapped_column(BigInteger, primary_key=True, nullable=False)
+    md: orm.Mapped[dict[str, Any]] = orm.mapped_column(JSONB, nullable=False)
 
 
 @dataclasses.dataclass
@@ -246,9 +250,11 @@ class TableSchemaVersionMd:
 class TableSchemaVersion(Base):
     __tablename__ = 'tableschemaversions'
 
-    tbl_id = sql.Column(UUID(as_uuid=True), ForeignKey('tables.id'), primary_key=True, nullable=False)
-    schema_version = sql.Column(BigInteger, primary_key=True, nullable=False)
-    md = sql.Column(JSONB, nullable=False)  # TableSchemaVersionMd
+    tbl_id: orm.Mapped[uuid.UUID] = orm.mapped_column(
+        UUID(as_uuid=True), ForeignKey('tables.id'), primary_key=True, nullable=False
+    )
+    schema_version: orm.Mapped[int] = orm.mapped_column(BigInteger, primary_key=True, nullable=False)
+    md: orm.Mapped[dict[str, Any]] = orm.mapped_column(JSONB, nullable=False)  # TableSchemaVersionMd
 
 
 @dataclasses.dataclass
@@ -271,7 +277,9 @@ class Function(Base):
 
     __tablename__ = 'functions'
 
-    id = sql.Column(UUID(as_uuid=True), primary_key=True, default=uuid.uuid4, nullable=False)
-    dir_id = sql.Column(UUID(as_uuid=True), ForeignKey('dirs.id'), nullable=True)
-    md = sql.Column(JSONB, nullable=False)  # FunctionMd
-    binary_obj = sql.Column(LargeBinary, nullable=True)
+    id: orm.Mapped[uuid.UUID] = orm.mapped_column(
+        UUID(as_uuid=True), primary_key=True, default=uuid.uuid4, nullable=False
+    )
+    dir_id: orm.Mapped[uuid.UUID] = orm.mapped_column(UUID(as_uuid=True), ForeignKey('dirs.id'), nullable=True)
+    md: orm.Mapped[dict[str, Any]] = orm.mapped_column(JSONB, nullable=False)  # FunctionMd
+    binary_obj: orm.Mapped[Optional[bytes]] = orm.mapped_column(LargeBinary, nullable=True)

pixeltable/plan.py

@@ -2,14 +2,15 @@ from __future__ import annotations
 
 import dataclasses
 import enum
+from textwrap import dedent
 from typing import Any, Iterable, Literal, Optional, Sequence
 from uuid import UUID
 
 import sqlalchemy as sql
 
 import pixeltable as pxt
-import pixeltable.exec as exec
-from pixeltable import catalog, exceptions as excs, exprs
+from pixeltable import catalog, exceptions as excs, exec, exprs
+from pixeltable.catalog import Column, TableVersionHandle
 from pixeltable.exec.sql_node import OrderByClause, OrderByItem, combine_order_by_clauses, print_order_by_clause
 
 
@@ -54,9 +55,9 @@ class JoinType(enum.Enum):
     def validated(cls, name: str, error_prefix: str) -> JoinType:
         try:
             return cls[name.upper()]
-        except KeyError:
-            val_strs = ', '.join(f'{s.lower()!r}' for s in cls.__members__.keys())
-            raise excs.Error(f'{error_prefix} must be one of: [{val_strs}]')
+        except KeyError as exc:
+            val_strs = ', '.join(f'{s.lower()!r}' for s in cls.__members__)
+            raise excs.Error(f'{error_prefix} must be one of: [{val_strs}]') from exc
 
 
 @dataclasses.dataclass
@@ -177,19 +178,21 @@ class Analyzer:
             )
 
         # check that Where clause and filter doesn't contain aggregates
-        if self.sql_where_clause is not None:
-            if any(_is_agg_fn_call(e) for e in self.sql_where_clause.subexprs(expr_class=exprs.FunctionCall)):
-                raise excs.Error(f'where() cannot contain aggregate functions: {self.sql_where_clause}')
-        if self.filter is not None:
-            if any(_is_agg_fn_call(e) for e in self.filter.subexprs(expr_class=exprs.FunctionCall)):
-                raise excs.Error(f'where() cannot contain aggregate functions: {self.filter}')
+        if self.sql_where_clause is not None and any(
+            _is_agg_fn_call(e) for e in self.sql_where_clause.subexprs(expr_class=exprs.FunctionCall)
+        ):
+            raise excs.Error(f'where() cannot contain aggregate functions: {self.sql_where_clause}')
+        if self.filter is not None and any(
+            _is_agg_fn_call(e) for e in self.filter.subexprs(expr_class=exprs.FunctionCall)
+        ):
+            raise excs.Error(f'where() cannot contain aggregate functions: {self.filter}')
 
         # check that grouping exprs don't contain aggregates and can be expressed as SQL (we perform sort-based
         # aggregation and rely on the SqlScanNode returning data in the correct order)
         for e in self.group_by_clause:
             if not self.sql_elements.contains(e):
                 raise excs.Error(f'Invalid grouping expression, needs to be expressible in SQL: {e}')
-            if e._contains(filter=lambda e: _is_agg_fn_call(e)):
+            if e._contains(filter=_is_agg_fn_call):
                 raise excs.Error(f'Grouping expression contains aggregate function: {e}')
 
     def _determine_agg_status(self, e: exprs.Expr, grouping_expr_ids: set[int]) -> tuple[bool, bool]:
@@ -207,7 +210,7 @@ class Analyzer:
             return True, False
         elif isinstance(e, exprs.Literal):
             return True, True
-        elif isinstance(e, exprs.ColumnRef) or isinstance(e, exprs.RowidRef):
+        elif isinstance(e, (exprs.ColumnRef, exprs.RowidRef)):
             # we already know that this isn't a grouping expr
             return False, True
         else:
@@ -275,14 +278,19 @@ class Planner:
         cls, tbl: catalog.TableVersion, rows: list[dict[str, Any]], ignore_errors: bool
     ) -> exec.ExecNode:
         """Creates a plan for TableVersion.insert()"""
-        assert not tbl.is_view()
+        assert not tbl.is_view
         # stored_cols: all cols we need to store, incl computed cols (and indices)
         stored_cols = [c for c in tbl.cols_by_id.values() if c.is_stored]
         assert len(stored_cols) > 0  # there needs to be something to store
+
+        cls.__check_valid_columns(tbl, stored_cols, 'inserted into')
+
         row_builder = exprs.RowBuilder([], stored_cols, [])
 
         # create InMemoryDataNode for 'rows'
-        plan: exec.ExecNode = exec.InMemoryDataNode(tbl, rows, row_builder, tbl.next_rowid)
+        plan: exec.ExecNode = exec.InMemoryDataNode(
+            TableVersionHandle(tbl.id, tbl.effective_version), rows, row_builder, tbl.next_rowid
+        )
 
         media_input_col_info = [
             exprs.ColumnSlotIdx(col_ref.col, col_ref.slot_idx)
@@ -318,7 +326,7 @@ class Planner:
     def create_df_insert_plan(
         cls, tbl: catalog.TableVersion, df: 'pxt.DataFrame', ignore_errors: bool
     ) -> exec.ExecNode:
-        assert not tbl.is_view()
+        assert not tbl.is_view
         plan = df._create_query_plan()  # ExecNode constructed by the DataFrame
 
         # Modify the plan RowBuilder to register the output columns
@@ -363,7 +371,7 @@ class Planner:
         """
         # retrieve all stored cols and all target exprs
         assert isinstance(tbl, catalog.TableVersionPath)
-        target = tbl.tbl_version  # the one we need to update
+        target = tbl.tbl_version.get()  # the one we need to update
         updated_cols = list(update_targets.keys())
         if len(recompute_targets) > 0:
             recomputed_cols = set(recompute_targets)
@@ -374,11 +382,14 @@ class Planner:
             recomputed_cols.update(idx_val_cols)
             # we only need to recompute stored columns (unstored ones are substituted away)
             recomputed_cols = {c for c in recomputed_cols if c.is_stored}
-        recomputed_base_cols = {col for col in recomputed_cols if col.tbl == target}
+
+        cls.__check_valid_columns(tbl.tbl_version.get(), recomputed_cols, 'updated in')
+
+        recomputed_base_cols = {col for col in recomputed_cols if col.tbl == tbl.tbl_version}
         copied_cols = [
             col
             for col in target.cols_by_id.values()
-            if col.is_stored and not col in updated_cols and not col in recomputed_base_cols
+            if col.is_stored and col not in updated_cols and col not in recomputed_base_cols
         ]
         select_list: list[exprs.Expr] = [exprs.ColumnRef(col) for col in copied_cols]
         select_list.extend(update_targets.values())
@@ -398,7 +409,25 @@ class Planner:
         for i, col in enumerate(all_base_cols):
             plan.row_builder.add_table_column(col, select_list[i].slot_idx)
         recomputed_user_cols = [c for c in recomputed_cols if c.name is not None]
-        return plan, [f'{c.tbl.name}.{c.name}' for c in updated_cols + recomputed_user_cols], recomputed_user_cols
+        return plan, [f'{c.tbl.get().name}.{c.name}' for c in updated_cols + recomputed_user_cols], recomputed_user_cols
+
+    @classmethod
+    def __check_valid_columns(
+        cls, tbl: catalog.TableVersion, cols: Iterable[Column], op_name: Literal['inserted into', 'updated in']
+    ) -> None:
+        for col in cols:
+            if col.value_expr is not None and not col.value_expr.is_valid:
+                raise excs.Error(
+                    dedent(
+                        f"""
+                        Data cannot be {op_name} the table {tbl.name!r},
+                        because the column {col.name!r} is currently invalid:
+                        {{validation_error}}
+                        """
+                    )
+                    .strip()
+                    .format(validation_error=col.value_expr.validation_error)
+                )
 
     @classmethod
     def create_batch_update_plan(
@@ -417,7 +446,7 @@ class Planner:
         - list of user-visible columns that are being recomputed
         """
         assert isinstance(tbl, catalog.TableVersionPath)
-        target = tbl.tbl_version  # the one we need to update
+        target = tbl.tbl_version.get()  # the one we need to update
         sa_key_cols: list[sql.Column] = []
         key_vals: list[tuple] = []
         if len(rowids) > 0:
@@ -440,7 +469,7 @@ class Planner:
         copied_cols = [
             col
             for col in target.cols_by_id.values()
-            if col.is_stored and not col in updated_cols and not col in recomputed_base_cols
+            if col.is_stored and col not in updated_cols and col not in recomputed_base_cols
         ]
         select_list: list[exprs.Expr] = [exprs.ColumnRef(col) for col in copied_cols]
         select_list.extend(exprs.ColumnRef(col) for col in updated_cols)
@@ -507,11 +536,11 @@ class Planner:
             - list of columns that are being recomputed
         """
         assert isinstance(view, catalog.TableVersionPath)
-        assert view.is_view()
-        target = view.tbl_version  # the one we need to update
+        assert view.is_view
+        target = view.tbl_version.get()  # the one we need to update
         # retrieve all stored cols and all target exprs
         recomputed_cols = set(recompute_targets.copy())
-        copied_cols = [col for col in target.cols_by_id.values() if col.is_stored and not col in recomputed_cols]
+        copied_cols = [col for col in target.cols_by_id.values() if col.is_stored and col not in recomputed_cols]
         select_list: list[exprs.Expr] = [exprs.ColumnRef(col) for col in copied_cols]
         # resolve recomputed exprs to stored columns in the base
         recomputed_exprs = [
@@ -551,13 +580,13 @@ class Planner:
             - number of materialized values per row
         """
         assert isinstance(view, catalog.TableVersionPath)
-        assert view.is_view()
+        assert view.is_view
         # things we need to materialize as DataRows:
         # 1. stored computed cols
         # - iterator columns are effectively computed, just not with a value_expr
         # - we can ignore stored non-computed columns because they have a default value that is supplied directly by
         #   the store
-        target = view.tbl_version  # the one we need to populate
+        target = view.tbl_version.get()  # the one we need to populate
         stored_cols = [c for c in target.cols_by_id.values() if c.is_stored]
         # 2. for component views: iterator args
         iterator_args = [target.iterator_args] if target.iterator_args is not None else []
@@ -585,8 +614,8 @@ class Planner:
             exact_version_only=view.get_bases() if propagates_insert else [],
         )
         exec_ctx = plan.ctx
-        if target.is_component_view():
-            plan = exec.ComponentIterationNode(target, plan)
+        if target.is_component_view:
+            plan = exec.ComponentIterationNode(view.tbl_version, plan)
         if len(view_output_exprs) > 0:
             plan = exec.ExprEvalNode(
                 row_builder, output_exprs=view_output_exprs, input_exprs=base_output_exprs, input=plan
@@ -639,11 +668,12 @@ class Planner:
     @classmethod
     def _is_contained_in(cls, l1: Iterable[exprs.Expr], l2: Iterable[exprs.Expr]) -> bool:
         """Returns True if l1 is contained in l2"""
-        s1, s2 = set(e.id for e in l1), set(e.id for e in l2)
-        return s1 <= s2
+        return {e.id for e in l1} <= {e.id for e in l2}
 
     @classmethod
-    def _insert_prefetch_node(cls, tbl_id: UUID, row_builder: exprs.RowBuilder, input: exec.ExecNode) -> exec.ExecNode:
+    def _insert_prefetch_node(
+        cls, tbl_id: UUID, row_builder: exprs.RowBuilder, input_node: exec.ExecNode
+    ) -> exec.ExecNode:
         """Returns a CachePrefetchNode into the plan if needed, otherwise returns input"""
         # we prefetch external files for all media ColumnRefs, even those that aren't part of the dependencies
         # of output_exprs: if unstored iterator columns are present, we might need to materialize ColumnRefs that
@@ -652,10 +682,10 @@ class Planner:
             e for e in list(row_builder.unique_exprs) if isinstance(e, exprs.ColumnRef) and e.col_type.is_media_type()
         ]
         if len(media_col_refs) == 0:
-            return input
+            return input_node
         # we need to prefetch external files for media column types
         file_col_info = [exprs.ColumnSlotIdx(e.col, e.slot_idx) for e in media_col_refs]
-        prefetch_node = exec.CachePrefetchNode(tbl_id, file_col_info, input)
+        prefetch_node = exec.CachePrefetchNode(tbl_id, file_col_info, input_node)
         return prefetch_node
 
     @classmethod
@@ -668,7 +698,7 @@ class Planner:
         order_by_clause: Optional[list[tuple[exprs.Expr, bool]]] = None,
         limit: Optional[exprs.Expr] = None,
         ignore_errors: bool = False,
-        exact_version_only: Optional[list[catalog.TableVersion]] = None,
+        exact_version_only: Optional[list[catalog.TableVersionHandle]] = None,
     ) -> exec.ExecNode:
         """Return plan for executing a query.
         Updates 'select_list' in place to make it executable.
@@ -714,7 +744,7 @@ class Planner:
         eval_ctx: exprs.RowBuilder.EvalCtx,
         limit: Optional[exprs.Expr] = None,
         with_pk: bool = False,
-        exact_version_only: Optional[list[catalog.TableVersion]] = None,
+        exact_version_only: Optional[list[catalog.TableVersionHandle]] = None,
     ) -> exec.ExecNode:
         """
         Create plan to materialize eval_ctx.
@@ -752,13 +782,11 @@ class Planner:
         )
         if analyzer.filter is not None:
             candidates.extend(
-                exprs.Expr.subexprs(analyzer.filter, filter=lambda e: sql_elements.contains(e), traverse_matches=False)
+                exprs.Expr.subexprs(analyzer.filter, filter=sql_elements.contains, traverse_matches=False)
             )
         if is_python_agg and analyzer.group_by_clause is not None:
             candidates.extend(
-                exprs.Expr.list_subexprs(
-                    analyzer.group_by_clause, filter=lambda e: sql_elements.contains(e), traverse_matches=False
-                )
+                exprs.Expr.list_subexprs(analyzer.group_by_clause, filter=sql_elements.contains, traverse_matches=False)
             )
         # not isinstance(...): we don't want to materialize Literals via a Select
         sql_exprs = exprs.ExprSet(e for e in candidates if not isinstance(e, exprs.Literal))

pixeltable/share/__init__.py

@@ -1 +1,3 @@
+# ruff: noqa: F401
+
 from .publish import publish_snapshot

pixeltable/share/packager.py

@@ -66,13 +66,15 @@ class TablePackager:
                 'tables': [
                     {
                         'table_id': str(t._tbl_version.id),
-                        # These are temporary; will replace with a better solution once the concurrency changes to catalog have
-                        # been merged
-                        'table_md': dataclasses.asdict(t._tbl_version._create_tbl_md()),
+                        # These are temporary; will replace with a better solution once the concurrency
+                        # changes to catalog have been merged
+                        'table_md': dataclasses.asdict(t._tbl_version.get()._create_tbl_md()),
                         'table_version_md': dataclasses.asdict(
-                            t._tbl_version._create_version_md(datetime.now().timestamp())
+                            t._tbl_version.get()._create_version_md(datetime.now().timestamp())
+                        ),
+                        'table_schema_version_md': dataclasses.asdict(
+                            t._tbl_version.get()._create_schema_version_md(0)
                         ),
-                        'table_schema_version_md': dataclasses.asdict(t._tbl_version._create_schema_version_md(0)),
                     }
                     for t in (table, *table._bases)
                 ]
@@ -91,11 +93,12 @@ class TablePackager:
         with open(self.tmp_dir / 'metadata.json', 'w', encoding='utf8') as fp:
             json.dump(self.md, fp)
         self.iceberg_catalog = sqlite_catalog(self.tmp_dir / 'warehouse')
-        ancestors = (self.table, *self.table._bases)
-        for t in ancestors:
-            _logger.info(f"Exporting table '{t._path}'.")
-            self.__export_table(t)
-        _logger.info(f'Building archive.')
+        with Env.get().begin_xact():
+            ancestors = (self.table, *self.table._bases)
+            for t in ancestors:
+                _logger.info(f"Exporting table '{t._path}'.")
+                self.__export_table(t)
+        _logger.info('Building archive.')
         bundle_path = self.__build_tarball()
         _logger.info(f'Packaging complete: {bundle_path}')
         return bundle_path
@@ -117,7 +120,7 @@ class TablePackager:
         # to get the column types, since we'll be substituting `fileurl`s for media columns.
         actual_col_types: list[ts.ColumnType] = []
 
-        for col_name, col in t._tbl_version.cols_by_name.items():
+        for col_name, col in t._tbl_version.get().cols_by_name.items():
             if not col.is_stored:
                 continue
             if col.col_type.is_media_type():
@@ -150,7 +153,7 @@ class TablePackager:
         """
         Iceberg tables must have a namespace, which cannot be the empty string, so we prepend `pxt` to the table path.
         """
-        parent_path = table._parent._path
+        parent_path = table._parent()._path()
         if len(parent_path) == 0:
             return 'pxt'
         else:

pixeltable/share/publish.py

@@ -1,16 +1,14 @@
-import dataclasses
 import os
 import sys
 import urllib.parse
 import urllib.request
-from datetime import datetime
 from pathlib import Path
 
 import requests
 from tqdm import tqdm
 
 import pixeltable as pxt
-from pixeltable import exceptions as excs, metadata
+from pixeltable import exceptions as excs
 from pixeltable.env import Env
 from pixeltable.utils import sha256sum
 
@@ -46,7 +44,7 @@ def publish_snapshot(dest_tbl_uri: str, src_tbl: pxt.Table) -> str:
     else:
         raise excs.Error(f'Unsupported destination: {destination_uri}')
 
-    Env.get().console_logger.info(f'Finalizing snapshot ...')
+    Env.get().console_logger.info('Finalizing snapshot ...')
 
     finalize_request_json = {
         'upload_id': upload_id,
@@ -83,7 +81,7 @@ def _upload_bundle_to_s3(bundle: Path, parsed_location: urllib.parse.ParseResult
     upload_args = {'ChecksumAlgorithm': 'SHA256'}
 
     progress_bar = tqdm(
-        desc=f'Uploading',
+        desc='Uploading',
         total=bundle.stat().st_size,
         unit='B',
         unit_scale=True,