datachain 0.24.1__tar.gz → 0.24.2__tar.gz

{datachain-0.24.1 → datachain-0.24.2}/.pre-commit-config.yaml

@@ -24,7 +24,7 @@ repos:
       - id: trailing-whitespace
         exclude: '^LICENSES/'
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: 'v0.12.0'
+    rev: 'v0.12.1'
     hooks:
       - id: ruff
         args: [--fix, --exit-non-zero-on-fix]
@@ -35,7 +35,7 @@ repos:
       - id: codespell
         additional_dependencies: ["tomli"]
   - repo: https://github.com/macisamuele/language-formatters-pre-commit-hooks
-    rev: v2.14.0
+    rev: v2.15.0
     hooks:
       - id: pretty-format-toml
         args: [--autofix, --no-sort]

{datachain-0.24.1 → datachain-0.24.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datachain
-Version: 0.24.1
+Version: 0.24.2
 Summary: Wrangle unstructured AI data at scale
 Author-email: Dmitry Petrov <support@dvc.org>
 License-Expression: Apache-2.0

{datachain-0.24.1 → datachain-0.24.2}/src/datachain/data_storage/warehouse.py

@@ -218,7 +218,7 @@ class AbstractWarehouse(ABC, Serializable):
         limit = query._limit
         paginated_query = query.limit(page_size)
 
-        offset = 0
+        offset = query._offset or 0
         num_yielded = 0
 
         # Ensure we're using a thread-local connection
@@ -234,13 +234,13 @@ class AbstractWarehouse(ABC, Serializable):
                 # Cursor results are not thread-safe, so we convert them to a list
                 results = list(wh.dataset_rows_select(paginated_query.offset(offset)))
 
-                processed = False
+                processed = 0
                 for row in results:
-                    processed = True
+                    processed += 1
                     yield row
                     num_yielded += 1
 
-                if not processed:
+                if processed < page_size:
                     break  # no more results
                 offset += page_size
 
@@ -343,6 +343,8 @@ class AbstractWarehouse(ABC, Serializable):
         if (id_col := get_query_id_column(query)) is None:
             raise RuntimeError("sys__id column not found in query")
 
+        query = query._clone().offset(None).limit(None).order_by(None)
+
         if is_batched:
             for batch in ids:
                 yield list(self.dataset_rows_select(query.where(id_col.in_(batch))))

{datachain-0.24.1 → datachain-0.24.2}/src/datachain/query/dataset.py

@@ -11,6 +11,7 @@ from collections.abc import Generator, Iterable, Iterator, Sequence
 from copy import copy
 from functools import wraps
 from secrets import token_hex
+from types import GeneratorType
 from typing import (
     TYPE_CHECKING,
     Any,
@@ -557,8 +558,8 @@ class UDFStep(Step, ABC):
         """
         assert self.partition_by is not None
 
-        if isinstance(self.partition_by, Sequence):
-            list_partition_by = self.partition_by
+        if isinstance(self.partition_by, (list, tuple, GeneratorType)):
+            list_partition_by = list(self.partition_by)
         else:
             list_partition_by = [self.partition_by]
 
@@ -575,7 +576,10 @@ class UDFStep(Step, ABC):
             f.dense_rank().over(order_by=partition_by).label(PARTITION_COLUMN_ID),
         ]
         self.catalog.warehouse.db.execute(
-            tbl.insert().from_select(cols, query.with_only_columns(*cols))
+            tbl.insert().from_select(
+                cols,
+                query.offset(None).limit(None).with_only_columns(*cols),
+            )
         )
 
         return tbl
@@ -601,13 +605,10 @@ class UDFStep(Step, ABC):
         if self.partition_by is not None:
             partition_tbl = self.create_partitions_table(query)
             temp_tables.append(partition_tbl.name)
-
-            subq = query.subquery()
-            query = (
-                sqlalchemy.select(*subq.c)
-                .outerjoin(partition_tbl, partition_tbl.c.sys__id == subq.c.sys__id)
-                .add_columns(*partition_columns())
-            )
+            query = query.outerjoin(
+                partition_tbl,
+                partition_tbl.c.sys__id == query.selected_columns.sys__id,
+            ).add_columns(*partition_columns())
 
         query, tables = self.process_input_query(query)
         temp_tables.extend(t.name for t in tables)

{datachain-0.24.1 → datachain-0.24.2}/src/datachain.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: datachain
-Version: 0.24.1
+Version: 0.24.2
 Summary: Wrangle unstructured AI data at scale
 Author-email: Dmitry Petrov <support@dvc.org>
 License-Expression: Apache-2.0

{datachain-0.24.1 → datachain-0.24.2}/tests/func/test_datachain.py

@@ -2324,3 +2324,72 @@ def test_agg(catalog_tmpfile, parallel):
         ],
         "parent",
     )
+
+
+@pytest.mark.parametrize("parallel", [1, 2])
+@pytest.mark.parametrize(
+    "offset,limit,files",
+    [
+        (None, 1000, [f"file{i:02d}" for i in range(100)]),
+        (None, 3, ["file00", "file01", "file02"]),
+        (0, 3, ["file00", "file01", "file02"]),
+        (97, 1000, ["file97", "file98", "file99"]),
+        (1, 2, ["file01", "file02"]),
+        (50, 3, ["file50", "file51", "file52"]),
+        (None, 0, []),
+        (50, 0, []),
+    ],
+)
+def test_agg_offset_limit(catalog_tmpfile, parallel, offset, limit, files):
+    def process(filename: list[str]) -> Iterator[tuple[str, int]]:
+        yield filename[0], len(filename)
+
+    ds = dc.read_values(
+        filename=[f"file{i:02d}" for i in range(100)],
+        value=list(range(100)),
+        session=catalog_tmpfile.session,
+    )
+    if offset is not None:
+        ds = ds.offset(offset)
+    if limit is not None:
+        ds = ds.limit(limit)
+    ds = (
+        ds.settings(parallel=parallel)
+        .agg(
+            process,
+            output={"filename": str, "count": int},
+            partition_by="filename",
+        )
+        .save("my-ds")
+    )
+
+    records = list(ds.to_records())
+    assert len(records) == len(files)
+    assert all(row["count"] == 1 for row in records)
+    assert sorted(row["filename"] for row in records) == sorted(files)
+
+
+@pytest.mark.parametrize("parallel", [1, 2])
+@pytest.mark.parametrize("sample", [0, 1, 3, 10, 50, 100])
+def test_agg_sample(catalog_tmpfile, parallel, sample):
+    def process(filename: list[str]) -> Iterator[tuple[str, int]]:
+        yield filename[0], len(filename)
+
+    ds = (
+        dc.read_values(
+            filename=[f"file{i:02d}" for i in range(100)],
+            session=catalog_tmpfile.session,
+        )
+        .sample(sample)
+        .settings(parallel=parallel)
+        .agg(
+            process,
+            output={"filename": str, "count": int},
+            partition_by="filename",
+        )
+        .save("my-ds")
+    )
+
+    records = list(ds.to_records())
+    assert len(records) == sample
+    assert all(row["count"] == 1 for row in records)