arize-phoenix 7.8.0__py3-none-any.whl → 7.8.1__py3-none-any.whl

{arize_phoenix-7.8.0.dist-info → arize_phoenix-7.8.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arize-phoenix
-Version: 7.8.0
+Version: 7.8.1
 Summary: AI Observability and Evaluation
 Project-URL: Documentation, https://docs.arize.com/phoenix/
 Project-URL: Issues, https://github.com/Arize-ai/phoenix/issues

{arize_phoenix-7.8.0.dist-info → arize_phoenix-7.8.1.dist-info}/RECORD

@@ -6,7 +6,7 @@ phoenix/exceptions.py,sha256=n2L2KKuecrdflB9MsCdAYCiSEvGJptIsfRkXMoJle7A,169
 phoenix/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 phoenix/services.py,sha256=kpW1WL0kiB8XJsO6XycvZVJ-lBkNoenhQ7atCvBoSe8,5365
 phoenix/settings.py,sha256=ht-0oN-sMV6SPXrk7Tu1EZlngpAYkGNLYPhO8DyrdQI,661
-phoenix/version.py,sha256=vZw91QDi0kVrQT2yAQVzBUsHnFXKxQOJTwxJLeLNYNo,22
+phoenix/version.py,sha256=q9HCYvd_KKYPNhrFJZNCqDLkLzTKEm20j2_jzE7u-Lw,22
 phoenix/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/core/embedding_dimension.py,sha256=zKGbcvwOXgLf-yrJBpQyKtd-LEOPRKHnUToyAU8Owis,87
 phoenix/core/model.py,sha256=qBFraOtmwCCnWJltKNP18DDG0mULXigytlFsa6YOz6k,4837
@@ -102,7 +102,7 @@ phoenix/server/api/utils.py,sha256=quCBRcusc6PUq9tJq7M8PgwFZp7nXgVAxtbw8feribY,8
 phoenix/server/api/dataloaders/__init__.py,sha256=nt5Lbs34q05GSvNCT2IGrkRW0kg65L8h3bZelzgpkYg,3909
 phoenix/server/api/dataloaders/annotation_summaries.py,sha256=2sHmIDX7n8tuPeBTs9bMKtlMKWn_Ph9awTZqmwn2Owc,5505
 phoenix/server/api/dataloaders/average_experiment_run_latency.py,sha256=GLFoFAbztOH-0FVzzZ77mATIO63UcjB50j3qXiNi-tE,1811
-phoenix/server/api/dataloaders/dataset_example_revisions.py,sha256=YG2VSWOkM7q_-RO8OwE8bHYX8d3_OH34SeaD4pvEt0w,5115
+phoenix/server/api/dataloaders/dataset_example_revisions.py,sha256=xF7M2dg3UmjhdCrscnztCIBBI0cg3RF48IIqvilpc18,4623
 phoenix/server/api/dataloaders/dataset_example_spans.py,sha256=z_MFquqAcJ9wat7BBp7MVeJ9BYuu2EZEdaog52iWDno,1390
 phoenix/server/api/dataloaders/document_evaluation_summaries.py,sha256=9fdROnzp-mymggHwNvpRkCk93LUFxxLy55-j3HP_2HY,5565
 phoenix/server/api/dataloaders/document_evaluations.py,sha256=KCnCItJ2DQOCHvxFp1KK2AStPN1akGlAGOEzEAu6c6I,1246
@@ -337,9 +337,9 @@ phoenix/utilities/project.py,sha256=auVpARXkDb-JgeX5f2aStyFIkeKvGwN9l7qrFeJMVxI,
 phoenix/utilities/re.py,sha256=x8Xbk-Wa6qDMAtUd_7JtZvKtrYEuMY-bchB0n163_5c,2006
 phoenix/utilities/span_store.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/utilities/template_formatters.py,sha256=gh9PJD6WEGw7TEYXfSst1UR4pWWwmjxMLrDVQ_CkpkQ,2779
-arize_phoenix-7.8.0.dist-info/METADATA,sha256=EM8prb9RNrEFDt2ZmeTjMnwammFcOhmf-LMhDooHnYU,23052
-arize_phoenix-7.8.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-arize_phoenix-7.8.0.dist-info/entry_points.txt,sha256=Pgpn8Upxx9P8z8joPXZWl2LlnAlGc3gcQoVchb06X1Q,94
-arize_phoenix-7.8.0.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
-arize_phoenix-7.8.0.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
-arize_phoenix-7.8.0.dist-info/RECORD,,
+arize_phoenix-7.8.1.dist-info/METADATA,sha256=saLGcoxnbaN2fat0opLf_ayMiIPtYHf7LgmOW-4RBjk,23052
+arize_phoenix-7.8.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+arize_phoenix-7.8.1.dist-info/entry_points.txt,sha256=Pgpn8Upxx9P8z8joPXZWl2LlnAlGc3gcQoVchb06X1Q,94
+arize_phoenix-7.8.1.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
+arize_phoenix-7.8.1.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
+arize_phoenix-7.8.1.dist-info/RECORD,,

phoenix/server/api/dataloaders/dataset_example_revisions.py

@@ -1,6 +1,6 @@
 from typing import Optional, Union
 
-from sqlalchemy import and_, case, func, null, or_, select
+from sqlalchemy import Integer, case, func, or_, select, union
 from sqlalchemy.sql.expression import literal
 from strawberry.dataloader import DataLoader
 from typing_extensions import TypeAlias
@@ -20,74 +20,50 @@ class DatasetExampleRevisionsDataLoader(DataLoader[Key, Result]):
     def __init__(self, db: DbSessionFactory) -> None:
         super().__init__(
             load_fn=self._load_fn,
-            max_batch_size=200,  # needed to prevent the size of the query from getting too large
+            # Setting max_batch_size to prevent the size of the query from getting too large.
+            # The maximum number of terms is SQLITE_MAX_COMPOUND_SELECT which defaults to 500.
+            # This is needed because of the compound select query below used in transferring
+            # the input data to the database. SQLite in fact has better ways to transfer data,
+            # but unfortunately they're not made available in sqlalchemy yet.
+            max_batch_size=200,
         )
         self._db = db
 
     async def _load_fn(self, keys: list[Key]) -> list[Union[Result, NotFound]]:
-        example_and_version_ids = tuple(
-            set(
-                (example_id, version_id)
-                for example_id, version_id in keys
-                if version_id is not None
-            )
-        )
-        versionless_example_ids = tuple(
-            set(example_id for example_id, version_id in keys if version_id is None)
-        )
-        resolved_example_and_version_ids = (
-            (
+        # sqlalchemy has limited SQLite support for VALUES, so use UNION ALL instead.
+        # For details, see https://github.com/sqlalchemy/sqlalchemy/issues/7228
+        keys_subquery = union(
+            *(
                 select(
-                    models.DatasetExample.id.label("example_id"),
-                    models.DatasetVersion.id.label("version_id"),
-                )
-                .select_from(models.DatasetExample)
-                .join(
-                    models.DatasetVersion,
-                    onclause=literal(True),  # cross join
-                )
-                .where(
-                    or_(
-                        *(
-                            and_(
-                                models.DatasetExample.id == example_id,
-                                models.DatasetVersion.id == version_id,
-                            )
-                            for example_id, version_id in example_and_version_ids
-                        )
-                    )
+                    literal(example_id, Integer).label("example_id"),
+                    literal(version_id, Integer).label("version_id"),
                 )
+                for example_id, version_id in keys
             )
-            .union(
-                select(
-                    models.DatasetExample.id.label("example_id"), null().label("version_id")
-                ).where(models.DatasetExample.id.in_(versionless_example_ids))
-            )
-            .subquery()
-        )
+        ).subquery()
         revision_ids = (
             select(
-                resolved_example_and_version_ids.c.example_id,
-                resolved_example_and_version_ids.c.version_id,
+                keys_subquery.c.example_id,
+                keys_subquery.c.version_id,
                 func.max(models.DatasetExampleRevision.id).label("revision_id"),
             )
-            .select_from(resolved_example_and_version_ids)
+            .select_from(keys_subquery)
             .join(
                 models.DatasetExampleRevision,
-                onclause=resolved_example_and_version_ids.c.example_id
+                onclause=keys_subquery.c.example_id
                 == models.DatasetExampleRevision.dataset_example_id,
             )
             .where(
                 or_(
-                    resolved_example_and_version_ids.c.version_id.is_(None),
-                    models.DatasetExampleRevision.dataset_version_id
-                    <= resolved_example_and_version_ids.c.version_id,
+                    # This query gets the latest `revision_id` for each example:
+                    # - If `version_id` is NOT given, it finds the maximum `revision_id`.
+                    # - If `version_id` is given, it finds the highest `revision_id` whose
+                    #   `version_id` is less than or equal to the one specified.
+                    keys_subquery.c.version_id.is_(None),
+                    models.DatasetExampleRevision.dataset_version_id <= keys_subquery.c.version_id,
                 )
             )
-            .group_by(
-                resolved_example_and_version_ids.c.example_id,
-                resolved_example_and_version_ids.c.version_id,
-            )
+            .group_by(keys_subquery.c.example_id, keys_subquery.c.version_id)
         ).subquery()
         query = (
             select(

phoenix/version.py

@@ -1 +1 @@
-__version__ = "7.8.0"
+__version__ = "7.8.1"