arize-phoenix 4.15.0__py3-none-any.whl → 4.16.0__py3-none-any.whl

{arize_phoenix-4.15.0.dist-info → arize_phoenix-4.16.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: arize-phoenix
-Version: 4.15.0
+Version: 4.16.0
 Summary: AI Observability and Evaluation
 Project-URL: Documentation, https://docs.arize.com/phoenix/
 Project-URL: Issues, https://github.com/Arize-ai/phoenix/issues
@@ -72,6 +72,7 @@ Requires-Dist: anthropic; extra == 'dev'
 Requires-Dist: arize[autoembeddings,llm-evaluation]; extra == 'dev'
 Requires-Dist: asgi-lifespan; extra == 'dev'
 Requires-Dist: asyncpg; extra == 'dev'
+Requires-Dist: faker>=26.0.0; extra == 'dev'
 Requires-Dist: gcsfs; extra == 'dev'
 Requires-Dist: google-cloud-aiplatform>=1.3; extra == 'dev'
 Requires-Dist: hatch; extra == 'dev'

{arize_phoenix-4.15.0.dist-info → arize_phoenix-4.16.0.dist-info}/RECORD

@@ -5,7 +5,7 @@ phoenix/exceptions.py,sha256=n2L2KKuecrdflB9MsCdAYCiSEvGJptIsfRkXMoJle7A,169
 phoenix/py.typed,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
 phoenix/services.py,sha256=aTxhcOA1pZHB6U-B3TEcp6fqDF5oT0xCUvEUNMZVTUQ,5175
 phoenix/settings.py,sha256=cO-qgis_S27nHirTobYI9hHPfZH18R--WMmxNdsVUwc,273
-phoenix/version.py,sha256=n9j59G6rZwNZUyNaaFWTiRnQ1XtoroCAMfk4OtFuagc,23
+phoenix/version.py,sha256=jqZFarxQPRfyIPY1ojpY1Nm-BNzQbn-cIh6gBXR-dN8,23
 phoenix/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/core/embedding_dimension.py,sha256=zKGbcvwOXgLf-yrJBpQyKtd-LEOPRKHnUToyAU8Owis,87
 phoenix/core/model.py,sha256=km_a--PBHOuA337ClRw9xqhOHhrUT6Rl9pz_zV0JYkQ,4843
@@ -14,16 +14,21 @@ phoenix/core/model_schema_adapter.py,sha256=0Tm_Y_gV-WED8fKBCaFXAEFwE3CTEZS1dowq
 phoenix/db/README.md,sha256=IvKaZyf9ECbGBYYePaRhBveKZwDbxAc-c7BMxJYZh6Q,595
 phoenix/db/__init__.py,sha256=pDjEFXukHmJBM-1D8RjmXkvLsz85YWNxMQczt81ec3A,118
 phoenix/db/alembic.ini,sha256=p8DjVqGUs_tTx8oU56JP7qj-rMUebNFizItUSv_hPhs,3763
-phoenix/db/bulk_inserter.py,sha256=Hagqne0MiHXi9Y1uVsXpzIqNF9HpSZcmqv00TuHChTc,11285
+phoenix/db/bulk_inserter.py,sha256=kBgR-4kr8LnqrN03UfTtXYOhwMf7IxUks4OViGo84h8,16418
 phoenix/db/engines.py,sha256=R3btYTSOSd6BwRA59EmhhojL0HCQ7NnzFIXQrPYS0iU,4812
-phoenix/db/helpers.py,sha256=mTBhPzdy_aU9gD7hNzUZJkAnV77ko5CdaXyoWH3snPA,2982
+phoenix/db/helpers.py,sha256=2zSc4n5IJfu-CaOFoBfqTB35M1nTFcAc8tqLsNtF2Jw,3488
 phoenix/db/migrate.py,sha256=MuhtNWnR24riROvarvKfbRb4_D5xuQi6P760vBUKl1E,2270
 phoenix/db/models.py,sha256=7DBWbxY3cx3ve2P1I0kkDKXzlt04zEFJuRPJWsVpH-I,20422
 phoenix/db/insertion/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+phoenix/db/insertion/constants.py,sha256=oLtCd7KxC6nQlGkhUoguF5PlomEVoZTddxc-huD3f2A,70
 phoenix/db/insertion/dataset.py,sha256=_vxy5e6W5jEuvO2fMKbbNCn9JvHkwI4LRKk_10eKFVg,7171
+phoenix/db/insertion/document_annotation.py,sha256=vMP1OkQ0GOC57aHLTErwF6cu6qHdBkjQFXbCwUtYYAs,5409
 phoenix/db/insertion/evaluation.py,sha256=SoI85N3MYUSeNgjKa5WzFw14OfNjNTjExv-2m3sxaR8,6371
-phoenix/db/insertion/helpers.py,sha256=v8ONIWZu_x1XNWKmz3gNu3839LqlZbFv4gyAYrLKMts,3068
+phoenix/db/insertion/helpers.py,sha256=z_Wnckhdf-F7xadWgaAV5ScXnLft8EtaYJCuIkma2Vw,3486
 phoenix/db/insertion/span.py,sha256=T9jOW3lyWte-JGD7wlP2ZqtO0-V57d8z6U_TldnuGuk,5527
+phoenix/db/insertion/span_annotation.py,sha256=Aa33nhEsq6NlXbRZ46fl2rVVehityRtxi6NYDXzE9vU,4705
+phoenix/db/insertion/trace_annotation.py,sha256=eIza4k4ImOPu9iwlAt8gv-A0ESw5XLGEa4dKGVqNSxk,4765
+phoenix/db/insertion/types.py,sha256=H9o_LJ1DabMVyVv3NZYtmJ26WusyihwlzV7AA-TOgGY,8272
 phoenix/db/migrations/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/db/migrations/env.py,sha256=QbzB5zrRs6XQQmrYeUpuzeilcMlM-MsbaAgHHYcIHTI,3626
 phoenix/db/migrations/script.py.mako,sha256=MEqL-2qATlST9TAOeYgscMn1uy6HUS9NFvDgl93dMj8,635
@@ -33,7 +38,7 @@ phoenix/db/migrations/versions/cf03bd6bae1d_init.py,sha256=CbWT3ZTR0CZqeT3zWLoTW
 phoenix/experiments/__init__.py,sha256=6JGwgUd7xCbGpuHqYZlsmErmYvVgv7N_j43bn3dUqsk,123
 phoenix/experiments/functions.py,sha256=4XaOLE1Co9sW_yjM1sypQClmOLtt9kwoxmhIEJ3f_rk,32209
 phoenix/experiments/tracing.py,sha256=wVpt8Ie9WNPoi1djJdcrkwCokHdTO0bicXViLg3O-1Y,2831
-phoenix/experiments/types.py,sha256=VuvDCcvUGeHIQuXS_xpz7Jq5xHdt3qu-O_C7IQ3DvF8,23397
+phoenix/experiments/types.py,sha256=SQqI-8CqA-bDTVIOUC1NOMi9bjEcDSlcG7o3Li1IBFY,23415
 phoenix/experiments/utils.py,sha256=wLu5Kvt1b4a8rGPRWq5G8RQ9XSiV8fCIVm51zWBI3-g,758
 phoenix/experiments/evaluators/__init__.py,sha256=j63fi3fa3U7-itVPHa82GowhjQRU-wO6yhO34u_lhsA,714
 phoenix/experiments/evaluators/base.py,sha256=jAwJs-V7jCp2UBChL0S3813Xyd9GN4rU4IEhX0nkFGs,5549
@@ -60,7 +65,7 @@ phoenix/pointcloud/pointcloud.py,sha256=4zAIkKs2xOUbchpj4XDAV-iPMXrfAJ15TG6rlIYG
 phoenix/pointcloud/projectors.py,sha256=zO_RrtDYSv2rqVOfIP2_9Cv11Dc8EmcZR94xhFcBYPU,1057
 phoenix/pointcloud/umap_parameters.py,sha256=3UQSjrysVOvq2V4KNpTMqNqNiK0BsTZnPBHWZ4fyJtQ,1708
 phoenix/server/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-phoenix/server/app.py,sha256=DaYK3-2YJVB6AWE_oJ1pX96H20pap9D60DoCm20p12A,19041
+phoenix/server/app.py,sha256=gAmGzOAhGO05IBEnBNfOc8Suw15_0_h0OCp-9RP0udQ,19098
 phoenix/server/grpc_server.py,sha256=jllxDNkpLQxDkvej4RhTokobowbvydF-SU8gSw1MTCc,3378
 phoenix/server/main.py,sha256=dvjv3g8ANpkvSGCUN02S2Yse643Nlwrp_bj4iXBSVTE,11082
 phoenix/server/prometheus.py,sha256=j9DHB2fERuq_ZKmwVaqR-9wx5WcPPuU1Cm5Bhg5241Y,2996
@@ -124,8 +129,10 @@ phoenix/server/api/input_types/PatchAnnotationInput.py,sha256=NWhkcbcGNPwfOYsN3w
 phoenix/server/api/input_types/PatchDatasetExamplesInput.py,sha256=E86aBGXDBC83jiEGwV5rilnoeQf6eqCfZ0aAVeIt2VI,890
 phoenix/server/api/input_types/PatchDatasetInput.py,sha256=OURtTVY8Z_oFEDtKwT1LCMaOK5D4QYo5TVQ6mDrex-g,328
 phoenix/server/api/input_types/PerformanceMetricInput.py,sha256=fElsLTSEYYgGFGMYTEGcYid39tXUKFdV_JkdHavMcbA,591
+phoenix/server/api/input_types/SpanAnnotationSort.py,sha256=T5pAGzmh4MiJp9JMAzNDByFVTczfw02FH4WFWwFezyI,361
 phoenix/server/api/input_types/SpanSort.py,sha256=BqV0DjI4m3RKFbzsHN47XNhJfFmErFIYg64c8EJAVMo,6312
 phoenix/server/api/input_types/TimeRange.py,sha256=yzx-gxj8mDeGLft1FzU_x1MVEgIG5Pt6-f8PUVDgipQ,522
+phoenix/server/api/input_types/TraceAnnotationSort.py,sha256=BzwiUnMh2VsgQYnhDlbJ6ljHugqIS4YDUlYzvq_tl3o,365
 phoenix/server/api/input_types/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 phoenix/server/api/mutations/__init__.py,sha256=UKUAhD5NY-ZI7XONnRRkaHoFuuU3idmE4fk6Sjgy18M,776
 phoenix/server/api/mutations/auth.py,sha256=vPRFoj7J6PV6QeODewG4K0PhoOebS5AfMRpbi_wuhyQ,311
@@ -142,13 +149,13 @@ phoenix/server/api/routers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJ
 phoenix/server/api/routers/utils.py,sha256=M41BoH-fl37izhRuN2aX7lWm7jOC20A_3uClv9TVUUY,583
 phoenix/server/api/routers/v1/__init__.py,sha256=nb49zcOdAi3DSGuC9gUubN9Yri-o7-WFdlGak4jGuFw,1462
 phoenix/server/api/routers/v1/datasets.py,sha256=QG6QeyfPi1tfBvVoa3jpNOcwtR5ntNfd3IW9rzP735Q,36920
-phoenix/server/api/routers/v1/evaluations.py,sha256=xfBoieGa9bmkvomJFfveGIS6LcC4AgjS4OeoUsTlJWk,9484
+phoenix/server/api/routers/v1/evaluations.py,sha256=FSfz9MTi8s65F07abDXlb9-y97fDZSYbqsCXpimwO7g,12628
 phoenix/server/api/routers/v1/experiment_evaluations.py,sha256=xoVhU71U3c1QJSXiKsAa4yiH-UAqkyVc7uKPyIHjrYY,4682
 phoenix/server/api/routers/v1/experiment_runs.py,sha256=7qvLYgqH58nxqhTnJ0hPf0PBfVmZnbxquodSnEGoQxk,6059
 phoenix/server/api/routers/v1/experiments.py,sha256=AFJb8Pi6SDKbIbkDRWUtPjikZIoRhc7cFLWm40l3mPk,9817
 phoenix/server/api/routers/v1/pydantic_compat.py,sha256=FeK8oe2brqu-djsoqRxiKL4tw5cHmi89OHVfCFxYsAo,2890
-phoenix/server/api/routers/v1/spans.py,sha256=yMs3Sm6s6JLV2nlmpnQ-ZJ_In0aTY5M3uOlLi-Wzb9c,8967
-phoenix/server/api/routers/v1/traces.py,sha256=Zl_hGHd-4rA0tXegH_GVoN9Ij84vbPB8oHu28fzGHA8,8029
+phoenix/server/api/routers/v1/spans.py,sha256=-nDcr0AjynS-0U6A2nxEQiUTLoNyfzE-3NiFObyfyVQ,8516
+phoenix/server/api/routers/v1/traces.py,sha256=4k_57UICPfRcyQsdPBK-Xo_n6oZ1p41XUwXhZEQpXwE,7568
 phoenix/server/api/routers/v1/utils.py,sha256=xvl2v-BKUkqmFVMmgmmWGFKuRBTrUdoiAeT3mCYEE68,3086
 phoenix/server/api/types/Annotation.py,sha256=7Ym7iuVcbwHlw2yIRylz4nATAF_Cm-Z17qcjiooj1cc,751
 phoenix/server/api/types/AnnotatorKind.py,sha256=rPgGdbN1Gvc109sGQ_ZH-gfJbp93V9wlarzTEJNtUwI,236
@@ -193,10 +200,10 @@ phoenix/server/api/types/Retrieval.py,sha256=OhMK2ncjoyp5h1yjKhjlKpoTbQrMHuxmgSF
 phoenix/server/api/types/ScalarDriftMetricEnum.py,sha256=IUAcRPpgL41WdoIgK6cNk2Te38SspXGyEs-S1fY23_A,232
 phoenix/server/api/types/Segments.py,sha256=m2yoegrxA1Tn7ZAy1rMjjD1isc752MaAXMoffkBlvrM,2921
 phoenix/server/api/types/SortDir.py,sha256=OUpXhlCzCxPoXSDkJJygEs9Rw9pMymfaZUG5zPTrw4Y,152
-phoenix/server/api/types/Span.py,sha256=0Ka8usQ8RscoeqZfz_AEHpfI2BXy0Q7caIAyo8LBc38,14511
+phoenix/server/api/types/Span.py,sha256=PRVjcC79qyxz8FW-YT90rDgcW8dkVO9_5pXxUXeH31g,15448
 phoenix/server/api/types/SpanAnnotation.py,sha256=6b5G-b_OoRvDL2ayWk7MkbqarLK-F-pQMx21CpUuNGY,1168
 phoenix/server/api/types/TimeSeries.py,sha256=wjzuxHFqCey0O7Ys25qiXyuqXK8an-osyNWUE8A_8G4,5227
-phoenix/server/api/types/Trace.py,sha256=ep-mPexub1ijxAnBvc2KrGsNVXO2SfDR1WxqER2wcD8,2376
+phoenix/server/api/types/Trace.py,sha256=-nh3A-S_BlQK1VSSOTWqM85l-WwJsRHifxeDi0sFWZE,3246
 phoenix/server/api/types/TraceAnnotation.py,sha256=OW6A2zr1gomOuG0XQe55dk15XXX2DSM0DzatRbHWH5A,1256
 phoenix/server/api/types/UMAPPoints.py,sha256=5sOuruzM8saXa8C2XiyUfk2XPrkVGmhqKpclMYRw1dk,1656
 phoenix/server/api/types/ValidationResult.py,sha256=pHwdYk4J7SJ5xhlWWHg_6qWkfk4rjOx-bSkGHvkDE3Q,142
@@ -215,13 +222,13 @@ phoenix/server/static/apple-touch-icon-76x76.png,sha256=CT_xT12I0u2i0WU8JzBZBuOQ
 phoenix/server/static/apple-touch-icon.png,sha256=fOfpjqGpWYbJ0eAurKsyoZP1EAs6ZVooBJ_SGk2ZkDs,3801
 phoenix/server/static/favicon.ico,sha256=bY0vvCKRftemZfPShwZtE93DiiQdaYaozkPGwNFr6H8,34494
 phoenix/server/static/modernizr.js,sha256=mvK-XtkNqjOral-QvzoqsyOMECXIMu5BQwSVN_wcU9c,2564
-phoenix/server/static/.vite/manifest.json,sha256=9PYR8RFuxdsTxW7Ofs1P8Ehx2DktABRYGTx0jGtAX84,1929
-phoenix/server/static/assets/components-kGgeFkHp.js,sha256=xUfFn5d7Q4ZltPMMQYFCHfXwGTeclBTSKjRQID0Fz9c,165411
-phoenix/server/static/assets/index-BctFO6S7.js,sha256=Y3BkQU18R8sJnJAuq0cU7sVPFjDcqhskG0AmAtv5Q8k,6337
-phoenix/server/static/assets/pages-DabDCmVd.js,sha256=cAiGKj0vFph9REng2JjQAVOax5IhgdSxkEBVs8u1Hh0,433015
+phoenix/server/static/.vite/manifest.json,sha256=Y5Vxm1VUdmt2c5ANmtD1Z0r2Q1mnApb-UhzLnyJxtU0,1929
+phoenix/server/static/assets/components-Ci5kMOk5.js,sha256=WxnZ4voqivS6S9USAF9zhC5Q2AdvhaJTGFbkTwDLiUI,165385
+phoenix/server/static/assets/index-BQG5WVX7.js,sha256=7RjUXpSAeTK-Q_mLrs4cWscpGyhtSIMI0RhsNn6SwMg,6347
+phoenix/server/static/assets/pages-BrevprVW.js,sha256=GZiR--twaLN-wYV1jvIM-pJiuPbae-HReCjkXNZ5O_U,441239
 phoenix/server/static/assets/vendor-CP0b0YG0.js,sha256=hoxXvVgcHofQrVahK5Q4hAMDaQiH6GygbQvdiXlhQAo,1355423
 phoenix/server/static/assets/vendor-DxkFTwjz.css,sha256=nZrkr0u6NNElFGvpWHk9GTHeGoibCXCli1bE7mXZGZg,1816
-phoenix/server/static/assets/vendor-arizeai-B5Hti8OB.js,sha256=H1iwTiHB40STvEBQe1BnZpej-9slXsuOgEZvvVvnz3Y,296355
+phoenix/server/static/assets/vendor-arizeai-DTbiPGp6.js,sha256=-o-2f0ByoAbbFXyNrOogVNfdqmXKVMsuxhnuwTJqVLc,296355
 phoenix/server/static/assets/vendor-codemirror-DtdPDzrv.js,sha256=ngcX7xxWUANIvplBuEqkDowqRkknIoX3gXUeQk-jtMQ,503031
 phoenix/server/static/assets/vendor-recharts-A0DA1O99.js,sha256=Zp6oIbsFZPDN-M7VNovxGfJkg9XUISg1fM7eRMtHCaw,282859
 phoenix/server/static/assets/vendor-three-DwGkEfCM.js,sha256=0D12ZgKzfKCTSdSTKJBFR2RZO_xxeMXrqDp0AszZqHY,620972
@@ -248,7 +255,7 @@ phoenix/trace/trace_dataset.py,sha256=Wq89jJ4hYQ1Qt-Uj11ZNzKQYQeKmGY6NqWStQiiTlM
 phoenix/trace/utils.py,sha256=1SEQr37cdHOM0P3BdL1dszArj3Zm-VJQyb1BcJs_qO8,1833
 phoenix/trace/dsl/README.md,sha256=ihmP9zGUC5V-TDbzKla76LuyDqPDQIBUH2BORwxNI68,2902
 phoenix/trace/dsl/__init__.py,sha256=WIQIjJg362XD3s50OsPJJ0xbDsGp41bSv7vDllLrPuA,144
-phoenix/trace/dsl/filter.py,sha256=DTDERSAexxDbTy5QvC48NSKhnr2pfIRDZ4PS-s3ZW80,32642
+phoenix/trace/dsl/filter.py,sha256=9NwATCUOgJ4Pms8XsEcinROUuxZ9UW-ISV09o65Ms70,32600
 phoenix/trace/dsl/helpers.py,sha256=ULAhqWULPqYWCSNX7y50DVKIqfySx86nqb6hDvZPnVk,3896
 phoenix/trace/dsl/query.py,sha256=W0t-tiXh2WIVb96lzFAGQOQ-U46uKux78d4KL3rW-PE,30316
 phoenix/trace/langchain/__init__.py,sha256=F37GfD1pd5Kuw7R7iRUM1zXXpO8xEcycNZh5dwqBXNk,109
@@ -269,8 +276,8 @@ phoenix/utilities/logging.py,sha256=lDXd6EGaamBNcQxL4vP1au9-i_SXe0OraUDiJOcszSw,
 phoenix/utilities/project.py,sha256=8IJuMM4yUMoooPi37sictGj8Etu9rGmq6RFtc9848cQ,436
 phoenix/utilities/re.py,sha256=PDve_OLjRTM8yQQJHC8-n3HdIONi7aNils3ZKRZ5uBM,2045
 phoenix/utilities/span_store.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-arize_phoenix-4.15.0.dist-info/METADATA,sha256=dGRIIc2CR5BuCgVqrzfF5kJphORO_AmyoqhVX7mVrS0,11775
-arize_phoenix-4.15.0.dist-info/WHEEL,sha256=1yFddiXMmvYK7QYTqtRNtX66WJ0Mz8PYEiEUoOUUxRY,87
-arize_phoenix-4.15.0.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
-arize_phoenix-4.15.0.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
-arize_phoenix-4.15.0.dist-info/RECORD,,
+arize_phoenix-4.16.0.dist-info/METADATA,sha256=GEUUxZhBhXAU_6ikOQlpm6SN8-ECGptFZtdWKUQ553Y,11820
+arize_phoenix-4.16.0.dist-info/WHEEL,sha256=1yFddiXMmvYK7QYTqtRNtX66WJ0Mz8PYEiEUoOUUxRY,87
+arize_phoenix-4.16.0.dist-info/licenses/IP_NOTICE,sha256=JBqyyCYYxGDfzQ0TtsQgjts41IJoa-hiwDrBjCb9gHM,469
+arize_phoenix-4.16.0.dist-info/licenses/LICENSE,sha256=HFkW9REuMOkvKRACuwLPT0hRydHb3zNg-fdFt94td18,3794
+arize_phoenix-4.16.0.dist-info/RECORD,,

phoenix/db/bulk_inserter.py

@@ -1,26 +1,36 @@
 import asyncio
 import logging
-from asyncio import Queue
+from asyncio import Queue, as_completed
+from collections import defaultdict
 from dataclasses import dataclass, field
 from datetime import datetime, timezone
+from functools import singledispatchmethod
 from itertools import islice
 from time import perf_counter
 from typing import (
     Any,
     Awaitable,
     Callable,
+    DefaultDict,
+    Dict,
     Iterable,
     List,
+    Mapping,
     Optional,
     Set,
     Tuple,
+    Type,
     cast,
 )
 
 from cachetools import LRUCache
+from sqlalchemy import Select, select
 from typing_extensions import TypeAlias
 
 import phoenix.trace.v1 as pb
+from phoenix.db import models
+from phoenix.db.insertion.constants import DEFAULT_RETRY_ALLOWANCE, DEFAULT_RETRY_DELAY_SEC
+from phoenix.db.insertion.document_annotation import DocumentAnnotationQueueInserter
 from phoenix.db.insertion.evaluation import (
     EvaluationInsertionEvent,
     InsertEvaluationError,
@@ -28,6 +38,9 @@ from phoenix.db.insertion.evaluation import (
 )
 from phoenix.db.insertion.helpers import DataManipulation, DataManipulationEvent
 from phoenix.db.insertion.span import SpanInsertionEvent, insert_span
+from phoenix.db.insertion.span_annotation import SpanAnnotationQueueInserter
+from phoenix.db.insertion.trace_annotation import TraceAnnotationQueueInserter
+from phoenix.db.insertion.types import Insertables, Precursors
 from phoenix.server.api.dataloaders import CacheForDataLoaders
 from phoenix.server.types import DbSessionFactory
 from phoenix.trace.schemas import Span
@@ -55,6 +68,8 @@ class BulkInserter:
         max_ops_per_transaction: int = 1000,
         max_queue_size: int = 1000,
         enable_prometheus: bool = False,
+        retry_delay_sec: float = DEFAULT_RETRY_DELAY_SEC,
+        retry_allowance: int = DEFAULT_RETRY_ALLOWANCE,
     ) -> None:
         """
         :param db: A function to initiate a new database session.
@@ -81,6 +96,9 @@ class BulkInserter:
         self._last_updated_at_by_project: LRUCache[ProjectRowId, datetime] = LRUCache(maxsize=100)
         self._cache_for_dataloaders = cache_for_dataloaders
         self._enable_prometheus = enable_prometheus
+        self._retry_delay_sec = retry_delay_sec
+        self._retry_allowance = retry_allowance
+        self._queue_inserters = _QueueInserters(db, self._retry_delay_sec, self._retry_allowance)
 
     def last_updated_at(self, project_rowid: Optional[ProjectRowId] = None) -> Optional[datetime]:
         if isinstance(project_rowid, ProjectRowId):
@@ -90,6 +108,7 @@ class BulkInserter:
     async def __aenter__(
         self,
     ) -> Tuple[
+        Callable[[Any], Awaitable[None]],
         Callable[[Span, str], Awaitable[None]],
         Callable[[pb.Evaluation], Awaitable[None]],
         Callable[[DataManipulation], None],
@@ -98,6 +117,7 @@ class BulkInserter:
         self._operations = Queue(maxsize=self._max_queue_size)
         self._task = asyncio.create_task(self._bulk_insert())
         return (
+            self._enqueue,
             self._queue_span,
             self._queue_evaluation,
             self._enqueue_operation,
@@ -109,6 +129,9 @@ class BulkInserter:
             self._task.cancel()
             self._task = None
 
+    async def _enqueue(self, *items: Any) -> None:
+        await self._queue_inserters.enqueue(*items)
+
     def _enqueue_operation(self, operation: DataManipulation) -> None:
         cast("Queue[DataManipulation]", self._operations).put_nowait(operation)
 
@@ -124,7 +147,17 @@ class BulkInserter:
         assert isinstance(self._operations, Queue)
         spans_buffer, evaluations_buffer = None, None
         # start first insert immediately if the inserter has not run recently
-        while self._running or not self._operations.empty() or self._spans or self._evaluations:
+        while (
+            self._running
+            or not self._queue_inserters.empty
+            or not self._operations.empty()
+            or self._spans
+            or self._evaluations
+        ):
+            if not self._queue_inserters.empty:
+                if inserted_ids := await self._queue_inserters.insert():
+                    for project_rowid in await self._get_project_rowids(inserted_ids):
+                        self._last_updated_at_by_project[project_rowid] = datetime.now(timezone.utc)
             if self._operations.empty() and not (self._spans or self._evaluations):
                 await asyncio.sleep(self._sleep)
                 continue
@@ -244,3 +277,97 @@ class BulkInserter:
                     BULK_LOADER_EXCEPTIONS.inc()
                 logger.exception("Failed to insert evaluations")
         return transaction_result
+
+    async def _get_project_rowids(
+        self,
+        inserted_ids: Mapping[Type[models.Base], List[int]],
+    ) -> Set[int]:
+        ans: Set[int] = set()
+        if not inserted_ids:
+            return ans
+        stmt: Select[Tuple[int]]
+        for table, ids in inserted_ids.items():
+            if not ids:
+                continue
+            if issubclass(table, models.SpanAnnotation):
+                stmt = (
+                    select(models.Project.id)
+                    .join(models.Trace)
+                    .join_from(models.Trace, models.Span)
+                    .join_from(models.Span, models.SpanAnnotation)
+                    .where(models.SpanAnnotation.id.in_(ids))
+                )
+            elif issubclass(table, models.DocumentAnnotation):
+                stmt = (
+                    select(models.Project.id)
+                    .join(models.Trace)
+                    .join_from(models.Trace, models.Span)
+                    .join_from(models.Span, models.DocumentAnnotation)
+                    .where(models.DocumentAnnotation.id.in_(ids))
+                )
+            elif issubclass(table, models.TraceAnnotation):
+                stmt = (
+                    select(models.Project.id)
+                    .join(models.Trace)
+                    .join_from(models.Trace, models.TraceAnnotation)
+                    .where(models.TraceAnnotation.id.in_(ids))
+                )
+            else:
+                continue
+            async with self._db() as session:
+                project_rowids = [_ async for _ in await session.stream_scalars(stmt)]
+                ans.update(project_rowids)
+        return ans
+
+
+class _QueueInserters:
+    def __init__(
+        self,
+        db: DbSessionFactory,
+        retry_delay_sec: float = DEFAULT_RETRY_DELAY_SEC,
+        retry_allowance: int = DEFAULT_RETRY_ALLOWANCE,
+    ) -> None:
+        self._db = db
+        args = (db, retry_delay_sec, retry_allowance)
+        self._span_annotations = SpanAnnotationQueueInserter(*args)
+        self._trace_annotations = TraceAnnotationQueueInserter(*args)
+        self._document_annotations = DocumentAnnotationQueueInserter(*args)
+        self._queues = (
+            self._span_annotations,
+            self._trace_annotations,
+            self._document_annotations,
+        )
+
+    async def insert(self) -> Dict[Type[models.Base], List[int]]:
+        ans: DefaultDict[Type[models.Base], List[int]] = defaultdict(list)
+        for coro in as_completed([q.insert() for q in self._queues]):
+            table, inserted_ids = await coro
+            if inserted_ids:
+                ans[table].extend(inserted_ids)
+        return ans
+
+    @property
+    def empty(self) -> bool:
+        return all(q.empty for q in self._queues)
+
+    async def enqueue(self, *items: Any) -> None:
+        for item in items:
+            await self._enqueue(item)
+
+    @singledispatchmethod
+    async def _enqueue(self, item: Any) -> None: ...
+
+    @_enqueue.register(Precursors.SpanAnnotation)
+    @_enqueue.register(Insertables.SpanAnnotation)
+    async def _(self, item: Precursors.SpanAnnotation) -> None:
+        await self._span_annotations.enqueue(item)
+
+    @_enqueue.register(Precursors.TraceAnnotation)
+    @_enqueue.register(Insertables.TraceAnnotation)
+    async def _(self, item: Precursors.TraceAnnotation) -> None:
+        await self._trace_annotations.enqueue(item)
+
+    @_enqueue.register(Precursors.DocumentAnnotation)
+    @_enqueue.register(Insertables.DocumentAnnotation)
+    async def _(self, item: Precursors.DocumentAnnotation) -> None:
+        await self._document_annotations.enqueue(item)

phoenix/db/helpers.py

@@ -1,5 +1,5 @@
 from enum import Enum
-from typing import Any, Optional, Tuple
+from typing import Any, Callable, Hashable, Iterable, List, Optional, Set, Tuple, TypeVar
 
 from openinference.semconv.trace import (
     OpenInferenceSpanKindValues,
@@ -80,3 +80,25 @@ def get_project_names_for_experiments(*experiment_ids: int) -> Select[Tuple[Opti
         .where(models.Experiment.id.in_(set(experiment_ids)))
         .where(models.Experiment.project_name.isnot(None))
     )
+
+
+_AnyT = TypeVar("_AnyT")
+_KeyT = TypeVar("_KeyT", bound=Hashable)
+
+
+def dedup(
+    items: Iterable[_AnyT],
+    key: Callable[[_AnyT], _KeyT],
+) -> List[_AnyT]:
+    """
+    Discard subsequent duplicates after the first appearance in `items`.
+    """
+    ans = []
+    seen: Set[_KeyT] = set()
+    for item in items:
+        if (k := key(item)) in seen:
+            continue
+        else:
+            ans.append(item)
+            seen.add(k)
+    return ans

phoenix/db/insertion/constants.py

@@ -0,0 +1,2 @@
+DEFAULT_RETRY_DELAY_SEC: float = 60
+DEFAULT_RETRY_ALLOWANCE: int = 10

phoenix/db/insertion/document_annotation.py

@@ -0,0 +1,157 @@
+from datetime import datetime
+from typing import Any, List, Mapping, NamedTuple, Optional, Tuple
+
+from sqlalchemy import Row, Select, and_, select, tuple_
+from sqlalchemy.ext.asyncio import AsyncSession
+from typing_extensions import TypeAlias
+
+from phoenix.db import models
+from phoenix.db.helpers import dedup, num_docs_col
+from phoenix.db.insertion.types import (
+    Insertables,
+    Postponed,
+    Precursors,
+    QueueInserter,
+    Received,
+)
+
+_Name: TypeAlias = str
+_SpanId: TypeAlias = str
+_SpanRowId: TypeAlias = int
+_DocumentPosition: TypeAlias = int
+_AnnoRowId: TypeAlias = int
+_NumDocs: TypeAlias = int
+
+_Key: TypeAlias = Tuple[_Name, _SpanId, _DocumentPosition]
+_UniqueBy: TypeAlias = Tuple[_Name, _SpanRowId, _DocumentPosition]
+_Existing: TypeAlias = Tuple[
+    _SpanRowId,
+    _SpanId,
+    _NumDocs,
+    Optional[_AnnoRowId],
+    Optional[_Name],
+    Optional[_DocumentPosition],
+    Optional[datetime],
+]
+
+
+class DocumentAnnotationQueueInserter(
+    QueueInserter[
+        Precursors.DocumentAnnotation,
+        Insertables.DocumentAnnotation,
+        models.DocumentAnnotation,
+    ],
+    table=models.DocumentAnnotation,
+    unique_by=("name", "span_rowid", "document_position"),
+):
+    async def _partition(
+        self,
+        session: AsyncSession,
+        *parcels: Received[Precursors.DocumentAnnotation],
+    ) -> Tuple[
+        List[Received[Insertables.DocumentAnnotation]],
+        List[Postponed[Precursors.DocumentAnnotation]],
+        List[Received[Precursors.DocumentAnnotation]],
+    ]:
+        to_insert: List[Received[Insertables.DocumentAnnotation]] = []
+        to_postpone: List[Postponed[Precursors.DocumentAnnotation]] = []
+        to_discard: List[Received[Precursors.DocumentAnnotation]] = []
+
+        stmt = self._select_existing(*map(_key, parcels))
+        existing: List[Row[_Existing]] = [_ async for _ in await session.stream(stmt)]
+        existing_spans: Mapping[str, _SpanAttr] = {
+            e.span_id: _SpanAttr(e.span_rowid, e.num_docs) for e in existing
+        }
+        existing_annos: Mapping[_Key, _AnnoAttr] = {
+            (e.name, e.span_id, e.document_position): _AnnoAttr(e.span_rowid, e.id, e.updated_at)
+            for e in existing
+            if e.id is not None
+            and e.name is not None
+            and e.document_position is not None
+            and e.updated_at is not None
+        }
+
+        for p in parcels:
+            if (anno := existing_annos.get(_key(p))) is not None:
+                if p.received_at <= anno.updated_at:
+                    to_discard.append(p)
+                else:
+                    to_insert.append(
+                        Received(
+                            received_at=p.received_at,
+                            item=p.item.as_insertable(
+                                span_rowid=anno.span_rowid,
+                                id_=anno.id_,
+                            ),
+                        )
+                    )
+            elif (span := existing_spans.get(p.item.span_id)) is not None:
+                if 0 <= p.item.document_position < span.num_docs:
+                    to_insert.append(
+                        Received(
+                            received_at=p.received_at,
+                            item=p.item.as_insertable(
+                                span_rowid=span.span_rowid,
+                            ),
+                        )
+                    )
+                else:
+                    to_discard.append(p)
+            elif isinstance(p, Postponed):
+                if p.retries_left > 1:
+                    to_postpone.append(p.postpone(p.retries_left - 1))
+                else:
+                    to_discard.append(p)
+            elif isinstance(p, Received):
+                to_postpone.append(p.postpone(self._retry_allowance))
+            else:
+                to_discard.append(p)
+
+        assert len(to_insert) + len(to_postpone) + len(to_discard) == len(parcels)
+        to_insert = dedup(sorted(to_insert, key=_time, reverse=True), _unique_by)[::-1]
+        return to_insert, to_postpone, to_discard
+
+    def _select_existing(self, *keys: _Key) -> Select[_Existing]:
+        anno = self.table
+        span = (
+            select(models.Span.id, models.Span.span_id, num_docs_col(self._db.dialect))
+            .where(models.Span.span_id.in_({span_id for _, span_id, *_ in keys}))
+            .cte()
+        )
+        onclause = and_(
+            span.c.id == anno.span_rowid,
+            anno.name.in_({name for name, *_ in keys}),
+            tuple_(anno.name, span.c.span_id, anno.document_position).in_(keys),
+        )
+        return select(
+            span.c.id.label("span_rowid"),
+            span.c.span_id,
+            span.c.num_docs,
+            anno.id,
+            anno.name,
+            anno.document_position,
+            anno.updated_at,
+        ).outerjoin_from(span, anno, onclause)
+
+
+class _SpanAttr(NamedTuple):
+    span_rowid: _SpanRowId
+    num_docs: _NumDocs
+
+
+class _AnnoAttr(NamedTuple):
+    span_rowid: _SpanRowId
+    id_: _AnnoRowId
+    updated_at: datetime
+
+
+def _key(p: Received[Precursors.DocumentAnnotation]) -> _Key:
+    return p.item.obj.name, p.item.span_id, p.item.document_position
+
+
+def _unique_by(p: Received[Insertables.DocumentAnnotation]) -> _UniqueBy:
+    return p.item.obj.name, p.item.span_rowid, p.item.document_position
+
+
+def _time(p: Received[Any]) -> datetime:
+    return p.received_at

phoenix/db/insertion/helpers.py

@@ -20,6 +20,7 @@ from sqlalchemy.ext.asyncio import AsyncSession
 from sqlalchemy.sql.elements import KeyedColumnElement
 from typing_extensions import TypeAlias, assert_never
 
+from phoenix.db import models
 from phoenix.db.helpers import SupportedSQLDialect
 from phoenix.db.models import Base
 
@@ -93,3 +94,15 @@ def _clean(
             yield "metadata", v
         else:
             yield k, v
+
+
+def as_kv(obj: models.Base) -> Iterator[Tuple[str, Any]]:
+    for k, c in obj.__table__.c.items():
+        if k in ["created_at", "updated_at"]:
+            continue
+        k = "metadata_" if k == "metadata" else k
+        v = getattr(obj, k, None)
+        if c.primary_key and v is None:
+            # postgresql disallows None for primary key
+            continue
+        yield k, v