Skip to content

prefect.tasks

Module containing the base workflow task class and decorator - for most use cases, using the @task decorator is preferred.

Task

Bases: Generic[P, R]

A Prefect task definition.

Note

We recommend using the @task decorator for most use-cases.

Wraps a function with an entrypoint to the Prefect engine. Calling this class within a flow function creates a new task run.

To preserve the input and output types, we use the generic type variables P and R for "Parameters" and "Returns" respectively.

Parameters:

Name Type Description Default
fn Callable[P, R]

The function defining the task.

required
name Optional[str]

An optional name for the task; if not provided, the name will be inferred from the given function.

None
description Optional[str]

An optional string description for the task.

None
tags Optional[Iterable[str]]

An optional set of tags to be associated with runs of this task. These tags are combined with any tags defined by a prefect.tags context at task runtime.

None
version Optional[str]

An optional string specifying the version of this task definition

None
cache_key_fn Optional[Callable[[TaskRunContext, Dict[str, Any]], Optional[str]]]

An optional callable that, given the task run context and call parameters, generates a string key; if the key matches a previous completed state, that state result will be restored instead of running the task again.

None
cache_expiration Optional[timedelta]

An optional amount of time indicating how long cached states for this task should be restorable; if not provided, cached states will never expire.

None
task_run_name Optional[Union[Callable[[], str], str]]

An optional name to distinguish runs of this task; this name can be provided as a string template with the task's keyword arguments as variables, or a function that returns a string.

None
retries Optional[int]

An optional number of times to retry on task run failure.

None
retry_delay_seconds Optional[Union[float, int, List[float], Callable[[int], List[float]]]]

Optionally configures how long to wait before retrying the task after failure. This is only applicable if retries is nonzero. This setting can either be a number of seconds, a list of retry delays, or a callable that, given the total number of retries, generates a list of retry delays. If a number of seconds, that delay will be applied to all retries. If a list, each retry will wait for the corresponding delay before retrying. When passing a callable or a list, the number of configured retry delays cannot exceed 50.

None
retry_jitter_factor Optional[float]

An optional factor that defines the factor to which a retry can be jittered in order to avoid a "thundering herd".

None
persist_result Optional[bool]

An optional toggle indicating whether the result of this task should be persisted to result storage. Defaults to None, which indicates that Prefect should choose whether the result should be persisted depending on the features being used.

None
result_storage Optional[ResultStorage]

An optional block to use to persist the result of this task. Defaults to the value set in the flow the task is called in.

None
result_storage_key Optional[str]

An optional key to store the result in storage at when persisted. Defaults to a unique identifier.

None
result_serializer Optional[ResultSerializer]

An optional serializer to use to serialize the result of this task for persistence. Defaults to the value set in the flow the task is called in.

None
timeout_seconds Union[int, float, None]

An optional number of seconds indicating a maximum runtime for the task. If the task exceeds this runtime, it will be marked as failed.

None
log_prints Optional[bool]

If set, print statements in the task will be redirected to the Prefect logger for the task run. Defaults to None, which indicates that the value from the flow should be used.

False
refresh_cache Optional[bool]

If set, cached results for the cache key are not used. Defaults to None, which indicates that a cached result from a previous execution with matching cache key is used.

None
on_failure Optional[List[Callable[[Task, TaskRun, State], None]]]

An optional list of callables to run when the task enters a failed state.

None
on_completion Optional[List[Callable[[Task, TaskRun, State], None]]]

An optional list of callables to run when the task enters a completed state.

None
retry_condition_fn Optional[Callable[[Task, TaskRun, State], bool]]

An optional callable run when a task run returns a Failed state. Should return True if the task should continue to its retry policy (e.g. retries=3), and False if the task should end as failed. Defaults to None, indicating the task should always continue to its retry policy.

None
viz_return_value Optional[Any]

An optional value to return when the task dependency tree is visualized.

None
Source code in prefect/tasks.py
 160
 161
 162
 163
 164
 165
 166
 167
 168
 169
 170
 171
 172
 173
 174
 175
 176
 177
 178
 179
 180
 181
 182
 183
 184
 185
 186
 187
 188
 189
 190
 191
 192
 193
 194
 195
 196
 197
 198
 199
 200
 201
 202
 203
 204
 205
 206
 207
 208
 209
 210
 211
 212
 213
 214
 215
 216
 217
 218
 219
 220
 221
 222
 223
 224
 225
 226
 227
 228
 229
 230
 231
 232
 233
 234
 235
 236
 237
 238
 239
 240
 241
 242
 243
 244
 245
 246
 247
 248
 249
 250
 251
 252
 253
 254
 255
 256
 257
 258
 259
 260
 261
 262
 263
 264
 265
 266
 267
 268
 269
 270
 271
 272
 273
 274
 275
 276
 277
 278
 279
 280
 281
 282
 283
 284
 285
 286
 287
 288
 289
 290
 291
 292
 293
 294
 295
 296
 297
 298
 299
 300
 301
 302
 303
 304
 305
 306
 307
 308
 309
 310
 311
 312
 313
 314
 315
 316
 317
 318
 319
 320
 321
 322
 323
 324
 325
 326
 327
 328
 329
 330
 331
 332
 333
 334
 335
 336
 337
 338
 339
 340
 341
 342
 343
 344
 345
 346
 347
 348
 349
 350
 351
 352
 353
 354
 355
 356
 357
 358
 359
 360
 361
 362
 363
 364
 365
 366
 367
 368
 369
 370
 371
 372
 373
 374
 375
 376
 377
 378
 379
 380
 381
 382
 383
 384
 385
 386
 387
 388
 389
 390
 391
 392
 393
 394
 395
 396
 397
 398
 399
 400
 401
 402
 403
 404
 405
 406
 407
 408
 409
 410
 411
 412
 413
 414
 415
 416
 417
 418
 419
 420
 421
 422
 423
 424
 425
 426
 427
 428
 429
 430
 431
 432
 433
 434
 435
 436
 437
 438
 439
 440
 441
 442
 443
 444
 445
 446
 447
 448
 449
 450
 451
 452
 453
 454
 455
 456
 457
 458
 459
 460
 461
 462
 463
 464
 465
 466
 467
 468
 469
 470
 471
 472
 473
 474
 475
 476
 477
 478
 479
 480
 481
 482
 483
 484
 485
 486
 487
 488
 489
 490
 491
 492
 493
 494
 495
 496
 497
 498
 499
 500
 501
 502
 503
 504
 505
 506
 507
 508
 509
 510
 511
 512
 513
 514
 515
 516
 517
 518
 519
 520
 521
 522
 523
 524
 525
 526
 527
 528
 529
 530
 531
 532
 533
 534
 535
 536
 537
 538
 539
 540
 541
 542
 543
 544
 545
 546
 547
 548
 549
 550
 551
 552
 553
 554
 555
 556
 557
 558
 559
 560
 561
 562
 563
 564
 565
 566
 567
 568
 569
 570
 571
 572
 573
 574
 575
 576
 577
 578
 579
 580
 581
 582
 583
 584
 585
 586
 587
 588
 589
 590
 591
 592
 593
 594
 595
 596
 597
 598
 599
 600
 601
 602
 603
 604
 605
 606
 607
 608
 609
 610
 611
 612
 613
 614
 615
 616
 617
 618
 619
 620
 621
 622
 623
 624
 625
 626
 627
 628
 629
 630
 631
 632
 633
 634
 635
 636
 637
 638
 639
 640
 641
 642
 643
 644
 645
 646
 647
 648
 649
 650
 651
 652
 653
 654
 655
 656
 657
 658
 659
 660
 661
 662
 663
 664
 665
 666
 667
 668
 669
 670
 671
 672
 673
 674
 675
 676
 677
 678
 679
 680
 681
 682
 683
 684
 685
 686
 687
 688
 689
 690
 691
 692
 693
 694
 695
 696
 697
 698
 699
 700
 701
 702
 703
 704
 705
 706
 707
 708
 709
 710
 711
 712
 713
 714
 715
 716
 717
 718
 719
 720
 721
 722
 723
 724
 725
 726
 727
 728
 729
 730
 731
 732
 733
 734
 735
 736
 737
 738
 739
 740
 741
 742
 743
 744
 745
 746
 747
 748
 749
 750
 751
 752
 753
 754
 755
 756
 757
 758
 759
 760
 761
 762
 763
 764
 765
 766
 767
 768
 769
 770
 771
 772
 773
 774
 775
 776
 777
 778
 779
 780
 781
 782
 783
 784
 785
 786
 787
 788
 789
 790
 791
 792
 793
 794
 795
 796
 797
 798
 799
 800
 801
 802
 803
 804
 805
 806
 807
 808
 809
 810
 811
 812
 813
 814
 815
 816
 817
 818
 819
 820
 821
 822
 823
 824
 825
 826
 827
 828
 829
 830
 831
 832
 833
 834
 835
 836
 837
 838
 839
 840
 841
 842
 843
 844
 845
 846
 847
 848
 849
 850
 851
 852
 853
 854
 855
 856
 857
 858
 859
 860
 861
 862
 863
 864
 865
 866
 867
 868
 869
 870
 871
 872
 873
 874
 875
 876
 877
 878
 879
 880
 881
 882
 883
 884
 885
 886
 887
 888
 889
 890
 891
 892
 893
 894
 895
 896
 897
 898
 899
 900
 901
 902
 903
 904
 905
 906
 907
 908
 909
 910
 911
 912
 913
 914
 915
 916
 917
 918
 919
 920
 921
 922
 923
 924
 925
 926
 927
 928
 929
 930
 931
 932
 933
 934
 935
 936
 937
 938
 939
 940
 941
 942
 943
 944
 945
 946
 947
 948
 949
 950
 951
 952
 953
 954
 955
 956
 957
 958
 959
 960
 961
 962
 963
 964
 965
 966
 967
 968
 969
 970
 971
 972
 973
 974
 975
 976
 977
 978
 979
 980
 981
 982
 983
 984
 985
 986
 987
 988
 989
 990
 991
 992
 993
 994
 995
 996
 997
 998
 999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
@PrefectObjectRegistry.register_instances
class Task(Generic[P, R]):
    """
    A Prefect task definition.

    !!! note
        We recommend using [the `@task` decorator][prefect.tasks.task] for most use-cases.

    Wraps a function with an entrypoint to the Prefect engine. Calling this class within a flow function
    creates a new task run.

    To preserve the input and output types, we use the generic type variables P and R for "Parameters" and
    "Returns" respectively.

    Args:
        fn: The function defining the task.
        name: An optional name for the task; if not provided, the name will be inferred
            from the given function.
        description: An optional string description for the task.
        tags: An optional set of tags to be associated with runs of this task. These
            tags are combined with any tags defined by a `prefect.tags` context at
            task runtime.
        version: An optional string specifying the version of this task definition
        cache_key_fn: An optional callable that, given the task run context and call
            parameters, generates a string key; if the key matches a previous completed
            state, that state result will be restored instead of running the task again.
        cache_expiration: An optional amount of time indicating how long cached states
            for this task should be restorable; if not provided, cached states will
            never expire.
        task_run_name: An optional name to distinguish runs of this task; this name can be provided
            as a string template with the task's keyword arguments as variables,
            or a function that returns a string.
        retries: An optional number of times to retry on task run failure.
        retry_delay_seconds: Optionally configures how long to wait before retrying the
            task after failure. This is only applicable if `retries` is nonzero. This
            setting can either be a number of seconds, a list of retry delays, or a
            callable that, given the total number of retries, generates a list of retry
            delays. If a number of seconds, that delay will be applied to all retries.
            If a list, each retry will wait for the corresponding delay before retrying.
            When passing a callable or a list, the number of configured retry delays
            cannot exceed 50.
        retry_jitter_factor: An optional factor that defines the factor to which a retry
            can be jittered in order to avoid a "thundering herd".
        persist_result: An optional toggle indicating whether the result of this task
            should be persisted to result storage. Defaults to `None`, which indicates
            that Prefect should choose whether the result should be persisted depending on
            the features being used.
        result_storage: An optional block to use to persist the result of this task.
            Defaults to the value set in the flow the task is called in.
        result_storage_key: An optional key to store the result in storage at when persisted.
            Defaults to a unique identifier.
        result_serializer: An optional serializer to use to serialize the result of this
            task for persistence. Defaults to the value set in the flow the task is
            called in.
        timeout_seconds: An optional number of seconds indicating a maximum runtime for
            the task. If the task exceeds this runtime, it will be marked as failed.
        log_prints: If set, `print` statements in the task will be redirected to the
            Prefect logger for the task run. Defaults to `None`, which indicates
            that the value from the flow should be used.
        refresh_cache: If set, cached results for the cache key are not used.
            Defaults to `None`, which indicates that a cached result from a previous
            execution with matching cache key is used.
        on_failure: An optional list of callables to run when the task enters a failed state.
        on_completion: An optional list of callables to run when the task enters a completed state.
        retry_condition_fn: An optional callable run when a task run returns a Failed state. Should
            return `True` if the task should continue to its retry policy (e.g. `retries=3`), and `False` if the task
            should end as failed. Defaults to `None`, indicating the task should always continue
            to its retry policy.
        viz_return_value: An optional value to return when the task dependency tree is visualized.
    """

    # NOTE: These parameters (types, defaults, and docstrings) should be duplicated
    #       exactly in the @task decorator
    def __init__(
        self,
        fn: Callable[P, R],
        name: Optional[str] = None,
        description: Optional[str] = None,
        tags: Optional[Iterable[str]] = None,
        version: Optional[str] = None,
        cache_key_fn: Optional[
            Callable[["TaskRunContext", Dict[str, Any]], Optional[str]]
        ] = None,
        cache_expiration: Optional[datetime.timedelta] = None,
        task_run_name: Optional[Union[Callable[[], str], str]] = None,
        retries: Optional[int] = None,
        retry_delay_seconds: Optional[
            Union[
                float,
                int,
                List[float],
                Callable[[int], List[float]],
            ]
        ] = None,
        retry_jitter_factor: Optional[float] = None,
        persist_result: Optional[bool] = None,
        result_storage: Optional[ResultStorage] = None,
        result_serializer: Optional[ResultSerializer] = None,
        result_storage_key: Optional[str] = None,
        cache_result_in_memory: bool = True,
        timeout_seconds: Union[int, float, None] = None,
        log_prints: Optional[bool] = False,
        refresh_cache: Optional[bool] = None,
        on_completion: Optional[List[Callable[["Task", TaskRun, State], None]]] = None,
        on_failure: Optional[List[Callable[["Task", TaskRun, State], None]]] = None,
        retry_condition_fn: Optional[Callable[["Task", TaskRun, State], bool]] = None,
        viz_return_value: Optional[Any] = None,
    ):
        # Validate if hook passed is list and contains callables
        hook_categories = [on_completion, on_failure]
        hook_names = ["on_completion", "on_failure"]
        for hooks, hook_name in zip(hook_categories, hook_names):
            if hooks is not None:
                if not hooks:
                    raise ValueError(f"Empty list passed for '{hook_name}'")
                try:
                    hooks = list(hooks)
                except TypeError:
                    raise TypeError(
                        f"Expected iterable for '{hook_name}'; got"
                        f" {type(hooks).__name__} instead. Please provide a list of"
                        f" hooks to '{hook_name}':\n\n"
                        f"@flow({hook_name}=[hook1, hook2])\ndef"
                        " my_flow():\n\tpass"
                    )

                for hook in hooks:
                    if not callable(hook):
                        raise TypeError(
                            f"Expected callables in '{hook_name}'; got"
                            f" {type(hook).__name__} instead. Please provide a list of"
                            f" hooks to '{hook_name}':\n\n"
                            f"@flow({hook_name}=[hook1, hook2])\ndef"
                            " my_flow():\n\tpass"
                        )

        if not callable(fn):
            raise TypeError("'fn' must be callable")

        self.description = description or inspect.getdoc(fn)
        update_wrapper(self, fn)
        self.fn = fn
        self.isasync = inspect.iscoroutinefunction(self.fn)

        if not name:
            if not hasattr(self.fn, "__name__"):
                self.name = type(self.fn).__name__
            else:
                self.name = self.fn.__name__
        else:
            self.name = name

        if task_run_name is not None:
            if not isinstance(task_run_name, str) and not callable(task_run_name):
                raise TypeError(
                    "Expected string or callable for 'task_run_name'; got"
                    f" {type(task_run_name).__name__} instead."
                )
        self.task_run_name = task_run_name

        self.version = version
        self.log_prints = log_prints

        raise_for_reserved_arguments(self.fn, ["return_state", "wait_for"])

        self.tags = set(tags if tags else [])

        self.task_key = _generate_task_key(self.fn)

        self.cache_key_fn = cache_key_fn
        self.cache_expiration = cache_expiration
        self.refresh_cache = refresh_cache

        # TaskRunPolicy settings
        # TODO: We can instantiate a `TaskRunPolicy` and add Pydantic bound checks to
        #       validate that the user passes positive numbers here

        self.retries = (
            retries if retries is not None else PREFECT_TASK_DEFAULT_RETRIES.value()
        )
        if retry_delay_seconds is None:
            retry_delay_seconds = PREFECT_TASK_DEFAULT_RETRY_DELAY_SECONDS.value()

        if callable(retry_delay_seconds):
            self.retry_delay_seconds = retry_delay_seconds(retries)
        else:
            self.retry_delay_seconds = retry_delay_seconds

        if isinstance(self.retry_delay_seconds, list) and (
            len(self.retry_delay_seconds) > 50
        ):
            raise ValueError("Can not configure more than 50 retry delays per task.")

        if retry_jitter_factor is not None and retry_jitter_factor < 0:
            raise ValueError("`retry_jitter_factor` must be >= 0.")

        self.retry_jitter_factor = retry_jitter_factor
        self.persist_result = persist_result
        self.result_storage = result_storage
        self.result_serializer = result_serializer
        self.result_storage_key = result_storage_key
        self.cache_result_in_memory = cache_result_in_memory
        self.timeout_seconds = float(timeout_seconds) if timeout_seconds else None
        self.on_completion = on_completion
        self.on_failure = on_failure

        # retry_condition_fn must be a callable or None. If it is neither, raise a TypeError
        if retry_condition_fn is not None and not (callable(retry_condition_fn)):
            raise TypeError(
                "Expected `retry_condition_fn` to be callable, got"
                f" {type(retry_condition_fn).__name__} instead."
            )

        self.retry_condition_fn = retry_condition_fn
        self.viz_return_value = viz_return_value

    def with_options(
        self,
        *,
        name: str = None,
        description: str = None,
        tags: Iterable[str] = None,
        cache_key_fn: Callable[
            ["TaskRunContext", Dict[str, Any]], Optional[str]
        ] = None,
        task_run_name: Optional[Union[Callable[[], str], str]] = None,
        cache_expiration: datetime.timedelta = None,
        retries: Optional[int] = NotSet,
        retry_delay_seconds: Union[
            float,
            int,
            List[float],
            Callable[[int], List[float]],
        ] = NotSet,
        retry_jitter_factor: Optional[float] = NotSet,
        persist_result: Optional[bool] = NotSet,
        result_storage: Optional[ResultStorage] = NotSet,
        result_serializer: Optional[ResultSerializer] = NotSet,
        result_storage_key: Optional[str] = NotSet,
        cache_result_in_memory: Optional[bool] = None,
        timeout_seconds: Union[int, float] = None,
        log_prints: Optional[bool] = NotSet,
        refresh_cache: Optional[bool] = NotSet,
        on_completion: Optional[
            List[Callable[["Task", TaskRun, State], Union[Awaitable[None], None]]]
        ] = None,
        on_failure: Optional[
            List[Callable[["Task", TaskRun, State], Union[Awaitable[None], None]]]
        ] = None,
        retry_condition_fn: Optional[Callable[["Task", TaskRun, State], bool]] = None,
        viz_return_value: Optional[Any] = None,
    ):
        """
        Create a new task from the current object, updating provided options.

        Args:
            name: A new name for the task.
            description: A new description for the task.
            tags: A new set of tags for the task. If given, existing tags are ignored,
                not merged.
            cache_key_fn: A new cache key function for the task.
            cache_expiration: A new cache expiration time for the task.
            task_run_name: An optional name to distinguish runs of this task; this name can be provided
                as a string template with the task's keyword arguments as variables,
                or a function that returns a string.
            retries: A new number of times to retry on task run failure.
            retry_delay_seconds: Optionally configures how long to wait before retrying
                the task after failure. This is only applicable if `retries` is nonzero.
                This setting can either be a number of seconds, a list of retry delays,
                or a callable that, given the total number of retries, generates a list
                of retry delays. If a number of seconds, that delay will be applied to
                all retries. If a list, each retry will wait for the corresponding delay
                before retrying. When passing a callable or a list, the number of
                configured retry delays cannot exceed 50.
            retry_jitter_factor: An optional factor that defines the factor to which a
                retry can be jittered in order to avoid a "thundering herd".
            persist_result: A new option for enabling or disabling result persistence.
            result_storage: A new storage type to use for results.
            result_serializer: A new serializer to use for results.
            result_storage_key: A new key for the persisted result to be stored at.
            timeout_seconds: A new maximum time for the task to complete in seconds.
            log_prints: A new option for enabling or disabling redirection of `print` statements.
            refresh_cache: A new option for enabling or disabling cache refresh.
            on_completion: A new list of callables to run when the task enters a completed state.
            on_failure: A new list of callables to run when the task enters a failed state.
            retry_condition_fn: An optional callable run when a task run returns a Failed state.
                Should return `True` if the task should continue to its retry policy, and `False`
                if the task should end as failed. Defaults to `None`, indicating the task should
                always continue to its retry policy.
            viz_return_value: An optional value to return when the task dependency tree is visualized.

        Returns:
            A new `Task` instance.

        Examples:

            Create a new task from an existing task and update the name

            >>> @task(name="My task")
            >>> def my_task():
            >>>     return 1
            >>>
            >>> new_task = my_task.with_options(name="My new task")

            Create a new task from an existing task and update the retry settings

            >>> from random import randint
            >>>
            >>> @task(retries=1, retry_delay_seconds=5)
            >>> def my_task():
            >>>     x = randint(0, 5)
            >>>     if x >= 3:  # Make a task that fails sometimes
            >>>         raise ValueError("Retry me please!")
            >>>     return x
            >>>
            >>> new_task = my_task.with_options(retries=5, retry_delay_seconds=2)

            Use a task with updated options within a flow

            >>> @task(name="My task")
            >>> def my_task():
            >>>     return 1
            >>>
            >>> @flow
            >>> my_flow():
            >>>     new_task = my_task.with_options(name="My new task")
            >>>     new_task()
        """
        return Task(
            fn=self.fn,
            name=name or self.name,
            description=description or self.description,
            tags=tags or copy(self.tags),
            cache_key_fn=cache_key_fn or self.cache_key_fn,
            cache_expiration=cache_expiration or self.cache_expiration,
            task_run_name=task_run_name,
            retries=retries if retries is not NotSet else self.retries,
            retry_delay_seconds=(
                retry_delay_seconds
                if retry_delay_seconds is not NotSet
                else self.retry_delay_seconds
            ),
            retry_jitter_factor=(
                retry_jitter_factor
                if retry_jitter_factor is not NotSet
                else self.retry_jitter_factor
            ),
            persist_result=(
                persist_result if persist_result is not NotSet else self.persist_result
            ),
            result_storage=(
                result_storage if result_storage is not NotSet else self.result_storage
            ),
            result_storage_key=(
                result_storage_key
                if result_storage_key is not NotSet
                else self.result_storage_key
            ),
            result_serializer=(
                result_serializer
                if result_serializer is not NotSet
                else self.result_serializer
            ),
            cache_result_in_memory=(
                cache_result_in_memory
                if cache_result_in_memory is not None
                else self.cache_result_in_memory
            ),
            timeout_seconds=(
                timeout_seconds if timeout_seconds is not None else self.timeout_seconds
            ),
            log_prints=(log_prints if log_prints is not NotSet else self.log_prints),
            refresh_cache=(
                refresh_cache if refresh_cache is not NotSet else self.refresh_cache
            ),
            on_completion=on_completion or self.on_completion,
            on_failure=on_failure or self.on_failure,
            retry_condition_fn=retry_condition_fn or self.retry_condition_fn,
            viz_return_value=viz_return_value or self.viz_return_value,
        )

    async def create_run(
        self,
        client: Optional[Union[PrefectClient, SyncPrefectClient]],
        parameters: Dict[str, Any] = None,
        flow_run_context: Optional[FlowRunContext] = None,
        parent_task_run_context: Optional[TaskRunContext] = None,
        wait_for: Optional[Iterable[PrefectFuture]] = None,
        extra_task_inputs: Optional[Dict[str, Set[TaskRunInput]]] = None,
    ) -> TaskRun:
        from prefect.utilities.engine import (
            _dynamic_key_for_task_run,
            _resolve_custom_task_run_name,
            collect_task_run_inputs,
        )

        if flow_run_context is None:
            flow_run_context = FlowRunContext.get()
        if parent_task_run_context is None:
            parent_task_run_context = TaskRunContext.get()
        if parameters is None:
            parameters = {}

        try:
            task_run_name = _resolve_custom_task_run_name(self, parameters)
        except TypeError:
            task_run_name = None

        if flow_run_context:
            dynamic_key = _dynamic_key_for_task_run(context=flow_run_context, task=self)
        else:
            dynamic_key = uuid4().hex

        # collect task inputs
        task_inputs = {
            k: await collect_task_run_inputs(v) for k, v in parameters.items()
        }

        # check if this task has a parent task run based on running in another
        # task run's existing context. A task run is only considered a parent if
        # it is in the same flow run (because otherwise presumably the child is
        # in a subflow, so the subflow serves as the parent) or if there is no
        # flow run
        if parent_task_run_context:
            # there is no flow run
            if not flow_run_context:
                task_inputs["__parents__"] = [
                    TaskRunResult(id=parent_task_run_context.task_run.id)
                ]
            # there is a flow run and the task run is in the same flow run
            elif (
                flow_run_context
                and parent_task_run_context.task_run.flow_run_id
                == flow_run_context.flow_run.id
            ):
                task_inputs["__parents__"] = [
                    TaskRunResult(id=parent_task_run_context.task_run.id)
                ]

        if wait_for:
            task_inputs["wait_for"] = await collect_task_run_inputs(wait_for)

        # Join extra task inputs
        for k, extras in (extra_task_inputs or {}).items():
            task_inputs[k] = task_inputs[k].union(extras)

        # create the task run
        task_run = client.create_task_run(
            task=self,
            name=task_run_name,
            flow_run_id=(
                getattr(flow_run_context.flow_run, "id", None)
                if flow_run_context and flow_run_context.flow_run
                else None
            ),
            dynamic_key=str(dynamic_key),
            state=Pending(),
            task_inputs=task_inputs,
            extra_tags=TagsContext.get().current_tags,
        )
        # the new engine uses sync clients but old engines use async clients
        if inspect.isawaitable(task_run):
            task_run = await task_run

        if flow_run_context and flow_run_context.flow_run:
            get_run_logger(flow_run_context).debug(
                f"Created task run {task_run.name!r} for task {self.name!r}"
            )
        else:
            logger.debug(f"Created task run {task_run.name!r} for task {self.name!r}")

        return task_run

    @overload
    def __call__(
        self: "Task[P, NoReturn]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> None:
        # `NoReturn` matches if a type can't be inferred for the function which stops a
        # sync function from matching the `Coroutine` overload
        ...

    @overload
    def __call__(
        self: "Task[P, T]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> T:
        ...

    @overload
    def __call__(
        self: "Task[P, T]",
        *args: P.args,
        return_state: Literal[True],
        **kwargs: P.kwargs,
    ) -> State[T]:
        ...

    def __call__(
        self,
        *args: P.args,
        return_state: bool = False,
        wait_for: Optional[Iterable[PrefectFuture]] = None,
        **kwargs: P.kwargs,
    ):
        """
        Run the task and return the result. If `return_state` is True returns
        the result is wrapped in a Prefect State which provides error handling.
        """
        from prefect.engine import enter_task_run_engine
        from prefect.task_engine import submit_autonomous_task_run_to_engine
        from prefect.task_runners import SequentialTaskRunner

        # Convert the call args/kwargs to a parameter dict
        parameters = get_call_parameters(self.fn, args, kwargs)

        return_type = "state" if return_state else "result"

        task_run_tracker = get_task_viz_tracker()
        if task_run_tracker:
            return track_viz_task(
                self.isasync, self.name, parameters, self.viz_return_value
            )

        if PREFECT_EXPERIMENTAL_ENABLE_NEW_ENGINE.value():
            from prefect.new_task_engine import run_task

            return run_task(
                task=self,
                parameters=parameters,
                wait_for=wait_for,
                return_type=return_type,
            )

        if (
            PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING.value()
            and not FlowRunContext.get()
        ):
            from prefect import get_client

            return submit_autonomous_task_run_to_engine(
                task=self,
                task_run=None,
                task_runner=SequentialTaskRunner(),
                parameters=parameters,
                return_type=return_type,
                client=get_client(),
            )
        entering_from_task_run = bool(TaskRunContext.get())

        return enter_task_run_engine(
            self,
            parameters=parameters,
            wait_for=wait_for,
            task_runner=SequentialTaskRunner(),
            return_type=return_type,
            mapped=False,
            entering_from_task_run=entering_from_task_run,
        )

    @overload
    def _run(
        self: "Task[P, NoReturn]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> PrefectFuture[None, Sync]:
        # `NoReturn` matches if a type can't be inferred for the function which stops a
        # sync function from matching the `Coroutine` overload
        ...

    @overload
    def _run(
        self: "Task[P, Coroutine[Any, Any, T]]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> Awaitable[State[T]]:
        ...

    @overload
    def _run(
        self: "Task[P, T]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> State[T]:
        ...

    def _run(
        self,
        *args: P.args,
        wait_for: Optional[Iterable[PrefectFuture]] = None,
        **kwargs: P.kwargs,
    ) -> Union[State, Awaitable[State]]:
        """
        Run the task and return the final state.
        """
        from prefect.engine import enter_task_run_engine
        from prefect.task_runners import SequentialTaskRunner

        # Convert the call args/kwargs to a parameter dict
        parameters = get_call_parameters(self.fn, args, kwargs)

        return enter_task_run_engine(
            self,
            parameters=parameters,
            wait_for=wait_for,
            return_type="state",
            task_runner=SequentialTaskRunner(),
            mapped=False,
        )

    @overload
    def submit(
        self: "Task[P, NoReturn]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> PrefectFuture[None, Sync]:
        # `NoReturn` matches if a type can't be inferred for the function which stops a
        # sync function from matching the `Coroutine` overload
        ...

    @overload
    def submit(
        self: "Task[P, Coroutine[Any, Any, T]]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> Awaitable[PrefectFuture[T, Async]]:
        ...

    @overload
    def submit(
        self: "Task[P, T]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> PrefectFuture[T, Sync]:
        ...

    @overload
    def submit(
        self: "Task[P, T]",
        *args: P.args,
        return_state: Literal[True],
        **kwargs: P.kwargs,
    ) -> State[T]:
        ...

    @overload
    def submit(
        self: "Task[P, T]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> TaskRun:
        ...

    @overload
    def submit(
        self: "Task[P, Coroutine[Any, Any, T]]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> Awaitable[TaskRun]:
        ...

    def submit(
        self,
        *args: Any,
        return_state: bool = False,
        wait_for: Optional[Iterable[PrefectFuture]] = None,
        **kwargs: Any,
    ) -> Union[PrefectFuture, Awaitable[PrefectFuture], TaskRun, Awaitable[TaskRun]]:
        """
        Submit a run of the task to the engine.

        If writing an async task, this call must be awaited.

        If called from within a flow function,

        Will create a new task run in the backing API and submit the task to the flow's
        task runner. This call only blocks execution while the task is being submitted,
        once it is submitted, the flow function will continue executing. However, note
        that the `SequentialTaskRunner` does not implement parallel execution for sync tasks
        and they are fully resolved on submission.

        Args:
            *args: Arguments to run the task with
            return_state: Return the result of the flow run wrapped in a
                Prefect State.
            wait_for: Upstream task futures to wait for before starting the task
            **kwargs: Keyword arguments to run the task with

        Returns:
            If `return_state` is False a future allowing asynchronous access to
                the state of the task
            If `return_state` is True a future wrapped in a Prefect State allowing asynchronous access to
                the state of the task

        Examples:

            Define a task

            >>> from prefect import task
            >>> @task
            >>> def my_task():
            >>>     return "hello"

            Run a task in a flow

            >>> from prefect import flow
            >>> @flow
            >>> def my_flow():
            >>>     my_task.submit()

            Wait for a task to finish

            >>> @flow
            >>> def my_flow():
            >>>     my_task.submit().wait()

            Use the result from a task in a flow

            >>> @flow
            >>> def my_flow():
            >>>     print(my_task.submit().result())
            >>>
            >>> my_flow()
            hello

            Run an async task in an async flow

            >>> @task
            >>> async def my_async_task():
            >>>     pass
            >>>
            >>> @flow
            >>> async def my_flow():
            >>>     await my_async_task.submit()

            Run a sync task in an async flow

            >>> @flow
            >>> async def my_flow():
            >>>     my_task.submit()

            Enforce ordering between tasks that do not exchange data
            >>> @task
            >>> def task_1():
            >>>     pass
            >>>
            >>> @task
            >>> def task_2():
            >>>     pass
            >>>
            >>> @flow
            >>> def my_flow():
            >>>     x = task_1.submit()
            >>>
            >>>     # task 2 will wait for task_1 to complete
            >>>     y = task_2.submit(wait_for=[x])

        """

        from prefect.engine import create_autonomous_task_run, enter_task_run_engine

        # Convert the call args/kwargs to a parameter dict
        parameters = get_call_parameters(self.fn, args, kwargs)
        return_type = "state" if return_state else "future"
        flow_run_context = FlowRunContext.get()

        task_viz_tracker = get_task_viz_tracker()
        if task_viz_tracker:
            raise VisualizationUnsupportedError(
                "`task.submit()` is not currently supported by `flow.visualize()`"
            )

        if PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING and not flow_run_context:
            create_autonomous_task_run_call = create_call(
                create_autonomous_task_run, task=self, parameters=parameters
            )
            if self.isasync:
                return from_async.wait_for_call_in_loop_thread(
                    create_autonomous_task_run_call
                )
            else:
                return from_sync.wait_for_call_in_loop_thread(
                    create_autonomous_task_run_call
                )
        if PREFECT_EXPERIMENTAL_ENABLE_NEW_ENGINE and flow_run_context:
            if self.isasync:
                return self._submit_async(
                    parameters=parameters,
                    flow_run_context=flow_run_context,
                    wait_for=wait_for,
                    return_state=return_state,
                )
            else:
                raise NotImplementedError(
                    "Submitting sync tasks with the new engine has not be implemented yet."
                )

        else:
            return enter_task_run_engine(
                self,
                parameters=parameters,
                wait_for=wait_for,
                return_type=return_type,
                task_runner=None,  # Use the flow's task runner
                mapped=False,
            )

    async def _submit_async(
        self,
        parameters: Dict[str, Any],
        flow_run_context: FlowRunContext,
        wait_for: Optional[Iterable[PrefectFuture]],
        return_state: bool,
    ):
        from prefect.new_task_engine import run_task_async

        task_runner = flow_run_context.task_runner

        task_run = await self.create_run(
            client=flow_run_context.client,
            flow_run_context=flow_run_context,
            parameters=parameters,
            wait_for=wait_for,
        )

        future = PrefectFuture(
            name=task_run.name,
            key=uuid4(),
            task_runner=task_runner,
            asynchronous=(self.isasync and flow_run_context.flow.isasync),
        )
        future.task_run = task_run
        flow_run_context.task_run_futures.append(future)
        await task_runner.submit(
            key=future.key,
            call=partial(
                run_task_async,
                task=self,
                task_run=task_run,
                parameters=parameters,
                wait_for=wait_for,
                return_type="state",
            ),
        )
        # TODO: I don't like this. Can we move responsibility for creating the future
        # and setting this anyio.Event to the task runner?
        future._submitted.set()

        if return_state:
            return await future.wait()
        else:
            return future

    @overload
    def map(
        self: "Task[P, NoReturn]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> List[PrefectFuture[None, Sync]]:
        # `NoReturn` matches if a type can't be inferred for the function which stops a
        # sync function from matching the `Coroutine` overload
        ...

    @overload
    def map(
        self: "Task[P, Coroutine[Any, Any, T]]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> Awaitable[List[PrefectFuture[T, Async]]]:
        ...

    @overload
    def map(
        self: "Task[P, T]",
        *args: P.args,
        **kwargs: P.kwargs,
    ) -> List[PrefectFuture[T, Sync]]:
        ...

    @overload
    def map(
        self: "Task[P, T]",
        *args: P.args,
        return_state: Literal[True],
        **kwargs: P.kwargs,
    ) -> List[State[T]]:
        ...

    def map(
        self,
        *args: Any,
        return_state: bool = False,
        wait_for: Optional[Iterable[PrefectFuture]] = None,
        **kwargs: Any,
    ) -> Any:
        """
        Submit a mapped run of the task to a worker.

        Must be called within a flow function. If writing an async task, this
        call must be awaited.

        Must be called with at least one iterable and all iterables must be
        the same length. Any arguments that are not iterable will be treated as
        a static value and each task run will receive the same value.

        Will create as many task runs as the length of the iterable(s) in the
        backing API and submit the task runs to the flow's task runner. This
        call blocks if given a future as input while the future is resolved. It
        also blocks while the tasks are being submitted, once they are
        submitted, the flow function will continue executing. However, note
        that the `SequentialTaskRunner` does not implement parallel execution
        for sync tasks and they are fully resolved on submission.

        Args:
            *args: Iterable and static arguments to run the tasks with
            return_state: Return a list of Prefect States that wrap the results
                of each task run.
            wait_for: Upstream task futures to wait for before starting the
                task
            **kwargs: Keyword iterable arguments to run the task with

        Returns:
            A list of futures allowing asynchronous access to the state of the
            tasks

        Examples:

            Define a task

            >>> from prefect import task
            >>> @task
            >>> def my_task(x):
            >>>     return x + 1

            Create mapped tasks

            >>> from prefect import flow
            >>> @flow
            >>> def my_flow():
            >>>     my_task.map([1, 2, 3])

            Wait for all mapped tasks to finish

            >>> @flow
            >>> def my_flow():
            >>>     futures = my_task.map([1, 2, 3])
            >>>     for future in futures:
            >>>         future.wait()
            >>>     # Now all of the mapped tasks have finished
            >>>     my_task(10)

            Use the result from mapped tasks in a flow

            >>> @flow
            >>> def my_flow():
            >>>     futures = my_task.map([1, 2, 3])
            >>>     for future in futures:
            >>>         print(future.result())
            >>> my_flow()
            2
            3
            4

            Enforce ordering between tasks that do not exchange data
            >>> @task
            >>> def task_1(x):
            >>>     pass
            >>>
            >>> @task
            >>> def task_2(y):
            >>>     pass
            >>>
            >>> @flow
            >>> def my_flow():
            >>>     x = task_1.submit()
            >>>
            >>>     # task 2 will wait for task_1 to complete
            >>>     y = task_2.map([1, 2, 3], wait_for=[x])

            Use a non-iterable input as a constant across mapped tasks
            >>> @task
            >>> def display(prefix, item):
            >>>    print(prefix, item)
            >>>
            >>> @flow
            >>> def my_flow():
            >>>     display.map("Check it out: ", [1, 2, 3])
            >>>
            >>> my_flow()
            Check it out: 1
            Check it out: 2
            Check it out: 3

            Use `unmapped` to treat an iterable argument as a constant
            >>> from prefect import unmapped
            >>>
            >>> @task
            >>> def add_n_to_items(items, n):
            >>>     return [item + n for item in items]
            >>>
            >>> @flow
            >>> def my_flow():
            >>>     return add_n_to_items.map(unmapped([10, 20]), n=[1, 2, 3])
            >>>
            >>> my_flow()
            [[11, 21], [12, 22], [13, 23]]
        """

        from prefect.engine import begin_task_map, enter_task_run_engine

        # Convert the call args/kwargs to a parameter dict; do not apply defaults
        # since they should not be mapped over
        parameters = get_call_parameters(self.fn, args, kwargs, apply_defaults=False)
        return_type = "state" if return_state else "future"

        task_viz_tracker = get_task_viz_tracker()
        if task_viz_tracker:
            raise VisualizationUnsupportedError(
                "`task.map()` is not currently supported by `flow.visualize()`"
            )

        if (
            PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING.value()
            and not FlowRunContext.get()
        ):
            map_call = create_call(
                begin_task_map,
                task=self,
                parameters=parameters,
                flow_run_context=None,
                wait_for=wait_for,
                return_type=return_type,
                task_runner=None,
                autonomous=True,
            )
            if self.isasync:
                return from_async.wait_for_call_in_loop_thread(map_call)
            else:
                return from_sync.wait_for_call_in_loop_thread(map_call)

        return enter_task_run_engine(
            self,
            parameters=parameters,
            wait_for=wait_for,
            return_type=return_type,
            task_runner=None,
            mapped=True,
        )

    def serve(self, task_runner: Optional[BaseTaskRunner] = None) -> "Task":
        """Serve the task using the provided task runner. This method is used to
        establish a websocket connection with the Prefect server and listen for
        submitted task runs to execute.

        Args:
            task_runner: The task runner to use for serving the task. If not provided,
                the default ConcurrentTaskRunner will be used.

        Examples:
            Serve a task using the default task runner
            >>> @task
            >>> def my_task():
            >>>     return 1

            >>> my_task.serve()
        """

        if not PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING:
            raise ValueError(
                "Task's `serve` method is an experimental feature and must be enabled with "
                "`prefect config set PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING=True`"
            )

        from prefect.task_server import serve

        serve(self, task_runner=task_runner)

map

Submit a mapped run of the task to a worker.

Must be called within a flow function. If writing an async task, this call must be awaited.

Must be called with at least one iterable and all iterables must be the same length. Any arguments that are not iterable will be treated as a static value and each task run will receive the same value.

Will create as many task runs as the length of the iterable(s) in the backing API and submit the task runs to the flow's task runner. This call blocks if given a future as input while the future is resolved. It also blocks while the tasks are being submitted, once they are submitted, the flow function will continue executing. However, note that the SequentialTaskRunner does not implement parallel execution for sync tasks and they are fully resolved on submission.

Parameters:

Name Type Description Default
*args Any

Iterable and static arguments to run the tasks with

()
return_state bool

Return a list of Prefect States that wrap the results of each task run.

False
wait_for Optional[Iterable[PrefectFuture]]

Upstream task futures to wait for before starting the task

None
**kwargs Any

Keyword iterable arguments to run the task with

{}

Returns:

Type Description
Any

A list of futures allowing asynchronous access to the state of the

Any

tasks

Define a task

>>> from prefect import task
>>> @task
>>> def my_task(x):
>>>     return x + 1

Create mapped tasks

>>> from prefect import flow
>>> @flow
>>> def my_flow():
>>>     my_task.map([1, 2, 3])

Wait for all mapped tasks to finish

>>> @flow
>>> def my_flow():
>>>     futures = my_task.map([1, 2, 3])
>>>     for future in futures:
>>>         future.wait()
>>>     # Now all of the mapped tasks have finished
>>>     my_task(10)

Use the result from mapped tasks in a flow

>>> @flow
>>> def my_flow():
>>>     futures = my_task.map([1, 2, 3])
>>>     for future in futures:
>>>         print(future.result())
>>> my_flow()
2
3
4

Enforce ordering between tasks that do not exchange data
>>> @task
>>> def task_1(x):
>>>     pass
>>>
>>> @task
>>> def task_2(y):
>>>     pass
>>>
>>> @flow
>>> def my_flow():
>>>     x = task_1.submit()
>>>
>>>     # task 2 will wait for task_1 to complete
>>>     y = task_2.map([1, 2, 3], wait_for=[x])

Use a non-iterable input as a constant across mapped tasks
>>> @task
>>> def display(prefix, item):
>>>    print(prefix, item)
>>>
>>> @flow
>>> def my_flow():
>>>     display.map("Check it out: ", [1, 2, 3])
>>>
>>> my_flow()
Check it out: 1
Check it out: 2
Check it out: 3

Use `unmapped` to treat an iterable argument as a constant
>>> from prefect import unmapped
>>>
>>> @task
>>> def add_n_to_items(items, n):
>>>     return [item + n for item in items]
>>>
>>> @flow
>>> def my_flow():
>>>     return add_n_to_items.map(unmapped([10, 20]), n=[1, 2, 3])
>>>
>>> my_flow()
[[11, 21], [12, 22], [13, 23]]
Source code in prefect/tasks.py
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
def map(
    self,
    *args: Any,
    return_state: bool = False,
    wait_for: Optional[Iterable[PrefectFuture]] = None,
    **kwargs: Any,
) -> Any:
    """
    Submit a mapped run of the task to a worker.

    Must be called within a flow function. If writing an async task, this
    call must be awaited.

    Must be called with at least one iterable and all iterables must be
    the same length. Any arguments that are not iterable will be treated as
    a static value and each task run will receive the same value.

    Will create as many task runs as the length of the iterable(s) in the
    backing API and submit the task runs to the flow's task runner. This
    call blocks if given a future as input while the future is resolved. It
    also blocks while the tasks are being submitted, once they are
    submitted, the flow function will continue executing. However, note
    that the `SequentialTaskRunner` does not implement parallel execution
    for sync tasks and they are fully resolved on submission.

    Args:
        *args: Iterable and static arguments to run the tasks with
        return_state: Return a list of Prefect States that wrap the results
            of each task run.
        wait_for: Upstream task futures to wait for before starting the
            task
        **kwargs: Keyword iterable arguments to run the task with

    Returns:
        A list of futures allowing asynchronous access to the state of the
        tasks

    Examples:

        Define a task

        >>> from prefect import task
        >>> @task
        >>> def my_task(x):
        >>>     return x + 1

        Create mapped tasks

        >>> from prefect import flow
        >>> @flow
        >>> def my_flow():
        >>>     my_task.map([1, 2, 3])

        Wait for all mapped tasks to finish

        >>> @flow
        >>> def my_flow():
        >>>     futures = my_task.map([1, 2, 3])
        >>>     for future in futures:
        >>>         future.wait()
        >>>     # Now all of the mapped tasks have finished
        >>>     my_task(10)

        Use the result from mapped tasks in a flow

        >>> @flow
        >>> def my_flow():
        >>>     futures = my_task.map([1, 2, 3])
        >>>     for future in futures:
        >>>         print(future.result())
        >>> my_flow()
        2
        3
        4

        Enforce ordering between tasks that do not exchange data
        >>> @task
        >>> def task_1(x):
        >>>     pass
        >>>
        >>> @task
        >>> def task_2(y):
        >>>     pass
        >>>
        >>> @flow
        >>> def my_flow():
        >>>     x = task_1.submit()
        >>>
        >>>     # task 2 will wait for task_1 to complete
        >>>     y = task_2.map([1, 2, 3], wait_for=[x])

        Use a non-iterable input as a constant across mapped tasks
        >>> @task
        >>> def display(prefix, item):
        >>>    print(prefix, item)
        >>>
        >>> @flow
        >>> def my_flow():
        >>>     display.map("Check it out: ", [1, 2, 3])
        >>>
        >>> my_flow()
        Check it out: 1
        Check it out: 2
        Check it out: 3

        Use `unmapped` to treat an iterable argument as a constant
        >>> from prefect import unmapped
        >>>
        >>> @task
        >>> def add_n_to_items(items, n):
        >>>     return [item + n for item in items]
        >>>
        >>> @flow
        >>> def my_flow():
        >>>     return add_n_to_items.map(unmapped([10, 20]), n=[1, 2, 3])
        >>>
        >>> my_flow()
        [[11, 21], [12, 22], [13, 23]]
    """

    from prefect.engine import begin_task_map, enter_task_run_engine

    # Convert the call args/kwargs to a parameter dict; do not apply defaults
    # since they should not be mapped over
    parameters = get_call_parameters(self.fn, args, kwargs, apply_defaults=False)
    return_type = "state" if return_state else "future"

    task_viz_tracker = get_task_viz_tracker()
    if task_viz_tracker:
        raise VisualizationUnsupportedError(
            "`task.map()` is not currently supported by `flow.visualize()`"
        )

    if (
        PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING.value()
        and not FlowRunContext.get()
    ):
        map_call = create_call(
            begin_task_map,
            task=self,
            parameters=parameters,
            flow_run_context=None,
            wait_for=wait_for,
            return_type=return_type,
            task_runner=None,
            autonomous=True,
        )
        if self.isasync:
            return from_async.wait_for_call_in_loop_thread(map_call)
        else:
            return from_sync.wait_for_call_in_loop_thread(map_call)

    return enter_task_run_engine(
        self,
        parameters=parameters,
        wait_for=wait_for,
        return_type=return_type,
        task_runner=None,
        mapped=True,
    )

serve

Serve the task using the provided task runner. This method is used to establish a websocket connection with the Prefect server and listen for submitted task runs to execute.

Parameters:

Name Type Description Default
task_runner Optional[BaseTaskRunner]

The task runner to use for serving the task. If not provided, the default ConcurrentTaskRunner will be used.

None

Examples:

Serve a task using the default task runner

>>> @task
>>> def my_task():
>>>     return 1
>>> my_task.serve()
Source code in prefect/tasks.py
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
def serve(self, task_runner: Optional[BaseTaskRunner] = None) -> "Task":
    """Serve the task using the provided task runner. This method is used to
    establish a websocket connection with the Prefect server and listen for
    submitted task runs to execute.

    Args:
        task_runner: The task runner to use for serving the task. If not provided,
            the default ConcurrentTaskRunner will be used.

    Examples:
        Serve a task using the default task runner
        >>> @task
        >>> def my_task():
        >>>     return 1

        >>> my_task.serve()
    """

    if not PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING:
        raise ValueError(
            "Task's `serve` method is an experimental feature and must be enabled with "
            "`prefect config set PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING=True`"
        )

    from prefect.task_server import serve

    serve(self, task_runner=task_runner)

submit

Submit a run of the task to the engine.

If writing an async task, this call must be awaited.

If called from within a flow function,

Will create a new task run in the backing API and submit the task to the flow's task runner. This call only blocks execution while the task is being submitted, once it is submitted, the flow function will continue executing. However, note that the SequentialTaskRunner does not implement parallel execution for sync tasks and they are fully resolved on submission.

Parameters:

Name Type Description Default
*args Any

Arguments to run the task with

()
return_state bool

Return the result of the flow run wrapped in a Prefect State.

False
wait_for Optional[Iterable[PrefectFuture]]

Upstream task futures to wait for before starting the task

None
**kwargs Any

Keyword arguments to run the task with

{}

Returns:

Type Description
Union[PrefectFuture, Awaitable[PrefectFuture], TaskRun, Awaitable[TaskRun]]

If return_state is False a future allowing asynchronous access to the state of the task

Union[PrefectFuture, Awaitable[PrefectFuture], TaskRun, Awaitable[TaskRun]]

If return_state is True a future wrapped in a Prefect State allowing asynchronous access to the state of the task

Define a task

>>> from prefect import task
>>> @task
>>> def my_task():
>>>     return "hello"

Run a task in a flow

>>> from prefect import flow
>>> @flow
>>> def my_flow():
>>>     my_task.submit()

Wait for a task to finish

>>> @flow
>>> def my_flow():
>>>     my_task.submit().wait()

Use the result from a task in a flow

>>> @flow
>>> def my_flow():
>>>     print(my_task.submit().result())
>>>
>>> my_flow()
hello

Run an async task in an async flow

>>> @task
>>> async def my_async_task():
>>>     pass
>>>
>>> @flow
>>> async def my_flow():
>>>     await my_async_task.submit()

Run a sync task in an async flow

>>> @flow
>>> async def my_flow():
>>>     my_task.submit()

Enforce ordering between tasks that do not exchange data
>>> @task
>>> def task_1():
>>>     pass
>>>
>>> @task
>>> def task_2():
>>>     pass
>>>
>>> @flow
>>> def my_flow():
>>>     x = task_1.submit()
>>>
>>>     # task 2 will wait for task_1 to complete
>>>     y = task_2.submit(wait_for=[x])
Source code in prefect/tasks.py
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
def submit(
    self,
    *args: Any,
    return_state: bool = False,
    wait_for: Optional[Iterable[PrefectFuture]] = None,
    **kwargs: Any,
) -> Union[PrefectFuture, Awaitable[PrefectFuture], TaskRun, Awaitable[TaskRun]]:
    """
    Submit a run of the task to the engine.

    If writing an async task, this call must be awaited.

    If called from within a flow function,

    Will create a new task run in the backing API and submit the task to the flow's
    task runner. This call only blocks execution while the task is being submitted,
    once it is submitted, the flow function will continue executing. However, note
    that the `SequentialTaskRunner` does not implement parallel execution for sync tasks
    and they are fully resolved on submission.

    Args:
        *args: Arguments to run the task with
        return_state: Return the result of the flow run wrapped in a
            Prefect State.
        wait_for: Upstream task futures to wait for before starting the task
        **kwargs: Keyword arguments to run the task with

    Returns:
        If `return_state` is False a future allowing asynchronous access to
            the state of the task
        If `return_state` is True a future wrapped in a Prefect State allowing asynchronous access to
            the state of the task

    Examples:

        Define a task

        >>> from prefect import task
        >>> @task
        >>> def my_task():
        >>>     return "hello"

        Run a task in a flow

        >>> from prefect import flow
        >>> @flow
        >>> def my_flow():
        >>>     my_task.submit()

        Wait for a task to finish

        >>> @flow
        >>> def my_flow():
        >>>     my_task.submit().wait()

        Use the result from a task in a flow

        >>> @flow
        >>> def my_flow():
        >>>     print(my_task.submit().result())
        >>>
        >>> my_flow()
        hello

        Run an async task in an async flow

        >>> @task
        >>> async def my_async_task():
        >>>     pass
        >>>
        >>> @flow
        >>> async def my_flow():
        >>>     await my_async_task.submit()

        Run a sync task in an async flow

        >>> @flow
        >>> async def my_flow():
        >>>     my_task.submit()

        Enforce ordering between tasks that do not exchange data
        >>> @task
        >>> def task_1():
        >>>     pass
        >>>
        >>> @task
        >>> def task_2():
        >>>     pass
        >>>
        >>> @flow
        >>> def my_flow():
        >>>     x = task_1.submit()
        >>>
        >>>     # task 2 will wait for task_1 to complete
        >>>     y = task_2.submit(wait_for=[x])

    """

    from prefect.engine import create_autonomous_task_run, enter_task_run_engine

    # Convert the call args/kwargs to a parameter dict
    parameters = get_call_parameters(self.fn, args, kwargs)
    return_type = "state" if return_state else "future"
    flow_run_context = FlowRunContext.get()

    task_viz_tracker = get_task_viz_tracker()
    if task_viz_tracker:
        raise VisualizationUnsupportedError(
            "`task.submit()` is not currently supported by `flow.visualize()`"
        )

    if PREFECT_EXPERIMENTAL_ENABLE_TASK_SCHEDULING and not flow_run_context:
        create_autonomous_task_run_call = create_call(
            create_autonomous_task_run, task=self, parameters=parameters
        )
        if self.isasync:
            return from_async.wait_for_call_in_loop_thread(
                create_autonomous_task_run_call
            )
        else:
            return from_sync.wait_for_call_in_loop_thread(
                create_autonomous_task_run_call
            )
    if PREFECT_EXPERIMENTAL_ENABLE_NEW_ENGINE and flow_run_context:
        if self.isasync:
            return self._submit_async(
                parameters=parameters,
                flow_run_context=flow_run_context,
                wait_for=wait_for,
                return_state=return_state,
            )
        else:
            raise NotImplementedError(
                "Submitting sync tasks with the new engine has not be implemented yet."
            )

    else:
        return enter_task_run_engine(
            self,
            parameters=parameters,
            wait_for=wait_for,
            return_type=return_type,
            task_runner=None,  # Use the flow's task runner
            mapped=False,
        )

with_options

Create a new task from the current object, updating provided options.

Parameters:

Name Type Description Default
name str

A new name for the task.

None
description str

A new description for the task.

None
tags Iterable[str]

A new set of tags for the task. If given, existing tags are ignored, not merged.

None
cache_key_fn Callable[[TaskRunContext, Dict[str, Any]], Optional[str]]

A new cache key function for the task.

None
cache_expiration timedelta

A new cache expiration time for the task.

None
task_run_name Optional[Union[Callable[[], str], str]]

An optional name to distinguish runs of this task; this name can be provided as a string template with the task's keyword arguments as variables, or a function that returns a string.

None
retries Optional[int]

A new number of times to retry on task run failure.

NotSet
retry_delay_seconds Union[float, int, List[float], Callable[[int], List[float]]]

Optionally configures how long to wait before retrying the task after failure. This is only applicable if retries is nonzero. This setting can either be a number of seconds, a list of retry delays, or a callable that, given the total number of retries, generates a list of retry delays. If a number of seconds, that delay will be applied to all retries. If a list, each retry will wait for the corresponding delay before retrying. When passing a callable or a list, the number of configured retry delays cannot exceed 50.

NotSet
retry_jitter_factor Optional[float]

An optional factor that defines the factor to which a retry can be jittered in order to avoid a "thundering herd".

NotSet
persist_result Optional[bool]

A new option for enabling or disabling result persistence.

NotSet
result_storage Optional[ResultStorage]

A new storage type to use for results.

NotSet
result_serializer Optional[ResultSerializer]

A new serializer to use for results.

NotSet
result_storage_key Optional[str]

A new key for the persisted result to be stored at.

NotSet
timeout_seconds Union[int, float]

A new maximum time for the task to complete in seconds.

None
log_prints Optional[bool]

A new option for enabling or disabling redirection of print statements.

NotSet
refresh_cache Optional[bool]

A new option for enabling or disabling cache refresh.

NotSet
on_completion Optional[List[Callable[[Task, TaskRun, State], Union[Awaitable[None], None]]]]

A new list of callables to run when the task enters a completed state.

None
on_failure Optional[List[Callable[[Task, TaskRun, State], Union[Awaitable[None], None]]]]

A new list of callables to run when the task enters a failed state.

None
retry_condition_fn Optional[Callable[[Task, TaskRun, State], bool]]

An optional callable run when a task run returns a Failed state. Should return True if the task should continue to its retry policy, and False if the task should end as failed. Defaults to None, indicating the task should always continue to its retry policy.

None
viz_return_value Optional[Any]

An optional value to return when the task dependency tree is visualized.

None

Returns:

Type Description

A new Task instance.

Create a new task from an existing task and update the name

>>> @task(name="My task")
>>> def my_task():
>>>     return 1
>>>
>>> new_task = my_task.with_options(name="My new task")

Create a new task from an existing task and update the retry settings

>>> from random import randint
>>>
>>> @task(retries=1, retry_delay_seconds=5)
>>> def my_task():
>>>     x = randint(0, 5)
>>>     if x >= 3:  # Make a task that fails sometimes
>>>         raise ValueError("Retry me please!")
>>>     return x
>>>
>>> new_task = my_task.with_options(retries=5, retry_delay_seconds=2)

Use a task with updated options within a flow

>>> @task(name="My task")
>>> def my_task():
>>>     return 1
>>>
>>> @flow
>>> my_flow():
>>>     new_task = my_task.with_options(name="My new task")
>>>     new_task()
Source code in prefect/tasks.py
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
def with_options(
    self,
    *,
    name: str = None,
    description: str = None,
    tags: Iterable[str] = None,
    cache_key_fn: Callable[
        ["TaskRunContext", Dict[str, Any]], Optional[str]
    ] = None,
    task_run_name: Optional[Union[Callable[[], str], str]] = None,
    cache_expiration: datetime.timedelta = None,
    retries: Optional[int] = NotSet,
    retry_delay_seconds: Union[
        float,
        int,
        List[float],
        Callable[[int], List[float]],
    ] = NotSet,
    retry_jitter_factor: Optional[float] = NotSet,
    persist_result: Optional[bool] = NotSet,
    result_storage: Optional[ResultStorage] = NotSet,
    result_serializer: Optional[ResultSerializer] = NotSet,
    result_storage_key: Optional[str] = NotSet,
    cache_result_in_memory: Optional[bool] = None,
    timeout_seconds: Union[int, float] = None,
    log_prints: Optional[bool] = NotSet,
    refresh_cache: Optional[bool] = NotSet,
    on_completion: Optional[
        List[Callable[["Task", TaskRun, State], Union[Awaitable[None], None]]]
    ] = None,
    on_failure: Optional[
        List[Callable[["Task", TaskRun, State], Union[Awaitable[None], None]]]
    ] = None,
    retry_condition_fn: Optional[Callable[["Task", TaskRun, State], bool]] = None,
    viz_return_value: Optional[Any] = None,
):
    """
    Create a new task from the current object, updating provided options.

    Args:
        name: A new name for the task.
        description: A new description for the task.
        tags: A new set of tags for the task. If given, existing tags are ignored,
            not merged.
        cache_key_fn: A new cache key function for the task.
        cache_expiration: A new cache expiration time for the task.
        task_run_name: An optional name to distinguish runs of this task; this name can be provided
            as a string template with the task's keyword arguments as variables,
            or a function that returns a string.
        retries: A new number of times to retry on task run failure.
        retry_delay_seconds: Optionally configures how long to wait before retrying
            the task after failure. This is only applicable if `retries` is nonzero.
            This setting can either be a number of seconds, a list of retry delays,
            or a callable that, given the total number of retries, generates a list
            of retry delays. If a number of seconds, that delay will be applied to
            all retries. If a list, each retry will wait for the corresponding delay
            before retrying. When passing a callable or a list, the number of
            configured retry delays cannot exceed 50.
        retry_jitter_factor: An optional factor that defines the factor to which a
            retry can be jittered in order to avoid a "thundering herd".
        persist_result: A new option for enabling or disabling result persistence.
        result_storage: A new storage type to use for results.
        result_serializer: A new serializer to use for results.
        result_storage_key: A new key for the persisted result to be stored at.
        timeout_seconds: A new maximum time for the task to complete in seconds.
        log_prints: A new option for enabling or disabling redirection of `print` statements.
        refresh_cache: A new option for enabling or disabling cache refresh.
        on_completion: A new list of callables to run when the task enters a completed state.
        on_failure: A new list of callables to run when the task enters a failed state.
        retry_condition_fn: An optional callable run when a task run returns a Failed state.
            Should return `True` if the task should continue to its retry policy, and `False`
            if the task should end as failed. Defaults to `None`, indicating the task should
            always continue to its retry policy.
        viz_return_value: An optional value to return when the task dependency tree is visualized.

    Returns:
        A new `Task` instance.

    Examples:

        Create a new task from an existing task and update the name

        >>> @task(name="My task")
        >>> def my_task():
        >>>     return 1
        >>>
        >>> new_task = my_task.with_options(name="My new task")

        Create a new task from an existing task and update the retry settings

        >>> from random import randint
        >>>
        >>> @task(retries=1, retry_delay_seconds=5)
        >>> def my_task():
        >>>     x = randint(0, 5)
        >>>     if x >= 3:  # Make a task that fails sometimes
        >>>         raise ValueError("Retry me please!")
        >>>     return x
        >>>
        >>> new_task = my_task.with_options(retries=5, retry_delay_seconds=2)

        Use a task with updated options within a flow

        >>> @task(name="My task")
        >>> def my_task():
        >>>     return 1
        >>>
        >>> @flow
        >>> my_flow():
        >>>     new_task = my_task.with_options(name="My new task")
        >>>     new_task()
    """
    return Task(
        fn=self.fn,
        name=name or self.name,
        description=description or self.description,
        tags=tags or copy(self.tags),
        cache_key_fn=cache_key_fn or self.cache_key_fn,
        cache_expiration=cache_expiration or self.cache_expiration,
        task_run_name=task_run_name,
        retries=retries if retries is not NotSet else self.retries,
        retry_delay_seconds=(
            retry_delay_seconds
            if retry_delay_seconds is not NotSet
            else self.retry_delay_seconds
        ),
        retry_jitter_factor=(
            retry_jitter_factor
            if retry_jitter_factor is not NotSet
            else self.retry_jitter_factor
        ),
        persist_result=(
            persist_result if persist_result is not NotSet else self.persist_result
        ),
        result_storage=(
            result_storage if result_storage is not NotSet else self.result_storage
        ),
        result_storage_key=(
            result_storage_key
            if result_storage_key is not NotSet
            else self.result_storage_key
        ),
        result_serializer=(
            result_serializer
            if result_serializer is not NotSet
            else self.result_serializer
        ),
        cache_result_in_memory=(
            cache_result_in_memory
            if cache_result_in_memory is not None
            else self.cache_result_in_memory
        ),
        timeout_seconds=(
            timeout_seconds if timeout_seconds is not None else self.timeout_seconds
        ),
        log_prints=(log_prints if log_prints is not NotSet else self.log_prints),
        refresh_cache=(
            refresh_cache if refresh_cache is not NotSet else self.refresh_cache
        ),
        on_completion=on_completion or self.on_completion,
        on_failure=on_failure or self.on_failure,
        retry_condition_fn=retry_condition_fn or self.retry_condition_fn,
        viz_return_value=viz_return_value or self.viz_return_value,
    )

exponential_backoff

A task retry backoff utility that configures exponential backoff for task retries. The exponential backoff design matches the urllib3 implementation.

Parameters:

Name Type Description Default
backoff_factor float

the base delay for the first retry, subsequent retries will increase the delay time by powers of 2.

required

Returns:

Type Description
Callable[[int], List[float]]

a callable that can be passed to the task constructor

Source code in prefect/tasks.py
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
def exponential_backoff(backoff_factor: float) -> Callable[[int], List[float]]:
    """
    A task retry backoff utility that configures exponential backoff for task retries.
    The exponential backoff design matches the urllib3 implementation.

    Arguments:
        backoff_factor: the base delay for the first retry, subsequent retries will
            increase the delay time by powers of 2.

    Returns:
        a callable that can be passed to the task constructor
    """

    def retry_backoff_callable(retries: int) -> List[float]:
        # no more than 50 retry delays can be configured on a task
        retries = min(retries, 50)

        return [backoff_factor * max(0, 2**r) for r in range(retries)]

    return retry_backoff_callable

task

Decorator to designate a function as a task in a Prefect workflow.

This decorator may be used for asynchronous or synchronous functions.

Parameters:

Name Type Description Default
name str

An optional name for the task; if not provided, the name will be inferred from the given function.

None
description str

An optional string description for the task.

None
tags Iterable[str]

An optional set of tags to be associated with runs of this task. These tags are combined with any tags defined by a prefect.tags context at task runtime.

None
version str

An optional string specifying the version of this task definition

None
cache_key_fn Callable[[TaskRunContext, Dict[str, Any]], Optional[str]]

An optional callable that, given the task run context and call parameters, generates a string key; if the key matches a previous completed state, that state result will be restored instead of running the task again.

None
cache_expiration timedelta

An optional amount of time indicating how long cached states for this task should be restorable; if not provided, cached states will never expire.

None
task_run_name Optional[Union[Callable[[], str], str]]

An optional name to distinguish runs of this task; this name can be provided as a string template with the task's keyword arguments as variables, or a function that returns a string.

None
retries int

An optional number of times to retry on task run failure

None
retry_delay_seconds Union[float, int, List[float], Callable[[int], List[float]]]

Optionally configures how long to wait before retrying the task after failure. This is only applicable if retries is nonzero. This setting can either be a number of seconds, a list of retry delays, or a callable that, given the total number of retries, generates a list of retry delays. If a number of seconds, that delay will be applied to all retries. If a list, each retry will wait for the corresponding delay before retrying. When passing a callable or a list, the number of configured retry delays cannot exceed 50.

None
retry_jitter_factor Optional[float]

An optional factor that defines the factor to which a retry can be jittered in order to avoid a "thundering herd".

None
persist_result Optional[bool]

An optional toggle indicating whether the result of this task should be persisted to result storage. Defaults to None, which indicates that Prefect should choose whether the result should be persisted depending on the features being used.

None
result_storage Optional[ResultStorage]

An optional block to use to persist the result of this task. Defaults to the value set in the flow the task is called in.

None
result_storage_key Optional[str]

An optional key to store the result in storage at when persisted. Defaults to a unique identifier.

None
result_serializer Optional[ResultSerializer]

An optional serializer to use to serialize the result of this task for persistence. Defaults to the value set in the flow the task is called in.

None
timeout_seconds Union[int, float]

An optional number of seconds indicating a maximum runtime for the task. If the task exceeds this runtime, it will be marked as failed.

None
log_prints Optional[bool]

If set, print statements in the task will be redirected to the Prefect logger for the task run. Defaults to None, which indicates that the value from the flow should be used.

None
refresh_cache Optional[bool]

If set, cached results for the cache key are not used. Defaults to None, which indicates that a cached result from a previous execution with matching cache key is used.

None
on_failure Optional[List[Callable[[Task, TaskRun, State], None]]]

An optional list of callables to run when the task enters a failed state.

None
on_completion Optional[List[Callable[[Task, TaskRun, State], None]]]

An optional list of callables to run when the task enters a completed state.

None
retry_condition_fn Optional[Callable[[Task, TaskRun, State], bool]]

An optional callable run when a task run returns a Failed state. Should return True if the task should continue to its retry policy (e.g. retries=3), and False if the task should end as failed. Defaults to None, indicating the task should always continue to its retry policy.

None
viz_return_value Any

An optional value to return when the task dependency tree is visualized.

None

Returns:

Type Description

A callable Task object which, when called, will submit the task for execution.

Examples:

Define a simple task

>>> @task
>>> def add(x, y):
>>>     return x + y

Define an async task

>>> @task
>>> async def add(x, y):
>>>     return x + y

Define a task with tags and a description

>>> @task(tags={"a", "b"}, description="This task is empty but its my first!")
>>> def my_task():
>>>     pass

Define a task with a custom name

>>> @task(name="The Ultimate Task")
>>> def my_task():
>>>     pass

Define a task that retries 3 times with a 5 second delay between attempts

>>> from random import randint
>>>
>>> @task(retries=3, retry_delay_seconds=5)
>>> def my_task():
>>>     x = randint(0, 5)
>>>     if x >= 3:  # Make a task that fails sometimes
>>>         raise ValueError("Retry me please!")
>>>     return x

Define a task that is cached for a day based on its inputs

>>> from prefect.tasks import task_input_hash
>>> from datetime import timedelta
>>>
>>> @task(cache_key_fn=task_input_hash, cache_expiration=timedelta(days=1))
>>> def my_task():
>>>     return "hello"
Source code in prefect/tasks.py
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
def task(
    __fn=None,
    *,
    name: str = None,
    description: str = None,
    tags: Iterable[str] = None,
    version: str = None,
    cache_key_fn: Callable[["TaskRunContext", Dict[str, Any]], Optional[str]] = None,
    cache_expiration: datetime.timedelta = None,
    task_run_name: Optional[Union[Callable[[], str], str]] = None,
    retries: int = None,
    retry_delay_seconds: Union[
        float,
        int,
        List[float],
        Callable[[int], List[float]],
    ] = None,
    retry_jitter_factor: Optional[float] = None,
    persist_result: Optional[bool] = None,
    result_storage: Optional[ResultStorage] = None,
    result_storage_key: Optional[str] = None,
    result_serializer: Optional[ResultSerializer] = None,
    cache_result_in_memory: bool = True,
    timeout_seconds: Union[int, float] = None,
    log_prints: Optional[bool] = None,
    refresh_cache: Optional[bool] = None,
    on_completion: Optional[List[Callable[["Task", TaskRun, State], None]]] = None,
    on_failure: Optional[List[Callable[["Task", TaskRun, State], None]]] = None,
    retry_condition_fn: Optional[Callable[["Task", TaskRun, State], bool]] = None,
    viz_return_value: Any = None,
):
    """
    Decorator to designate a function as a task in a Prefect workflow.

    This decorator may be used for asynchronous or synchronous functions.

    Args:
        name: An optional name for the task; if not provided, the name will be inferred
            from the given function.
        description: An optional string description for the task.
        tags: An optional set of tags to be associated with runs of this task. These
            tags are combined with any tags defined by a `prefect.tags` context at
            task runtime.
        version: An optional string specifying the version of this task definition
        cache_key_fn: An optional callable that, given the task run context and call
            parameters, generates a string key; if the key matches a previous completed
            state, that state result will be restored instead of running the task again.
        cache_expiration: An optional amount of time indicating how long cached states
            for this task should be restorable; if not provided, cached states will
            never expire.
        task_run_name: An optional name to distinguish runs of this task; this name can be provided
            as a string template with the task's keyword arguments as variables,
            or a function that returns a string.
        retries: An optional number of times to retry on task run failure
        retry_delay_seconds: Optionally configures how long to wait before retrying the
            task after failure. This is only applicable if `retries` is nonzero. This
            setting can either be a number of seconds, a list of retry delays, or a
            callable that, given the total number of retries, generates a list of retry
            delays. If a number of seconds, that delay will be applied to all retries.
            If a list, each retry will wait for the corresponding delay before retrying.
            When passing a callable or a list, the number of configured retry delays
            cannot exceed 50.
        retry_jitter_factor: An optional factor that defines the factor to which a retry
            can be jittered in order to avoid a "thundering herd".
        persist_result: An optional toggle indicating whether the result of this task
            should be persisted to result storage. Defaults to `None`, which indicates
            that Prefect should choose whether the result should be persisted depending on
            the features being used.
        result_storage: An optional block to use to persist the result of this task.
            Defaults to the value set in the flow the task is called in.
        result_storage_key: An optional key to store the result in storage at when persisted.
            Defaults to a unique identifier.
        result_serializer: An optional serializer to use to serialize the result of this
            task for persistence. Defaults to the value set in the flow the task is
            called in.
        timeout_seconds: An optional number of seconds indicating a maximum runtime for
            the task. If the task exceeds this runtime, it will be marked as failed.
        log_prints: If set, `print` statements in the task will be redirected to the
            Prefect logger for the task run. Defaults to `None`, which indicates
            that the value from the flow should be used.
        refresh_cache: If set, cached results for the cache key are not used.
            Defaults to `None`, which indicates that a cached result from a previous
            execution with matching cache key is used.
        on_failure: An optional list of callables to run when the task enters a failed state.
        on_completion: An optional list of callables to run when the task enters a completed state.
        retry_condition_fn: An optional callable run when a task run returns a Failed state. Should
            return `True` if the task should continue to its retry policy (e.g. `retries=3`), and `False` if the task
            should end as failed. Defaults to `None`, indicating the task should always continue
            to its retry policy.
        viz_return_value: An optional value to return when the task dependency tree is visualized.

    Returns:
        A callable `Task` object which, when called, will submit the task for execution.

    Examples:
        Define a simple task

        >>> @task
        >>> def add(x, y):
        >>>     return x + y

        Define an async task

        >>> @task
        >>> async def add(x, y):
        >>>     return x + y

        Define a task with tags and a description

        >>> @task(tags={"a", "b"}, description="This task is empty but its my first!")
        >>> def my_task():
        >>>     pass

        Define a task with a custom name

        >>> @task(name="The Ultimate Task")
        >>> def my_task():
        >>>     pass

        Define a task that retries 3 times with a 5 second delay between attempts

        >>> from random import randint
        >>>
        >>> @task(retries=3, retry_delay_seconds=5)
        >>> def my_task():
        >>>     x = randint(0, 5)
        >>>     if x >= 3:  # Make a task that fails sometimes
        >>>         raise ValueError("Retry me please!")
        >>>     return x

        Define a task that is cached for a day based on its inputs

        >>> from prefect.tasks import task_input_hash
        >>> from datetime import timedelta
        >>>
        >>> @task(cache_key_fn=task_input_hash, cache_expiration=timedelta(days=1))
        >>> def my_task():
        >>>     return "hello"
    """

    if __fn:
        return cast(
            Task[P, R],
            Task(
                fn=__fn,
                name=name,
                description=description,
                tags=tags,
                version=version,
                cache_key_fn=cache_key_fn,
                cache_expiration=cache_expiration,
                task_run_name=task_run_name,
                retries=retries,
                retry_delay_seconds=retry_delay_seconds,
                retry_jitter_factor=retry_jitter_factor,
                persist_result=persist_result,
                result_storage=result_storage,
                result_storage_key=result_storage_key,
                result_serializer=result_serializer,
                cache_result_in_memory=cache_result_in_memory,
                timeout_seconds=timeout_seconds,
                log_prints=log_prints,
                refresh_cache=refresh_cache,
                on_completion=on_completion,
                on_failure=on_failure,
                retry_condition_fn=retry_condition_fn,
                viz_return_value=viz_return_value,
            ),
        )
    else:
        return cast(
            Callable[[Callable[P, R]], Task[P, R]],
            partial(
                task,
                name=name,
                description=description,
                tags=tags,
                version=version,
                cache_key_fn=cache_key_fn,
                cache_expiration=cache_expiration,
                task_run_name=task_run_name,
                retries=retries,
                retry_delay_seconds=retry_delay_seconds,
                retry_jitter_factor=retry_jitter_factor,
                persist_result=persist_result,
                result_storage=result_storage,
                result_storage_key=result_storage_key,
                result_serializer=result_serializer,
                cache_result_in_memory=cache_result_in_memory,
                timeout_seconds=timeout_seconds,
                log_prints=log_prints,
                refresh_cache=refresh_cache,
                on_completion=on_completion,
                on_failure=on_failure,
                retry_condition_fn=retry_condition_fn,
                viz_return_value=viz_return_value,
            ),
        )

task_input_hash

A task cache key implementation which hashes all inputs to the task using a JSON or cloudpickle serializer. If any arguments are not JSON serializable, the pickle serializer is used as a fallback. If cloudpickle fails, this will return a null key indicating that a cache key could not be generated for the given inputs.

Parameters:

Name Type Description Default
context TaskRunContext

the active TaskRunContext

required
arguments Dict[str, Any]

a dictionary of arguments to be passed to the underlying task

required

Returns:

Type Description
Optional[str]

a string hash if hashing succeeded, else None

Source code in prefect/tasks.py
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
def task_input_hash(
    context: "TaskRunContext", arguments: Dict[str, Any]
) -> Optional[str]:
    """
    A task cache key implementation which hashes all inputs to the task using a JSON or
    cloudpickle serializer. If any arguments are not JSON serializable, the pickle
    serializer is used as a fallback. If cloudpickle fails, this will return a null key
    indicating that a cache key could not be generated for the given inputs.

    Arguments:
        context: the active `TaskRunContext`
        arguments: a dictionary of arguments to be passed to the underlying task

    Returns:
        a string hash if hashing succeeded, else `None`
    """
    return hash_objects(
        # We use the task key to get the qualified name for the task and include the
        # task functions `co_code` bytes to avoid caching when the underlying function
        # changes
        context.task.task_key,
        context.task.fn.__code__.co_code.hex(),
        arguments,
    )