mlrun 1.7.0rc14__py3-none-any.whl → 1.7.0rc22__py3-none-any.whl

mlrun/feature_store/retrieval/spark_merger.py

@@ -12,11 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
+
 import pandas as pd
 import semver
 
 import mlrun
+from mlrun.datastore.sources import ParquetSource
 from mlrun.datastore.targets import get_offline_target
+from mlrun.utils.helpers import additional_filters_warning
 
 from ...runtimes import RemoteSparkRuntime
 from ...runtimes.sparkjob import Spark3Runtime
@@ -225,7 +228,12 @@ class SparkFeatureMerger(BaseMerger):
         start_time=None,
         end_time=None,
         time_column=None,
+        additional_filters=None,
     ):
+        mlrun.utils.helpers.additional_filters_warning(
+            additional_filters, self.__class__
+        )
+
         source_kwargs = {}
         if feature_set.spec.passthrough:
             if not feature_set.spec.source:
@@ -235,6 +243,7 @@ class SparkFeatureMerger(BaseMerger):
             source_kind = feature_set.spec.source.kind
             source_path = feature_set.spec.source.path
             source_kwargs.update(feature_set.spec.source.attributes)
+            source_kwargs.pop("additional_filters", None)
         else:
             target = get_offline_target(feature_set)
             if not target:
@@ -248,12 +257,19 @@ class SparkFeatureMerger(BaseMerger):
         # entity_timestamp_column is from a specific feature set (can't be entity timestamp)
         source_driver = mlrun.datastore.sources.source_kind_to_driver[source_kind]
 
+        if source_driver != ParquetSource:
+            additional_filters_warning(additional_filters, source_driver)
+            additional_filters = None
+        additional_filters_dict = (
+            {"additional_filters": additional_filters} if additional_filters else {}
+        )
         source = source_driver(
             name=self.vector.metadata.name,
             path=source_path,
             time_field=time_column,
             start_time=start_time,
             end_time=end_time,
+            **additional_filters_dict,
             **source_kwargs,
         )
 

mlrun/frameworks/__init__.py

@@ -12,5 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 #
+"""
+MLRun provides a quick and easy integration into your code with mlrun.frameworks: a collection of sub-modules
+for the most commonly used machine and deep learning frameworks, providing features such as automatic logging,
+model management, and distributed training.
+"""
+
 # flake8: noqa  - this is until we take care of the F401 violations with respect to __all__ & sphinx
 from .parallel_coordinates import compare_db_runs, compare_run_objects

mlrun/frameworks/_dl_common/loggers/tensorboard_logger.py

@@ -547,9 +547,9 @@ class TensorboardLogger(Logger, Generic[DLTypes.WeightType]):
                     "inputs",
                     "parameters",
                 ]:
-                    text += "\n  * **{}**: {}".format(
-                        property_name.capitalize(),
-                        self._markdown_print(value=property_value, tabs=2),
+                    text += (
+                        f"\n  * **{property_name.capitalize()}**: "
+                        f"{self._markdown_print(value=property_value, tabs=2)}"
                     )
         else:
             for property_name, property_value in self._extract_epoch_results().items():
@@ -614,13 +614,8 @@ class TensorboardLogger(Logger, Generic[DLTypes.WeightType]):
         :return: The generated link.
         """
         return (
-            '<a href="{}/{}/{}/jobs/monitor/{}/overview" target="_blank">{}</a>'.format(
-                config.resolve_ui_url(),
-                config.ui.projects_prefix,
-                context.project,
-                context.uid,
-                link_text,
-            )
+            f'<a href="{config.resolve_ui_url()}/{config.ui.projects_prefix}/{context.project}'
+            f'/jobs/monitor/{context.uid}/overview" target="_blank">{link_text}</a>'
         )
 
     @staticmethod
@@ -653,13 +648,13 @@ class TensorboardLogger(Logger, Generic[DLTypes.WeightType]):
         if isinstance(value, list):
             if len(value) == 0:
                 return ""
-            text = "\n" + yaml.dump(value)
+            text = "\n" + yaml.safe_dump(value)
             text = "  \n".join(["  " * tabs + line for line in text.splitlines()])
             return text
         if isinstance(value, dict):
             if len(value) == 0:
                 return ""
-            text = yaml.dump(value)
+            text = yaml.safe_dump(value)
             text = "  \n".join(
                 ["  " * tabs + "- " + line for line in text.splitlines()]
             )

mlrun/frameworks/parallel_coordinates.py

@@ -295,7 +295,7 @@ def compare_db_runs(
     iter=False,
     start_time_from: datetime = None,
     hide_identical: bool = True,
-    exclude: list = [],
+    exclude: list = None,
     show=None,
     colorscale: str = "Blues",
     filename=None,
@@ -332,6 +332,7 @@ def compare_db_runs(
         **query_args,
     )
 
+    exclude = exclude or []
     runs_df = _runs_list_to_df(runs_list)
     plot_as_html = gen_pcp_plot(
         runs_df,

mlrun/frameworks/tf_keras/__init__.py

@@ -18,6 +18,7 @@ from typing import Any, Union
 from tensorflow import keras
 
 import mlrun
+import mlrun.common.constants as mlrun_constants
 
 from .callbacks import MLRunLoggingCallback, TensorboardLoggingCallback
 from .mlrun_interface import TFKerasMLRunInterface
@@ -126,7 +127,9 @@ def apply_mlrun(
     # # Use horovod:
     if use_horovod is None:
         use_horovod = (
-            context.labels.get("kind", "") == "mpijob" if context is not None else False
+            context.labels.get(mlrun_constants.MLRunInternalLabels.kind, "") == "mpijob"
+            if context is not None
+            else False
         )
 
     # Create a model handler:

mlrun/k8s_utils.py

@@ -141,17 +141,6 @@ def verify_label_key(key: str):
     if not key:
         raise mlrun.errors.MLRunInvalidArgumentError("label key cannot be empty")
 
-    mlrun.utils.helpers.verify_field_regex(
-        f"project.metadata.labels.'{key}'",
-        key,
-        mlrun.utils.regex.k8s_character_limit,
-    )
-
-    if key.startswith("k8s.io/") or key.startswith("kubernetes.io/"):
-        raise mlrun.errors.MLRunInvalidArgumentError(
-            "Labels cannot start with 'k8s.io/' or 'kubernetes.io/'"
-        )
-
     parts = key.split("/")
     if len(parts) == 1:
         name = parts[0]
@@ -173,12 +162,22 @@ def verify_label_key(key: str):
             "Label key can only contain one '/'"
         )
 
+    mlrun.utils.helpers.verify_field_regex(
+        f"project.metadata.labels.'{key}'",
+        name,
+        mlrun.utils.regex.k8s_character_limit,
+    )
     mlrun.utils.helpers.verify_field_regex(
         f"project.metadata.labels.'{key}'",
         name,
         mlrun.utils.regex.qualified_name,
     )
 
+    if key.startswith("k8s.io/") or key.startswith("kubernetes.io/"):
+        raise mlrun.errors.MLRunInvalidArgumentError(
+            "Labels cannot start with 'k8s.io/' or 'kubernetes.io/'"
+        )
+
 
 def verify_label_value(value, label_key):
     mlrun.utils.helpers.verify_field_regex(

mlrun/launcher/base.py

@@ -18,10 +18,11 @@ import os
 import uuid
 from typing import Any, Callable, Optional, Union
 
+import mlrun_pipelines.common.ops
+
 import mlrun.common.schemas
 import mlrun.config
 import mlrun.errors
-import mlrun.kfpops
 import mlrun.lists
 import mlrun.model
 import mlrun.runtimes
@@ -390,7 +391,7 @@ class BaseLauncher(abc.ABC):
             return
 
         if result and runtime.kfp and err is None:
-            mlrun.kfpops.write_kfpmeta(result)
+            mlrun_pipelines.common.ops.write_kfpmeta(result)
 
         self._log_track_results(runtime.is_child, result, run)
 
@@ -403,7 +404,7 @@ class BaseLauncher(abc.ABC):
             )
             if (
                 run.status.state
-                in mlrun.runtimes.constants.RunStates.error_and_abortion_states()
+                in mlrun.common.runtimes.constants.RunStates.error_and_abortion_states()
             ):
                 if runtime._is_remote and not runtime.is_child:
                     logger.error(

mlrun/launcher/client.py

@@ -16,6 +16,7 @@ from typing import Optional
 
 import IPython
 
+import mlrun.common.constants as mlrun_constants
 import mlrun.errors
 import mlrun.launcher.base as launcher
 import mlrun.lists
@@ -69,13 +70,14 @@ class ClientBaseLauncher(launcher.BaseLauncher, abc.ABC):
     def _store_function(
         runtime: "mlrun.runtimes.BaseRuntime", run: "mlrun.run.RunObject"
     ):
-        run.metadata.labels["kind"] = runtime.kind
+        run.metadata.labels[mlrun_constants.MLRunInternalLabels.kind] = runtime.kind
         mlrun.runtimes.utils.enrich_run_labels(
-            run.metadata.labels, [mlrun.runtimes.constants.RunLabels.owner]
+            run.metadata.labels, [mlrun.common.runtimes.constants.RunLabels.owner]
         )
         if run.spec.output_path:
             run.spec.output_path = run.spec.output_path.replace(
-                "{{run.user}}", run.metadata.labels["owner"]
+                "{{run.user}}",
+                run.metadata.labels[mlrun_constants.MLRunInternalLabels.owner],
             )
         db = runtime._get_db()
         if db and runtime.kind != "handler":

mlrun/launcher/local.py

@@ -15,6 +15,7 @@ import os
 import pathlib
 from typing import Callable, Optional, Union
 
+import mlrun.common.constants as mlrun_constants
 import mlrun.common.schemas.schedule
 import mlrun.errors
 import mlrun.launcher.client as launcher
@@ -68,6 +69,7 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
         notifications: Optional[list[mlrun.model.Notification]] = None,
         returns: Optional[list[Union[str, dict[str, str]]]] = None,
         state_thresholds: Optional[dict[str, int]] = None,
+        reset_on_run: Optional[bool] = None,
     ) -> "mlrun.run.RunObject":
         # do not allow local function to be scheduled
         if self._is_run_local and schedule is not None:
@@ -87,6 +89,7 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
                 name=name,
                 workdir=workdir,
                 handler=handler,
+                reset_on_run=reset_on_run,
             )
 
         # sanity check
@@ -132,8 +135,13 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
         runtime: "mlrun.runtimes.BaseRuntime",
         run: Optional[Union["mlrun.run.RunTemplate", "mlrun.run.RunObject"]] = None,
     ):
-        if "V3IO_USERNAME" in os.environ and "v3io_user" not in run.metadata.labels:
-            run.metadata.labels["v3io_user"] = os.environ.get("V3IO_USERNAME")
+        if (
+            "V3IO_USERNAME" in os.environ
+            and mlrun_constants.MLRunInternalLabels.v3io_user not in run.metadata.labels
+        ):
+            run.metadata.labels[mlrun_constants.MLRunInternalLabels.v3io_user] = (
+                os.environ.get("V3IO_USERNAME")
+            )
 
         # store function object in db unless running from within a run pod
         if not runtime.is_child:
@@ -206,6 +214,7 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
         name: Optional[str] = "",
         workdir: Optional[str] = "",
         handler: Optional[str] = None,
+        reset_on_run: Optional[bool] = None,
     ):
         project = project or runtime.metadata.project
         function_name = name or runtime.metadata.name
@@ -244,6 +253,7 @@ class ClientLocalLauncher(launcher.ClientBaseLauncher):
             fn.spec.build = runtime.spec.build
 
         run.spec.handler = handler
+        run.spec.reset_on_run = reset_on_run
         return fn
 
     @staticmethod

mlrun/launcher/remote.py

@@ -17,6 +17,7 @@ from typing import Optional, Union
 import pandas as pd
 import requests
 
+import mlrun.common.constants as mlrun_constants
 import mlrun.common.schemas.schedule
 import mlrun.db
 import mlrun.errors
@@ -58,6 +59,7 @@ class ClientRemoteLauncher(launcher.ClientBaseLauncher):
         notifications: Optional[list[mlrun.model.Notification]] = None,
         returns: Optional[list[Union[str, dict[str, str]]]] = None,
         state_thresholds: Optional[dict[str, int]] = None,
+        reset_on_run: Optional[bool] = None,
     ) -> "mlrun.run.RunObject":
         self.enrich_runtime(runtime, project)
         run = self._create_run_object(task)
@@ -100,8 +102,13 @@ class ClientRemoteLauncher(launcher.ClientBaseLauncher):
         if runtime.verbose:
             logger.info(f"runspec:\n{run.to_yaml()}")
 
-        if "V3IO_USERNAME" in os.environ and "v3io_user" not in run.metadata.labels:
-            run.metadata.labels["v3io_user"] = os.environ.get("V3IO_USERNAME")
+        if (
+            "V3IO_USERNAME" in os.environ
+            and mlrun_constants.MLRunInternalLabels.v3io_user not in run.metadata.labels
+        ):
+            run.metadata.labels[mlrun_constants.MLRunInternalLabels.v3io_user] = (
+                os.environ.get("V3IO_USERNAME")
+            )
 
         logger.info(
             "Storing function",

mlrun/lists.py

@@ -21,7 +21,7 @@ import mlrun.frameworks
 from .artifacts import Artifact, dict_to_artifact
 from .config import config
 from .render import artifacts_to_html, runs_to_html
-from .utils import flatten, get_artifact_target, get_in, is_legacy_artifact
+from .utils import flatten, get_artifact_target, get_in
 
 list_header = [
     "project",
@@ -29,12 +29,14 @@ list_header = [
     "iter",
     "start",
     "state",
+    "kind",
     "name",
     "labels",
     "inputs",
     "parameters",
     "results",
     "artifacts",
+    "artifact_uris",
     "error",
 ]
 
@@ -56,12 +58,14 @@ class RunList(list):
                 get_in(run, "metadata.iteration", ""),
                 get_in(run, "status.start_time", ""),
                 get_in(run, "status.state", ""),
+                get_in(run, "step_kind", get_in(run, "kind", "")),
                 get_in(run, "metadata.name", ""),
                 get_in(run, "metadata.labels", ""),
                 get_in(run, "spec.inputs", ""),
                 get_in(run, "spec.parameters", ""),
                 get_in(run, "status.results", ""),
                 get_in(run, "status.artifacts", []),
+                get_in(run, "status.artifact_uris", {}),
                 get_in(run, "status.error", ""),
             ]
             if extend_iterations and iterations:
@@ -184,7 +188,7 @@ class ArtifactList(list):
             "uri": ["uri", "uri"],
         }
         for artifact in self:
-            fields_index = 0 if is_legacy_artifact(artifact) else 1
+            fields_index = 1
             row = [get_in(artifact, v[fields_index], "") for k, v in head.items()]
             artifact_uri = dict_to_artifact(artifact).uri
             last_index = len(row) - 1

mlrun/model.py

@@ -27,13 +27,14 @@ from typing import Any, Optional, Union
 import pydantic.error_wrappers
 
 import mlrun
+import mlrun.common.constants as mlrun_constants
 import mlrun.common.schemas.notification
+import mlrun.utils.regex
 
 from .utils import (
     dict_to_json,
     dict_to_yaml,
     get_artifact_target,
-    is_legacy_artifact,
     logger,
     template_artifact_path,
 )
@@ -682,10 +683,14 @@ class Notification(ModelObj):
 
     def __init__(
         self,
-        kind=None,
+        kind: mlrun.common.schemas.notification.NotificationKind = (
+            mlrun.common.schemas.notification.NotificationKind.slack
+        ),
         name=None,
         message=None,
-        severity=None,
+        severity: mlrun.common.schemas.notification.NotificationSeverity = (
+            mlrun.common.schemas.notification.NotificationSeverity.INFO
+        ),
         when=None,
         condition=None,
         secret_params=None,
@@ -694,12 +699,10 @@ class Notification(ModelObj):
         sent_time=None,
         reason=None,
     ):
-        self.kind = kind or mlrun.common.schemas.notification.NotificationKind.slack
+        self.kind = kind
         self.name = name or ""
         self.message = message or ""
-        self.severity = (
-            severity or mlrun.common.schemas.notification.NotificationSeverity.INFO
-        )
+        self.severity = severity
         self.when = when or ["completed"]
         self.condition = condition or ""
         self.secret_params = secret_params or {}
@@ -769,7 +772,10 @@ class RunMetadata(ModelObj):
     def is_workflow_runner(self):
         if not self.labels:
             return False
-        return self.labels.get("job-type", "") == "workflow-runner"
+        return (
+            self.labels.get(mlrun_constants.MLRunInternalLabels.job_type, "")
+            == "workflow-runner"
+        )
 
 
 class HyperParamStrategies:
@@ -866,6 +872,7 @@ class RunSpec(ModelObj):
         returns=None,
         notifications=None,
         state_thresholds=None,
+        reset_on_run=None,
     ):
         # A dictionary of parsing configurations that will be read from the inputs the user set. The keys are the inputs
         # keys (parameter names) and the values are the type hint given in the input keys after the colon.
@@ -902,6 +909,7 @@ class RunSpec(ModelObj):
         self.allow_empty_resources = allow_empty_resources
         self._notifications = notifications or []
         self.state_thresholds = state_thresholds or {}
+        self.reset_on_run = reset_on_run
 
     def _serialize_field(
         self, struct: dict, field_name: str = None, strip: bool = False
@@ -1208,6 +1216,7 @@ class RunStatus(ModelObj):
         ui_url=None,
         reason: str = None,
         notifications: dict[str, Notification] = None,
+        artifact_uris: dict[str, str] = None,
     ):
         self.state = state or "created"
         self.status_text = status_text
@@ -1222,6 +1231,8 @@ class RunStatus(ModelObj):
         self.ui_url = ui_url
         self.reason = reason
         self.notifications = notifications or {}
+        # Artifact key -> URI mapping, since the full artifacts are not stored in the runs DB table
+        self.artifact_uris = artifact_uris or {}
 
     def is_failed(self) -> Optional[bool]:
         """
@@ -1435,11 +1446,14 @@ class RunObject(RunTemplate):
             unknown_error = ""
             if (
                 self.status.state
-                in mlrun.runtimes.constants.RunStates.abortion_states()
+                in mlrun.common.runtimes.constants.RunStates.abortion_states()
             ):
                 unknown_error = "Run was aborted"
 
-            elif self.status.state in mlrun.runtimes.constants.RunStates.error_states():
+            elif (
+                self.status.state
+                in mlrun.common.runtimes.constants.RunStates.error_states()
+            ):
                 unknown_error = "Unknown error"
 
             return (
@@ -1477,7 +1491,7 @@ class RunObject(RunTemplate):
             outputs = {k: v for k, v in self.status.results.items()}
         if self.status.artifacts:
             for a in self.status.artifacts:
-                key = a["key"] if is_legacy_artifact(a) else a["metadata"]["key"]
+                key = a["metadata"]["key"]
                 outputs[key] = get_artifact_target(a, self.metadata.project)
         return outputs
 
@@ -1520,7 +1534,10 @@ class RunObject(RunTemplate):
 
     def state(self):
         """current run state"""
-        if self.status.state in mlrun.runtimes.constants.RunStates.terminal_states():
+        if (
+            self.status.state
+            in mlrun.common.runtimes.constants.RunStates.terminal_states()
+        ):
             return self.status.state
         self.refresh()
         return self.status.state or "unknown"
@@ -1534,8 +1551,10 @@ class RunObject(RunTemplate):
             iter=self.metadata.iteration,
         )
         if run:
-            self.status = RunStatus.from_dict(run.get("status", {}))
-            self.status.from_dict(run.get("status", {}))
+            run_status = run.get("status", {})
+            # Artifacts are not stored in the DB, so we need to preserve them here
+            run_status["artifacts"] = self.status.artifacts
+            self.status = RunStatus.from_dict(run_status)
             return self
 
     def show(self):
@@ -1582,7 +1601,7 @@ class RunObject(RunTemplate):
         last_pull_log_time = None
         logs_enabled = show_logs is not False
         state = self.state()
-        if state not in mlrun.runtimes.constants.RunStates.terminal_states():
+        if state not in mlrun.common.runtimes.constants.RunStates.terminal_states():
             logger.info(
                 f"run {self.metadata.name} is not completed yet, waiting for it to complete",
                 current_state=state,
@@ -1592,7 +1611,8 @@ class RunObject(RunTemplate):
             if (
                 logs_enabled
                 and logs_interval
-                and state not in mlrun.runtimes.constants.RunStates.terminal_states()
+                and state
+                not in mlrun.common.runtimes.constants.RunStates.terminal_states()
                 and (
                     last_pull_log_time is None
                     or (datetime.now() - last_pull_log_time).seconds > logs_interval
@@ -1601,7 +1621,7 @@ class RunObject(RunTemplate):
                 last_pull_log_time = datetime.now()
                 state, offset = self.logs(watch=False, offset=offset)
 
-            if state in mlrun.runtimes.constants.RunStates.terminal_states():
+            if state in mlrun.common.runtimes.constants.RunStates.terminal_states():
                 if logs_enabled and logs_interval:
                     self.logs(watch=False, offset=offset)
                 break
@@ -1613,7 +1633,10 @@ class RunObject(RunTemplate):
                 )
         if logs_enabled and not logs_interval:
             self.logs(watch=False)
-        if raise_on_failure and state != mlrun.runtimes.constants.RunStates.completed:
+        if (
+            raise_on_failure
+            and state != mlrun.common.runtimes.constants.RunStates.completed
+        ):
             raise mlrun.errors.MLRunRuntimeError(
                 f"Task {self.metadata.name} did not complete (state={state})"
             )
@@ -1629,9 +1652,12 @@ class RunObject(RunTemplate):
 
     @staticmethod
     def parse_uri(uri: str) -> tuple[str, str, str, str]:
-        uri_pattern = (
-            r"^(?P<project>.*)@(?P<uid>.*)\#(?P<iteration>.*?)(:(?P<tag>.*))?$"
-        )
+        """Parse the run's uri
+
+        :param uri: run uri in the format of <project>@<uid>#<iteration>[:tag]
+        :return: project, uid, iteration, tag
+        """
+        uri_pattern = mlrun.utils.regex.run_uri_pattern
         match = re.match(uri_pattern, uri)
         if not match:
             raise ValueError(

mlrun/model_monitoring/__init__.py

@@ -15,7 +15,7 @@
 # flake8: noqa  - this is until we take care of the F401 violations with respect to __all__ & sphinx
 # for backwards compatibility
 
-from .db import get_store_object
+from .db import get_store_object, get_tsdb_connector
 from .helpers import get_stream_path
 from .model_endpoint import ModelEndpoint
 from .tracking_policy import TrackingPolicy