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

mlrun/utils/helpers.py

@@ -26,7 +26,7 @@ import sys
 import typing
 import warnings
 from datetime import datetime, timezone
-from importlib import import_module
+from importlib import import_module, reload
 from os import path
 from types import ModuleType
 from typing import Any, Optional
@@ -39,7 +39,7 @@ import pandas
 import semver
 import yaml
 from dateutil import parser
-from deprecated import deprecated
+from mlrun_pipelines.models import PipelineRun
 from pandas._libs.tslibs.timestamps import Timedelta, Timestamp
 from yaml.representer import RepresenterError
 
@@ -76,19 +76,6 @@ class OverwriteBuildParamsWarning(FutureWarning):
     pass
 
 
-# TODO: remove in 1.7.0
-@deprecated(
-    version="1.5.0",
-    reason="'parse_versioned_object_uri' will be removed from this file in 1.7.0, use "
-    "'mlrun.common.helpers.parse_versioned_object_uri' instead",
-    category=FutureWarning,
-)
-def parse_versioned_object_uri(uri: str, default_project: str = ""):
-    return mlrun.common.helpers.parse_versioned_object_uri(
-        uri=uri, default_project=default_project
-    )
-
-
 class StorePrefix:
     """map mlrun store objects to prefixes"""
 
@@ -119,14 +106,9 @@ class StorePrefix:
 
 
 def get_artifact_target(item: dict, project=None):
-    if is_legacy_artifact(item):
-        db_key = item.get("db_key")
-        project_str = project or item.get("project")
-        tree = item.get("tree")
-    else:
-        db_key = item["spec"].get("db_key")
-        project_str = project or item["metadata"].get("project")
-        tree = item["metadata"].get("tree")
+    db_key = item["spec"].get("db_key")
+    project_str = project or item["metadata"].get("project")
+    tree = item["metadata"].get("tree")
 
     kind = item.get("kind")
     if kind in ["dataset", "model", "artifact"] and db_key:
@@ -135,11 +117,15 @@ def get_artifact_target(item: dict, project=None):
             target = f"{target}@{tree}"
         return target
 
-    return (
-        item.get("target_path")
-        if is_legacy_artifact(item)
-        else item["spec"].get("target_path")
-    )
+    return item["spec"].get("target_path")
+
+
+# TODO: left for migrations testing purposes. Remove in 1.8.0.
+def is_legacy_artifact(artifact):
+    if isinstance(artifact, dict):
+        return "metadata" not in artifact
+    else:
+        return not hasattr(artifact, "metadata")
 
 
 logger = create_logger(config.log_level, config.log_formatter, "mlrun", sys.stdout)
@@ -195,8 +181,12 @@ def verify_field_regex(
             )
             if mode == mlrun.common.schemas.RegexMatchModes.all:
                 if raise_on_failure:
+                    if len(field_name) > max_chars:
+                        field_name = field_name[:max_chars] + "...truncated"
+                    if len(field_value) > max_chars:
+                        field_value = field_value[:max_chars] + "...truncated"
                     raise mlrun.errors.MLRunInvalidArgumentError(
-                        f"Field '{field_name[:max_chars]}' is malformed. '{field_value[:max_chars]}' "
+                        f"Field '{field_name}' is malformed. '{field_value}' "
                         f"does not match required pattern: {pattern}"
                     )
                 return False
@@ -669,7 +659,7 @@ def parse_artifact_uri(uri, default_project=""):
         [3] = tag
         [4] = tree
     """
-    uri_pattern = r"^((?P<project>.*)/)?(?P<key>.*?)(\#(?P<iteration>.*?))?(:(?P<tag>.*?))?(@(?P<tree>.*))?$"
+    uri_pattern = mlrun.utils.regex.artifact_uri_pattern
     match = re.match(uri_pattern, uri)
     if not match:
         raise ValueError(
@@ -801,34 +791,6 @@ def gen_html_table(header, rows=None):
     return style + '<table class="tg">\n' + out + "</table>\n\n"
 
 
-def new_pipe_metadata(
-    artifact_path: str = None,
-    cleanup_ttl: int = None,
-    op_transformers: list[typing.Callable] = None,
-):
-    from kfp.dsl import PipelineConf
-
-    def _set_artifact_path(task):
-        from kubernetes import client as k8s_client
-
-        task.add_env_variable(
-            k8s_client.V1EnvVar(name="MLRUN_ARTIFACT_PATH", value=artifact_path)
-        )
-        return task
-
-    conf = PipelineConf()
-    cleanup_ttl = cleanup_ttl or int(config.kfp_ttl)
-
-    if cleanup_ttl:
-        conf.set_ttl_seconds_after_finished(cleanup_ttl)
-    if artifact_path:
-        conf.add_op_transformer(_set_artifact_path)
-    if op_transformers:
-        for op_transformer in op_transformers:
-            conf.add_op_transformer(op_transformer)
-    return conf
-
-
 def _convert_python_package_version_to_image_tag(version: typing.Optional[str]):
     return (
         version.replace("+", "-").replace("0.0.0-", "") if version is not None else None
@@ -1015,17 +977,27 @@ def get_ui_url(project, uid=None):
     return url
 
 
+def get_model_endpoint_url(project, model_name, model_endpoint_id):
+    url = ""
+    if mlrun.mlconf.resolve_ui_url():
+        url = f"{mlrun.mlconf.resolve_ui_url()}/{mlrun.mlconf.ui.projects_prefix}/{project}/models"
+        if model_name:
+            url += f"/model-endpoints/{model_name}/{model_endpoint_id}/overview"
+    return url
+
+
 def get_workflow_url(project, id=None):
     url = ""
     if mlrun.mlconf.resolve_ui_url():
-        url = "{}/{}/{}/jobs/monitor-workflows/workflow/{}".format(
-            mlrun.mlconf.resolve_ui_url(), mlrun.mlconf.ui.projects_prefix, project, id
+        url = (
+            f"{mlrun.mlconf.resolve_ui_url()}/{mlrun.mlconf.ui.projects_prefix}"
+            f"/{project}/jobs/monitor-workflows/workflow/{id}"
         )
     return url
 
 
 def are_strings_in_exception_chain_messages(
-    exception: Exception, strings_list=list[str]
+    exception: Exception, strings_list: list[str]
 ) -> bool:
     while exception is not None:
         if any([string in str(exception) for string in strings_list]):
@@ -1047,16 +1019,35 @@ def create_class(pkg_class: str):
     return class_
 
 
-def create_function(pkg_func: str):
+def create_function(pkg_func: str, reload_modules: bool = False):
     """Create a function from a package.module.function string
 
     :param pkg_func:  full function location,
                       e.g. "sklearn.feature_selection.f_classif"
+    :param reload_modules: reload the function again.
     """
     splits = pkg_func.split(".")
     pkg_module = ".".join(splits[:-1])
     cb_fname = splits[-1]
     pkg_module = __import__(pkg_module, fromlist=[cb_fname])
+
+    if reload_modules:
+        # Even though the function appears in the modules list, we need to reload
+        # the code again because it may have changed
+        try:
+            logger.debug("Reloading module", module=pkg_func)
+            _reload(
+                pkg_module,
+                max_recursion_depth=mlrun.mlconf.function.spec.reload_max_recursion_depth,
+            )
+        except Exception as exc:
+            logger.warning(
+                "Failed to reload module. Not all associated modules can be reloaded, import them manually."
+                "Or, with Jupyter, restart the Python kernel.",
+                module=pkg_func,
+                err=mlrun.errors.err_to_str(exc),
+            )
+
     function_ = getattr(pkg_module, cb_fname)
     return function_
 
@@ -1114,8 +1105,14 @@ def get_class(class_name, namespace=None):
     return class_object
 
 
-def get_function(function, namespace):
-    """return function callable object from function name string"""
+def get_function(function, namespaces, reload_modules: bool = False):
+    """Return function callable object from function name string
+
+    :param function: path to the function ([class_name::]function)
+    :param namespaces: one or list of namespaces/modules to search the function in
+    :param reload_modules: reload the function again
+    :return: function handler (callable)
+    """
     if callable(function):
         return function
 
@@ -1124,12 +1121,12 @@ def get_function(function, namespace):
         if not function.endswith(")"):
             raise ValueError('function expression must start with "(" and end with ")"')
         return eval("lambda event: " + function[1:-1], {}, {})
-    function_object = _search_in_namespaces(function, namespace)
+    function_object = _search_in_namespaces(function, namespaces)
     if function_object is not None:
         return function_object
 
     try:
-        function_object = create_function(function)
+        function_object = create_function(function, reload_modules)
     except (ImportError, ValueError) as exc:
         raise ImportError(
             f"state/function init failed, handler '{function}' not found"
@@ -1138,18 +1135,24 @@ def get_function(function, namespace):
 
 
 def get_handler_extended(
-    handler_path: str, context=None, class_args: dict = {}, namespaces=None
+    handler_path: str,
+    context=None,
+    class_args: dict = None,
+    namespaces=None,
+    reload_modules: bool = False,
 ):
-    """get function handler from [class_name::]handler string
+    """Get function handler from [class_name::]handler string
 
     :param handler_path:  path to the function ([class_name::]handler)
     :param context:       MLRun function/job client context
     :param class_args:    optional dict of class init kwargs
     :param namespaces:    one or list of namespaces/modules to search the handler in
+    :param reload_modules: reload the function again
     :return: function handler (callable)
     """
+    class_args = class_args or {}
     if "::" not in handler_path:
-        return get_function(handler_path, namespaces)
+        return get_function(handler_path, namespaces, reload_modules)
 
     splitted = handler_path.split("::")
     class_path = splitted[0].strip()
@@ -1224,7 +1227,7 @@ def calculate_dataframe_hash(dataframe: pandas.DataFrame):
     return hashlib.sha1(pandas.util.hash_pandas_object(dataframe).values).hexdigest()
 
 
-def template_artifact_path(artifact_path, project, run_uid="project"):
+def template_artifact_path(artifact_path, project, run_uid=None):
     """
     Replace {{run.uid}} with the run uid and {{project}} with the project name in the artifact path.
     If no run uid is provided, the word `project` will be used instead as it is assumed to be a project
@@ -1232,6 +1235,7 @@ def template_artifact_path(artifact_path, project, run_uid="project"):
     """
     if not artifact_path:
         return artifact_path
+    run_uid = run_uid or "project"
     artifact_path = artifact_path.replace("{{run.uid}}", run_uid)
     artifact_path = _fill_project_path_template(artifact_path, project)
     return artifact_path
@@ -1291,13 +1295,6 @@ def str_to_timestamp(time_str: str, now_time: Timestamp = None):
     return Timestamp(time_str)
 
 
-def is_legacy_artifact(artifact):
-    if isinstance(artifact, dict):
-        return "metadata" not in artifact
-    else:
-        return not hasattr(artifact, "metadata")
-
-
 def is_link_artifact(artifact):
     if isinstance(artifact, dict):
         return (
@@ -1307,7 +1304,7 @@ def is_link_artifact(artifact):
         return artifact.kind == mlrun.common.schemas.ArtifactCategories.link.value
 
 
-def format_run(run: dict, with_project=False) -> dict:
+def format_run(run: PipelineRun, with_project=False) -> dict:
     fields = [
         "id",
         "name",
@@ -1344,17 +1341,17 @@ def format_run(run: dict, with_project=False) -> dict:
     # pipelines are yet to populate the status or workflow has failed
     # as observed https://jira.iguazeng.com/browse/ML-5195
     # set to unknown to ensure a status is returned
-    if run["status"] is None:
-        run["status"] = inflection.titleize(mlrun.runtimes.constants.RunStates.unknown)
+    if run.get("status", None) is None:
+        run["status"] = inflection.titleize(
+            mlrun.common.runtimes.constants.RunStates.unknown
+        )
 
     return run
 
 
 def get_in_artifact(artifact: dict, key, default=None, raise_on_missing=False):
     """artifact can be dict or Artifact object"""
-    if is_legacy_artifact(artifact):
-        return artifact.get(key, default)
-    elif key == "kind":
+    if key == "kind":
         return artifact.get(key, default)
     else:
         for block in ["metadata", "spec", "status"]:
@@ -1596,3 +1593,80 @@ def get_serving_spec():
             )
     spec = json.loads(data)
     return spec
+
+
+def additional_filters_warning(additional_filters, class_name):
+    if additional_filters and any(additional_filters):
+        mlrun.utils.logger.warn(
+            f"additional_filters parameter is not supported in {class_name},"
+            f" parameter has been ignored."
+        )
+
+
+def validate_component_version_compatibility(
+    component_name: typing.Literal["iguazio", "nuclio"], *min_versions: str
+):
+    """
+    :param component_name: Name of the component to validate compatibility for.
+    :param min_versions: Valid minimum version(s) required, assuming no 2 versions has equal major and minor.
+    """
+    parsed_min_versions = [
+        semver.VersionInfo.parse(min_version) for min_version in min_versions
+    ]
+    parsed_current_version = None
+    component_current_version = None
+    try:
+        if component_name == "iguazio":
+            component_current_version = mlrun.mlconf.igz_version
+            parsed_current_version = mlrun.mlconf.get_parsed_igz_version()
+
+            if parsed_current_version:
+                # ignore pre-release and build metadata, as iguazio version always has them, and we only care about the
+                # major, minor, and patch versions
+                parsed_current_version = semver.VersionInfo.parse(
+                    f"{parsed_current_version.major}.{parsed_current_version.minor}.{parsed_current_version.patch}"
+                )
+        if component_name == "nuclio":
+            component_current_version = mlrun.mlconf.nuclio_version
+            parsed_current_version = semver.VersionInfo.parse(
+                mlrun.mlconf.nuclio_version
+            )
+        if not parsed_current_version:
+            return True
+    except ValueError:
+        # only log when version is set but invalid
+        if component_current_version:
+            logger.warning(
+                "Unable to parse current version, assuming compatibility",
+                component_name=component_name,
+                current_version=component_current_version,
+                min_versions=min_versions,
+            )
+        return True
+
+    parsed_min_versions.sort(reverse=True)
+    for parsed_min_version in parsed_min_versions:
+        if parsed_current_version < parsed_min_version:
+            return False
+    return True
+
+
+def format_alert_summary(
+    alert: mlrun.common.schemas.AlertConfig, event_data: mlrun.common.schemas.Event
+) -> str:
+    result = alert.summary.replace("{{project}}", alert.project)
+    result = result.replace("{{name}}", alert.name)
+    result = result.replace("{{entity}}", event_data.entity.ids[0])
+    return result
+
+
+def _reload(module, max_recursion_depth):
+    """Recursively reload modules."""
+    if max_recursion_depth <= 0:
+        return
+
+    reload(module)
+    for attribute_name in dir(module):
+        attribute = getattr(module, attribute_name)
+        if type(attribute) is ModuleType:
+            _reload(attribute, max_recursion_depth - 1)

mlrun/utils/logger.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 
 import logging
+import typing
 from enum import Enum
 from sys import stdout
 from traceback import format_exception
@@ -92,7 +93,25 @@ class HumanReadableFormatter(_BaseFormatter):
 
 class HumanReadableExtendedFormatter(HumanReadableFormatter):
     def format(self, record) -> str:
-        more = self._resolve_more(record)
+        more = ""
+        record_with = self._record_with(record)
+        if record_with:
+
+            def _format_value(val):
+                formatted_val = (
+                    val
+                    if isinstance(val, str)
+                    else str(orjson.loads(self._json_dump(val)))
+                )
+                return (
+                    formatted_val.replace("\n", "\n\t\t")
+                    if len(formatted_val) < 4096
+                    else repr(formatted_val)
+                )
+
+            more = "\n\t" + "\n\t".join(
+                [f"{key}: {_format_value(val)}" for key, val in record_with.items()]
+            )
         return (
             "> "
             f"{self.formatTime(record, self.datefmt)} "
@@ -221,14 +240,27 @@ class FormatterKinds(Enum):
     JSON = "json"
 
 
-def create_formatter_instance(formatter_kind: FormatterKinds) -> logging.Formatter:
+def resolve_formatter_by_kind(
+    formatter_kind: FormatterKinds,
+) -> type[
+    typing.Union[HumanReadableFormatter, HumanReadableExtendedFormatter, JSONFormatter]
+]:
     return {
-        FormatterKinds.HUMAN: HumanReadableFormatter(),
-        FormatterKinds.HUMAN_EXTENDED: HumanReadableExtendedFormatter(),
-        FormatterKinds.JSON: JSONFormatter(),
+        FormatterKinds.HUMAN: HumanReadableFormatter,
+        FormatterKinds.HUMAN_EXTENDED: HumanReadableExtendedFormatter,
+        FormatterKinds.JSON: JSONFormatter,
     }[formatter_kind]
 
 
+def create_test_logger(name: str = "mlrun", stream: IO[str] = stdout) -> Logger:
+    return create_logger(
+        level="debug",
+        formatter_kind=FormatterKinds.HUMAN_EXTENDED.name,
+        name=name,
+        stream=stream,
+    )
+
+
 def create_logger(
     level: Optional[str] = None,
     formatter_kind: str = FormatterKinds.HUMAN.name,
@@ -243,11 +275,11 @@ def create_logger(
     logger_instance = Logger(level, name=name, propagate=False)
 
     # resolve formatter
-    formatter_instance = create_formatter_instance(
+    formatter_instance = resolve_formatter_by_kind(
         FormatterKinds(formatter_kind.lower())
     )
 
     # set handler
-    logger_instance.set_handler("default", stream or stdout, formatter_instance)
+    logger_instance.set_handler("default", stream or stdout, formatter_instance())
 
     return logger_instance

mlrun/utils/notifications/notification/__init__.py

@@ -51,14 +51,19 @@ class NotificationTypes(str, enum.Enum):
             self.console: [self.ipython],
         }.get(self, [])
 
+    @classmethod
+    def local(cls) -> list[str]:
+        return [
+            cls.console,
+            cls.ipython,
+        ]
+
     @classmethod
     def all(cls) -> list[str]:
-        return list(
-            [
-                cls.console,
-                cls.git,
-                cls.ipython,
-                cls.slack,
-                cls.webhook,
-            ]
-        )
+        return [
+            cls.console,
+            cls.git,
+            cls.ipython,
+            cls.slack,
+            cls.webhook,
+        ]

mlrun/utils/notifications/notification/base.py

@@ -77,7 +77,7 @@ class NotificationBase:
                 return f"[{severity}] {message}"
             return (
                 f"[{severity}] {message} for project {alert.project} "
-                f"UID {event_data.entity.id}. Value {event_data.value}"
+                f"UID {event_data.entity.ids[0]}. Values {event_data.value_dict}"
             )
 
         if not runs:

mlrun/utils/notifications/notification/slack.py

@@ -32,6 +32,7 @@ class SlackNotification(NotificationBase):
         "completed": ":smiley:",
         "running": ":man-running:",
         "error": ":x:",
+        "skipped": ":zzz:",
     }
 
     async def push(
@@ -135,8 +136,16 @@ class SlackNotification(NotificationBase):
         line = [
             self._get_slack_row(f":bell: {alert.name} alert has occurred"),
             self._get_slack_row(f"*Project:*\n{alert.project}"),
-            self._get_slack_row(f"*UID:*\n{event_data.entity.id}"),
+            self._get_slack_row(f"*ID:*\n{event_data.entity.ids[0]}"),
         ]
+
+        if alert.summary:
+            line.append(
+                self._get_slack_row(
+                    f"*Summary:*\n{mlrun.utils.helpers.format_alert_summary(alert, event_data)}"
+                )
+            )
+
         if event_data.value_dict:
             data_lines = []
             for key, value in event_data.value_dict.items():
@@ -144,32 +153,50 @@ class SlackNotification(NotificationBase):
             data_text = "\n".join(data_lines)
             line.append(self._get_slack_row(f"*Event data:*\n{data_text}"))
 
-        if url := mlrun.utils.helpers.get_ui_url(alert.project, event_data.entity.id):
-            line.append(self._get_slack_row(f"*Overview:*\n<{url}|*Job overview*>"))
+        if (
+            event_data.entity.kind == mlrun.common.schemas.alert.EventEntityKind.JOB
+        ):  # JOB entity
+            uid = event_data.value_dict.get("uid")
+            url = mlrun.utils.helpers.get_ui_url(alert.project, uid)
+            overview_type = "Job overview"
+        else:  # MODEL entity
+            model_name = event_data.value_dict.get("model")
+            model_endpoint_id = event_data.value_dict.get("model_endpoint_id")
+            url = mlrun.utils.helpers.get_model_endpoint_url(
+                alert.project, model_name, model_endpoint_id
+            )
+            overview_type = "Model endpoint"
+
+        line.append(self._get_slack_row(f"*Overview:*\n<{url}|*{overview_type}*>"))
 
         return line
 
     def _get_run_line(self, run: dict) -> dict:
         meta = run["metadata"]
         url = mlrun.utils.helpers.get_ui_url(meta.get("project"), meta.get("uid"))
-        if url:
+
+        # Only show the URL if the run is not a function (serving or mlrun function)
+        kind = run.get("step_kind")
+        state = run["status"].get("state", "")
+        if state != "skipped" and (url and not kind or kind == "run"):
             line = f'<{url}|*{meta.get("name")}*>'
         else:
             line = meta.get("name")
-        state = run["status"].get("state", "")
+        if kind:
+            line = f'{line} *({run.get("step_kind", run.get("kind", ""))})*'
         line = f'{self.emojis.get(state, ":question:")}  {line}'
         return self._get_slack_row(line)
 
     def _get_run_result(self, run: dict) -> dict:
         state = run["status"].get("state", "")
         if state == "error":
-            error_status = run["status"].get("error", "")
+            error_status = run["status"].get("error", "") or state
             result = f"*{error_status}*"
         else:
             result = mlrun.utils.helpers.dict_to_str(
                 run["status"].get("results", {}), ", "
             )
-        return self._get_slack_row(result or "None")
+        return self._get_slack_row(result or state)
 
     @staticmethod
     def _get_slack_row(text: str) -> dict:

mlrun/utils/notifications/notification/webhook.py

@@ -57,7 +57,7 @@ class WebhookNotification(NotificationBase):
             request_body["alert"] = alert.dict()
             if event_data:
                 request_body["value"] = event_data.value_dict
-                request_body["id"] = event_data.entity.id
+                request_body["id"] = event_data.entity.ids[0]
 
         if custom_html:
             request_body["custom_html"] = custom_html