# Copyright 2023 Iguazio
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import ast
import datetime
import os
import random
import sys
import time
import warnings
from collections import Counter
from copy import copy
from typing import Any, Optional, Union
from urllib.parse import urlparse
import pandas as pd
from mergedeep import merge
import mlrun
import mlrun.utils.helpers
from mlrun.config import config
from mlrun.datastore.snowflake_utils import (
get_snowflake_password,
get_snowflake_spark_options,
)
from mlrun.datastore.utils import transform_list_filters_to_tuple
from mlrun.model import DataSource, DataTarget, DataTargetBase, TargetPathObject
from mlrun.utils import logger, now_date
from mlrun.utils.helpers import to_parquet
from mlrun.utils.v3io_clients import get_frames_client
from .. import errors
from ..data_types import ValueType
from ..platforms.iguazio import parse_path, split_path
from .datastore_profile import datastore_profile_read
from .spark_utils import spark_session_update_hadoop_options
from .utils import (
_generate_sql_query_with_time_filter,
filter_df_start_end_time,
select_columns_from_df,
)
class TargetTypes:
csv = "csv"
parquet = "parquet"
nosql = "nosql"
redisnosql = "redisnosql"
tsdb = "tsdb"
stream = "stream"
kafka = "kafka"
dataframe = "dataframe"
custom = "custom"
sql = "sql"
snowflake = "snowflake"
@staticmethod
def all():
return [
TargetTypes.csv,
TargetTypes.parquet,
TargetTypes.nosql,
TargetTypes.redisnosql,
TargetTypes.tsdb,
TargetTypes.stream,
TargetTypes.kafka,
TargetTypes.dataframe,
TargetTypes.custom,
TargetTypes.sql,
TargetTypes.snowflake,
]
def generate_target_run_id():
return f"{round(time.time() * 1000)}_{random.randint(0, 999)}"
def write_spark_dataframe_with_options(spark_options, df, mode, write_format=None):
non_hadoop_spark_options = spark_session_update_hadoop_options(
df.sql_ctx.sparkSession, spark_options
)
if write_format:
df.write.format(write_format).mode(mode).save(**non_hadoop_spark_options)
else:
df.write.mode(mode).save(**non_hadoop_spark_options)
def default_target_names():
targets = mlrun.mlconf.feature_store.default_targets
return [target.strip() for target in targets.split(",")]
def get_default_targets(offline_only=False):
"""initialize the default feature set targets list"""
return [
DataTargetBase(target, name=str(target), partitioned=(target == "parquet"))
for target in default_target_names()
if not offline_only or not target == "nosql"
]
def update_targets_run_id_for_ingest(overwrite, targets, targets_in_status):
run_id = generate_target_run_id()
for target in targets:
if overwrite or target.name not in targets_in_status.keys():
target.run_id = run_id
else:
target.run_id = targets_in_status[target.name].run_id
def get_default_prefix_for_target(kind):
data_prefixes = mlrun.mlconf.feature_store.data_prefixes
data_prefix = getattr(data_prefixes, kind, None)
if not data_prefix:
data_prefix = data_prefixes.default
return data_prefix
def get_default_prefix_for_source(kind):
return get_default_prefix_for_target(kind)
def validate_target_paths_for_engine(
targets, engine, source: Union[DataSource, pd.DataFrame]
):
"""Validating that target paths are suitable for the required engine.
validate for single file targets only (parquet and csv).
spark:
cannot be a single file path (e.g - ends with .csv or .pq)
storey:
if csv - must be a single file.
if parquet - in case of partitioned it must be a directory,
else can be both single file or directory
pandas:
if source contains chunksize attribute - path must be a directory
else if parquet - if partitioned(=True) - path must be a directory
else - path must be a single file
:param targets: list of data target objects
:param engine: name of the processing engine (storey, pandas, or spark), defaults to storey
:param source: source dataframe or other sources (e.g. parquet source see:
:py:class:`~mlrun.datastore.ParquetSource` and other classes in
mlrun.datastore with suffix Source)
"""
for base_target in targets:
if hasattr(base_target, "kind") and (
base_target.kind == TargetTypes.parquet
or base_target.kind == TargetTypes.csv
):
target = get_target_driver(base_target)
is_single_file = target.is_single_file()
if engine == "spark" and is_single_file:
raise mlrun.errors.MLRunInvalidArgumentError(
f"spark CSV/Parquet targets must be directories, got path:'{target.path}'"
)
elif engine == "pandas":
# check if source is DataSource (not DataFrame) and if contains chunk size
if isinstance(source, DataSource) and source.attributes.get(
"chunksize"
):
if is_single_file:
raise mlrun.errors.MLRunInvalidArgumentError(
"pandas CSV/Parquet targets must be a directory "
f"for a chunked source, got path:'{target.path}'"
)
elif target.kind == TargetTypes.parquet and target.partitioned:
if is_single_file:
raise mlrun.errors.MLRunInvalidArgumentError(
"partitioned Parquet target for pandas engine must be a directory, "
f"got path:'{target.path}'"
)
elif not is_single_file:
raise mlrun.errors.MLRunInvalidArgumentError(
"When using a non chunked source, "
f"pandas CSV/Parquet targets must be a single file, got path:'{target.path}'"
)
elif not engine or engine == "storey":
if target.kind == TargetTypes.csv and not is_single_file:
raise mlrun.errors.MLRunInvalidArgumentError(
f"CSV target for storey engine must be a single file, got path:'{target.path}'"
)
elif (
target.kind == TargetTypes.parquet
and target.partitioned
and is_single_file
):
raise mlrun.errors.MLRunInvalidArgumentError(
f"partitioned Parquet target for storey engine must be a directory, got path:'{target.path}'"
)
def validate_target_list(targets):
"""Check that no target overrides another target in the list (name/path)"""
if not targets:
return
targets_by_kind_name = [kind for kind in targets if isinstance(kind, str)]
no_name_target_types_count = Counter(
[
target.kind
for target in targets
if hasattr(target, "name") and hasattr(target, "kind") and not target.name
]
+ targets_by_kind_name
)
target_types_requiring_name = [
target_type
for target_type, target_type_count in no_name_target_types_count.items()
if target_type_count > 1
]
if target_types_requiring_name:
raise mlrun.errors.MLRunInvalidArgumentError(
"Only one default name per target type is allowed (please "
f"specify name for {target_types_requiring_name} target)"
)
target_names_count = Counter(
[target.name for target in targets if hasattr(target, "name") and target.name]
)
targets_with_same_name = [
target_name
for target_name, target_name_count in target_names_count.items()
if target_name_count > 1
]
if targets_with_same_name:
raise mlrun.errors.MLRunInvalidArgumentError(
"Each target must have a unique name (more than one target with "
f"those names found {targets_with_same_name})"
)
no_path_target_types_count = Counter(
[
target.kind
for target in targets
if hasattr(target, "path") and hasattr(target, "kind") and not target.path
]
+ targets_by_kind_name
)
target_types_requiring_path = [
target_type
for target_type, target_type_count in no_path_target_types_count.items()
if target_type_count > 1
]
if target_types_requiring_path:
raise mlrun.errors.MLRunInvalidArgumentError(
"Only one default path per target type is allowed (please specify "
f"path for {target_types_requiring_path} target)"
)
target_paths_count = Counter(
[target.path for target in targets if hasattr(target, "path") and target.path]
)
targets_with_same_path = [
target_path
for target_path, target_path_count in target_paths_count.items()
if target_path_count > 1
]
if targets_with_same_path:
raise mlrun.errors.MLRunInvalidArgumentError(
"Each target must have a unique path (more than one target "
f"with those names found {targets_with_same_path})"
)
def validate_target_placement(graph, final_step, targets):
if final_step or graph.is_empty():
return True
for target in targets:
if not target.after_step:
raise mlrun.errors.MLRunInvalidArgumentError(
"writer step location is undetermined due to graph branching"
", set the target .after_step attribute or the graph .final_step"
)
def add_target_steps(graph, resource, targets, to_df=False, final_step=None):
"""add the target steps to the graph"""
targets = targets or []
key_columns = resource.spec.entities
timestamp_key = resource.spec.timestamp_key
features = resource.spec.features
table = None
for target in targets:
# if fset is in passthrough mode, ingest skips writing the data to offline targets
if resource.spec.passthrough and kind_to_driver[target.kind].is_offline:
continue
driver = get_target_driver(target, resource)
table = driver.get_table_object() or table
driver.update_resource_status()
if target.after_step:
target.attributes["infer_columns_from_data"] = True
driver.add_writer_step(
graph,
target.after_step or final_step,
features=features if not target.after_step else None,
key_columns=key_columns,
timestamp_key=timestamp_key,
featureset_status=resource.status,
)
if to_df:
# add dataframe target, will return a dataframe
driver = DFTarget()
driver.add_writer_step(
graph,
final_step,
features=features,
key_columns=key_columns,
timestamp_key=timestamp_key,
)
return table
offline_lookup_order = [TargetTypes.parquet, TargetTypes.csv]
online_lookup_order = [TargetTypes.nosql]
def get_offline_target(featureset, name=None):
"""return an optimal offline feature set target"""
# todo: take status, start_time and lookup order into account
offline_targets = [
target
for target in featureset.status.targets
if kind_to_driver[target.kind].is_offline
]
target = None
if name:
target = next((t for t in offline_targets if t.name == name), None)
else:
for kind in offline_lookup_order:
target = next((t for t in offline_targets if t.kind == kind), None)
if target:
break
if target is None and offline_targets:
target = offline_targets[0]
if target:
return get_target_driver(target, featureset)
return None
def get_online_target(resource, name=None):
"""return an optimal online feature set target"""
# todo: take lookup order into account
for target in resource.status.targets:
if name and target.name != name:
continue
driver = kind_to_driver[target.kind]
if driver.is_online:
return get_target_driver(target, resource)
return None
def get_target_driver(target_spec, resource=None):
if isinstance(target_spec, dict):
target_spec = DataTargetBase.from_dict(target_spec)
driver_class = kind_to_driver[target_spec.kind]
return driver_class.from_spec(target_spec, resource)
class BaseStoreTarget(DataTargetBase):
"""base target storage driver, used to materialize feature set/vector data"""
kind = ""
is_table = False
suffix = ""
is_online = False
is_offline = False
support_spark = False
support_storey = False
support_pandas = False
support_append = False
def __init__(
self,
name: str = "",
path=None,
attributes: dict[str, str] = None,
after_step=None,
columns=None,
partitioned: bool = False,
key_bucketing_number: Optional[int] = None,
partition_cols: Optional[list[str]] = None,
time_partitioning_granularity: Optional[str] = None,
max_events: Optional[int] = None,
flush_after_seconds: Optional[int] = None,
storage_options: dict[str, str] = None,
schema: dict[str, Any] = None,
credentials_prefix=None,
):
super().__init__(
self.kind,
name,
path,
attributes,
after_step,
partitioned,
key_bucketing_number,
partition_cols,
time_partitioning_granularity,
max_events,
flush_after_seconds,
schema=schema,
credentials_prefix=credentials_prefix,
)
self.name = name or self.kind
self.path = str(path) if path is not None else None
self.after_step = after_step
self.attributes = attributes or {}
self.columns = columns or []
self.partitioned = partitioned
self.key_bucketing_number = key_bucketing_number
self.partition_cols = partition_cols
self.time_partitioning_granularity = time_partitioning_granularity
self.max_events = max_events
self.flush_after_seconds = flush_after_seconds
self.storage_options = storage_options
self.schema = schema or {}
self.credentials_prefix = credentials_prefix
if credentials_prefix:
warnings.warn(
"The 'credentials_prefix' parameter is deprecated and will be removed in "
"1.9.0. Please use datastore profiles instead.",
FutureWarning,
)
self._target = None
self._resource = None
self._secrets = {}
def _get_credential(self, key, default_value=None):
return mlrun.get_secret_or_env(
key,
secret_provider=self._secrets,
default=default_value,
prefix=self.credentials_prefix,
)
def _get_store_and_path(self):
credentials_prefix_secrets = (
{"CREDENTIALS_PREFIX": self.credentials_prefix}
if self.credentials_prefix
else None
)
store, resolved_store_path, url = mlrun.store_manager.get_or_create_store(
self.get_target_path(),
credentials_prefix_secrets,
)
return store, resolved_store_path, url
def _get_column_list(self, features, timestamp_key, key_columns, with_type=False):
result = []
if self.columns:
if with_type:
columns = set(self.columns)
for feature in features:
if feature.name in columns:
result.append((feature.name, feature.value_type))
else:
result = self.columns
elif features:
if with_type:
for feature in features:
result.append((feature.name, feature.value_type))
else:
result = list(features.keys())
if key_columns:
for key in reversed(key_columns):
if key not in result:
if with_type:
result.insert(0, (key, ValueType.STRING))
else:
result.insert(0, key)
if timestamp_key:
if with_type:
result = [(timestamp_key, ValueType.DATETIME)] + result
else:
result = [timestamp_key] + result
return result
def write_dataframe(
self,
df,
key_column=None,
timestamp_key=None,
chunk_id=0,
**kwargs,
) -> Optional[int]:
if hasattr(df, "rdd"):
options = self.get_spark_options(key_column, timestamp_key)
options.update(kwargs)
df = self.prepare_spark_df(df, key_column, timestamp_key, options)
write_format = options.pop("format", None)
write_spark_dataframe_with_options(
options, df, "overwrite", write_format=write_format
)
elif hasattr(df, "dask"):
dask_options = self.get_dask_options()
store, path_in_store, target_path = self._get_store_and_path()
storage_options = store.get_storage_options()
df = df.repartition(partition_size="100MB")
try:
if dask_options["format"] == "parquet":
df.to_parquet(
generate_path_with_chunk(self, chunk_id, target_path),
storage_options=storage_options,
)
elif dask_options["format"] == "csv":
df.to_csv(
generate_path_with_chunk(self, chunk_id, target_path),
storage_options=storage_options,
)
else:
raise NotImplementedError(
"Format for writing dask dataframe should be CSV or Parquet!"
)
except Exception as exc:
raise RuntimeError("Failed to write Dask Dataframe") from exc
else:
store, path_in_store, target_path = self._get_store_and_path()
target_path = generate_path_with_chunk(self, chunk_id, target_path)
file_system = store.filesystem
if (
file_system.protocol == "file"
# fsspec 2023.10.0 changed protocol from "file" to ("file", "local")
or isinstance(file_system.protocol, (tuple, list))
and "file" in file_system.protocol
):
dir = os.path.dirname(target_path)
if dir:
os.makedirs(dir, exist_ok=True)
target_df = df
partition_cols = None # single parquet file
if not mlrun.utils.helpers.is_parquet_file(target_path): # directory
partition_cols = []
if timestamp_key and (
self.partitioned or self.time_partitioning_granularity
):
target_df = df.copy(deep=False)
time_partitioning_granularity = self.time_partitioning_granularity
if not time_partitioning_granularity and self.partitioned:
time_partitioning_granularity = (
mlrun.utils.helpers.DEFAULT_TIME_PARTITIONING_GRANULARITY
)
for unit, fmt in [
("year", "%Y"),
("month", "%m"),
("day", "%d"),
("hour", "%H"),
("minute", "%M"),
]:
partition_cols.append(unit)
target_df[unit] = pd.DatetimeIndex(
target_df[timestamp_key]
).format(date_format=fmt)
if unit == time_partitioning_granularity:
break
# Partitioning will be performed on timestamp_key and then on self.partition_cols
# (We might want to give the user control on this order as additional functionality)
partition_cols += self.partition_cols or []
storage_options = store.get_storage_options()
if storage_options and self.storage_options:
storage_options = merge(storage_options, self.storage_options)
else:
storage_options = storage_options or self.storage_options
self._write_dataframe(
target_df,
storage_options,
target_path,
partition_cols=partition_cols,
**kwargs,
)
try:
return file_system.size(target_path)
except Exception:
return None
@staticmethod
def _write_dataframe(df, storage_options, target_path, partition_cols, **kwargs):
raise NotImplementedError()
def set_secrets(self, secrets):
self._secrets = secrets
def set_resource(self, resource):
self._resource = resource
@classmethod
def from_spec(cls, spec: DataTargetBase, resource=None):
"""create target driver from target spec or other target driver"""
driver = cls()
driver.name = spec.name
driver.path = spec.path
driver.attributes = spec.attributes
driver.schema = spec.schema
driver.credentials_prefix = spec.credentials_prefix
if hasattr(spec, "columns"):
driver.columns = spec.columns
if hasattr(spec, "_secrets"):
driver._secrets = spec._secrets
driver.partitioned = spec.partitioned
driver.key_bucketing_number = spec.key_bucketing_number
driver.partition_cols = spec.partition_cols
driver.time_partitioning_granularity = spec.time_partitioning_granularity
driver.max_events = spec.max_events
driver.flush_after_seconds = spec.flush_after_seconds
driver.storage_options = spec.storage_options
driver.credentials_prefix = spec.credentials_prefix
driver._resource = resource
driver.run_id = spec.run_id
driver.after_step = spec.after_step
return driver
def get_table_object(self):
"""get storey Table object"""
return None
def get_target_path(self):
path_object = self._target_path_object
project_name = self._resource.metadata.project if self._resource else None
return (
path_object.get_absolute_path(project_name=project_name)
if path_object
else None
)
def get_target_templated_path(self):
path_object = self._target_path_object
return path_object.get_templated_path() if path_object else None
@property
def _target_path_object(self):
"""return the actual/computed target path"""
is_single_file = hasattr(self, "is_single_file") and self.is_single_file()
if self._resource and self.path:
parsed_url = urlparse(self.path)
# When the URL consists only from scheme and endpoint and no path,
# make a default path for DS and redis targets.
# Also ignore KafkaTarget when it uses the ds scheme (no default path for KafkaTarget)
if (
not isinstance(self, KafkaTarget)
and parsed_url.scheme in ["ds", "redis", "rediss"]
and (not parsed_url.path or parsed_url.path == "/")
):
return TargetPathObject(
_get_target_path(
self,
self._resource,
self.run_id is not None,
netloc=parsed_url.netloc,
scheme=parsed_url.scheme,
),
self.run_id,
is_single_file,
)
return self.get_path() or (
TargetPathObject(
_get_target_path(self, self._resource, self.run_id is not None),
self.run_id,
is_single_file,
)
if self._resource
else None
)
def update_resource_status(self, status="", producer=None, size=None):
"""update the data target status"""
self._target = self._target or DataTarget(
self.kind, self.name, self.get_target_templated_path()
)
target = self._target
target.attributes = self.attributes
target.run_id = self.run_id
target.status = status or target.status or "created"
target.updated = now_date().isoformat()
target.size = size
target.producer = producer or target.producer
# Copy partitioning-related fields to the status, since these are needed if reading the actual data that
# is related to the specific target.
# TODO - instead of adding more fields to the status targets, we should consider changing any functionality
# that depends on "spec-fields" to use a merge between the status and the spec targets. One such place
# is the fset.to_dataframe() function.
target.partitioned = self.partitioned
target.key_bucketing_number = self.key_bucketing_number
target.partition_cols = self.partition_cols
target.time_partitioning_granularity = self.time_partitioning_granularity
target.credentials_prefix = self.credentials_prefix
self._resource.status.update_target(target)
return target
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
if not self.support_storey:
raise mlrun.errors.MLRunRuntimeError(
f"{type(self).__name__} does not support storey engine"
)
raise NotImplementedError()
def purge(self):
"""
Delete the files of the target.
Do not use this function directly from the sdk. Use FeatureSet.purge_targets.
"""
store, path_in_store, target_path = self._get_store_and_path()
if path_in_store not in ["", "/"]:
store.rm(path_in_store, recursive=True)
else:
raise mlrun.errors.MLRunInvalidArgumentError(
"Unable to delete target. Please Use purge_targets from FeatureSet object."
)
def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
"""return the target data as dataframe"""
if not self.support_pandas:
raise NotImplementedError()
mlrun.utils.helpers.additional_filters_warning(
additional_filters, self.__class__
)
return mlrun.get_dataitem(self.get_target_path()).as_df(
columns=columns,
df_module=df_module,
start_time=start_time,
end_time=end_time,
time_column=time_column,
**kwargs,
)
def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
# options used in spark.read.load(**options)
if not self.support_spark:
raise mlrun.errors.MLRunRuntimeError(
f"{type(self).__name__} does not support spark engine"
)
raise NotImplementedError()
def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
return df
def get_dask_options(self):
raise NotImplementedError()
@property
def source_spark_attributes(self) -> dict:
return {}
[docs]class ParquetTarget(BaseStoreTarget):
"""Parquet target storage driver, used to materialize feature set/vector data into parquet files.
:param name: optional, target name. By default will be called ParquetTarget
:param path: optional, Output path. Can be either a file or directory.
This parameter is forwarded as-is to pandas.DataFrame.to_parquet().
Default location v3io:///projects/{project}/FeatureStore/{name}/parquet/
:param attributes: optional, extra attributes for storey.ParquetTarget
:param after_step: optional, after what step in the graph to add the target
:param columns: optional, which columns from data to write
:param partitioned: optional, whether to partition the file, False by default,
if True without passing any other partition field, the data will be partitioned by /year/month/day/hour
:param key_bucketing_number: optional, None by default will not partition by key,
0 will partition by the key as is, any other number X will create X partitions and hash the keys to one of them
:param partition_cols: optional, name of columns from the data to partition by
:param time_partitioning_granularity: optional. the smallest time unit to partition the data by.
For example "hour" will yield partitions of the format /year/month/day/hour
:param max_events: optional. Maximum number of events to write at a time.
All events will be written on flow termination,
or after flush_after_seconds (if flush_after_seconds is set). Default 10k events
:param flush_after_seconds: optional. Maximum number of seconds to hold events before they are written.
All events will be written on flow termination, or after max_events are accumulated (if max_events is set).
Default 15 minutes
"""
kind = TargetTypes.parquet
is_offline = True
support_spark = True
support_storey = True
support_dask = True
support_pandas = True
support_append = True
def __init__(
self,
name: str = "",
path=None,
attributes: dict[str, str] = None,
after_step=None,
columns=None,
partitioned: bool = None,
key_bucketing_number: Optional[int] = None,
partition_cols: Optional[list[str]] = None,
time_partitioning_granularity: Optional[str] = None,
max_events: Optional[int] = 10000,
flush_after_seconds: Optional[int] = 900,
storage_options: dict[str, str] = None,
):
self.path = path
if partitioned is None:
partitioned = not self.is_single_file()
super().__init__(
name,
path,
attributes,
after_step,
columns,
partitioned,
key_bucketing_number,
partition_cols,
time_partitioning_granularity,
max_events=max_events,
flush_after_seconds=flush_after_seconds,
storage_options=storage_options,
)
if (
time_partitioning_granularity is not None
and time_partitioning_granularity
not in mlrun.utils.helpers.LEGAL_TIME_UNITS
):
raise errors.MLRunInvalidArgumentError(
f"time_partitioning_granularity parameter must be one of "
f"{','.join(mlrun.utils.helpers.LEGAL_TIME_UNITS)}, "
f"not {time_partitioning_granularity}."
)
@staticmethod
def _write_dataframe(df, storage_options, target_path, partition_cols, **kwargs):
# In order to save the DataFrame in parquet format, all of the column names must be strings:
df.columns = [str(column) for column in df.columns.tolist()]
to_parquet(
df,
target_path,
partition_cols=partition_cols,
storage_options=storage_options,
**kwargs,
)
[docs] def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
if self.attributes.get("infer_columns_from_data"):
column_list = None
else:
column_list = self._get_column_list(
features=features,
timestamp_key=timestamp_key,
key_columns=None,
with_type=True,
)
# need to extract types from features as part of column list
partition_cols = None
if self.key_bucketing_number is not None:
partition_cols = [("$key", self.key_bucketing_number)]
if self.partition_cols:
partition_cols = partition_cols or []
partition_cols.extend(self.partition_cols)
time_partitioning_granularity = self.time_partitioning_granularity
if self.partitioned and all(
value is None
for value in [
time_partitioning_granularity,
self.key_bucketing_number,
self.partition_cols,
]
):
time_partitioning_granularity = (
mlrun.utils.helpers.DEFAULT_TIME_PARTITIONING_GRANULARITY
)
if time_partitioning_granularity is not None:
partition_cols = partition_cols or []
for time_unit in mlrun.utils.helpers.LEGAL_TIME_UNITS:
partition_cols.append(f"${time_unit}")
if time_unit == time_partitioning_granularity:
break
target_path = self.get_target_path()
if not self.partitioned and not mlrun.utils.helpers.is_parquet_file(
target_path
):
partition_cols = []
tuple_key_columns = []
for key_column in key_columns:
tuple_key_columns.append((key_column.name, key_column.value_type))
step = graph.add_step(
name=self.name or "ParquetTarget",
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.ParquetStoreyTarget",
path=target_path,
columns=column_list,
index_cols=tuple_key_columns,
partition_cols=partition_cols,
time_field=timestamp_key,
max_events=self.max_events,
flush_after_seconds=self.flush_after_seconds,
update_last_written=featureset_status.update_last_written_for_target,
**self.attributes,
)
original_to_dict = step.to_dict
def delete_update_last_written(*arg, **kargs):
result = original_to_dict(*arg, **kargs)
result["class_args"].pop("update_last_written", None)
return result
# update_last_written is not serializable (ML-5108)
step.to_dict = delete_update_last_written
[docs] def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
partition_cols = []
if timestamp_key:
time_partitioning_granularity = self.time_partitioning_granularity
if (
not time_partitioning_granularity
and self.partitioned
and not self.partition_cols
):
time_partitioning_granularity = (
mlrun.utils.helpers.DEFAULT_TIME_PARTITIONING_GRANULARITY
)
if time_partitioning_granularity:
for unit in mlrun.utils.helpers.LEGAL_TIME_UNITS:
partition_cols.append(unit)
if unit == time_partitioning_granularity:
break
store, path, url = self._get_store_and_path()
spark_options = store.get_spark_options()
spark_options.update(
{
"path": store.spark_url + path,
"format": "parquet",
}
)
for partition_col in self.partition_cols or []:
partition_cols.append(partition_col)
if partition_cols:
spark_options["partitionBy"] = partition_cols
return spark_options
[docs] def get_dask_options(self):
return {"format": "parquet"}
[docs] def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
"""return the target data as dataframe"""
result = mlrun.get_dataitem(self.get_target_path()).as_df(
columns=columns,
df_module=df_module,
format="parquet",
start_time=start_time,
end_time=end_time,
time_column=time_column,
additional_filters=transform_list_filters_to_tuple(additional_filters),
**kwargs,
)
if not columns:
drop_cols = []
if self.time_partitioning_granularity:
for col in mlrun.utils.helpers.LEGAL_TIME_UNITS:
drop_cols.append(col)
if col == self.time_partitioning_granularity:
break
elif (
self.partitioned
and not self.partition_cols
and not self.key_bucketing_number
):
drop_cols = mlrun.utils.helpers.DEFAULT_TIME_PARTITIONS
if drop_cols:
# if these columns aren't present for some reason, that's no reason to fail
result.drop(columns=drop_cols, inplace=True, errors="ignore")
return result
[docs] def is_single_file(self):
return mlrun.utils.helpers.is_parquet_file(self.path)
[docs] def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
# If partitioning by time, add the necessary columns
if (
timestamp_key
and isinstance(spark_options, dict)
and "partitionBy" in spark_options
):
from pyspark.sql.functions import (
dayofmonth,
hour,
minute,
month,
second,
year,
)
time_unit_to_op = {
"year": year,
"month": month,
"day": dayofmonth,
"hour": hour,
"minute": minute,
"second": second,
}
timestamp_col = df[timestamp_key]
for partition in spark_options["partitionBy"]:
if partition not in df.columns and partition in time_unit_to_op:
op = time_unit_to_op[partition]
df = df.withColumn(partition, op(timestamp_col))
return df
[docs]class CSVTarget(BaseStoreTarget):
kind = TargetTypes.csv
suffix = ".csv"
is_offline = True
support_spark = True
support_storey = True
support_pandas = True
@staticmethod
def _write_dataframe(df, storage_options, target_path, partition_cols, **kwargs):
# avoid writing the range index unless explicitly specified via kwargs
if isinstance(df.index, pd.RangeIndex):
kwargs["index"] = kwargs.get("index", False)
df.to_csv(target_path, storage_options=storage_options, **kwargs)
[docs] def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
column_list = self._get_column_list(
features=features, timestamp_key=timestamp_key, key_columns=key_columns
)
target_path = self.get_target_path()
graph.add_step(
name=self.name or "CSVTarget",
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.CSVStoreyTarget",
path=target_path,
columns=column_list,
header=True,
index_cols=key_columns,
**self.attributes,
)
[docs] def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
store, path, url = self._get_store_and_path()
spark_options = store.get_spark_options()
spark_options.update(
{
"path": store.spark_url + path,
"format": "csv",
"header": "true",
}
)
return spark_options
[docs] def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
import pyspark.sql.functions as funcs
for col_name, col_type in df.dtypes:
# covers TimestampType and TimestampNTZType, which was added in PySpark 3.4.0
if col_type.startswith("timestamp"):
# df.write.csv saves timestamps with millisecond precision, but we want microsecond precision
# for compatibility with storey.
df = df.withColumn(
col_name, funcs.date_format(col_name, "yyyy-MM-dd HH:mm:ss.SSSSSS")
)
return df
[docs] def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
mlrun.utils.helpers.additional_filters_warning(
additional_filters, self.__class__
)
df = super().as_df(
columns=columns,
df_module=df_module,
entities=entities,
format="csv",
start_time=start_time,
end_time=end_time,
time_column=time_column,
**kwargs,
)
if entities:
df.set_index(keys=entities, inplace=True)
return df
[docs] def is_single_file(self):
if self.path:
return self.path.endswith(".csv")
return True
class SnowflakeTarget(BaseStoreTarget):
"""
:param attributes: A dictionary of attributes for Snowflake connection; will be overridden by database parameters
if they exist.
:param url: Snowflake hostname, in the format: <account_name>.<region>.snowflakecomputing.com
:param user: Snowflake user for login
:param db_schema: Database schema
:param database: Database name
:param warehouse: Snowflake warehouse name
:param table_name: Snowflake table name
"""
support_spark = True
support_append = True
is_offline = True
kind = TargetTypes.snowflake
def __init__(
self,
name: str = "",
path=None,
attributes: dict[str, str] = None,
after_step=None,
columns=None,
partitioned: bool = False,
key_bucketing_number: Optional[int] = None,
partition_cols: Optional[list[str]] = None,
time_partitioning_granularity: Optional[str] = None,
max_events: Optional[int] = None,
flush_after_seconds: Optional[int] = None,
storage_options: dict[str, str] = None,
schema: dict[str, Any] = None,
credentials_prefix=None,
url: str = None,
user: str = None,
db_schema: str = None,
database: str = None,
warehouse: str = None,
table_name: str = None,
):
attributes = attributes or {}
if url:
attributes["url"] = url
if user:
attributes["user"] = user
if database:
attributes["database"] = database
if db_schema:
attributes["db_schema"] = db_schema
if warehouse:
attributes["warehouse"] = warehouse
if table_name:
attributes["table"] = table_name
super().__init__(
name,
path,
attributes,
after_step,
list(schema.keys()) if schema else columns,
partitioned,
key_bucketing_number,
partition_cols,
time_partitioning_granularity,
max_events=max_events,
flush_after_seconds=flush_after_seconds,
storage_options=storage_options,
schema=schema,
credentials_prefix=credentials_prefix,
)
def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
spark_options = get_snowflake_spark_options(self.attributes)
spark_options["dbtable"] = self.attributes.get("table")
return spark_options
def purge(self):
import snowflake.connector
missing = [
key
for key in ["database", "db_schema", "table", "url", "user", "warehouse"]
if self.attributes.get(key) is None
]
if missing:
raise mlrun.errors.MLRunRuntimeError(
f"Can't purge Snowflake target, "
f"some attributes are missing: {', '.join(missing)}"
)
account = self.attributes["url"].replace(".snowflakecomputing.com", "")
with snowflake.connector.connect(
account=account,
user=self.attributes["user"],
password=get_snowflake_password(),
warehouse=self.attributes["warehouse"],
) as snowflake_connector:
drop_statement = (
f"DROP TABLE IF EXISTS {self.attributes['database']}.{self.attributes['db_schema']}"
f".{self.attributes['table']}"
)
snowflake_connector.execute_string(drop_statement)
def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
raise mlrun.errors.MLRunRuntimeError(
f"{type(self).__name__} does not support pandas engine"
)
@property
def source_spark_attributes(self) -> dict:
keys = ["url", "user", "database", "db_schema", "warehouse"]
attributes = self.attributes or {}
snowflake_dict = {key: attributes.get(key) for key in keys}
table = attributes.get("table")
snowflake_dict["query"] = f"SELECT * from {table}" if table else None
return snowflake_dict
class NoSqlBaseTarget(BaseStoreTarget):
is_table = True
is_online = True
support_append = True
support_storey = True
writer_step_name = "base_name"
def __new__(cls, *args, **kwargs):
if cls is NoSqlBaseTarget:
raise TypeError(f"only children of '{cls.__name__}' may be instantiated")
return object.__new__(cls)
def get_table_object(self):
raise NotImplementedError()
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
table, column_list = self._get_table_and_columns(features, key_columns)
graph.add_step(
name=self.name or self.writer_step_name,
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.NoSqlStoreyTarget",
columns=column_list,
table=table,
**self.attributes,
)
def _get_table_and_columns(self, features, key_columns):
key_columns = list(key_columns.keys())
table = self._resource.uri
column_list = self._get_column_list(
features=features,
timestamp_key=None,
key_columns=key_columns,
with_type=True,
)
if not self.columns:
aggregate_features = (
[key for key, feature in features.items() if feature.aggregate]
if features
else []
)
column_list = [
col for col in column_list if col[0] not in aggregate_features
]
return table, column_list
def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
raise NotImplementedError()
def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
raise NotImplementedError()
def get_dask_options(self):
return {"format": "csv"}
def write_dataframe(
self, df, key_column=None, timestamp_key=None, chunk_id=0, **kwargs
):
if hasattr(df, "rdd"):
options = self.get_spark_options(key_column, timestamp_key)
options.update(kwargs)
df = self.prepare_spark_df(df)
write_format = options.pop("format", None)
write_spark_dataframe_with_options(
options, df, "overwrite", write_format=write_format
)
else:
# To prevent modification of the original dataframe and make sure
# that the last event of a key is the one being persisted
if len(df.index.names) and df.index.names[0] is not None:
df = df.groupby(df.index.names).last()
else:
df = df.copy(deep=False)
access_key = self._get_credential("V3IO_ACCESS_KEY")
store, path_in_store, target_path = self._get_store_and_path()
storage_options = store.get_storage_options()
access_key = storage_options.get("v3io_access_key", access_key)
_, path_with_container = parse_path(target_path)
container, path = split_path(path_with_container)
frames_client = get_frames_client(
token=access_key, address=config.v3io_framesd, container=container
)
frames_client.write("kv", path, df, index_cols=key_column, **kwargs)
[docs]class NoSqlTarget(NoSqlBaseTarget):
kind = TargetTypes.nosql
support_spark = True
writer_step_name = "NoSqlTarget"
[docs] def get_table_object(self):
from storey import Table, V3ioDriver
store, path_in_store, target_path = self._get_store_and_path()
endpoint, uri = parse_path(target_path)
storage_options = store.get_storage_options()
access_key = storage_options.get("v3io_access_key")
return Table(
uri,
V3ioDriver(webapi=endpoint or mlrun.mlconf.v3io_api, access_key=access_key),
flush_interval_secs=mlrun.mlconf.feature_store.flush_interval,
)
[docs] def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
store, path_in_store, target_path = self._get_store_and_path()
storage_options = store.get_storage_options()
store_access_key = storage_options.get("v3io_access_key")
env_access_key = self._secrets.get(
"V3IO_ACCESS_KEY", os.getenv("V3IO_ACCESS_KEY")
)
if store_access_key and env_access_key and store_access_key != env_access_key:
logger.warning(
"The Spark v3io connector does not support access_key parameterization."
"Spark will disregard the store-provided key."
)
spark_options = {
"path": store.spark_url + path_in_store,
"format": "io.iguaz.v3io.spark.sql.kv",
}
if isinstance(key_column, list) and len(key_column) >= 1:
spark_options["key"] = key_column[0]
if len(key_column) > 2:
spark_options["sorting-key"] = "_spark_object_name"
if len(key_column) == 2:
spark_options["sorting-key"] = key_column[1]
else:
spark_options["key"] = key_column
if not overwrite:
spark_options["columnUpdate"] = True
return spark_options
[docs] def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
from pyspark.sql.functions import col
spark_udf_directory = os.path.dirname(os.path.abspath(__file__))
sys.path.append(spark_udf_directory)
try:
import spark_udf
df.rdd.context.addFile(spark_udf.__file__)
for col_name, col_type in df.dtypes:
if col_type.startswith("decimal("):
# V3IO does not support this level of precision
df = df.withColumn(col_name, col(col_name).cast("double"))
if len(key_columns) > 2:
return df.withColumn(
"_spark_object_name",
spark_udf.hash_and_concat_v3io_udf(
*[col(c) for c in key_columns[1:]]
),
)
finally:
sys.path.remove(spark_udf_directory)
return df
class RedisNoSqlTarget(NoSqlBaseTarget):
kind = TargetTypes.redisnosql
support_spark = True
writer_step_name = "RedisNoSqlTarget"
@staticmethod
def get_server_endpoint(path, credentials_prefix=None):
endpoint, uri = parse_path(path)
endpoint = endpoint or mlrun.mlconf.redis.url
if endpoint.startswith("ds://"):
datastore_profile = datastore_profile_read(endpoint)
if not datastore_profile:
raise ValueError(f"Failed to load datastore profile '{endpoint}'")
if datastore_profile.type != "redis":
raise ValueError(
f"Trying to use profile of type '{datastore_profile.type}' as redis datastore"
)
endpoint = datastore_profile.url_with_credentials()
else:
parsed_endpoint = urlparse(endpoint)
if parsed_endpoint.username or parsed_endpoint.password:
raise mlrun.errors.MLRunInvalidArgumentError(
"Provide Redis username and password only via secrets"
)
credentials_prefix = credentials_prefix or mlrun.get_secret_or_env(
key="CREDENTIALS_PREFIX"
)
user = mlrun.get_secret_or_env(
"REDIS_USER", default="", prefix=credentials_prefix
)
password = mlrun.get_secret_or_env(
"REDIS_PASSWORD", default="", prefix=credentials_prefix
)
host = parsed_endpoint.hostname
port = parsed_endpoint.port if parsed_endpoint.port else "6379"
scheme = parsed_endpoint.scheme
if user or password:
endpoint = f"{scheme}://{user}:{password}@{host}:{port}"
else:
endpoint = f"{scheme}://{host}:{port}"
return endpoint, uri
def get_table_object(self):
from storey import Table
from storey.redis_driver import RedisDriver
endpoint, uri = self.get_server_endpoint(
self.get_target_path(), self.credentials_prefix
)
return Table(
uri,
RedisDriver(redis_url=endpoint, key_prefix="/"),
flush_interval_secs=mlrun.mlconf.feature_store.flush_interval,
)
def get_spark_options(self, key_column=None, timestamp_key=None, overwrite=True):
endpoint, uri = self.get_server_endpoint(
self.get_target_path(), self.credentials_prefix
)
parsed_endpoint = urlparse(endpoint)
store, path_in_store, path = self._get_store_and_path()
return {
"key.column": "_spark_object_name",
"table": "{" + path_in_store,
"format": "org.apache.spark.sql.redis",
"host": parsed_endpoint.hostname,
"port": parsed_endpoint.port,
"user": parsed_endpoint.username if parsed_endpoint.username else None,
"auth": parsed_endpoint.password if parsed_endpoint.password else None,
}
def prepare_spark_df(self, df, key_columns, timestamp_key=None, spark_options=None):
from pyspark.sql.functions import col
spark_udf_directory = os.path.dirname(os.path.abspath(__file__))
sys.path.append(spark_udf_directory)
try:
import spark_udf
df.rdd.context.addFile(spark_udf.__file__)
df = df.withColumn(
"_spark_object_name",
spark_udf.hash_and_concat_redis_udf(*[col(c) for c in key_columns]),
)
finally:
sys.path.remove(spark_udf_directory)
return df
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
table, column_list = self._get_table_and_columns(features, key_columns)
graph.add_step(
path=self.get_target_path(),
name=self.name or self.writer_step_name,
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.RedisNoSqlStoreyTarget",
columns=column_list,
table=table,
credentials_prefix=self.credentials_prefix,
**self.attributes,
)
[docs]class StreamTarget(BaseStoreTarget):
kind = TargetTypes.stream
is_table = False
is_online = False
support_spark = False
support_storey = True
support_append = True
[docs] def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
column_list = self._get_column_list(
features=features, timestamp_key=timestamp_key, key_columns=key_columns
)
stream_path = self.get_target_path()
if not stream_path:
raise mlrun.errors.MLRunInvalidArgumentError("StreamTarget requires a path")
graph.add_step(
name=self.name or "StreamTarget",
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.StreamStoreyTarget",
columns=column_list,
stream_path=stream_path,
**self.attributes,
)
class KafkaTarget(BaseStoreTarget):
"""
Kafka target storage driver, used to write data into kafka topics.
example::
# define target
kafka_target = KafkaTarget(
name="kafka", path="my_topic", brokers="localhost:9092"
)
# ingest
stocks_set.ingest(stocks, [kafka_target])
:param name: target name
:param path: topic name e.g. "my_topic"
:param after_step: optional, after what step in the graph to add the target
:param columns: optional, which columns from data to write
:param bootstrap_servers: Deprecated. Use the brokers parameter instead
:param producer_options: additional configurations for kafka producer
:param brokers: kafka broker as represented by a host:port pair, or a list of kafka brokers, e.g.
"localhost:9092", or ["kafka-broker-1:9092", "kafka-broker-2:9092"]
"""
kind = TargetTypes.kafka
is_table = False
is_online = False
support_spark = False
support_storey = True
support_append = True
def __init__(
self,
*args,
bootstrap_servers=None,
producer_options=None,
brokers=None,
**kwargs,
):
attrs = {}
# TODO: Remove this in 1.9.0
if bootstrap_servers:
if brokers:
raise mlrun.errors.MLRunInvalidArgumentError(
"KafkaTarget cannot be created with both the 'brokers' parameter and the deprecated "
"'bootstrap_servers' parameter. Please use 'brokers' only."
)
warnings.warn(
"'bootstrap_servers' parameter is deprecated in 1.7.0 and will be removed in 1.9.0, "
"use 'brokers' instead.",
FutureWarning,
)
brokers = bootstrap_servers
if brokers:
attrs["brokers"] = brokers
if producer_options is not None:
attrs["producer_options"] = producer_options
super().__init__(*args, attributes=attrs, **kwargs)
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
column_list = self._get_column_list(
features=features, timestamp_key=timestamp_key, key_columns=key_columns
)
path = self.get_target_path()
if not path:
raise mlrun.errors.MLRunInvalidArgumentError("KafkaTarget requires a path")
graph.add_step(
name=self.name or "KafkaTarget",
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.KafkaStoreyTarget",
columns=column_list,
path=path,
attributes=self.attributes,
)
def purge(self):
pass
class TSDBTarget(BaseStoreTarget):
kind = TargetTypes.tsdb
is_table = False
is_online = False
support_spark = False
support_storey = True
support_append = True
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
endpoint, uri = parse_path(self.get_target_path())
if not timestamp_key:
raise mlrun.errors.MLRunInvalidArgumentError(
"feature set timestamp_key must be specified for TSDBTarget writer"
)
column_list = self._get_column_list(
features=features, timestamp_key=None, key_columns=key_columns
)
graph.add_step(
name=self.name or "TSDBTarget",
class_name="mlrun.datastore.storeytargets.TSDBStoreyTarget",
after=after,
graph_shape="cylinder",
path=uri,
time_col=timestamp_key,
index_cols=key_columns,
columns=column_list,
**self.attributes,
)
def write_dataframe(
self, df, key_column=None, timestamp_key=None, chunk_id=0, **kwargs
):
access_key = self._secrets.get("V3IO_ACCESS_KEY", os.getenv("V3IO_ACCESS_KEY"))
new_index = []
if timestamp_key:
new_index.append(timestamp_key)
if key_column:
if isinstance(key_column, str):
key_column = [key_column]
new_index.extend(key_column)
store, path_in_store, target_path = self._get_store_and_path()
storage_options = store.get_storage_options()
access_key = storage_options.get("v3io_access_key", access_key)
_, path_with_container = parse_path(target_path)
container, path = split_path(path_with_container)
frames_client = get_frames_client(
token=access_key,
address=config.v3io_framesd,
container=container,
)
frames_client.write(
"tsdb", path, df, index_cols=new_index if new_index else None, **kwargs
)
class CustomTarget(BaseStoreTarget):
kind = "custom"
is_table = False
is_online = False
support_spark = False
support_storey = True
support_pandas = True
def __init__(
self,
class_name: str,
name: str = "",
after_step=None,
**attributes,
):
attributes = attributes or {}
attributes["class_name"] = class_name
super().__init__(name, "", attributes, after_step=after_step)
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
attributes = copy(self.attributes)
class_name = attributes.pop("class_name")
graph.add_step(
name=self.name,
after=after,
graph_shape="cylinder",
class_name=class_name,
**attributes,
)
class DFTarget(BaseStoreTarget):
kind = TargetTypes.dataframe
support_storey = True
support_pandas = True
def __init__(self, *args, name="dataframe", **kwargs):
self._df = None
super().__init__(*args, name=name, **kwargs)
def set_df(self, df):
self._df = df
def update_resource_status(self, status="", producer=None, size=None):
pass
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
# todo: column filter
graph.add_step(
name=self.name or "WriteToDataFrame",
after=after,
graph_shape="cylinder",
class_name="storey.ReduceToDataFrame",
index=key_columns,
insert_key_column_as=key_columns,
insert_time_column_as=timestamp_key,
)
def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
mlrun.utils.helpers.additional_filters_warning(
additional_filters, self.__class__
)
return select_columns_from_df(
filter_df_start_end_time(
self._df,
time_column=time_column,
start_time=start_time,
end_time=end_time,
),
columns,
)
class SQLTarget(BaseStoreTarget):
kind = TargetTypes.sql
is_online = True
support_spark = False
support_storey = True
support_pandas = True
def __init__(
self,
name: str = "",
path=None,
attributes: dict[str, str] = None,
after_step=None,
partitioned: bool = False,
key_bucketing_number: Optional[int] = None,
partition_cols: Optional[list[str]] = None,
time_partitioning_granularity: Optional[str] = None,
max_events: Optional[int] = None,
flush_after_seconds: Optional[int] = None,
storage_options: dict[str, str] = None,
db_url: str = None,
table_name: str = None,
schema: dict[str, Any] = None,
primary_key_column: str = "",
if_exists: str = "append",
create_table: bool = False,
# create_according_to_data: bool = False,
varchar_len: int = 50,
parse_dates: list[str] = None,
):
"""
Write to SqlDB as output target for a flow.
example::
db_url = "sqlite:///stockmarket.db"
schema = {'time': datetime.datetime, 'ticker': str,
'bid': float, 'ask': float, 'ind': int}
target = SqlDBTarget(table_name=f'{name}-target', db_url=db_url, create_table=True,
schema=schema, primary_key_column=key)
:param name:
:param path:
:param attributes:
:param after_step:
:param partitioned:
:param key_bucketing_number:
:param partition_cols:
:param time_partitioning_granularity:
:param max_events:
:param flush_after_seconds:
:param storage_options:
:param db_url: url string connection to sql database.
If not set, the MLRUN_SQL__URL environment variable will
be used.
:param table_name: the name of the table to access,
from the current database
:param schema: the schema of the table (must pass when
create_table=True)
:param primary_key_column: the primary key of the table (must pass always)
:param if_exists: {'fail', 'replace', 'append'}, default 'append'
- fail: If table exists, do nothing.
- replace: If table exists, drop it, recreate it, and insert data.
- append: If table exists, insert data. Create if does not exist.
:param create_table: pass True if you want to create new table named by
table_name with schema on current database.
:param create_according_to_data: (not valid)
:param varchar_len : the defalut len of the all the varchar column (using if needed to create the table).
:param parse_dates : all the field to be parsed as timestamp.
"""
create_according_to_data = False # TODO: open for user
db_url = db_url or mlrun.mlconf.sql.url
if db_url is None or table_name is None:
attr = {}
else:
# check for table existence and acts according to the user input
self._primary_key_column = primary_key_column
attr = {
"table_name": table_name,
"db_path": db_url,
"create_according_to_data": create_according_to_data,
"if_exists": if_exists,
"parse_dates": parse_dates,
"varchar_len": varchar_len,
}
path = (
f"mlrunSql://@{db_url}//@{table_name}"
f"//@{str(create_according_to_data)}//@{if_exists}//@{primary_key_column}//@{create_table}"
)
if attributes:
attributes.update(attr)
else:
attributes = attr
super().__init__(
name,
path,
attributes,
after_step,
list(schema.keys()) if schema else None,
partitioned,
key_bucketing_number,
partition_cols,
time_partitioning_granularity,
max_events=max_events,
flush_after_seconds=flush_after_seconds,
storage_options=storage_options,
schema=schema,
)
def get_table_object(self):
from storey import SQLDriver, Table
(db_path, table_name, _, _, primary_key, _) = self._parse_url()
try:
primary_key = ast.literal_eval(primary_key)
except Exception:
pass
return Table(
f"{db_path}/{table_name}",
SQLDriver(db_path=db_path, primary_key=primary_key),
flush_interval_secs=mlrun.mlconf.feature_store.flush_interval,
)
def add_writer_step(
self,
graph,
after,
features,
key_columns=None,
timestamp_key=None,
featureset_status=None,
):
key_columns = list(key_columns.keys())
column_list = self._get_column_list(
features=features, timestamp_key=timestamp_key, key_columns=key_columns
)
table = self._resource.uri
self._create_sql_table()
for step in graph.steps.values():
if step.class_name == "storey.AggregateByKey":
raise mlrun.errors.MLRunRuntimeError(
"SQLTarget does not support aggregation step"
)
graph.add_step(
name=self.name or "SqlTarget",
after=after,
graph_shape="cylinder",
class_name="mlrun.datastore.storeytargets.NoSqlStoreyTarget",
columns=column_list,
header=True,
table=table,
index_cols=key_columns,
**self.attributes,
)
def as_df(
self,
columns=None,
df_module=None,
entities=None,
start_time=None,
end_time=None,
time_column=None,
additional_filters=None,
**kwargs,
):
try:
import sqlalchemy
except (ModuleNotFoundError, ImportError) as exc:
self._raise_sqlalchemy_import_error(exc)
mlrun.utils.helpers.additional_filters_warning(
additional_filters, self.__class__
)
db_path, table_name, _, _, _, _ = self._parse_url()
engine = sqlalchemy.create_engine(db_path)
parse_dates: Optional[list[str]] = self.attributes.get("parse_dates")
with engine.connect() as conn:
query, parse_dates = _generate_sql_query_with_time_filter(
table_name=table_name,
engine=engine,
time_column=time_column,
parse_dates=parse_dates,
start_time=start_time,
end_time=end_time,
)
df = pd.read_sql(
query,
con=conn,
parse_dates=parse_dates,
columns=columns,
)
if self._primary_key_column:
df.set_index(self._primary_key_column, inplace=True)
return df
def write_dataframe(
self, df, key_column=None, timestamp_key=None, chunk_id=0, **kwargs
):
try:
import sqlalchemy
except (ModuleNotFoundError, ImportError) as exc:
self._raise_sqlalchemy_import_error(exc)
self._create_sql_table()
if hasattr(df, "rdd"):
raise ValueError("Spark is not supported")
else:
(
db_path,
table_name,
create_according_to_data,
if_exists,
primary_key,
_,
) = self._parse_url()
create_according_to_data = bool(create_according_to_data)
engine = sqlalchemy.create_engine(
db_path,
)
connection = engine.connect()
if create_according_to_data:
# todo : create according to first row.
pass
df.to_sql(table_name, connection, if_exists=if_exists)
def _parse_url(self):
path = self.path[len("mlrunSql:///") :]
return path.split("//@")
def purge(self):
pass
def _create_sql_table(self):
(
db_path,
table_name,
create_according_to_data,
if_exists,
primary_key,
create_table,
) = self._parse_url()
try:
import sqlalchemy
except (ModuleNotFoundError, ImportError) as exc:
self._raise_sqlalchemy_import_error(exc)
try:
primary_key = ast.literal_eval(primary_key)
primary_key_for_check = primary_key
except Exception:
primary_key_for_check = [primary_key]
engine = sqlalchemy.create_engine(db_path)
with engine.connect() as conn:
metadata = sqlalchemy.MetaData()
table_exists = engine.dialect.has_table(conn, table_name)
if not table_exists and not create_table:
raise ValueError(f"Table named {table_name} is not exist")
elif not table_exists and create_table:
type_to_sql_type = {
int: sqlalchemy.Integer,
str: sqlalchemy.String(self.attributes.get("varchar_len")),
datetime.datetime: sqlalchemy.dialects.mysql.DATETIME(fsp=6),
pd.Timestamp: sqlalchemy.dialects.mysql.DATETIME(fsp=6),
bool: sqlalchemy.Boolean,
float: sqlalchemy.Float,
datetime.timedelta: sqlalchemy.Interval,
pd.Timedelta: sqlalchemy.Interval,
}
# creat new table with the given name
columns = []
for col, col_type in self.schema.items():
col_type_sql = type_to_sql_type.get(col_type)
if col_type_sql is None:
raise TypeError(
f"'{col_type}' unsupported type for column '{col}'"
)
columns.append(
sqlalchemy.Column(
col,
col_type_sql,
primary_key=(col in primary_key_for_check),
)
)
sqlalchemy.Table(table_name, metadata, *columns)
metadata.create_all(engine)
if_exists = "append"
self.path = (
f"mlrunSql://@{db_path}//@{table_name}"
f"//@{str(create_according_to_data)}//@{if_exists}//@{primary_key}//@{create_table}"
)
conn.close()
@staticmethod
def _raise_sqlalchemy_import_error(exc):
raise mlrun.errors.MLRunMissingDependencyError(
"Using 'SQLTarget' requires sqlalchemy package. Use pip install mlrun[sqlalchemy] to install it."
) from exc
kind_to_driver = {
TargetTypes.parquet: ParquetTarget,
TargetTypes.csv: CSVTarget,
TargetTypes.nosql: NoSqlTarget,
TargetTypes.redisnosql: RedisNoSqlTarget,
TargetTypes.dataframe: DFTarget,
TargetTypes.stream: StreamTarget,
TargetTypes.kafka: KafkaTarget,
TargetTypes.tsdb: TSDBTarget,
TargetTypes.custom: CustomTarget,
TargetTypes.sql: SQLTarget,
TargetTypes.snowflake: SnowflakeTarget,
}
def _get_target_path(driver, resource, run_id_mode=False, netloc=None, scheme=""):
"""return the default target path given the resource and target kind"""
kind = driver.kind
suffix = driver.suffix
if not suffix:
if (
kind == ParquetTarget.kind
and resource.kind == mlrun.common.schemas.ObjectKind.feature_vector
):
suffix = ".parquet"
kind_prefix = (
"sets"
if resource.kind == mlrun.common.schemas.ObjectKind.feature_set
else "vectors"
)
name = resource.metadata.name
project = resource.metadata.project or mlrun.mlconf.default_project
default_kind_name = kind
if scheme == "ds":
# "dsnosql" is not an actual target like Parquet or Redis; rather, it serves
# as a placeholder that can be used in any specified target
default_kind_name = "dsnosql"
if scheme == "redis" or scheme == "rediss":
default_kind_name = TargetTypes.redisnosql
netloc = netloc or ""
data_prefix = get_default_prefix_for_target(default_kind_name).format(
ds_profile_name=netloc, # In case of ds profile, set its the name
authority=netloc, # In case of redis, replace {authority} with netloc
project=project,
kind=kind,
name=name,
)
if scheme == "rediss":
data_prefix = data_prefix.replace("redis://", "rediss://", 1)
# todo: handle ver tag changes, may need to copy files?
if not run_id_mode:
version = resource.metadata.tag
name = f"{name}-{version or 'latest'}"
return f"{data_prefix}/{kind_prefix}/{name}{suffix}"
def generate_path_with_chunk(target, chunk_id, path):
if path is None:
return ""
prefix, suffix = os.path.splitext(path)
if chunk_id and not target.partitioned and not target.time_partitioning_granularity:
return f"{prefix}/{chunk_id:0>4}{suffix}"
return path
