Implementation:Evidentlyai Evidently UI Backport

Knowledge Sources	Evidentlyai_Evidently
Domains	UI, Compatibility, Metrics, Testing, Reporting
Last Updated	2026-02-14 12:00 GMT

Overview

Provides a compatibility layer that converts v2 metric results, data definitions, and snapshots into their v1 legacy equivalents, enabling the v2 reporting system to work with the legacy UI and workspace infrastructure.

Description

The UI backport module bridges the gap between Evidently's v2 metric/report system and the legacy v1 UI system. The v1 UI expects specific data structures for rendering dashboards and storing snapshots, so this module provides conversion functions and adapter classes.

Metric Result Adapters: Several adapter classes wrap v2 metric results in v1-compatible models:

MetricResultV2Adapter: Base adapter storing widget dictionaries
SingleValueV1: Adapts SingleValue (float/int/str)
ByLabelValueV1: Adapts ByLabelValue (per-label values)
ByLabelCountValueV1: Adapts ByLabelCountValue (per-label counts and shares)
CountValueV1: Adapts CountValue (count and share)
MeanStdValueV1: Adapts MeanStdValue (mean and standard deviation)
PresetMetricValueV1: Adapts preset metric widgets

Metric Adapters: MetricV2Adapter wraps a v2 Metric as a v1 Metric and carries its fingerprint for deduplication. MetricV2PresetAdapter handles preset metrics with a simple ID. Both have corresponding renderers that extract widget information from stored results.

Test Adapters: TestV2Adapter wraps a v2 BoundTest as a v1 Test, and TestV2Parameters provides empty v1-compatible test parameters. TestsConfig stores test configuration data as a v2 metric for serialization.

Conversion Functions:

metric_result_v2_to_v1(): Converts a v2 MetricResult to its v1 equivalent, dispatching by type
metric_v2_to_v1(): Wraps a v2 Metric in a MetricV2Adapter
data_definition_v2_to_v1(): Converts a v2 DataDefinition to a v1 DataDefinition, mapping column types
snapshot_v2_to_v1(): The main conversion function. Converts an entire v2 Snapshot to a v1 Snapshot, including all metrics, metric results, tests, test results, and data definitions. It also processes descriptors and handles test widget generation.

Dashboard Panel: SingleValueDashboardPanel extends DashboardPanelV2 to render single-value metric time series as line plots in the legacy dashboard.

Main function: Includes a main() function demonstrating the v2-to-v1 conversion workflow, creating demo snapshots and adding them to a workspace.

Usage

This module is used internally whenever v2 reports or snapshots need to be displayed in the legacy UI or stored in the legacy workspace. The primary entry point is snapshot_v2_to_v1(), which is called by Project.add_snapshot_async() when a v2 snapshot is provided.

Code Reference

Source Location

Repository: Evidentlyai_Evidently
File: src/evidently/ui/backport.py

Signature

# Metric result adapters
class MetricResultV2Adapter(MetricResultV1):
    widget: List[dict]

class SingleValueV1(MetricResultV2Adapter):
    value: Union[float, int, str]

class ByLabelValueV1(MetricResultV2Adapter):
    values: Dict[Label, Union[float, int, bool, str]]

class ByLabelCountValueV1(MetricResultV2Adapter):
    counts: Dict[Label, int]
    shares: Dict[Label, float]

class CountValueV1(MetricResultV2Adapter):
    count: int
    share: float

class MeanStdValueV1(MetricResultV2Adapter):
    mean: float
    std: float

# Conversion functions
def metric_result_v2_to_v1(metric_result: MetricResultV2, ignore_widget: bool = False) -> MetricResultV1: ...
def metric_v2_to_v1(metric: MetricV2) -> MetricV1: ...
def data_definition_v2_to_v1(dd: DataDefinition, reference_present: bool) -> DataDefinitionV1: ...
def snapshot_v2_to_v1(snapshot: SnapshotV2) -> SnapshotV1: ...

# Metric adapters
class MetricV2Adapter(MetricV1[MetricResultV2Adapter]):
    metric: Union[MetricV2, dict]
    fingerprint: Fingerprint

class MetricV2PresetAdapter(MetricV1[MetricResultV2Adapter]):
    id: str

# Test adapters
class TestV2Adapter(TestV1):
    test: BoundTest

# Dashboard
class SingleValueDashboardPanel(DashboardPanelV2):
    metric_id: str

Import

from evidently.ui.backport import snapshot_v2_to_v1
from evidently.ui.backport import metric_result_v2_to_v1
from evidently.ui.backport import metric_v2_to_v1
from evidently.ui.backport import data_definition_v2_to_v1
from evidently.ui.backport import MetricV2Adapter, MetricV2PresetAdapter
from evidently.ui.backport import SingleValueDashboardPanel

I/O Contract

snapshot_v2_to_v1

Parameter	Type	Description
snapshot	`SnapshotV2`	A v2 snapshot from the new reporting system

Output	Type	Description
return	`SnapshotV1`	A v1 snapshot compatible with the legacy UI

metric_result_v2_to_v1

Parameter	Type	Description
metric_result	`MetricResultV2`	A v2 metric result (SingleValue, ByLabelValue, etc.)
ignore_widget	`bool`	If True, skip widget generation (default: False)

Output	Type	Description
return	`MetricResultV1`	A v1 metric result adapter with widget data

data_definition_v2_to_v1

Parameter	Type	Description
dd	`DataDefinition`	V2 data definition with column type lists
reference_present	`bool`	Whether reference data is present

Output	Type	Description
return	`DataDefinitionV1`	V1 data definition with column mapping

Usage Examples

from evidently.ui.backport import snapshot_v2_to_v1

# Convert a v2 snapshot for legacy UI storage
from evidently.core.report import Report, Snapshot as SnapshotV2
from evidently.metrics import MeanValue

report = Report([MeanValue(column="col")])
snapshot_v2 = report.run(dataset, None)

# Convert to v1 format for legacy workspace
snapshot_v1 = snapshot_v2_to_v1(snapshot_v2)

# The v1 snapshot can now be added to a legacy workspace project
project.add_snapshot(snapshot_v1)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment