feat(nimbus): Use metric-config-parser for feature monitoring configs (#9229)

* feat(nimbus): Use metric-config-parser for feature monitoring configs

- Replace yaml + tomllib with NimbusFeatureMonitoringSpec from metric-config-parser
- Change is_events_stream bool to type string (event_stream / metrics / clients_daily)
- Add --config-path CLI option to point generator at metric-hub checkout
- Configs now live in metric-hub/nimbus_feature_monitoring/ as TOML files

Relates to EXP-6732

* feat(nimbus): Nimbus feature monitoring

* feat(nimbus): Nimbus feature monitoring

* feat(nimbus): Nimbus feature monitoring

* feat(nimbus): bump mozilla-metric-config-parser to 2026.4.3

* fix(nimbus): update query template to use source_table.type instead of is_events_stream

* feat(nimbus): remove --config-repo option, auto-clone metric-hub via from_github_repo

* revert(nimbus): remove telemetry-alerts from bqetl_marketing_analysis email list

* feat(nimbus): use ConfigCollection.from_github_repo().featmon_configs instead of FeatmonSpec

* fix(nimbus): use feat.nimbus_slug() for feature name to match actual Nimbus experiment slug

* fix(nimbus): rename unused feat_name to _feat_name

* fix(nimbus): fix filter_labels -> label_in and ratios -> feature.ratios in query template

* fix(nimbus): update featmon to use FeatmonConfig.spec.* accessors and fix stale urlbar view

- Update __init__.py to iterate list[FeatmonConfig] and access dataset/data_sources/features via .spec
- Restore urlbar_events_daily view to point at _v2 (was inadvertently left at _v1 due to branch staleness)

* fix(nimbus): revert unrelated formatting changes in __init__.py

* feat(nimbus): update mozilla-metric-config-parser to 2026.4.3 in requirements.txt

* fix(nimbus): rename event_stream to events_stream, add ratios from config, add tests

* fix(nimbus): update mozilla-metric-config-parser to 2026.4.4

* refactor(nimbus): follow *Spec/*Configuration pattern for SourceTable

Because - SourceTable re-declared fields (table_name, type, analysis_unit_id)
that already exist on SourceTableSpec, causing duplication.

This commit - renames SourceTable to SourceTableConfiguration, adds a
spec: SourceTableSpec field, and delegates name/table_name/type/
analysis_unit_id/time_partition_field via properties. Rendering-only
context (project, dataset, dimensions) stays as direct fields.

* fix(nimbus): move tests to tests/sql_generators and rename spec field

Because - tests in sql_generators/ don't get picked up by the test runner;
spec field name was too generic.

This commit - moves test_generate_queries.py to tests/sql_generators/
nimbus_feature_monitoring/ and renames SourceTableConfiguration.spec
to source_table_spec per reviewer suggestion.

* fix(nimbus): remove unused pytest import and apply black formatting to tests

---------

Co-authored-by: Yashika Khurana <yashikakhurana@Yashikas-MBP.home.local>
Co-authored-by: Yashika Khurana <yashikakhurana@Yashikas-MacBook-Pro.local>

Author: 3 people

requirements.in

@@ -29,7 +29,7 @@ mdx_truly_sane_lists==1.3
 mkdocs==1.6.1
 mkdocs-material==9.7.1
 mkdocs-awesome-pages-plugin==2.10.1
-mozilla-metric-config-parser==2026.1.1
+mozilla-metric-config-parser==2026.4.4
 mozilla-schema-generator==0.5.1
 mypy==1.15.0
 openai==2.6.1

requirements.txt

@@ -1257,8 +1257,8 @@ more-itertools==10.4.0 \
     --hash=sha256:0f7d9f83a0a8dcfa8a2694a770590d98a67ea943e3d9f5298309a484758c4e27 \
     --hash=sha256:fe0e63c4ab068eac62410ab05cccca2dc71ec44ba8ef29916a0090df061cf923
     # via jaraco-classes
-mozilla-metric-config-parser==2026.1.1 \
-    --hash=sha256:0e74d8a453a839cd5eb82799655ddc1d2c79c461ce099375749c23b6ffdb4903
+mozilla-metric-config-parser==2026.4.4 \
+    --hash=sha256:76e64aa99636cb325e8105bb5364cf5b4dbc793ef78a810c0c8a9c94e9667f31
     # via -r requirements.in
 mozilla-nimbus-schemas==2024.9.3 \
     --hash=sha256:5e1196637308fb0241a209eee027a436e986fc6220ad416eebb06b7162e226bc \

sql_generators/nimbus_feature_monitoring/__init__.py

@@ -1,12 +1,12 @@
 """Experiment monitoring materialized view generation."""
 
-import os
 from dataclasses import dataclass
 from pathlib import Path
 from typing import Any
 
 import click
-import yaml
+from metric_config_parser.config import ConfigCollection
+from metric_config_parser.featmon import SourceTableSpec
 from jinja2 import Environment, FileSystemLoader
 
 from bigquery_etl.cli.utils import use_cloud_function_option
@@ -39,22 +39,40 @@ def __post_init__(self):
 
 
 @dataclass
-class SourceTable:
-    name: str
+class SourceTableConfiguration:
+    """Wraps a SourceTableSpec with rendering context needed to generate SQL."""
+
+    source_table_spec: SourceTableSpec
     project: str
     dataset: str
-    table_name: str
     dimensions: list[Dimension]
-    analysis_unit_id: str = "client_info.client_id"
-    time_partition_field: str = "submission_timestamp"
-    is_events_stream: bool = False
 
     def __post_init__(self):
         self.last_dimensions = [d for d in self.dimensions if d.aggregator == "last"]
         self.non_last_dimensions = [
             d for d in self.dimensions if d.aggregator != "last"
         ]
 
+    @property
+    def name(self) -> str:
+        return self.source_table_spec.name
+
+    @property
+    def table_name(self) -> str:
+        return self.source_table_spec.table_name
+
+    @property
+    def analysis_unit_id(self) -> str:
+        return self.source_table_spec.analysis_unit_id
+
+    @property
+    def type(self) -> str:
+        return self.source_table_spec.type
+
+    @property
+    def time_partition_field(self) -> str:
+        return "submission_timestamp"
+
 
 @dataclass
 class Metric:
@@ -68,7 +86,7 @@ def __post_init__(self):
         self.is_labeled = isinstance(self, LabeledMetric)
 
     @classmethod
-    def from_(_, source_is_events_stream=False, **kwargs) -> "Metric":
+    def from_(_, source_type="metrics", **kwargs) -> "Metric":
         data_type = kwargs["data_type"]
 
         if (_key := "field") not in kwargs:
@@ -86,7 +104,7 @@ def from_(_, source_is_events_stream=False, **kwargs) -> "Metric":
         if (_key := "ping_aggregator") not in kwargs:
             if data_type in ("boolean", "labeled_boolean"):
                 kwargs[_key] = "logical_or"
-            elif data_type == "event" and source_is_events_stream:
+            elif data_type == "event" and source_type == "events_stream":
                 kwargs[_key] = "countif"
             else:
                 kwargs[_key] = "sum"
@@ -118,7 +136,7 @@ class Feature:
     project: str
     dataset: str
     metrics_by_source: dict[str, list[Metric]]
-    ratios: list[list[str, str]]
+    ratios: list[list[str]]
 
     def all_metrics(self):
         return [
@@ -139,37 +157,33 @@ def generate_queries(project, path, write_dir):
     metadata_template = env.get_template("metadata.yaml")
     view_template = env.get_template("view.sql")
     schema = (template_dir / "schema.yaml").read_text()
-    for app_config_path in path.glob("*.yaml"):
-        app_config = yaml.safe_load(app_config_path.read_text())
-        dataset = app_config["dataset"]
+    for app_config in ConfigCollection.from_github_repo().featmon_configs:
+        dataset = app_config.spec.dataset
         source_tables = {}
-        for source_name, source in app_config["source_tables"].items():
-            if source is None:
-                source = {}
+        for source_name, source in app_config.spec.data_sources.items():
             dimensions = []
-            for dim_name, dim in source.pop("dimensions", {}).items():
+            for dim_name, dim in source.dimensions.items():
                 if dim is None:
                     dim = {}
+                else:
+                    dim = dict(dim)
                 dimensions.append(
                     Dimension(
                         name=dim_name,
                         field=dim.pop("field", dim_name),
                         **dim,
                     )
                 )
-            source_tables[source_name] = SourceTable(
-                name=source_name,
-                project=source.pop("project", project),
-                dataset=source.pop("dataset", dataset),
+            source_tables[source_name] = SourceTableConfiguration(
+                source_table_spec=source,
+                project=project,
+                dataset=dataset,
                 dimensions=dimensions,
-                **source,
             )
         features = []
-        for feat_name, feat in app_config["features"].items():
+        for _feat_name, feat in app_config.spec.features.items():
             metrics_by_source = {}
-            for source_name, source_metrics in feat.pop(
-                "metrics_by_source", {}
-            ).items():
+            for source_name, source_metrics in feat.metrics_by_source.items():
                 metrics_by_source[source_name] = metrics = []
                 for data_type, data_type_metrics in source_metrics.items():
                     if data_type == "event":
@@ -185,9 +199,9 @@ def generate_queries(project, path, write_dir):
                                         event_name=metric.pop(
                                             "event_name", metric_name
                                         ),
-                                        source_is_events_stream=source_tables[
+                                        source_type=source_tables[
                                             source_name
-                                        ].is_events_stream,
+                                        ].type,
                                         **metric,
                                     )
                                 )
@@ -202,12 +216,11 @@ def generate_queries(project, path, write_dir):
                             )
             features.append(
                 Feature(
-                    name=feat_name,
-                    project=feat.pop("project", project),
-                    dataset=feat.pop("dataset", f"{dataset}_derived"),
-                    ratios=feat.pop("ratios", []),
+                    name=feat.nimbus_slug(),
+                    project=project,
+                    dataset=f"{dataset}_derived",
+                    ratios=feat.ratios,
                     metrics_by_source=metrics_by_source,
-                    **feat,
                 )
             )
 

sql_generators/nimbus_feature_monitoring/templates/firefox_desktop.yaml

@@ -58,7 +58,7 @@ rollouts_cte AS (
             (
               SELECT {{ metric.label_aggregator }}(value)
               FROM UNNEST({{metric.field}})
-              {% if metric.filter_labels %}
+              {% if metric.label_in %}
                 WHERE key IN (
                   {% for label in metric.label_in %}
                     "{{ label }}"{{ "," if not loop.last }}
@@ -67,7 +67,7 @@ rollouts_cte AS (
               {% endif %}
             )
           {% elif metric.data_type == "event" %}
-            {% if source_table.is_events_stream %}
+            {% if source_table.type == "events_stream" %}
               event_category = "{{ metric.category }}"
               AND event_name = "{{ metric.event_name }}"
             {% else %}
@@ -89,7 +89,7 @@ rollouts_cte AS (
       , UNNEST(
         ARRAY_CONCAT(
           COALESCE(
-            {% if source_table.is_events_stream %}
+            {% if source_table.type == "events_stream" %}
               UNNEST(JSON_KEYS(experiments, 1))
             {% else %}
               ARRAY(
@@ -198,7 +198,7 @@ FROM
 UNPIVOT (
   value FOR metric IN
   (
-    {% for numerator, denominator in ratios %}
+    {% for numerator, denominator in feature.ratios %}
       ratio_{{numerator}}_to_{{denominator}},
     {% endfor %}
     {% for metric, aggregator in feature.all_metrics() %}