feat(metrics): wiki on unified metric-results

src/backend/services/analytics/src/domain/metric_definitions/builtin.rs

@@ -192,6 +192,15 @@ pub const BUILTIN_SOURCES: &[BuiltinSource] = &[
         ],
         dimensions: &[],
     },
+    BuiltinSource {
+        source: SourceSeed {
+            key: "wiki",
+            kind: SourceKind::ManagedObservation,
+            source_ref: "wiki_metric_observations",
+        },
+        measures: &["pages_created", "edits", "pages_edited", "comments"],
+        dimensions: &[],
+    },
 ];
 
 pub const BUILTIN_METRICS: &[MetricSeed] = &[
@@ -1383,6 +1392,96 @@ pub const BUILTIN_METRICS: &[MetricSeed] = &[
         }],
         dimensions: &[],
     },
+    MetricSeed {
+        metric_key: "wiki.pages_created",
+        source_key: "wiki",
+        label: "Pages created",
+        description: Some("Wiki pages authored"),
+        explanation: Some(
+            "Wiki pages the person created during the period, counted on the \
+             page's creation date.",
+        ),
+        unit: Some("pages"),
+        format: MetricFormat::Integer,
+        direction: MetricDirection::HigherIsBetter,
+        entity_type: EntityType::Person,
+        computation: SeedComputation::Sum,
+        transform: None,
+        peer_cohort_key: Some(CohortKey::OrgUnit),
+        inputs: &[InputSeed {
+            input_role: MetricInputRole::Value,
+            measure_key: "pages_created",
+        }],
+        dimensions: &[],
+    },
+    MetricSeed {
+        metric_key: "wiki.edits",
+        source_key: "wiki",
+        label: "Page edits",
+        description: Some("Wiki edit sessions"),
+        explanation: Some(
+            "Logical wiki edits the person made during the period. Consecutive \
+             saves of the same page within a short window count as one edit, so \
+             autosaves do not inflate the number.",
+        ),
+        unit: Some("edits"),
+        format: MetricFormat::Integer,
+        direction: MetricDirection::HigherIsBetter,
+        entity_type: EntityType::Person,
+        computation: SeedComputation::Sum,
+        transform: None,
+        peer_cohort_key: Some(CohortKey::OrgUnit),
+        inputs: &[InputSeed {
+            input_role: MetricInputRole::Value,
+            measure_key: "edits",
+        }],
+        dimensions: &[],
+    },
+    MetricSeed {
+        metric_key: "wiki.pages_edited",
+        source_key: "wiki",
+        label: "Pages edited",
+        description: Some("Distinct wiki pages edited"),
+        explanation: Some(
+            "Distinct wiki pages the person edited during the period, counted \
+             per day the page was touched.",
+        ),
+        unit: Some("pages"),
+        format: MetricFormat::Integer,
+        direction: MetricDirection::HigherIsBetter,
+        entity_type: EntityType::Person,
+        computation: SeedComputation::Sum,
+        transform: None,
+        peer_cohort_key: Some(CohortKey::OrgUnit),
+        inputs: &[InputSeed {
+            input_role: MetricInputRole::Value,
+            measure_key: "pages_edited",
+        }],
+        dimensions: &[],
+    },
+    MetricSeed {
+        metric_key: "wiki.comments",
+        source_key: "wiki",
+        label: "Comments received",
+        description: Some("Comments on the person's wiki pages"),
+        explanation: Some(
+            "Comments and replies other people left on wiki pages the person \
+             authored — a signal of how much their documentation is read and \
+             discussed.",
+        ),
+        unit: Some("comments"),
+        format: MetricFormat::Integer,
+        direction: MetricDirection::HigherIsBetter,
+        entity_type: EntityType::Person,
+        computation: SeedComputation::Sum,
+        transform: None,
+        peer_cohort_key: Some(CohortKey::OrgUnit),
+        inputs: &[InputSeed {
+            input_role: MetricInputRole::Value,
+            measure_key: "comments",
+        }],
+        dimensions: &[],
+    },
 ];
 
 #[cfg(test)]

src/ingestion/dbt/tests/gold/assert_wiki_observations_entity_id_shape.sql

@@ -0,0 +1,14 @@
+-- Build-integrity check (untagged → error severity under `dbt build`).
+-- Unified entity ids for persons are lowercased emails; the runtime and the
+-- cohort view join on exact string equality, so an empty, mixed-case, or
+-- non-email id (an unresolved wiki account leaking past the staging email
+-- gate) silently drops the person from every surface.
+SELECT
+    entity_id,
+    measure_key,
+    count() AS row_count
+FROM {{ ref('wiki_metric_observations') }}
+WHERE entity_id = ''
+   OR entity_id != lower(entity_id)
+   OR entity_id NOT LIKE '%@%'
+GROUP BY entity_id, measure_key

src/ingestion/dbt/tests/gold/assert_wiki_observations_nonnegative.sql

@@ -0,0 +1,10 @@
+-- Build-integrity check (untagged → error severity under `dbt build`).
+-- Every wiki measure is a count (pages, edit sessions, distinct pages,
+-- comments) — non-negative by construction. A negative value is a
+-- regression in the gold model, not a data condition.
+SELECT
+    measure_key,
+    count() AS row_count
+FROM {{ ref('wiki_metric_observations') }}
+WHERE value < 0
+GROUP BY measure_key

src/ingestion/dbt/tests/gold/assert_wiki_observations_subject_key_shape.sql

@@ -0,0 +1,11 @@
+-- Build-integrity check (untagged → error severity under `dbt build`).
+-- The wiki family has no distinct-count measure, so the model is a single
+-- value UNION branch that stamps subject_key = NULL on every row. A
+-- non-NULL subject_key means a measure branch drifted into a distinct-count
+-- shape — nothing downstream would count it, so it must never appear.
+SELECT
+    measure_key,
+    countIf(subject_key IS NOT NULL) AS subject_rows
+FROM {{ ref('wiki_metric_observations') }}
+GROUP BY measure_key
+HAVING subject_rows > 0

src/ingestion/dbt/tests/gold/assert_wiki_observations_unique_grain.sql

@@ -0,0 +1,16 @@
+-- Build-integrity check (untagged → error severity under `dbt build`).
+-- All four wiki measures are day-grain sums: exactly one row per (tenant,
+-- entity, date, measure, dimensions, subject). A duplicate means FINAL dedup
+-- regressed on a class read or the engagement join fanned out, silently
+-- inflating the sums.
+SELECT
+    tenant_id,
+    entity_id,
+    metric_date,
+    measure_key,
+    dimensions,
+    subject_key,
+    count() AS row_count
+FROM {{ ref('wiki_metric_observations') }}
+GROUP BY tenant_id, entity_id, metric_date, measure_key, dimensions, subject_key
+HAVING count() > 1

src/ingestion/gold/schema.yml

@@ -375,6 +375,76 @@ models:
         tests:
           - not_null
 
+  - name: wiki_metric_observations
+    description: >
+      Source measure observations for the unified metrics runtime, wiki
+      family. Day-grain sums over the wiki class contracts: page creation
+      from the page snapshot, edit sessions and distinct pages edited from
+      the per-author activity rollup, and comments received attributed to
+      the page author. No event or distinct-count measures, so subject_key
+      is NULL on every row and the model is a single value UNION branch.
+      The column set is the same published contract the analytics service
+      schema validator probes (`OBSERVATION_COLUMNS`); changing it requires
+      a coordinated backend change.
+    columns:
+      - name: tenant_id
+        description: "Tenant isolation field"
+        tests:
+          - not_null
+      - name: source_key
+        description: "Logical source key registered in the metric registry"
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values: ['wiki']
+      - name: entity_type
+        description: "Measured entity type"
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values: ['person']
+      - name: entity_id
+        description: >
+          Normalized entity identifier: the author's lowercased email as
+          resolved in the wiki staging models. Accounts that do not resolve
+          to an email are excluded upstream.
+        tests:
+          - not_null
+      - name: metric_date
+        description: >
+          Observation date: page creation date for pages_created, edit day
+          for edits/pages_edited, comment day for comments.
+        tests:
+          - not_null
+      - name: observed_at
+        description: "Reserved for point-in-time semantics. NULL for all wiki measures."
+      - name: measure_key
+        description: "Source measure key registered in the metric registry"
+        tests:
+          - not_null
+          - accepted_values:
+              arguments:
+                values:
+                  - pages_created
+                  - edits
+                  - pages_edited
+                  - comments
+      - name: value
+        description: >
+          Measure value. Rows are emitted only when the source provides a
+          value, so value is never NULL; the column stays Nullable(Float64)
+          in the published contract.
+        tests:
+          - not_null
+      - name: subject_key
+        description: "Unused by this family (no distinct-count measure); always NULL."
+      - name: dimensions
+        description: >
+          Array(Tuple(key, value, label)). Empty for every wiki measure — the
+          family carries no breakdown dimension.
+
   - name: metric_entity_cohorts_current
     description: >
       Current cohort membership per entity for peer comparison. One row per

src/ingestion/gold/wiki_metric_observations.sql

@@ -0,0 +1,142 @@
+{{ config(
+    materialized='table',
+    engine='MergeTree',
+    order_by=['source_key', 'measure_key', 'entity_id', 'metric_date'],
+    schema='insight',
+    alias='wiki_metric_observations',
+    tags=['gold']
+) }}
+
+-- Source measure observations for the unified metrics runtime, wiki family.
+-- Reads the wiki class contracts only (class_wiki_pages, class_wiki_activity,
+-- class_wiki_engagement); every measure is emitted through the shape macros
+-- in macros/metric_observation_measures.sql. No dimensions: wiki sources
+-- (Confluence, Outline) feed one undifferentiated family.
+--
+-- Materialized as a sorted table: the pipeline runs once per dbt build —
+-- the only time the silver inputs can have changed — and the ordering key
+-- mirrors the runtime's filter shape (source_key, measure_key, entity_id,
+-- metric_date), so single-measure queries read index-pruned ranges.
+--
+-- Grain per measure:
+--   day-grain sums (creation date, page author):   pages_created (1/page —
+--                    the page object's own author/created_at are the
+--                    canonical creation facts; a version-derived proxy
+--                    undercounts imported pages)
+--   day-grain sums (edit date, version author):    edits (logical edit
+--                    sessions — autosave bursts collapsed in silver, see
+--                    class_wiki_activity), pages_edited (distinct pages
+--                    touched that day)
+--   day-grain sums (comment date, page author):    comments (engagement
+--                    RECEIVED on the person's pages — footer + inline +
+--                    replies; the commenter is deliberately not the entity,
+--                    see class_wiki_engagement's page-centric design note)
+--
+-- Attribution: entity_id = lower(author_email); only email-shaped keys pass.
+-- Confluence resolves emails through the Jira directory join in staging and
+-- yields NULL on tenants without Jira — those rows are excluded as
+-- unmatchable rather than carried as dead entities (cohorts and API requests
+-- address people by email). Outline resolves from its own user stream.
+--
+-- Memory shape: every class read keeps FINAL (ReplacingMergeTree dedup) over
+-- a pruned column set. class_wiki_activity is already (author, day) grain,
+-- so its sums are near-free. The single join in the model attributes
+-- page-day comment rollups to the page author: engagement ⋈ pages on
+-- (tenant_id, source_id, page_id) — source_id is part of the key so a
+-- page_id colliding across two wiki instances of one tenant cannot fan out.
+--
+-- Peer measurability (who enters a metric's peer pool) is decided HERE, by
+-- row emission — the runtime never fabricates zeros. All four measures are
+-- engagement-gated: a row exists only where the source recorded authorship,
+-- editing, or received comments. Rostered-but-inactive people take no
+-- standing rather than dragging peer medians toward zero.
+
+WITH
+-- Page -> author attribution for the comment join. No created_at gate:
+-- comment attribution needs only the page's author, so a page snapshot
+-- without a creation timestamp still receives its comments.
+pages AS (
+    SELECT
+        tenant_id,
+        source_id,
+        page_id,
+        lower(author_email) AS entity_id
+    FROM {{ ref('class_wiki_pages') }} FINAL
+    WHERE author_email LIKE '%@%'
+),
+-- Dated branch for pages_created; the creation date is the metric date, so
+-- only here does a missing created_at exclude the page.
+page_creations AS (
+    SELECT
+        tenant_id,
+        lower(author_email) AS entity_id,
+        toDate(created_at) AS metric_date,
+        CAST([] AS Array(Tuple(key String, value String, label Nullable(String)))) AS no_dimensions
+    FROM {{ ref('class_wiki_pages') }} FINAL
+    WHERE author_email LIKE '%@%'
+      AND created_at IS NOT NULL
+),
+activity AS (
+    SELECT
+        tenant_id,
+        lower(author_email) AS entity_id,
+        day AS metric_date,
+        total_edits,
+        pages_edited,
+        CAST([] AS Array(Tuple(key String, value String, label Nullable(String)))) AS no_dimensions
+    FROM {{ ref('class_wiki_activity') }} FINAL
+    WHERE author_email LIKE '%@%'
+      AND day IS NOT NULL
+),
+engagement AS (
+    SELECT
+        e.tenant_id AS tenant_id,
+        p.entity_id AS entity_id,
+        e.day AS metric_date,
+        e.total_comments AS total_comments,
+        CAST([] AS Array(Tuple(key String, value String, label Nullable(String)))) AS no_dimensions
+    FROM (
+        SELECT
+            tenant_id,
+            source_id,
+            page_id,
+            day,
+            total_comments
+        FROM {{ ref('class_wiki_engagement') }} FINAL
+        WHERE day IS NOT NULL
+    ) AS e
+    INNER JOIN pages AS p
+        ON e.tenant_id = p.tenant_id
+       AND e.source_id = p.source_id
+       AND e.page_id = p.page_id
+),
+value_measures AS (
+    {{ sum_measure('pages_created', 'page_creations', '1', 'no_dimensions') }}
+
+    UNION ALL
+
+    {{ sum_measure('edits', 'activity', 'total_edits', 'no_dimensions') }}
+
+    UNION ALL
+
+    {{ sum_measure('pages_edited', 'activity', 'pages_edited', 'no_dimensions') }}
+
+    UNION ALL
+
+    {{ sum_measure('comments', 'engagement', 'total_comments', 'no_dimensions') }}
+)
+SELECT
+    assumeNotNull(tenant_id) AS tenant_id,
+    'wiki' AS source_key,
+    'person' AS entity_type,
+    assumeNotNull(entity_id) AS entity_id,
+    assumeNotNull(metric_date) AS metric_date,
+    CAST(NULL AS Nullable(DateTime64(3))) AS observed_at,
+    measure_key,
+    value,
+    CAST(NULL AS Nullable(String)) AS subject_key,
+    dimensions
+FROM value_measures
+WHERE tenant_id IS NOT NULL
+  AND entity_id IS NOT NULL
+  AND metric_date IS NOT NULL