feat: extend enricher to cover extra fields [CM-1220] (#4186)

Signed-off-by: Mouad BANI <mouad-mb@outlook.com>

Author: mbani01

backend/src/osspckgs/migrations/V1780996561__repo_activity_snapshot.sql

@@ -0,0 +1,30 @@
+ALTER TABLE repos ADD COLUMN IF NOT EXISTS security_policy_enabled boolean;
+ALTER TABLE repos ADD COLUMN IF NOT EXISTS security_file_enabled boolean;
+ALTER TABLE repos ADD COLUMN IF NOT EXISTS snapshot_at timestamptz;
+
+CREATE TABLE IF NOT EXISTS repo_activity_snapshot (
+  repo_id                                   bigint PRIMARY KEY REFERENCES repos(id) ON DELETE CASCADE,
+  snapshot_at                               timestamptz NOT NULL,
+  window_months                             int NOT NULL DEFAULT 12,
+  -- commit activity
+  commits_last_12m                          int,
+  commits_last_6m                           int,
+  commits_prior_6m                          int,
+  -- PR health
+  prs_opened_last_12m                       int,
+  prs_merged_last_12m                       int,
+  prs_closed_unmerged_12m                   int,
+  pr_median_time_to_merge_hours             int,
+  pr_median_time_to_first_response_hours    int,
+  -- issue health
+  issues_opened_last_12m                    int,
+  issues_closed_last_12m                    int,
+  issues_opened_last_6m                     int,
+  issues_opened_prior_6m                    int,
+  issues_open_now                           int,
+  issue_median_time_to_close_hours          int,
+  issue_median_time_to_first_response_hours int
+);
+
+CREATE INDEX IF NOT EXISTS repo_activity_snapshot_snapshot_at_idx
+  ON repo_activity_snapshot (snapshot_at);

services/apps/packages_worker/src/enricher/computeMedians.ts

@@ -0,0 +1,105 @@
+const MS_PER_HOUR = 1000 * 60 * 60
+
+function median(values: number[]): number | null {
+  if (values.length === 0) return null
+  const sorted = [...values].sort((a, b) => a - b)
+  const middleIndex = Math.floor(sorted.length / 2)
+  return sorted.length % 2 === 0
+    ? (sorted[middleIndex - 1] + sorted[middleIndex]) / 2
+    : sorted[middleIndex]
+}
+
+function hoursBetween(startDate: string, endDate: string): number {
+  return (new Date(endDate).getTime() - new Date(startDate).getTime()) / MS_PER_HOUR
+}
+
+function toIntHours(hours: number | null): number | null {
+  return hours != null ? Math.round(hours) : null
+}
+
+// Shapes mirror exactly what GitHub GraphQL returns for comments/reviews nodes
+export interface ResponseNode {
+  createdAt: string
+  author: { login: string } | null
+}
+
+export interface PrNode {
+  createdAt: string
+  mergedAt: string | null
+  author: { login: string } | null
+  comments: { nodes: ResponseNode[] }
+  reviews: { nodes: ResponseNode[] }
+}
+
+export interface IssueNode {
+  createdAt: string
+  closedAt: string | null
+  author: { login: string } | null
+  comments: { nodes: ResponseNode[] }
+}
+
+function firstNonAuthorResponseHours(
+  itemCreatedAt: string,
+  authorLogin: string | null,
+  responses: ResponseNode[],
+): number | null {
+  const firstResponse = responses.find(
+    (response) => response.author?.login && response.author.login !== authorLogin,
+  )
+  return firstResponse ? hoursBetween(itemCreatedAt, firstResponse.createdAt) : null
+}
+
+export function computePrMedians(prs: PrNode[]): {
+  medianTimeToMergeHours: number | null
+  medianTimeToFirstResponseHours: number | null
+} {
+  const mergeHours: number[] = []
+  const firstResponseHours: number[] = []
+
+  for (const pr of prs) {
+    if (pr.mergedAt != null) {
+      mergeHours.push(hoursBetween(pr.createdAt, pr.mergedAt))
+    }
+
+    const allResponses = [...pr.comments.nodes, ...pr.reviews.nodes].sort(
+      (left, right) => new Date(left.createdAt).getTime() - new Date(right.createdAt).getTime(),
+    )
+    const responseHours = firstNonAuthorResponseHours(
+      pr.createdAt,
+      pr.author?.login ?? null,
+      allResponses,
+    )
+    if (responseHours != null) firstResponseHours.push(responseHours)
+  }
+
+  return {
+    medianTimeToMergeHours: toIntHours(median(mergeHours)),
+    medianTimeToFirstResponseHours: toIntHours(median(firstResponseHours)),
+  }
+}
+
+export function computeIssueMedians(issues: IssueNode[]): {
+  medianTimeToCloseHours: number | null
+  medianTimeToFirstResponseHours: number | null
+} {
+  const closeHours: number[] = []
+  const firstResponseHours: number[] = []
+
+  for (const issue of issues) {
+    if (issue.closedAt != null) {
+      closeHours.push(hoursBetween(issue.createdAt, issue.closedAt))
+    }
+
+    const responseHours = firstNonAuthorResponseHours(
+      issue.createdAt,
+      issue.author?.login ?? null,
+      issue.comments.nodes,
+    )
+    if (responseHours != null) firstResponseHours.push(responseHours)
+  }
+
+  return {
+    medianTimeToCloseHours: toIntHours(median(closeHours)),
+    medianTimeToFirstResponseHours: toIntHours(median(firstResponseHours)),
+  }
+}