chore: remove repos from integrations.settings column for github and github-nango integrations (CM-1005)

backend/src/database/migrations/V1772043927__removeReposFromGithubSettings.sql

@@ -0,0 +1,30 @@
+-- Backup settings before any modifications
+CREATE TABLE integration.integrations_settings_backup_02_28_2026 AS
+SELECT id, settings FROM integrations;
+
+-- Strip repos from orgs in settings for github and github-nango integrations
+-- Repos now live in public.repositories table and are populated into API responses
+-- via the compatibility layer in integrationRepository._populateRelations
+UPDATE integrations
+SET settings = jsonb_set(
+  settings,
+  '{orgs}',
+  (
+    SELECT coalesce(jsonb_agg(
+      org - 'repos'
+    ), '[]'::jsonb)
+    FROM jsonb_array_elements(settings->'orgs') org
+  )
+)
+WHERE platform IN ('github', 'github-nango')
+  AND settings->'orgs' IS NOT NULL
+  AND "deletedAt" IS NULL
+  AND status != 'mapping';
+
+-- Also clean up top-level repos/unavailableRepos if present
+UPDATE integrations
+SET settings = settings - 'repos' - 'unavailableRepos'
+WHERE platform IN ('github', 'github-nango')
+  AND (settings ? 'repos' OR settings ? 'unavailableRepos')
+  AND "deletedAt" IS NULL
+  AND status != 'mapping';

backend/src/database/repositories/integrationRepository.ts

@@ -9,7 +9,10 @@ import {
   fetchGlobalIntegrationsStatusCount,
   fetchGlobalNotConnectedIntegrations,
   fetchGlobalNotConnectedIntegrationsCount,
+  getNangoMappingsForIntegration,
 } from '@crowd/data-access-layer/src/integrations'
+import { SequelizeQueryExecutor } from '@crowd/data-access-layer/src/queryExecutor'
+import { getReposGroupedByOrg } from '@crowd/data-access-layer/src/repositories'
 import { IntegrationRunState, PlatformType } from '@crowd/types'
 
 import SequelizeFilterUtils from '../utils/sequelizeFilterUtils'
@@ -585,24 +588,45 @@ class IntegrationRepository {
 
     const output = record.get({ plain: true })
 
-    // For github-nango integrations, populate settings.nangoMapping from the dedicated table
-    // so the API contract remains unchanged for frontend consumers
+    const qx = new SequelizeQueryExecutor(record.sequelize)
+
+    // For github-nango integrations, populate settings.nangoMapping from dedicated table
     if (output.platform === PlatformType.GITHUB_NANGO) {
-      const rows = await record.sequelize.query(
-        `SELECT "connectionId", owner, "repoName" FROM integration.nango_mapping WHERE "integrationId" = :integrationId`,
-        {
-          replacements: { integrationId: output.id },
-          type: QueryTypes.SELECT,
-        },
-      )
+      const nangoMapping = await getNangoMappingsForIntegration(qx, output.id)
+      if (Object.keys(nangoMapping).length > 0) {
+        output.settings = { ...output.settings, nangoMapping }
+      }
+    }
 
-      if (rows.length > 0) {
-        const nangoMapping: Record<string, { owner: string; repoName: string }> = {}
-        for (const row of rows as { connectionId: string; owner: string; repoName: string }[]) {
-          nangoMapping[row.connectionId] = { owner: row.owner, repoName: row.repoName }
+    // For both github and github-nango, populate orgs[].repos from repositories table
+    if (
+      (output.platform === PlatformType.GITHUB || output.platform === PlatformType.GITHUB_NANGO) &&
+      output.settings?.orgs?.length > 0
+    ) {
+      const reposByOrg = await getReposGroupedByOrg(qx, output.id)
+
+      // Only overwrite orgs[].repos from the repositories table if there are rows.
+      // During the 'mapping' phase (legacy github connect), repos live in settings
+      // before being written to the repositories table via mapGithubRepos.
+      if (Object.keys(reposByOrg).length > 0) {
+        output.settings = {
+          ...output.settings,
+          orgs: output.settings.orgs.map((org) => ({
+            ...org,
+            repos: (reposByOrg[org.name] || []).map((r) => ({
+              url: r.url,
+              name: r.name,
+              owner: r.owner,
+              forkedFrom: r.forkedFrom,
+              updatedAt: r.updatedAt,
+            })),
+          })),
         }
-        output.settings = { ...output.settings, nangoMapping }
       }
+
+      // Strip legacy top-level keys that may still exist in the DB column
+      delete output.settings.repos
+      delete output.settings.unavailableRepos
     }
 
     return output

backend/src/services/collectionService.ts

@@ -27,6 +27,7 @@ import {
 } from '@crowd/data-access-layer/src/collections'
 import { fetchIntegrationsForSegment } from '@crowd/data-access-layer/src/integrations'
 import { QueryFilter } from '@crowd/data-access-layer/src/query'
+import { getReposForGithubIntegration } from '@crowd/data-access-layer/src/repositories'
 import {
   ICreateRepositoryGroup,
   IRepositoryGroup,
@@ -533,13 +534,8 @@ export class CollectionService extends LoggerBase {
     return listRepositoryGroups(qx, { insightsProjectId })
   }
 
-  static isSingleRepoOrg(orgs: GithubIntegrationSettings['orgs']): boolean {
-    return (
-      Array.isArray(orgs) &&
-      orgs.length === 1 &&
-      Array.isArray(orgs[0]?.repos) &&
-      orgs[0].repos.length === 1
-    )
+  static isSingleRepoOrg(orgs: GithubIntegrationSettings['orgs'], repoCount: number): boolean {
+    return Array.isArray(orgs) && orgs.length === 1 && repoCount === 1
   }
 
   /**
@@ -613,13 +609,12 @@ export class CollectionService extends LoggerBase {
       const settings = githubIntegration.settings as GithubIntegrationSettings
 
       // The orgs must have at least one repo
-      if (
-        !settings?.orgs ||
-        !Array.isArray(settings.orgs) ||
-        settings.orgs.length === 0 ||
-        !Array.isArray(settings.orgs[0].repos) ||
-        settings.orgs[0].repos.length === 0
-      ) {
+      if (!settings?.orgs || !Array.isArray(settings.orgs) || settings.orgs.length === 0) {
+        return null
+      }
+
+      const repos = await getReposForGithubIntegration(qx, githubIntegration.id)
+      if (repos.length === 0) {
         return null
       }
 
@@ -633,11 +628,8 @@ export class CollectionService extends LoggerBase {
         return null
       }
 
-      const details = CollectionService.isSingleRepoOrg(settings.orgs)
-        ? await GithubIntegrationService.findRepoDetails(
-            mainOrg.name,
-            settings.orgs[0].repos[0].name,
-          )
+      const details = CollectionService.isSingleRepoOrg(settings.orgs, repos.length)
+        ? await GithubIntegrationService.findRepoDetails(mainOrg.name, repos[0].name)
         : {
             ...(await GithubIntegrationService.findOrgDetails(mainOrg.name)),
             topics: mainOrg.topics,