use default branch instead of head for git show operation

Author: brendan-kellam

packages/backend/src/github.ts

@@ -28,6 +28,7 @@ export type OctokitRepository = {
     stargazers_count?: number,
     watchers_count?: number,
     subscribers_count?: number,
+    default_branch?: string,
     forks_count?: number,
     archived?: boolean,
     topics?: string[],

packages/backend/src/repoCompileUtils.test.ts

@@ -6,6 +6,7 @@ vi.mock('./git.js', () => ({
     isPathAValidGitRepoRoot: vi.fn(),
     getOriginUrl: vi.fn(),
     isUrlAValidGitRepo: vi.fn(),
+    getLocalDefaultBranch: vi.fn(),
 }));
 
 // Mock the glob module

packages/backend/src/repoCompileUtils.ts

@@ -15,7 +15,7 @@ import { BitbucketConnectionConfig, GerritConnectionConfig, GiteaConnectionConfi
 import { ProjectVisibility } from "azure-devops-node-api/interfaces/CoreInterfaces.js";
 import path from 'path';
 import { glob } from 'glob';
-import { getOriginUrl, isPathAValidGitRepoRoot, isUrlAValidGitRepo } from './git.js';
+import { getLocalDefaultBranch, getOriginUrl, isPathAValidGitRepoRoot, isUrlAValidGitRepo } from './git.js';
 import assert from 'assert';
 import GitUrlParse from 'git-url-parse';
 import { RepoMetadata } from '@sourcebot/shared';
@@ -118,6 +118,7 @@ export const createGitHubRepoRecord = ({
         cloneUrl: cloneUrl.toString(),
         webUrl: repo.html_url,
         name: repoName,
+        defaultBranch: repo.default_branch,
         displayName: repoDisplayName,
         imageUrl: repo.owner.avatar_url,
         isFork: repo.fork,
@@ -185,6 +186,7 @@ export const compileGitlabConfig = async (
             cloneUrl: cloneUrl.toString(),
             webUrl: projectUrl,
             name: repoName,
+            defaultBranch: project.default_branch,
             displayName: repoDisplayName,
             imageUrl: avatarUrl,
             isFork: isFork,
@@ -257,6 +259,7 @@ export const compileGiteaConfig = async (
             webUrl: repo.html_url,
             name: repoName,
             displayName: repoDisplayName,
+            defaultBranch: repo.default_branch,
             imageUrl: repo.owner?.avatar_url,
             isFork: repo.fork!,
             isPublic: isPublic,
@@ -339,6 +342,10 @@ export const compileGerritConfig = async (
             webUrl: webUrl,
             name: repoName,
             displayName: repoDisplayName,
+            // @note: the gerrit api doesn't return the default branch (without a seperate query).
+            // Instead, the default branch will be set once the repo is cloned.
+            // @see: repoIndexManager.ts
+            defaultBranch: undefined,
             isFork: false,
             isArchived: false,
             org: {
@@ -444,6 +451,7 @@ export const compileBitbucketConfig = async (
         const repoName = path.join(repoNameRoot, displayName);
         const cloneUrl = getCloneUrl(repo);
         const webUrl = getWebUrl(repo);
+        const defaultBranch = isServer ? (repo as BitbucketServerRepository).defaultBranch : (repo as BitbucketCloudRepository).mainbranch?.name;
 
         const record: RepoData = {
             external_id: externalId,
@@ -453,6 +461,7 @@ export const compileBitbucketConfig = async (
             webUrl: webUrl,
             name: repoName,
             displayName: displayName,
+            defaultBranch,
             isFork: isFork,
             isPublic: isPublic,
             isArchived: isArchived,
@@ -557,6 +566,8 @@ export const compileGenericGitHostConfig_file = async (
 
         const remoteUrl = GitUrlParse(origin);
 
+        const defaultBranch = await getLocalDefaultBranch({ path: repoPath });
+
         // @note: matches the naming here:
         // https://github.com/sourcebot-dev/zoekt/blob/main/gitindex/index.go#L293
         // Go's url.URL.Host includes the port if present (even default ports like 443),
@@ -573,6 +584,7 @@ export const compileGenericGitHostConfig_file = async (
             cloneUrl: `file://${repoPath}`,
             name: repoName,
             displayName: repoName,
+            defaultBranch,
             isFork: false,
             isArchived: false,
             org: {
@@ -612,7 +624,6 @@ export const compileGenericGitHostConfig_file = async (
     }
 }
 
-
 export const compileGenericGitHostConfig_url = async (
     config: GenericGitHostConnectionConfig,
     connectionId: number,
@@ -645,6 +656,10 @@ export const compileGenericGitHostConfig_url = async (
         cloneUrl: remoteUrl.toString(),
         name: repoName,
         displayName: repoName,
+        // @note: we can't determine the default branch from the remote url.
+        // Instead, the default branch will be set once the repo is cloned.
+        // @see: repoIndexManager.ts
+        defaultBranch: undefined,
         isFork: false,
         isArchived: false,
         org: {
@@ -719,6 +734,7 @@ export const compileAzureDevOpsConfig = async (
             webUrl: webUrl,
             name: repoName,
             displayName: repoDisplayName,
+            defaultBranch: repo.defaultBranch,
             imageUrl: null,
             isFork: !!repo.isFork,
             isArchived: false,

packages/backend/src/repoIndexManager.ts

@@ -498,6 +498,7 @@ export class RepoIndexManager {
                 });
 
                 const pushedAt = await getLatestCommitTimestamp({ path: repoPath });
+                const defaultBranch = await getLocalDefaultBranch({ path: repoPath });
 
                 const jobMetadata = repoIndexingJobMetadataSchema.parse(jobData.metadata);
 
@@ -511,6 +512,13 @@ export class RepoIndexManager {
                             ...(jobData.repo.metadata as RepoMetadata),
                             indexedRevisions: jobMetadata.indexedRevisions,
                         } satisfies RepoMetadata,
+                        // @note: always update the default branch. While this field can be set
+                        // during connection syncing, by setting it here we ensure that a) the
+                        // default branch is as up to date as possible (since repo indexing happens
+                        // more frequently than connection syncing) and b) for hosts where it is
+                        // impossible to determine the default branch from the host's API
+                        // (e.g., generic git url), we still set the default branch here.
+                        defaultBranch: defaultBranch,
                     }
                 });
 

packages/db/prisma/migrations/20260131013553_add_default_branch_to_repo_table/migration.sql

@@ -0,0 +1,2 @@
+-- AlterTable
+ALTER TABLE "Repo" ADD COLUMN     "defaultBranch" TEXT;

packages/db/prisma/schema.prisma

@@ -59,6 +59,7 @@ model Repo {
   webUrl                String?
   connections           RepoToConnection[]
   imageUrl              String?
+  defaultBranch         String?
 
   permittedAccounts  AccountToRepoPermission[]
   permissionSyncJobs RepoPermissionSyncJob[]

packages/web/src/features/search/fileSourceApi.ts

@@ -22,7 +22,9 @@ export const getFileSource = async ({ path: filePath, repo: repoName, ref }: Fil
         const { path: repoPath } = getRepoPath(repo);
         const git = simpleGit().cwd(repoPath);
 
-        const gitRef = ref ?? 'HEAD';
+        const gitRef = ref ??
+            repo.defaultBranch ??
+            'HEAD';
 
         let source: string;
         try {