(fix) Always upsert metadata.gitConfig into the repo s.t., it is synced

Author: brendan-kellam

packages/backend/src/git.ts

@@ -1,21 +1,15 @@
 import { simpleGit, SimpleGitProgressEvent } from 'simple-git';
 
-export const cloneRepository = async (cloneURL: string, path: string, gitConfig?: Record<string, string>, onProgress?: (event: SimpleGitProgressEvent) => void) => {
+export const cloneRepository = async (cloneURL: string, path: string, onProgress?: (event: SimpleGitProgressEvent) => void) => {
     const git = simpleGit({
         progress: onProgress,
     });
-
-    const configParams = Object.entries(gitConfig ?? {}).flatMap(
-        ([key, value]) => ['--config', `${key}=${value}`]
-    );
-
     try {
         await git.clone(
             cloneURL,
             path,
             [
                 "--bare",
-                ...configParams
             ]
         );
 
@@ -48,6 +42,26 @@ export const fetchRepository = async (path: string, onProgress?: (event: SimpleG
     }
 }
 
+/**
+ * Applies the gitConfig to the repo at the given path. Note that this will
+ * override the values for any existing keys, and append new values for keys
+ * that do not exist yet. It will _not_ remove any existing keys that are not
+ * present in gitConfig.
+ */
+export const upsertGitConfig = async (path: string, gitConfig: Record<string, string>, onProgress?: (event: SimpleGitProgressEvent) => void) => {
+    const git = simpleGit({
+        progress: onProgress,
+    }).cwd(path);
+
+    try {
+        for (const [key, value] of Object.entries(gitConfig)) {
+            await git.addConfig(key, value);
+        }
+    } catch (error) {
+        throw new Error(`Failed to set git config ${path}`);
+    }
+}
+
 export const getBranches = async (path: string) => {
     const git = simpleGit();
     const branches = await git.cwd({

packages/backend/src/repoManager.ts

@@ -5,7 +5,7 @@ import { Connection, PrismaClient, Repo, RepoToConnection, RepoIndexingStatus, S
 import { GithubConnectionConfig, GitlabConnectionConfig, GiteaConnectionConfig } from '@sourcebot/schemas/v3/connection.type';
 import { AppContext, Settings, RepoMetadata } from "./types.js";
 import { getRepoPath, getTokenFromConfig, measure, getShardPrefix } from "./utils.js";
-import { cloneRepository, fetchRepository } from "./git.js";
+import { cloneRepository, fetchRepository, upsertGitConfig } from "./git.js";
 import { existsSync, readdirSync, promises } from 'fs';
 import { indexGitRepository } from "./zoekt.js";
 import { PromClient } from './promClient.js';
@@ -240,7 +240,7 @@ export class RepoManager implements IRepoManager {
                 }
             }
 
-            const { durationMs } = await measure(() => cloneRepository(cloneUrl.toString(), repoPath, metadata.gitConfig, ({ method, stage, progress }) => {
+            const { durationMs } = await measure(() => cloneRepository(cloneUrl.toString(), repoPath, ({ method, stage, progress }) => {
                 this.logger.debug(`git.${method} ${stage} stage ${progress}% complete for ${repo.id}`)
             }));
             cloneDuration_s = durationMs / 1000;
@@ -249,6 +249,13 @@ export class RepoManager implements IRepoManager {
             this.logger.info(`Cloned ${repo.id} in ${cloneDuration_s}s`);
         }
 
+        // Regardless of clone or fetch, always upsert the git config for the repo.
+        // This ensures that the git config is always up to date for whatever we
+        // have in the DB.
+        if (metadata.gitConfig) {
+            await upsertGitConfig(repoPath, metadata.gitConfig);
+        }
+
         this.logger.info(`Indexing ${repo.id}...`);
         const { durationMs } = await measure(() => indexGitRepository(repo, this.settings, this.ctx));
         const indexDuration_s = durationMs / 1000;

packages/backend/src/zoekt.ts

@@ -57,7 +57,17 @@ export const indexGitRepository = async (repo: Repo, settings: Settings, ctx: Ap
         revisions = revisions.slice(0, 64);
     }
 
-    const command = `zoekt-git-index -allow_missing_branches -index ${ctx.indexPath} -max_trigram_count ${settings.maxTrigramCount} -file_limit ${settings.maxFileSize} -branches ${revisions.join(',')} -tenant_id ${repo.orgId} -shard_prefix ${shardPrefix} ${repoPath}`;
+    const command = [
+        'zoekt-git-index',
+        '-allow_missing_branches',
+        `-index ${ctx.indexPath}`,
+        `-max_trigram_count ${settings.maxTrigramCount}`,
+        `-file_limit ${settings.maxFileSize}`,
+        `-branches ${revisions.join(',')}`,
+        `-tenant_id ${repo.orgId}`,
+        `-shard_prefix ${shardPrefix}`,
+        repoPath
+    ].join(' ');
 
     return new Promise<{ stdout: string, stderr: string }>((resolve, reject) => {
         exec(command, (error, stdout, stderr) => {