fix(search): Fix issue with filtering on generic git repo indexed from http/https (#742)

* fix: add gitConfig for generic git host repos with URL to fix file access and search

For generic git host repos with HTTP/HTTPS URLs, the gitConfig metadata was
missing. This caused issues because:
1. After cloning, remote.origin.url is unset for security reasons
2. Zoekt uses gitConfig (specifically zoekt.name) to identify repos
3. Without gitConfig, zoekt couldn't properly identify the repo name

This fix adds proper gitConfig metadata (zoekt.name, zoekt.web-url, etc.) to
compileGenericGitHostConfig_url, similar to how other code hosts handle it.

Fixes SOU-218

Co-authored-by: michael <michael@sourcebot.dev>

* changelog

---------

Co-authored-by: Cursor Agent <cursoragent@cursor.com>

Author: msukkari

CHANGELOG.md

@@ -7,6 +7,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+- Fixed issue with filtering on generic git repo indexed from http/https [#742](https://github.com/sourcebot-dev/sourcebot/pull/742)
+
 ## [4.10.10] - 2026-01-16
 
 ### Added

packages/backend/src/repoCompileUtils.test.ts

@@ -1,23 +1,25 @@
 import { expect, test, vi, describe, beforeEach, afterEach } from 'vitest';
-import { compileGenericGitHostConfig_file } from './repoCompileUtils';
+import { compileGenericGitHostConfig_file, compileGenericGitHostConfig_url } from './repoCompileUtils';
 
 // Mock the git module
 vi.mock('./git.js', () => ({
     isPathAValidGitRepoRoot: vi.fn(),
     getOriginUrl: vi.fn(),
+    isUrlAValidGitRepo: vi.fn(),
 }));
 
 // Mock the glob module
 vi.mock('glob', () => ({
     glob: vi.fn(),
 }));
 
-import { isPathAValidGitRepoRoot, getOriginUrl } from './git.js';
+import { isPathAValidGitRepoRoot, getOriginUrl, isUrlAValidGitRepo } from './git.js';
 import { glob } from 'glob';
 
 const mockedGlob = vi.mocked(glob);
 const mockedIsPathAValidGitRepoRoot = vi.mocked(isPathAValidGitRepoRoot);
 const mockedGetOriginUrl = vi.mocked(getOriginUrl);
+const mockedIsUrlAValidGitRepo = vi.mocked(isUrlAValidGitRepo);
 
 describe('compileGenericGitHostConfig_file', () => {
     beforeEach(() => {
@@ -122,3 +124,71 @@ describe('compileGenericGitHostConfig_file', () => {
         expect(result.warnings[0]).toContain('not a git repository');
     });
 });
+
+describe('compileGenericGitHostConfig_url', () => {
+    beforeEach(() => {
+        vi.clearAllMocks();
+    });
+
+    afterEach(() => {
+        vi.resetAllMocks();
+    });
+
+    test('should return warning when url is not a valid git repo', async () => {
+        mockedIsUrlAValidGitRepo.mockResolvedValue(false);
+
+        const config = {
+            type: 'git' as const,
+            url: 'https://example.com/not-a-repo',
+        };
+
+        const result = await compileGenericGitHostConfig_url(config, 1);
+
+        expect(result.repoData).toHaveLength(0);
+        expect(result.warnings).toHaveLength(1);
+        expect(result.warnings[0]).toContain('not a git repository');
+    });
+
+    test('should successfully compile with gitConfig when valid git repo url is found', async () => {
+        mockedIsUrlAValidGitRepo.mockResolvedValue(true);
+
+        const config = {
+            type: 'git' as const,
+            url: 'https://git.kernel.org/pub/scm/bluetooth/bluez.git',
+        };
+
+        const result = await compileGenericGitHostConfig_url(config, 1);
+
+        expect(result.repoData).toHaveLength(1);
+        expect(result.warnings).toHaveLength(0);
+        expect(result.repoData[0].cloneUrl).toBe('https://git.kernel.org/pub/scm/bluetooth/bluez.git');
+        expect(result.repoData[0].name).toBe('git.kernel.org/pub/scm/bluetooth/bluez');
+        
+        // Verify gitConfig is set properly (this is the key fix for SOU-218)
+        const metadata = result.repoData[0].metadata as { gitConfig?: Record<string, string> };
+        expect(metadata.gitConfig).toBeDefined();
+        expect(metadata.gitConfig!['zoekt.name']).toBe('git.kernel.org/pub/scm/bluetooth/bluez');
+        expect(metadata.gitConfig!['zoekt.web-url']).toBe('https://git.kernel.org/pub/scm/bluetooth/bluez.git');
+        expect(metadata.gitConfig!['zoekt.display-name']).toBe('git.kernel.org/pub/scm/bluetooth/bluez');
+        expect(metadata.gitConfig!['zoekt.archived']).toBe('0');
+        expect(metadata.gitConfig!['zoekt.fork']).toBe('0');
+        expect(metadata.gitConfig!['zoekt.public']).toBe('1');
+    });
+
+    test('should handle url with trailing .git correctly', async () => {
+        mockedIsUrlAValidGitRepo.mockResolvedValue(true);
+
+        const config = {
+            type: 'git' as const,
+            url: 'https://github.com/test/repo.git',
+        };
+
+        const result = await compileGenericGitHostConfig_url(config, 1);
+
+        expect(result.repoData).toHaveLength(1);
+        expect(result.repoData[0].name).toBe('github.com/test/repo');
+        
+        const metadata = result.repoData[0].metadata as { gitConfig?: Record<string, string> };
+        expect(metadata.gitConfig!['zoekt.name']).toBe('github.com/test/repo');
+    });
+});

packages/backend/src/repoCompileUtils.ts

@@ -609,9 +609,17 @@ export const compileGenericGitHostConfig_url = async (
             }
         },
         metadata: {
+            gitConfig: {
+                'zoekt.name': repoName,
+                'zoekt.web-url': remoteUrl.toString(),
+                'zoekt.archived': marshalBool(false),
+                'zoekt.fork': marshalBool(false),
+                'zoekt.public': marshalBool(true),
+                'zoekt.display-name': repoName,
+            },
             branches: config.revisions?.branches ?? undefined,
             tags: config.revisions?.tags ?? undefined,
-        }
+        } satisfies RepoMetadata,
     };
 
     return {