headquarter/openspec/changes/archive/2026-05-19-smart-git-url-parsing/design.md

Smart Git URL Parsing - Design

Architecture

User pastes URL → Frontend validates → Backend validates → Clone repo
                    ↓                      ↓
            Show suggestions         Parse & suggest

URL Detection Logic

Patterns to Detect

1. GitHub/GitLab/Bitbucket browser URLs

   • https://github.com/owner/repo/tree/branch-name
   • https://github.com/owner/repo/blob/branch/path/to/file
   • https://github.com/owner/repo/pull/123
   • https://gitlab.com/owner/repo/-/tree/branch
   • https://bitbucket.org/owner/repo/src/branch/

2. URLs with query parameters

   • https://github.com/owner/repo?tab=readme-ov-file
   • https://github.com/owner/repo.git?branch=develop

3. Valid clone URLs (should pass through)

   • https://github.com/owner/repo.git
   • git@github.com:owner/repo.git
   • https://github.com/owner/repo (without .git)

URL Parsing Rules

def extract_base_repo_url(url: str) -> str | None:
    """Extract base repository URL from a browser/git URL.
    
    Examples:
        https://github.com/user/repo/tree/main → https://github.com/user/repo.git
        https://github.com/user/repo.git → https://github.com/user/repo.git
        git@github.com:user/repo.git → git@github.com:user/repo.git
        https://gitlab.com/user/repo/-/blob/main/README.md → https://gitlab.com/user/repo.git
    
    Returns None if URL doesn't match known patterns.
    """
    ...

Algorithm:

1. Remove query parameters
2. Detect host (github.com, gitlab.com, bitbucket.org, etc.)
3. For GitHub: Remove /tree/*, /blob/*, /pull/*, /issues/* paths
4. For GitLab: Remove /-/tree/*, /-/blob/* paths
5. For Bitbucket: Remove /src/* paths
6. Ensure .git suffix
7. Return cleaned URL or None

API Changes

POST /repositories (enhanced)

Request Body:

{
  "project_id": "uuid",
  "name": "my-repo",
  "remote_url": "https://github.com/user/repo/tree/main",
  "is_mirror": false
}

New Response for Non-Repo URLs (422):

{
  "detail": "URL appears to be a browser URL, not a git clone URL",
  "suggested_url": "https://github.com/user/repo.git",
  "original_url": "https://github.com/user/repo/tree/main",
  "needs_confirmation": true
}

New Endpoint: POST /repositories/parse-url

Request:

{
  "url": "https://github.com/user/repo/tree/main"
}

Response:

{
  "original_url": "https://github.com/user/repo/tree/main",
  "base_url": "https://github.com/user/repo.git",
  "is_valid_repo_url": false,
  "needs_parsing": true,
  "message": "This looks like a browser URL. Did you mean to clone https://github.com/user/repo.git?"
}

Frontend Flow

Repository Creation Dialog (Enhanced)

1. User pastes URL
2. Frontend calls /repositories/parse-url (debounced)
3. If URL needs parsing:
   • Show yellow warning indicator
   • Display: "This looks like a browser URL"
   • Show suggested URL with "Use this instead" button
   • Allow user to proceed with original URL anyway
4. If URL is valid:
   • Show green checkmark
   • Proceed normally
5. User clicks "Create"
6. If backend returns 422 with suggestion:
   • Show confirmation dialog with suggested URL
   • Options: "Use suggested URL", "Use original", "Cancel"

UI Components

URLInput Component:

• Input field with validation status icon
• Shows inline suggestions when URL is detected as browser URL
• Green/yellow/red border based on validation

URLCorrectionDialog Component:

• Modal dialog for confirming URL correction
• Shows before/after comparison
• Clear action buttons

Implementation Order

1. Backend utilities - URL parsing functions with tests
2. Backend endpoint - /repositories/parse-url
3. Backend validation - Enhanced POST /repositories with suggestion response
4. Frontend URL input - Enhanced input with validation feedback
5. Frontend dialog - Confirmation dialog for URL corrections
6. Integration - Wire up parse-url endpoint to frontend
7. Tests - Unit tests for URL parsing, integration tests for flow

Error Handling

Invalid URLs

• Completely malformed URLs: Return 400 with clear message
• Unsupported hosts: Return 400 with "Unsupported git host"
• Private repos (auth needed): Return 401/403 with auth instructions
• Non-existent repos: Return 404 (from git clone failure)

Clone Failures

• Network issues: Retry with exponential backoff
• Auth required: Prompt for credentials
• Large repos: Show progress indicator
• Timeout: Increase timeout for large repos