Files
mtc-project-sync-script/CLAUDE.md

88 lines
3.8 KiB
Markdown

# Project Sync Toolkit
A bash toolkit for synchronizing projects between remote servers and local development machines.
## Overview
The main script `sync-project.sh` downloads a project from a remote server via tar archive and generates helper scripts for ongoing synchronization. It's designed to be symlinked from a projects directory so multiple projects can share the same toolkit.
## Architecture
```
sync-project.sh # Main orchestrator - generates all helper scripts
sync-git-project.sh # Full project setup via git clone (--git mode)
shared.sh # Shared functions and variables
MANUAL # CLI help text (displayed by -h flag)
lazygit # Binary uploaded to server for git operations
scripts/ # Helper script templates
local-remote.sh # Push files to server (rsync)
remote-local.sh # Pull files from server (rsync)
ssh.sh # SSH access with --db, --git, --todo flags
sftp-watch.sh # Continuous sync using watchexec
watch-build.sh # Build and deploy automation
split-conflicts.sh # Git conflict resolution helper
find_image_size.php # Image aspect ratio analyzer
```
## How It Works
### Default Mode (tar sync)
1. User runs `./sync-project.sh` from their projects directory (via symlink)
2. Script prompts for: server, password, path (default ~/public_leo), local folder
3. Creates tar archive on server, downloads and extracts locally
4. Generates helper scripts with embedded credentials (from templates in `scripts/`)
5. On full sync: syncs terminal info, uploads lazygit, updates .gitignore
### Git Mode (`--git`)
1. User runs `./sync-project.sh --git`
2. Script prompts for: server, password, path, local folder
3. Clones repository via git over SSH
4. Performs full project setup (12 steps):
- Clone repository, check for .npmrc
- Copy Docker env file (.docker/config/.env.local → .env)
- Clean Docker volumes (stop/remove containers and volumes)
- Run docker-setup.sh
- Copy server .env to .server.env
- Modify local .env (APP_NAME, APP_KEY from server)
- Create .env.prod.local with production credentials
- Add helper scripts (ssh.sh)
- Upload lazygit to server
- SSH key management (generate or download staging.key)
- Add staging.key to SSH agent
- Database setup (db:wipe + make db-sync)
## Modes
- **Default**: Full setup with all scripts, lazygit, terminfo sync
- **--git**: Full project setup via git clone with Docker, env, SSH keys, and database
- **--sftp-only**: Helper scripts only, no lazygit or .gitignore updates
- **--files-only**: Quick sync, excludes node_modules/vendor, no scripts generated
## Key Conventions
- All SSH/SCP commands use `$SSH_OPTS` for consistent options (currently: auto-accept new hosts)
- Paths starting with `~` are resolved to full paths via SSH before use
- Helper scripts get credentials injected as variables at the top, then template content appended
- Conflict prevention uses `.remote-in-progress` flag file
- Terminal detection checks `$TERM` for "ghostty" or "kitty" substrings
## Shared Functions (shared.sh)
- `pause_for_user <message>` - Display message and wait for Enter key
- `get_env_value <file> <key>` - Extract value from .env file (strips quotes, CRLF)
- `set_env_value <file> <key> <value>` - Add or update value in .env file
- `add_to_gitignore <entry>` - Add entry to .gitignore if not present
## Generated Script Structure
Each helper script is built by:
1. Writing a header with `#!/bin/bash`, `set -e`, and credential variables
2. Appending the corresponding template from `scripts/` directory
## Testing
- `-h` flag should display MANUAL and exit
- Path resolution requires valid SSH credentials
- Terminfo sync only runs on full sync (no flags) and requires infocmp
- See `verification.md` for comprehensive test cases for `--git` mode