# AGENTS.md — Image Server Project

## Project Overview

A FastAPI-based image server that serves files from directories or ZIP archives using hashed paths for secure, randomized access. Supports password-protected ZIPs via HTTP Basic Auth. Provides a full-screen image viewer with navigation controls and auto-advance.

## Language & Runtime

- **Language**: Python 3.13
- **Framework**: FastAPI 0.128.0+
- **Server**: Uvicorn 0.40.0+
- **Package manager**: UV

## Build, Lint, and Test Commands

### Running the Server

```bash
# Run with a directory
uv run main.py /path/to/files

# Run with a ZIP archive
uv run main.py /path/to/archive.zip

# Run with a glob pattern
uv run main.py "path/to/zips/*.zip"
```

### Testing

```bash
# Run all tests
uv run pytest

# Run a single test
uv run pytest tests/test_file_indexer.py::TestClass::test_method

# Run tests matching a pattern
uv run pytest -k "test_name_pattern"
```

### Linting

```bash
# Install ruff (fast linter)
uv add ruff

# Run linting
ruff check .

# Auto-fix issues
ruff check --fix .
```

### Formatting

Always format code with black after implementing every feature:

```bash
# Install black
uv add black

# Format all code
uv run black .
```

### Type Checking

```bash
# Install mypy
uv add mypy

# Run type checking
uv run mypy main.py
```

## Project Structure

```
image_server/
├── main.py              # Main application — all server logic, routing, indexing
├── frontend.html        # Single-page HTML template for the image viewer UI
├── pyproject.toml       # Project configuration (dependencies, metadata)
├── uv.lock              # UV lockfile for reproducible builds
├── README.md            # User-facing documentation
├── AGENTS.md            # This file — agent/dev documentation
└── tests/               # Test suite
    ├── conftest.py      # Pytest fixtures and configuration
    ├── test_file_indexer.py    # Tests for FileIndexer class
    ├── test_zip_indexer.py     # Tests for ZipFileIndexer class
    ├── test_encrypted_zip.py   # Tests for password-protected ZIP handling
    ├── test_endpoints.py       # Tests for API endpoints
    ├── test_navigation.py      # Tests for navigation logic
    └── testfile.zip            # Test fixture ZIP file
```

## Code Architecture

### Entry Point (`main.py`)

All code lives in `main.py`. The app is initialized via `initialize_server()` which is called from `__main__`.

### Global State

```python
file_mapping = {}   # dict[str, str] — hash -> filepath/filename (merged across all indexers)
indexers = []       # list[FileIndexer] — all active indexers
```

### Indexer Classes

#### `FileIndexer` (base class)

Indexes files from a directory tree.

- **`__init__(path: str, salt: str | None)`** — Initializes with a path and optional salt
- **`salt` (property)** — Lazily generates a random salt if not provided
- **`_hash_path(filepath: str) -> str`** — SHA-256 hash of `filepath + salt`
- **`_index() -> dict[str, str]`** — Walks directory tree, builds hash -> filepath mapping
- **`get_file_by_hash(file_hash: str)`** — Yields file content bytes (generator)
- **`get_filename_by_hash(file_hash: str) -> str | None`** — Returns filepath by hash

#### `ZipFileIndexer(FileIndexer)`

Indexes files from a ZIP archive. Extends `FileIndexer`.

- **`_open_zip()`** — Opens the ZIP file for reading
- **`_index() -> dict[str, str]`** — Indexes files within the ZIP (hash -> internal filename)
- **`_needs_password(filename: str) -> bool`** — Lazy detection of password protection (cached)
- **`is_file_encrypted(file_hash: str) -> bool`** — Check if a file requires a password
- **`get_file_by_hash(file_hash: str, password: bytes | None)`** — Yields file content, handles encryption
- **`get_filename_by_hash(file_hash: str) -> str | None`** — Returns internal ZIP filename by hash

#### `INDEXER_MAP`

```python
INDEXER_MAP = {".zip": ZipFileIndexer}
```

Maps file extensions to indexer classes. Extend this dict to support new archive formats.

### Helper Functions

- **`initialize_server(args: argparse.Namespace)`** — Bootstraps the server: creates indexers, populates `file_mapping`
- **`_find_indexer_for_hash(file_hash: str)`** — Finds which indexer owns a given hash
- **`_get_zip_password(file_hash: str, request: Request) -> bytes | None`** — Extracts password from HTTP Basic Auth header
- **`_raise_unauthorized()`** — Raises HTTP 401 with `WWW-Authenticate: Basic` header
- **`_get_random_hash() -> str`** — Returns a random hash from the indexed files
- **`_build_url(file_hash, order, delay) -> str`** — Builds a URL with optional query params
- **`_get_navigation_data(file_hash, order)`** — Returns dict with `file_hash`, `next_hash`, `prev_hash`, `filename`
- **`_render_page(navigation_data, ...)`** — Renders `frontend.html` template with substituted values

### Routes

| Route | Handler | Description |
|-------|---------|-------------|
| `GET /` | `root()` | Redirect to random file |
| `GET /{file_hash}` | `hash_page()` | Serve image page with navigation |
| `GET /api/{file_hash}/data` | `get_file_data()` | Serve raw file data |
| `GET /api/health` | `health_check()` | Health check (status + file count) |

### Frontend Template (`frontend.html`)

Single HTML file using `string.Template` substitution. Placeholders:
- `$img_url` — URL to fetch the image data
- `$image_click_url` — URL navigated to on image click
- `$next_url` / `$prev_url` — Navigation URLs
- `$filename` — Display filename
- `$extra_meta` — Extra `<meta>` tags (e.g., auto-refresh)
- `$play_button` — Play/pause button HTML

## Code Style Guidelines

### Imports

- Standard library imports first, then third-party, then local
- Group imports by type: stdlib, third-party, local
- Use absolute imports for local modules

### Formatting Style

- Use 4 spaces for indentation (no tabs)
- Maximum line length: 88 characters (black default)
- Use blank lines to separate logical sections (2 blank lines between top-level definitions)
- One blank line between method definitions within a class
- **Always run `black .` after completing every feature**

### Types

- Use type hints for all function arguments and return values
- Use Python 3.13+ typing features where appropriate

### Naming Conventions

- **Classes**: `PascalCase` (e.g., `FileIndexer`, `ZipFileIndexer`)
- **Functions/variables**: `snake_case` (e.g., `get_file_by_hash`, `file_mapping`)
- **Constants**: `UPPER_SNAKE_CASE` (e.g., `INDEXER_MAP`)
- **Private methods/attributes**: Prefix with underscore (e.g., `_index`, `_salt`)
- Use descriptive, full words for names (avoid abbreviations except well-known ones)

### Error Handling

- Use exceptions for exceptional cases, not flow control
- Raise `HTTPException` for HTTP-level errors in FastAPI endpoints
- Provide meaningful error messages

### Async/Await

- Use `async def` for FastAPI endpoint handlers
- Use regular functions for synchronous operations (file I/O, hashing)
- Do not mix sync and async improperly

### File Operations

- Use `pathlib.Path` for path manipulations
- Use context managers (`with` statements) for file operations
- Handle both directory and archive (ZIP) file sources

### Documentation

- Use docstrings for public classes and functions
- Follow Google-style docstring format

### Best Practices

- Keep functions small and focused (single responsibility)
- Use properties for computed attributes with caching
- Avoid global state where possible; use dependency injection
- Use dataclasses or attrs for simple data containers
- Follow PEP 8 style guidelines

## Common Tasks

### Adding a New Endpoint

1. Add the route handler in `main.py`
2. Use appropriate HTTP method decorator (`@app.get`, `@app.post`, etc.)
3. Add type hints and docstrings
4. Handle errors with `HTTPException`

### Adding a New File Indexer

1. Create a new class inheriting from `FileIndexer`
2. Override the `_index` method to populate `self._file_mapping`
3. Override `get_file_by_hash` and `get_filename_by_hash` as needed
4. Register in `INDEXER_MAP` dictionary with the file extension as key

### Modifying the Frontend

1. Edit `frontend.html` for HTML/CSS/JS changes
2. Add new `string.Template` placeholders as needed
3. Update `_render_page()` to pass new values
4. Ensure navigation URLs are built correctly with `_build_url()`

### Running the Server in Development

```bash
# Direct run (rebuilds on each start)
uv run main.py /path/to/files

# With custom options
uv run main.py /path/to/files --host 127.0.0.1 --port 3000 --salt mysecret
```