Metadata-Version: 2.4
Name: argparse-boost
Version: 0.2.0
Summary: Extension module for argparse
Author: yakimka
License: MIT
Project-URL: repository, https://github.com/yakimka/argparse-boost
Platform: Linux
Classifier: Development Status :: 3 - Alpha
Classifier: Intended Audience :: Developers
Classifier: Operating System :: OS Independent
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Programming Language :: Python :: 3.14
Classifier: Programming Language :: Python :: Implementation :: PyPy
Requires-Python: <4.0.0,>=3.10
Description-Content-Type: text/markdown
License-File: LICENSE
Dynamic: license-file

# argparse-boost

[![Build Status](https://github.com/yakimka/argparse_boost/actions/workflows/workflow-ci.yml/badge.svg?branch=main&event=push)](https://github.com/yakimka/argparse_boost/actions/workflows/workflow-ci.yml)
[![Codecov](https://codecov.io/gh/yakimka/argparse-boost/branch/main/graph/badge.svg)](https://codecov.io/gh/yakimka/argparse-boost)
[![PyPI - Version](https://img.shields.io/pypi/v/argparse-boost.svg)](https://pypi.org/project/argparse-boost/)
[![PyPI - Python Version](https://img.shields.io/pypi/pyversions/argparse-boost)](https://pypi.org/project/picodi/)
[![PyPI - Downloads](https://img.shields.io/pypi/dm/argparse-boost)](https://pypi.org/project/picodi/)

argparse-boost is a library for building CLI applications with automatic argument parsing from
dataclasses and environment variable support.

## Installation

```bash
pip install argparse-boost
```

## How to use

### Quick Start: Multi-Command CLI

Build multi-command CLIs (like `git`, `docker`, `kubectl`) with automatic command discovery:

**Project structure:**

```
myapp/
└── cli/
    ├── __main__.py
    └── greet.py
```

**myapp/cli/__main__.py:**

```python
from argparse_boost import Config, setup_cli
from myapp import cli


def main() -> None:
    config = Config(
        app_name="myapp",
        env_prefix="MYAPP_",
    )
    setup_cli(
        config=config,
        description="My CLI application",
        commands_package=cli,
    )


if __name__ == "__main__":
    main()
```

**myapp/cli/greet.py:**

```python
from dataclasses import dataclass


@dataclass(kw_only=True)
class GreetConfig:
    name: str
    greeting: str = "Hello"


def main(args: GreetConfig) -> None:
    """Greet someone with a custom message."""
    print(f"{args.greeting}, {args.name}!")
```

**Usage:**

```bash
python -m myapp.cli greet --name World --greeting Hi
# Output: Hi, World!

python -m myapp.cli greet --name Alice
# Output: Hello, Alice!

python -m myapp.cli --help
# Shows available commands

python -m myapp.cli greet --help
# Shows arguments for greet command
```

The library automatically:

- Discovers commands in the `cli/` directory
- Parses arguments from the dataclass type hint in `main()`
- Generates help text from dataclass fields
- Supports environment variables with the specified prefix

### Writing Commands with Dataclass Auto-Parsing

The easiest way to write a command is to use a dataclass type hint in the `main()` function.
Arguments are automatically parsed from the dataclass fields:

```python
# myapp/cli/deploy.py
from dataclasses import dataclass, field
from typing import Annotated
from argparse_boost import Help


@dataclass(kw_only=True)
class DeployConfig:
    environment: Annotated[str, Help("Target environment (dev/staging/prod)")]
    version: Annotated[str, Help("Version to deploy")]
    dry_run: Annotated[bool, Help("Simulate deployment without making changes")] = False
    services: Annotated[list[str], Help("Services to deploy")] = field(
        default_factory=list
    )


def main(args: DeployConfig) -> None:
    """Deploy application to the specified environment."""
    if args.dry_run:
        print(f"[DRY RUN] Would deploy {args.version} to {args.environment}")
    else:
        print(f"Deploying {args.version} to {args.environment}...")

    if args.services:
        print(f"Services: {', '.join(args.services)}")
```

**Usage:**

```bash
python -m myapp.cli deploy --environment prod --version 1.2.3
python -m myapp.cli deploy --environment staging --version 1.2.4 --dry-run true
python -m myapp.cli deploy --environment dev --version 1.2.5 --services "api,worker"
```

This approach gives you:

- Type-safe argument parsing
- Automatic validation (required fields, type checking)
- Auto-generated help text with descriptions
- Support for complex types (lists, dicts, nested dataclasses)

### Writing Commands with Manual Parser Setup

For more control over argument parsing, use `setup_parser()` and work with `argparse.Namespace`:

```python
import argparse


def setup_parser(parser: argparse.ArgumentParser) -> None:
    """Configure arguments for this command."""
    parser.add_argument(
        "--environment",
        required=True,
        choices=["dev", "staging", "prod"],
        help="Target environment",
    )
    parser.add_argument(
        "--version",
        required=True,
        help="Version to deploy",
    )
    parser.add_argument(
        "--force",
        action="store_true",
        help="Force deployment even if validation fails",
    )


def main(args: argparse.Namespace) -> None:
    """Deploy application to the specified environment."""
    print(f"Deploying {args.version} to {args.environment}")
    if args.force:
        print("Force mode enabled - skipping validation")
```

This approach is useful when you need:

- Custom argument behavior (choices, actions, mutual exclusivity)
- Integration with existing argparse code
- Arguments that don't map cleanly to dataclass fields

**Async command support:**
Commands can also be async - the library automatically handles them with `asyncio.run()`:

```python
import asyncio


async def main() -> None:
    """Async command example."""
    await asyncio.sleep(1)
```

### Configuration Management (Without CLI)

You can use argparse-boost for configuration management in your project, completely separate from
CLI applications. This is useful for web apps, services, or any Python project that needs type-safe
configuration from environment variables:

```python
from dataclasses import dataclass, field
from argparse_boost import construct_dataclass, Config


@dataclass(kw_only=True)
class DatabaseConfig:
    host: str
    port: int = 5432
    user: str = "postgres"
    password: str
    pool_size: int = 10


@dataclass(kw_only=True)
class AppConfig:
    debug: bool = False
    db: DatabaseConfig = field(default_factory=DatabaseConfig)
    allowed_hosts: list[str] = field(default_factory=list)


# Load configuration from environment variables
app_config = construct_dataclass(
    AppConfig,
    config=Config(env_prefix="APP_"),
)

# Now use your type-safe config
print(f"Connecting to {app_config.db.host}:{app_config.db.port}")
print(f"Debug mode: {app_config.debug}")
```

**Environment variables:**

```bash
APP_DEBUG=true
APP_DB_HOST=localhost
APP_DB_PORT=5433
APP_DB_PASSWORD=secret
APP_DB_POOL_SIZE=20
APP_ALLOWED_HOSTS=example.com,api.example.com
```

The configuration is loaded with full type conversion:

- `APP_DEBUG=true` → `config.debug = True`
- `APP_DB_PORT=5433` → `config.db.port = 5433`
- `APP_ALLOWED_HOSTS=a,b` → `config.allowed_hosts = ["a", "b"]`

This approach gives you:

- Type-safe configuration with dataclasses
- Automatic type conversion from environment variables
- Support for nested configuration (flattened to ENV vars)
- No CLI overhead - just configuration management

### Core Features

- **[Multi-Command CLI Framework](#quick-start-multi-command-cli)** - Build git-like CLIs with
  automatic command discovery
- **[Dataclass Auto-Parsing](#writing-commands-with-dataclass-auto-parsing)** - Type-safe argument
  parsing from dataclass type hints
- **[Configuration Management](#configuration-management-without-cli)** - Load type-safe config from
  environment variables
- **[Environment Variable Support](#environment-variable-support)** - Auto-merge ENV vars with CLI
  args (CLI takes priority)
- **[Nested Dataclasses](#nested-dataclasses)** - Automatic flattening with intuitive naming
- **[Custom Parsers](#custom-parsers)** - Extend parsing with custom functions via
  `Annotated[T, Parser(func)]`
- **[Type-Safe Parsing](#advanced-types)** - Support for int, float, str, bool, list, dict, Optional
- **[Custom Help Text](#custom-help-text)** - Enhanced help formatting with defaults and ENV var
  display


### Environment Variable Support

Commands automatically read environment variables before parsing CLI arguments. CLI arguments always
take priority over environment variables. The `env_prefix` specified in `setup_main()` is used for
all commands.

Environment variable naming:

- CLI option `--host` → ENV var `MYAPP_HOST` (with prefix from setup_main)
- CLI option `--port` → ENV var `MYAPP_PORT`
- Multi-word options: `--log-level` → `MYAPP_LOG_LEVEL` (dashes become underscores)

**Example:**

```python
# myapp/cli/connect.py
from dataclasses import dataclass


@dataclass(kw_only=True)
class ConnectConfig:
    host: str
    port: int = 5432
    log_level: str = "INFO"


def main(args: ConnectConfig) -> None:
    """Connect to a server with logging."""
    print(f"Connecting to {args.host}:{args.port}")
    print(f"Log level: {args.log_level}")
```

**Usage:**

```bash
# Using environment variables only
MYAPP_HOST=localhost MYAPP_LOG_LEVEL=DEBUG python -m myapp.cli connect

# CLI overrides ENV
MYAPP_HOST=localhost python -m myapp.cli connect --host production.example.com --port 443
# Result: host='production.example.com', port=443

# Mix both
MYAPP_LOG_LEVEL=DEBUG python -m myapp.cli connect --host db.example.com
# Result: host='db.example.com', log_level='DEBUG'
```

Environment variables are displayed in the help output when they are currently set:

```bash
MYAPP_HOST=localhost python -m myapp.cli connect --help
# Shows: --host (env: MYAPP_HOST=localhost)
```

### Nested Dataclasses

Nested dataclasses are automatically flattened into CLI arguments with intuitive naming:

```python
# myapp/cli/migrate.py
from dataclasses import dataclass, field


@dataclass(kw_only=True)
class DatabaseConfig:
    host: str
    port: int = 5432
    use_ssl: bool = False


@dataclass(kw_only=True)
class MigrateConfig:
    migration_name: str
    db: DatabaseConfig = field(default_factory=lambda: DatabaseConfig(host="localhost"))


def main(args: MigrateConfig) -> None:
    """Run database migrations."""
    ssl_status = "with SSL" if args.db.use_ssl else "without SSL"
    print(f"Running migration '{args.migration_name}'")
    print(f"Connecting to {args.db.host}:{args.db.port} {ssl_status}")
```

Generated CLI arguments:

- `--migration-name` → `args.migration_name`
- `--db-host` → `args.db.host`
- `--db-port` → `args.db.port`
- `--db-use-ssl` → `args.db.use_ssl`

Environment variables (with MYAPP prefix from setup_main):

- `MYAPP_MIGRATION_NAME`
- `MYAPP_DB_HOST`
- `MYAPP_DB_PORT`
- `MYAPP_DB_USE_SSL`

**Usage:**

```bash
# Using CLI arguments
python -m myapp.cli migrate --migration-name add_users --db-host postgres.local --db-port 5433

# Using environment variables
MYAPP_DB_HOST=postgres.local MYAPP_DB_USE_SSL=true python -m myapp.cli migrate --migration-name add_users
```

### Custom Parsers

Use custom parsing functions for specialized types via `Annotated`:

```python
# myapp/cli/analyze.py
from dataclasses import dataclass
from typing import Annotated
from argparse_boost import Parser


def parse_percentage(value: str) -> float:
    """Parse percentage string like '75%' to float 0.75"""
    if isinstance(value, float):
        return value
    return float(value.rstrip("%")) / 100


@dataclass(kw_only=True)
class AnalyzeConfig:
    dataset: str
    threshold: Annotated[float, Parser(parse_percentage)] = 0.5
    min_confidence: Annotated[float, Parser(parse_percentage)] = 0.8


def main(args: AnalyzeConfig) -> None:
    """Analyze dataset with custom threshold."""
    print(f"Analyzing {args.dataset}")
    print(f"Threshold: {args.threshold:.2%}")
    print(f"Min confidence: {args.min_confidence:.2%}")
```

**Usage:**

```bash
python -m myapp.cli analyze --dataset users.csv --threshold 75% --min-confidence 90%
# Output:
# Analyzing users.csv
# Threshold: 75.00%
# Min confidence: 90.00%
```

Custom parsers are useful for:

- Parsing duration strings (e.g., "1h30m" → seconds)
- Parsing file sizes (e.g., "10MB" → bytes)
- Parsing percentages, ratios, or custom formats
- Validating and transforming complex inputs

### Help Text

Add descriptive help text to fields using `Annotated` with `Help`:

```python
# myapp/cli/backup.py
from dataclasses import dataclass
from typing import Annotated
from argparse_boost import Help


@dataclass(kw_only=True)
class BackupConfig:
    source: Annotated[str, Help("Source directory to backup")]
    destination: Annotated[str, Help("Destination directory for backup")]
    timeout: Annotated[int, Help("Backup timeout in seconds")] = 3600
    retries: Annotated[int, Help("Number of retry attempts on failure")] = 3
    compress: Annotated[bool, Help("Compress backup files")] = True


def main(args: BackupConfig) -> None:
    """Create a backup of the source directory."""
    compression = "compressed" if args.compress else "uncompressed"
    print(f"Backing up {args.source} to {args.destination} ({compression})")
    print(f"Timeout: {args.timeout}s, Retries: {args.retries}")
```

**View help:**

```bash
python -m myapp.cli backup --help
# Shows:
#   --source             Source directory to backup (Required)
#   --destination        Destination directory for backup (Required)
#   --timeout            Backup timeout in seconds (Default: 3600)
#   --retries            Number of retry attempts on failure (Default: 3)
#   --compress           Compress backup files (Default: True)
```

Help text is automatically enhanced with default values and "Required" indicators for fields without
defaults.

### Advanced Types

#### Lists

Lists are parsed from comma-separated values:

```python
# myapp/cli/tag.py
from dataclasses import dataclass, field


@dataclass(kw_only=True)
class TagConfig:
    resource: str
    tags: list[str] = field(default_factory=list)
    allowed_ports: list[int] = field(default_factory=list)


def main(args: TagConfig) -> None:
    """Tag a resource with metadata."""
    print(f"Tagging resource: {args.resource}")
    if args.tags:
        print(f"Tags: {', '.join(args.tags)}")
    if args.allowed_ports:
        print(f"Allowed ports: {', '.join(map(str, args.allowed_ports))}")
```

**Usage:**

```bash
# Using CLI arguments
python -m myapp.cli tag --resource server-01 --tags "web,api,prod" --allowed-ports "80,443,8080"
# Output:
# Tagging resource: server-01
# Tags: web, api, prod
# Allowed ports: 80, 443, 8080

# Using environment variables
MYAPP_RESOURCE=server-02 MYAPP_TAGS="database,backup" python -m myapp.cli tag
```

#### Dictionaries

Dictionaries are parsed from comma-separated key-value pairs using `:` or `=`:

```python
# myapp/cli/configure.py
from dataclasses import dataclass, field


@dataclass(kw_only=True)
class ConfigureConfig:
    service: str
    limits: dict[str, int] = field(default_factory=dict)
    settings: dict[str, str] = field(default_factory=dict)


def main(args: ConfigureConfig) -> None:
    """Configure service with limits and settings."""
    print(f"Configuring {args.service}")
    if args.limits:
        for key, value in args.limits.items():
            print(f"  Limit {key}: {value}")
    if args.settings:
        for key, value in args.settings.items():
            print(f"  Setting {key}: {value}")
```

**Usage:**

```bash
# Using colon separator
python -m myapp.cli configure --service api --limits "daily:100,monthly:3000"

# Using equals separator
python -m myapp.cli configure --service worker --settings "timeout=30,retries=3"
```

#### Optional Types

Optional fields are supported via `T | None` or `Optional[T]`:

```python
# myapp/cli/process.py
from dataclasses import dataclass


@dataclass(kw_only=True)
class ProcessConfig:
    input_file: str
    output_file: str | None = None
    max_retries: int | None = 3
    format: str | None = None


def main(args: ProcessConfig) -> None:
    """Process a file with optional parameters."""
    print(f"Processing {args.input_file}")
    if args.output_file:
        print(f"Output to: {args.output_file}")
    else:
        print("Output to: stdout")

    print(f"Max retries: {args.max_retries}")
    if args.format:
        print(f"Format: {args.format}")
```

**Usage:**

```bash
# Only required field
python -m myapp.cli process --input-file data.csv
# Output: stdout, max_retries: 3, format: None

# With optional fields
python -m myapp.cli process --input-file data.csv --output-file result.json --format json
```

### Programmatic Usage

For more control, use the low-level API:

```python
from dataclasses import dataclass
from argparse_boost import (
    BoostedArgumentParser,
    construct_dataclass,
    Config,
    dict_from_args,
)


@dataclass(kw_only=True)
class Settings:
    host: str
    port: int = 5432


# Option 1: Simple approach - automatic ENV loading
settings = construct_dataclass(
    Settings,
    config=Config(env_prefix="APP_"),
)

# Option 2: Manual approach with BoostedArgumentParser
parser = BoostedArgumentParser(prog="myapp", env_prefix="APP_")
parser.parse_arguments_from_dataclass(Settings)
args = parser.parse_args(["--host", "example.com"])

# Extract CLI arguments and construct with overrides
cli_data = dict_from_args(args, Settings)
settings = construct_dataclass(
    Settings,
    override=cli_data,
    config=Config(env_prefix="APP_"),
)
```

### API Reference

**Framework Functions:**

- `setup_main()` - Entry point for multi-command CLI applications with automatic command discovery

**Configuration Functions:**

- `construct_dataclass(dc_type, override, *, config)` - Construct dataclass from environment and overrides
- `env_for_dataclass(fields, config)` - Read ENV vars for specific fields (advanced use)
- `Config` - Configuration class with env_prefix, loaders, and other options

**Parser Classes:**

- `BoostedArgumentParser` - Extended ArgumentParser with ENV variable support
- `DefaultsHelpFormatter` - Help formatter showing defaults and required indicators

**Dataclass Parsing:**

- `dict_from_args(args, dataclass_type)` - Extract CLI args to flat dict

**Annotations:**

- `Parser(func)` - Custom parsing function for `Annotated` fields
- `Help(text)` - Custom help text for `Annotated` fields

**Exceptions:**

- `CliSetupError` - Base exception for CLI setup errors
- `FieldNameConflictError` - Raised when flattened field names collide
- `UnsupportedFieldTypeError` - Raised for unsupported field types

### Supported Types

| Type             | Example                   | CLI Input             | ENV Input               |
|------------------|---------------------------|-----------------------|-------------------------|
| `int`            | `count: int`              | `--count 42`          | `APP_COUNT=42`          |
| `float`          | `ratio: float`            | `--ratio 3.14`        | `APP_RATIO=3.14`        |
| `str`            | `name: str`               | `--name hello`        | `APP_NAME=hello`        |
| `bool`           | `debug: bool`             | `--debug true`        | `APP_DEBUG=yes`         |
| `list[T]`        | `tags: list[str]`         | `--tags a,b,c`        | `APP_TAGS=a,b,c`        |
| `dict[K,V]`      | `limits: dict[str,int]`   | `--limits a:1,b:2`    | `APP_LIMITS=a:1,b:2`    |
| `T \| None`      | `port: int \| None`       | `--port 80`           | `APP_PORT=80`           |
| Nested dataclass | `db: Database`            | `--db-host localhost` | `APP_DB_HOST=localhost` |
| Custom           | `Annotated[T, Parser(f)]` | Custom parsing        | Custom parsing          |

**Type constraints:**

- `list[T]` and `dict[K,V]`: T, K, V must be simple types (int, float, str, bool) or Optional simple
  types
- `Union` is only supported as `Optional[T]` (i.e., `T | None`)
- Generic unions like `str | int` are not supported

**Boolean parsing** accepts (case-insensitive):

- True: `true`, `yes`, `on`, `1`
- False: `false`, `no`, `off`, `0`

### Tips and Best Practices

1. **Use `field(default_factory=...)` for mutable defaults:**
   ```python
   from dataclasses import dataclass, field


   @dataclass(kw_only=True)
   class Settings:
       tags: list[str] = field(default_factory=list)  # Good
       # tags: list[str] = []  # Bad - mutable default
   ```

2. **Avoid field name conflicts with nested dataclasses:**
   ```python
   # This will raise FieldNameConflictError:
   @dataclass(kw_only=True)
   class Database:
       password: str


   @dataclass(kw_only=True)
   class Settings:
       db_password: str  # Conflicts with db.password when flattened
       db: Database
   ```

3. **Use `env_prefix` to scope environment variables:**
   ```python
   parser = BoostedArgumentParser(prog="myapp", env_prefix="MYAPP_")
   # Prevents conflicts with other apps' ENV vars
   ```

## Development

### Quick Start

1. Clone the repository:
   ```bash
   git clone <repository-url>
   ```
2. (?) Copy the example settings file and configure your settings:
   ```bash
   cp settings.example.yaml settings.yaml
   ```
3. Build the Docker images:
   ```bash
   docker-compose build
   ```
4. Install dependencies:
   ```bash
   make uv args="sync"
   ```
5. Start the service:
   ```bash
   docker-compose up
   ```

### Making Changes

1. List available `make` commands:
   ```bash
   make help
   ```
2. Check code style with:
   ```bash
   make lint
   ```
3. Run tests using:
   ```bash
   make test
   ```
4. Manage dependencies via uv:
   ```bash
   make uv args="<uv-args>"
   ```
    - For example: `make uv args="add picodi"`

5. For local CI debugging:
   ```bash
   make run-ci
   ```

#### Pre-commit Hooks

We use [pre-commit](https://pre-commit.com/) for linting and formatting:

- It runs inside a Docker container by default.
- Optionally, set up hooks locally:
  ```bash
  pre-commit install
  ```

#### Mypy

We use [mypy](https://mypy.readthedocs.io/en/stable/) for static type checking.

It is configured for strictly typed code, so you may need to add type hints to your code.
But don't be very strict, sometimes it's better to use `Any` type.

## License

[MIT](https://github.com/yakimka/argparse-boost/blob/main/LICENSE)

## Credits

This project was generated with [
`yakimka/cookiecutter-pyproject`](https://github.com/yakimka/cookiecutter-pyproject).
