Writing custom JSON parser

What This Skill Does

The DRY philosophy is a central tenet of b00t: YEI exist to contribute ONLY new and novel meaningful work. This skill helps you:

• Identify when code is being duplicated or reinvented
• Find existing libraries instead of writing new code
• Use Rust functionality via PyO3 rather than duplicate in Python
• Contribute upstream rather than maintain private forks
• Write lean, maintainable code with minimal dependencies

When It Activates

Activate this skill when you see:

• "implement [common functionality]"
• "create a [parser/​validator/​client]"
• "write code to [read/​parse/​validate] [format]"
• Any task that sounds like it might already exist in a library
• Code that duplicates existing Rust functionality
• Multiple implementations of the same logic

Core Principles

1. DRY: Don't Repeat Yourself

AVOID writing code for functionality that exists in libraries:

❌ Anti-pattern:

# Writing custom JSON parser
def parse_json(text):
    # 200 lines of parsing logic...

✅ DRY approach:

import json
data = json.loads(text)

2. NRtW: Never Reinvent the Wheel

SEARCH for existing solutions before coding:

# Search for Python packages
pip search [functionality]
# or
uv pip search [functionality]

# Check PyPI
https://pypi.org/​search/?q=[functionality]

# Check Rust crates
https://crates.io/​search?q=[functionality]

3. Leverage Rust via PyO3

USE Rust for heavy lifting, expose to Python:

❌ Anti-pattern:

# Duplicating Rust datum parsing in Python
def parse_datum_file(path: str) -> dict:
    with open(path) as f:
        toml_data = toml.load(f)
    # Validation logic...
    # Parsing logic...
    return processed_data

✅ DRY approach:

# Use Rust via PyO3
import b00t_py
datum = b00t_py.load_ai_model_datum("model-name", "~/​.dotfiles/​_b00t_")

Why? Rust implementation already exists, is faster, type-safe, and tested.

4. Contribute Upstream

FORK and PATCH forward, don't maintain private copies:

❌ Anti-pattern:

# Copy library code into project
cp -r /​path/​to/​library my_project/​vendored/
# Make private modifications

✅ DRY approach:

# Fork the library
gh repo fork upstream/​library

# Create patch
git checkout -b fix/​issue-123
# Make changes
git commit -m "fix: resolve issue #123"

# Submit PR
gh pr create --upstream

# Use your fork temporarily
# pyproject.toml
dependencies = [
    "library @ git+https://github.com/​you/​library@fix/​issue-123"
]

Decision Tree

Need to implement functionality?
    ↓
Does it already exist in a library?
    ├─ YES → Use the library (DRY)
    └─ NO ↓
           Is it standard functionality?
               ├─ YES → Search harder, it probably exists
               └─ NO ↓
                      Does similar Rust code exist in b00t?
                          ├─ YES → Expose via PyO3 (DRY)
                          └─ NO ↓
                                 Is this truly novel?
                                     ├─ YES → Implement (with tests!)
                                     └─ NO → Reconsider: use library

Examples

Finding Libraries

Task: Parse TOML files

# Search
pip search toml

# Results: tomli, tomlkit, pytoml
# Use established: tomli (or tomllib in Python 3.11+)

Task: Make HTTP requests

# DON'T: Write custom HTTP client
# DO: Use httpx or requests
pip install httpx

Task: Validate Pydantic models

# DON'T: Write custom validation
# DO: Use Pydantic's built-in validation
from pydantic import BaseModel, field_validator

Using Rust via PyO3

b00t Pattern: Rust does heavy lifting, Python uses it.

Datum Operations

❌ Duplicate (Anti-pattern):

# b00t_j0b_py/​datum_parser.py
import toml

class DatumParser:
    def load_provider(self, name: str):
        path = f"~/​.dotfiles/​_b00t_/​{name}.ai.toml"
        with open(os.path.expanduser(path)) as f:
            data = toml.load(f)
        # Validation...
        # Parsing...
        return data

✅ DRY (Use Rust):

# Use PyO3 bindings
import b00t_py

datum = b00t_py.load_ai_model_datum("model-name", "~/​.dotfiles/​_b00t_")

Why better?

• ✅ No duplication - single source of truth in Rust
• ✅ Type-safe - Rust ensures correctness
• ✅ Tested - Rust tests cover this
• ✅ Faster - Rust performance
• ✅ Maintainable - one codebase, not two

Environment Validation

❌ Duplicate:

def validate_provider_env(provider: str) -> bool:
    # Read datum
    # Parse required env vars
    # Check os.environ
    # Return result

✅ DRY:

import b00t_py

validation = b00t_py.check_provider_env("openrouter", "~/​.dotfiles/​_b00t_")
if not validation["available"]:
    print(f"Missing: {validation['missing_env_vars']}")

Contributing Upstream

Scenario: Bug in pydantic-ai library

❌ Anti-pattern:

# Copy code into project
cp -r site-packages/​pydantic_ai b00t_j0b_py/​vendored/
# Fix bug privately
# Now you maintain a fork forever

✅ DRY approach:

# Fork
gh repo fork pydantic/​pydantic-ai

# Fix and test
git checkout -b fix/​agent-validation-bug
# Make changes
pytest tests/
git commit -m "fix: agent validation for None values"

# Submit PR
gh pr create --title "fix: agent validation for None values"

# Temporarily use your fork
# pyproject.toml
dependencies = [
    "pydantic-ai @ git+https://github.com/​elasticdotventures/​pydantic-ai@fix/​agent-validation-bug"
]

# After PR merged, switch back to upstream
dependencies = [
    "pydantic-ai>=0.0.15"  # includes fix
]

Library Selection Criteria

When choosing a library:

✅ Good Signs

• ✅ Many stars (>1000 on GitHub)
• ✅ Active maintenance (commits in last month)
• ✅ Minimal open issues/​PRs
• ✅ Good documentation
• ✅ Permissive license (MIT, Apache, BSD)
• ✅ Used by major projects
• ✅ Type hints (Python) or strong types (Rust)
• ✅ Comprehensive tests
• ✅ Lively, polite community discussions

🚩 Red Flags

• 🚩 Abandoned (no commits in 1+ years)
• 🚩 Many unresolved issues
• 🚩 No tests
• 🚩 Copyleft license (GPL) for permissive projects
• 🚩 No type hints
• 🚩 Breaking changes without semver
• 🚩 Hostile maintainers

PyO3 Pattern

When to Use Rust (via PyO3)

Use Rust for:

• ✅ Performance-critical code
• ✅ Type-safe validation
• ✅ Complex parsing
• ✅ Shared logic between Rust and Python
• ✅ System-level operations

How to Expose Rust to Python

// b00t-py/​src/​lib.rs
use pyo3::prelude::*;

#[pyfunction]
fn my_function(py: Python<'_>, arg: &str) -> PyResult<String> {
    // Rust implementation
    Ok(format!("Processed: {}", arg))
}

#[pymodule]
fn b00t_py(_py: Python, m: &PyModule) -> PyResult<()> {
    m.add_function(wrap_pyfunction!(my_function, m)?)?;
    Ok(())
}

# Python usage
import b00t_py

result = b00t_py.my_function("test")

Code Review Checklist

Before writing code, ask:

1. ☐ Does this functionality exist in a library?
2. ☐ Does similar Rust code exist in b00t?
3. ☐ Can I use PyO3 to expose Rust instead?
4. ☐ Is this truly novel functionality?
5. ☐ Have I searched PyPI/​crates.io?
6. ☐ Have I checked existing b00t modules?

If all answers are "no", then implement.

Anti-Patterns to Avoid

1. Reinventing Standard Library

❌ Bad:

def read_json_file(path):
    with open(path) as f:
        return custom_json_parse(f.read())

✅ Good:

import json

def read_json_file(path):
    with open(path) as f:
        return json.load(f)

2. Duplicating Rust Logic

❌ Bad:

# Reimplementing datum validation in Python
class DatumValidator:
    def validate_env(self, provider): ...
    def parse_toml(self, path): ...

✅ Good:

# Use Rust via PyO3
import b00t_py
validation = b00t_py.check_provider_env(provider, path)

3. Private Forks

❌ Bad:

# Fork library, never contribute back
# Maintain private version forever

✅ Good:

# Fork, fix, PR upstream
# Use fork temporarily until merged
# Switch back to upstream after merge

4. Not Using Type Hints

❌ Bad:

def process_data(data):
    # No types, unclear what's expected
    return data.transform()

✅ Good:

from pydantic import BaseModel

def process_data(data: dict[str, Any]) -> ProcessedData:
    return ProcessedData(**data)

Benefits of DRY

1. Less code to maintain - fewer bugs, faster development
2. Better quality - libraries are tested by many users
3. Security updates - library maintainers handle CVEs
4. Performance - Rust is faster than Python for heavy tasks
5. Type safety - Rust ensures correctness
6. Community - contribute to ecosystem, get help

Related Skills

• datum-system: Uses Rust via PyO3 (DRY example)
• direnv-pattern: Uses existing direnv tool (DRY)
• justfile-usage: Uses casey/​just tool (DRY)

References

• CLAUDE.md - YEI MUST ALWAYS/​NEVER section
• b00t-py/​src/​lib.rs - PyO3 bindings example
• b00t-j0b-py/​pyproject.toml - Dependency management

Summary

DRY Philosophy:

• 🔍 Search for existing solutions first
• 📦 Use established libraries over custom code
• 🦀 Leverage Rust via PyO3 instead of duplicating in Python
• 🔄 Fork, fix, PR upstream - don't maintain private copies
• ✅ Write only novel, meaningful code
• 🧪 Test everything you write

Alignment: A lean hive is a happy hive. Finding and patching bugs in libraries is divine; committing buggy code is unforgivable.