src/researchlink/validators/output_validator.py

"""Output directory validation."""

from __future__ import annotations

from pathlib import Path

# Spec paper-module files a complete module must contain.
REQUIRED_FILES = [
    "paper.md",
    "metadata.json",
    "sources.json",
    "bibtex.bib",
    "summary.md",
    "claims.md",
    "related_work.md",
    "implementation.md",
    "reproduction.md",
    "study_notes.md",
    "review.md",
    "README.md",
    "generation-report.md",
]


def validate_paper_module(output_dir: Path) -> list[dict]:
    """
    Check that all required spec files exist in a paper module directory.
    Returns list of {ok: bool, message: str} dicts.
    """
    results = []

    if not output_dir.exists():
        return [{"ok": False, "message": f"Output directory does not exist: {output_dir}"}]

    for filename in REQUIRED_FILES:
        path = output_dir / filename
        if path.exists():
            results.append({"ok": True, "message": f"{filename} exists"})
        else:
            results.append({"ok": False, "message": f"Missing: {filename}"})

    # Check metadata.json / sources.json are valid JSON
    for name in ("metadata.json", "sources.json"):
        path = output_dir / name
        if path.exists():
            try:
                import json
                json.loads(path.read_text(encoding="utf-8"))
                results.append({"ok": True, "message": f"{name} is valid JSON"})
            except Exception as e:
                results.append({"ok": False, "message": f"{name} JSON parse error: {e}"})

    return results