core/content_sync.py

"""Mirror chapters from the SSOT (`AI_Ethic_HOT_chatbot/shared_content/`) into
this space's local `skills/case-corpus/references/` folder.

Runs at app startup. On HF Spaces the SSOT folder doesn't exist (it's local-only
design infrastructure, never deployed) — sync is a silent no-op and the deployed
copy is used as-is.

The flow keeps the SSOT as the single editable source: edit chapters in
`shared_content/` → next local `app.py` run mirrors any changes into
`skills/case-corpus/references/` (which is what the bot actually reads at runtime;
this is the only case-related content the bot sees) → `deploy.py` ships the
verified copy.
"""

import sys
from pathlib import Path
import shutil

from core.config_loader import BASE_DIR

# SSOT lives at AI_Ethic_HOT_chatbot/shared_content/, two folders up from this file.
SSOT_PATH = BASE_DIR.parent / "shared_content"
LOCAL_CHAPTERS_PATH = BASE_DIR / "skills" / "case-corpus" / "references"


def sync_chapters_from_ssot():
    """Mirror SSOT chapter files into LOCAL_CHAPTERS_PATH. Idempotent.

    - Copies any SSOT *.md whose bytes differ from the local copy (or that doesn't exist locally).
    - Removes any local *.md that no longer exists in SSOT.
    - Does NOT touch the README.md (we exclude it explicitly so each side can have its own notes).
    - Does nothing when SSOT_PATH is absent (HF Space scenario).
    """
    if not SSOT_PATH.exists():
        return

    LOCAL_CHAPTERS_PATH.mkdir(parents=True, exist_ok=True)

    ssot_files = {p.name: p for p in SSOT_PATH.glob("*.md") if p.name != "README.md"}
    local_files = {p.name: p for p in LOCAL_CHAPTERS_PATH.glob("*.md")}

    synced_changes = []

    # Copy SSOT → local for any file that's new or differs.
    for name, src in ssot_files.items():
        dst = local_files.get(name) or (LOCAL_CHAPTERS_PATH / name)
        if not dst.exists() or src.read_bytes() != dst.read_bytes():
            shutil.copy2(src, dst)
            synced_changes.append(f"+{name}" if not local_files.get(name) else f"~{name}")

    # Remove any local file that no longer exists in SSOT (pure mirror).
    for name in local_files.keys() - ssot_files.keys():
        local_files[name].unlink()
        synced_changes.append(f"-{name}")

    if synced_changes:
        print(f"[content_sync] {', '.join(synced_changes)}", file=sys.stderr)