pokemon/scripts/resync-vault-library-paths.py

#!/usr/bin/env python3
"""Resync each vault hack note's `library_path` to the actual on-disk ROM file.

The [Hack] filename standardization renamed ROMs on disk but left the vault
notes' `library_path` frontmatter pointing at old names/paths. This matches each
note to its real file (by normalized title, falling back to the old path's
basename) and rewrites `library_path` + any exact body copy of the old value.

Read-only by default; pass --apply to write. ROM listings are read from
/tmp/romsync/<plat>.txt (one filename per line), pulled from valhalla.
"""
import os, re, sys, glob

VAULT = r"C:\Users\MattC\Documents\Obsidian Vault"
HACKS = os.path.join(VAULT, "Pokémon ROM Hacks", "Hacks")
LISTS = os.path.join(os.path.dirname(__file__), "..", ".scrape", "romsync")
PLAT = {"GBA": "gba", "GBC": "gbc", "GB": "gb", "NDS": "nds", "3DS": "3ds", "PC": "windows"}
APPLY = "--apply" in sys.argv

def norm(s):
    s = os.path.splitext(s)[0].lower()
    s = re.sub(r"\[[^\]]*\]", "", s)      # drop [Hack], [2.0.5], [Full], ...
    s = re.sub(r"\([^)]*\)", "", s)       # drop (Gold inc), (Spanish), ...
    s = re.sub(r"pok[eé]?mon", "", s)     # drop the franchise prefix/word
    s = re.sub(r"[^a-z0-9]", "", s)       # keep bare digits (Dark Rising 2)
    return s

# ---- build per-platform file index: normkey -> [filenames] ----
idx = {}
for path in glob.glob(os.path.join(LISTS, "*.txt")):
    plat = os.path.splitext(os.path.basename(path))[0]
    idx[plat] = {}
    for line in open(path, encoding="utf-8"):
        fn = line.strip()
        if not fn or not re.search(r"\.(gba|gbc|gb|nds|3ds|cci|cxi|zip)$", fn, re.I):
            continue
        idx[plat].setdefault(norm(fn), []).append(fn)

def fm_get(text, key):
    m = re.search(rf"(?m)^{key}:\s*(.*)$", text)
    return m.group(1).strip().strip('"').strip("'") if m else None

FUZZY = "--fuzzy" in sys.argv   # also write the proposed fuzzy/cross-platform matches

def find_exact_all(key):
    """Exact normalized-key match across ALL platforms (catches wrong platform field)."""
    hits = []
    for p, files in idx.items():
        for fn in files.get(key, []):
            hits.append((p, fn))
    return hits

# Notes whose only fuzzy candidate is a different/ambiguous hack — held for manual
# review, never auto-applied even with --fuzzy.
FUZZY_HOLD = {
    "Bloody Fusion.md", "Too Many Types.md", "HeartGold Generation.md",
    "FireRed Reignited LeafGreen Regrown.md", "Kanto Complete.md",
    "Mystery Dungeon - Wigglytuff's Bizarre Adventure.md",
}

def find_fuzzy(nk):
    """One key is a substring of the other; require a [Hack] file; must be unique.
    Skip very short keys (e.g. a hack literally named 'Z') to avoid spurious hits."""
    if len(nk) < 4: return []
    hits = []
    for p, files in idx.items():
        for k, fns in files.items():
            if len(k) < 4 or not nk: continue
            if nk in k or k in nk:
                for fn in fns:
                    if "[hack]" in fn.lower():
                        hits.append((p, fn))
    return sorted(set(hits))

changed = unchanged = unmatched = ambiguous = 0
report = {"CHANGED": [], "UNMATCHED": [], "AMBIGUOUS": [], "FUZZY": []}

for md in sorted(glob.glob(os.path.join(HACKS, "*.md"))):
    text = open(md, encoding="utf-8").read()
    title = fm_get(text, "title") or os.path.splitext(os.path.basename(md))[0]
    platform = (fm_get(text, "platform") or "").upper()
    cur = fm_get(text, "library_path") or ""
    plat = PLAT.get(platform)
    note = os.path.basename(md)
    if not plat or plat not in idx:
        continue
    def write_path(newpath):
        if not APPLY: return
        new = re.sub(r"(?m)^(library_path:\s*).*$", lambda m: m.group(1) + newpath, text, count=1)
        # Update body copies ONLY when the old value was an emulation-rom path;
        # never touch /storage1/labdata/... served-files links.
        if cur and cur.startswith(("roms/", "/storage1/Emulation/")):
            new = new.replace(cur, newpath)
        open(md, "w", encoding="utf-8", newline="").write(new)

    nk = norm(title)
    # 1) exact match within the declared platform (auto-apply)
    cand = idx[plat].get(nk) or (idx[plat].get(norm(os.path.basename(cur))) if cur else None)
    if cand and len(cand) == 1:
        newpath = f"roms/{plat}/{cand[0]}"
        if cur == newpath: unchanged += 1; continue
        changed += 1; report["CHANGED"].append((note, cur, newpath)); write_path(newpath); continue
    if cand and len(cand) > 1:
        ambiguous += 1; report["AMBIGUOUS"].append((note, cand)); continue
    # 2) cross-platform exact, then fuzzy substring -> PROPOSAL (apply only with --fuzzy)
    prop = find_exact_all(nk) or find_fuzzy(nk)
    if len(prop) == 1:
        p2, fn = prop[0]; newpath = f"roms/{p2}/{fn}"
        if cur == newpath: unchanged += 1; continue
        report["FUZZY"].append((note, cur, newpath))
        if FUZZY and note not in FUZZY_HOLD: changed += 1; write_path(newpath)
        continue
    unmatched += 1; report["UNMATCHED"].append((note, platform, cur))

summary = (f"{'APPLIED' if APPLY else 'DRY-RUN'}{' +FUZZY' if FUZZY else ''}: {changed} changed, "
           f"{unchanged} already correct, {len(report['FUZZY'])} fuzzy proposals, "
           f"{unmatched} unmatched (no on-disk file — likely Wanted/gated), {ambiguous} ambiguous")
lines = [summary, ""]
for tag in ("AMBIGUOUS", "FUZZY", "CHANGED", "UNMATCHED"):
    rows = report[tag]
    if not rows: continue
    lines.append(f"--- {tag} ({len(rows)}) ---")
    for r in rows:
        if tag in ("CHANGED", "FUZZY"): lines.append(f"  {r[0]}\n      {r[1] or '—'}  ->  {r[2]}")
        elif tag == "AMBIGUOUS": lines.append(f"  {r[0]}  -> {r[1]}")
        else: lines.append(f"  {r[0]} [{r[1]}]  (cur: {r[2] or '—'})")
    lines.append("")
rep = os.path.join(os.path.dirname(__file__), "..", ".scrape", "resync-report.txt")
open(rep, "w", encoding="utf-8").write("\n".join(lines))
print(summary)
print("full report ->", os.path.normpath(rep))