chore(romhacks): catalog publisher tooling (analyze + render Obsidian notes)

2026-06-08 04:25:58 -05:00
parent ae57c7f661
commit 8224a8fddb
2 changed files with 248 additions and 0 deletions
@@ -0,0 +1,130 @@
+#!/usr/bin/env python3
+"""Analysis pass for the Obsidian catalog publisher (read-only).
+
+For every routed library/<slug>/ game, derive its catalog fields:
+  - name, platform (UPPERCASE), kind, base, version, status, generation,
+    library_path, source link, #guides, #art.
+Generation comes from which forum the thread lived in (exports/<channelId>/);
+base/version/status are parsed from the staged description.
+
+Emits one JSON object per line (so the caller can dedupe against the existing
+vault notes) plus a summary. Writes nothing outside stdout.
+"""
+import json
+import os
+import pathlib
+import re
+
+LIBRARY = pathlib.Path(os.environ.get("LIBRARY_DIR", "/library"))
+METADATA = pathlib.Path(os.environ.get("METADATA_DIR", "/metadata"))
+EXPORTS = pathlib.Path(os.environ.get("EXPORTS_DIR", "/exports"))
+
+CHANNEL_GEN = {
+    "1474174360926290105": "Gen I", "1474174532196761802": "Gen II",
+    "1474174608587489416": "Gen III", "1474174672210755714": "Gen IV",
+    "1474174736438132756": "Gen V", "1474174789680632084": "Gen VI",
+    "1474174859671240898": "Gen VII", "1474174950138183880": "Gen VIII",
+    "1474175004248903935": "Gen IX", "1474175244209098986": "Joiplay/RPGXP",
+    "1475605095805747351": "Unique",
+}
+PLAT_UP = {"gba": "GBA", "gbc": "GBC", "gb": "GB", "nds": "NDS",
+           "3ds": "3DS", "switch": "Switch", "pc": "PC", "patch": "Patch",
+           "unknown": "—"}
+
+
+def norm(name):
+    n = name.lower()
+    n = re.sub(r"pok[eé]mon", "", n)
+    n = re.sub(r"[^a-z0-9]+", "", n)
+    return n
+
+
+def gen_map():
+    m = {}
+    for chan_dir in (EXPORTS.iterdir() if EXPORTS.exists() else []):
+        if not chan_dir.is_dir():
+            continue
+        gen = CHANNEL_GEN.get(chan_dir.name)
+        if not gen:
+            continue
+        for jf in chan_dir.glob("*.json"):
+            try:
+                d = json.loads(jf.read_text(encoding="utf-8"))
+                nm = d.get("channel", {}).get("name")
+                if nm:
+                    m[norm(nm)] = gen
+            except Exception:
+                pass
+    return m
+
+
+def parse_desc(desc):
+    base = ver = status = "—"
+    m = re.search(r"based on\s*:?\s*([A-Za-z0-9 .&'+-]{2,40})", desc, re.I)
+    if m:
+        base = m.group(1).strip().rstrip(".").strip()
+    m = re.search(r"version\s*:?\s*([^\n]{1,30})", desc, re.I)
+    if m:
+        ver = m.group(1).strip()
+    low = desc.lower()
+    if "completed" in low or re.search(r"\bcomplete\b", low):
+        status = "Complete"
+    elif "beta" in low:
+        status = "Beta"
+    elif "ongoing" in low or "in development" in low:
+        status = "Ongoing"
+    return base, ver, status
+
+
+def main():
+    gens = gen_map()
+    rows = []
+    for folder in sorted(p for p in LIBRARY.iterdir() if p.is_dir()):
+        hj = folder / "handoff.json"
+        if not hj.exists():
+            continue
+        h = json.loads(hj.read_text(encoding="utf-8"))
+        name = h.get("name", folder.name)
+        slug = h.get("slug", folder.name)
+        plat = PLAT_UP.get(h.get("platform", "unknown"), "—")
+        desc = ""
+        mp = METADATA / slug / "metadata.json"
+        if mp.exists():
+            try:
+                desc = json.loads(mp.read_text(encoding="utf-8")).get("description", "")
+            except Exception:
+                pass
+        base, ver, status = parse_desc(desc)
+        gen = gens.get(norm(name), "—")
+        art = h.get("art", [])
+        guides = h.get("guides", [])
+        links = h.get("links", [])
+        artifact = h.get("artifact")
+        kind = h.get("kind")
+        lib_path = (f"/storage1/Emulation/roms/{h.get('platform')}/{artifact}"
+                    if kind == "console" and artifact else
+                    f"/storage1/labdata/romhacks/library/{slug}/")
+        rows.append({
+            "name": name, "slug": slug, "norm": norm(name), "platform": plat,
+            "kind": kind, "base": base, "version": ver, "status": status,
+            "generation": gen, "library_path": lib_path,
+            "source": links[0] if links else None, "n_links": len(links),
+            "n_guides": len(guides), "n_art": len(art), "artifact": artifact,
+        })
+
+    for r in rows:
+        print(json.dumps(r, ensure_ascii=False))
+
+    # summary to stderr-ish (still stdout, prefixed)
+    from collections import Counter
+    pc = Counter(r["platform"] for r in rows)
+    kc = Counter(r["kind"] for r in rows)
+    print("@@@SUMMARY", json.dumps({
+        "total": len(rows), "by_platform": dict(pc), "by_kind": dict(kc),
+        "with_base": sum(1 for r in rows if r["base"] != "—"),
+        "with_gen": sum(1 for r in rows if r["generation"] != "—"),
+    }, ensure_ascii=False))
+
+
+if __name__ == "__main__":
+    main()
@@ -0,0 +1,118 @@
+import json, re, pathlib, datetime
+
+rows = json.load(open("/tmp/catalog_new.json", encoding="utf-8"))
+META = pathlib.Path("/storage1/labdata/romhacks/metadata")
+OUT = pathlib.Path("/tmp/catalog_notes"); OUT.mkdir(exist_ok=True)
+TODAY = "2026-06-08"
+
+BASES = ["HeartGold","SoulSilver","FireRed","LeafGreen","Omega Ruby","Alpha Sapphire",
+         "Black 2","White 2","Emerald","Crystal","Platinum","Diamond","Pearl",
+         "Ruby","Sapphire","Gold","Silver","Black","White","Red","Blue","Yellow",
+         "Sun","Moon","Sword","Shield","Scarlet","Violet"]
+PLAT_TAG = {"GBA":"gba","GBC":"gbc","GB":"gb","NDS":"nds","3DS":"3ds","Switch":"switch","PC":"pc","Patch":"patch","—":"unknown"}
+
+def q(v):
+    return '"' + str(v).replace("\\", "\\\\").replace('"', '\\"') + '"'
+
+def clean_version(v):
+    v = re.sub(r"^[\s*:_-]+", "", str(v or "")).strip()
+    return v or "—"
+
+def strip_prefix(name):
+    return re.sub(r"^\s*pok[eé]mon\s+", "", name, flags=re.I).strip()
+
+def safe_filename(name):
+    return re.sub(r'[\\/:*?"<>|]+', " ", strip_prefix(name)).strip() or "Unknown"
+
+def detect_base(desc, parsed):
+    # Only trust an explicit "based on : <base>" segment — scanning the whole
+    # description gives false positives (e.g. "Black Market", "Red Gyarados").
+    m = re.search(r"based on\s*:?\s*(.{0,40})", desc, re.I)
+    if not m:
+        return None
+    seg = m.group(1)
+    for spaced, canon in [(r"fire\s*red","firered"),(r"leaf\s*green","leafgreen"),
+                          (r"heart\s*gold","heartgold"),(r"soul\s*silver","soulsilver"),
+                          (r"omega\s*ruby","omegaruby"),(r"alpha\s*sapphire","alphasapphire")]:
+        seg = re.sub(spaced, canon, seg, flags=re.I)
+    for b in BASES:
+        if re.search(r"\b" + re.escape(b) + r"\b", seg, re.I):
+            return b
+    return None
+
+def clean_desc(desc):
+    # drop the markdown download links already captured separately; keep prose
+    lines = []
+    for ln in desc.splitlines():
+        if re.search(r"\bdownload\b", ln, re.I) and "http" in ln.lower():
+            continue
+        lines.append(ln)
+    return "\n".join(lines).strip()
+
+def render(r):
+    name = r["name"]; title = "Pokémon " + strip_prefix(name)
+    desc = ""
+    mp = META / r["slug"] / "metadata.json"
+    if mp.exists():
+        try: desc = json.loads(mp.read_text(encoding="utf-8")).get("description","")
+        except Exception: pass
+    base = detect_base(desc, r["base"]) or "—"
+    ptag = PLAT_TAG.get(r["platform"],"unknown")
+    tags = ["hack", f"platform/{ptag}", f"status/{r['status'].lower()}" if r["status"]!="—" else None,
+            f"base/{base.lower().replace(' ','')}" if base!="—" else None,
+            f"kind/{r['kind']}", "source/discord"]
+    tags = [t for t in tags if t]
+    fm = [
+        "---",
+        f"title: {q(title)}",
+        f"platform: {q(r['platform'])}",
+        f"base: {q(base)}",
+        f"version: {q(clean_version(r['version']))}",
+        f"status: {q(r['status'])}",
+        "type: []",
+        f"generation: {q(r['generation'])}",
+        f"library_path: {q(r['library_path'])}",
+        f"source: {q(r['source'] or '—')}",
+        f"added: {TODAY}",
+        "play_status: Unplayed",
+        f"tags: [{', '.join(tags)}]",
+        "---",
+        "",
+        f"# {title}",
+        "",
+    ]
+    kindline = {"console":"Console ROM hack — in RomM.","fangame":"RPG-Maker / Joiplay fan-game (PC) — not in RomM.",
+                "patch":"Patch file — apply to a base ROM.","unknown":"Imported."}.get(r["kind"],"Imported.")
+    fm.append(f"> [!info] Auto-imported from the Discord catalog ({r['generation']}) on {TODAY}. {kindline}")
+    fm.append("")
+    cd = clean_desc(desc)
+    if cd:
+        fm += ["## Summary", "", cd, ""]
+    # download
+    links = []
+    if mp.exists():
+        try: links = json.loads(mp.read_text(encoding="utf-8")).get("links",[])
+        except Exception: pass
+    if links:
+        fm += ["## Download", ""] + [f"- {l}" for l in links] + [""]
+    fm += ["## Files", "",
+           f"- **Artifact:** `{r['artifact'] or 'n/a'}`",
+           f"- **Library:** `/storage1/labdata/romhacks/library/{r['slug']}/`"
+           + (f" — {r['n_guides']} guide/extra file(s)" if r['n_guides'] else "")]
+    fm.append("")
+    return safe_filename(name), "\n".join(fm)
+
+made = []
+for r in rows:
+    fn, body = render(r)
+    (OUT / (fn + ".md")).write_text(body, encoding="utf-8")
+    made.append((fn, r))
+print("rendered", len(made), "notes to /tmp/catalog_notes/")
+
+# show 2 samples: a console GBA hack and a fangame
+for want in ("console","fangame"):
+    for fn, r in made:
+        if r["kind"] == want:
+            print("\n========== SAMPLE (%s): %s.md ==========" % (want, fn))
+            print((OUT / (fn + ".md")).read_text(encoding="utf-8")[:1400])
+            break