import zipfile, io, json, tomllib, re, pickle, os, glob
def lenient_json(b):
    s=b.decode("utf-8-sig",errors="replace")
    try: return json.loads(s)
    except Exception:
        s2=re.sub(r'//[^\n]*','',s); s2=re.sub(r',(\s*[}\]])',r'\1',s2)
        try: return json.loads(s2)
        except Exception: return None
mods={}   # modid -> dict(jar, version, deps, nested)
files=[]  # (jar, path, json) for data recipe/tags + lang
lang={}; models=set(); errors=[]
def walk(z, jarname, nested=False):
    names=z.namelist()
    for t in ("META-INF/neoforge.mods.toml","META-INF/mods.toml"):
        if t in names:
            raw=z.read(t).decode("utf-8",errors="replace")
            try: m=tomllib.loads(raw)
            except Exception as e: errors.append((jarname,"toml",str(e))); m={}
            for md in m.get("mods",[]):
                mid=md.get("modId"); ver=str(md.get("version"))
                if ver.startswith("${"):
                    mf=z.read("META-INF/MANIFEST.MF").decode(errors="replace") if "META-INF/MANIFEST.MF" in names else ""
                    mm=re.search(r'Implementation-Version: *(\S+)',mf); ver=mm.group(1) if mm else ver
                deps=m.get("dependencies",{}).get(mid,[])
                mods.setdefault(mid,[]).append(dict(jar=jarname,version=ver,deps=deps,nested=nested,file=t))
            break
    for n in names:
        if n.startswith("META-INF/jarjar/") and n.endswith(".jar"):
            try: walk(zipfile.ZipFile(io.BytesIO(z.read(n))), jarname+"!"+n.split("/")[-1], True)
            except Exception as e: errors.append((jarname,n,str(e)))
        elif n.startswith("data/") and n.endswith(".json"):
            p=n.split("/")
            if len(p)>=4 and (p[2] in("recipe","recipes") or (p[2]=="tags" and p[3] in("item","items","block","blocks"))):
                j=lenient_json(z.read(n))
                if j is None: errors.append((jarname,n,"badjson"))
                files.append((jarname,n,j))
        elif n.startswith("assets/") and n.endswith("lang/en_us.json"):
            j=lenient_json(z.read(n))
            if isinstance(j,dict):
                for k,v in j.items():
                    if k.startswith(("item.","block.")) and k.count(".")==2: lang.setdefault(k,[]).append((jarname,v))
        elif n.startswith("assets/") and "/models/item/" in n and n.endswith(".json"):
            p=n.split("/"); models.add(p[1]+":"+"/".join(p[4:])[:-5])
jars=["base_minecraft.jar","base_neoforge.jar"]+sorted(glob.glob("jars/*.jar"))
for j in jars: walk(zipfile.ZipFile(j), os.path.basename(j))
pickle.dump(dict(mods=mods,files=files,lang=lang,models=models,errors=errors),open("data.pkl","wb"))
print(len(mods),len(files),len(lang),len(models),len(errors))
for e in errors[:20]: print(e)