GaCy-GO/tools/export_scene_metadata.py

"""Export complete RuntimeFloor group captures from a Client Player.log.

The capture must include FLOOR/GROUP/END records from a read-only Client
GameCoreConfigLoader probe. Inputs and Client revision are explicit arguments.
"""
import argparse
import hashlib
import json
import pathlib
import re

parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("log", type=pathlib.Path)
parser.add_argument("--assembly", type=pathlib.Path, required=True)
parser.add_argument("--client-version", required=True)
parser.add_argument("--output", type=pathlib.Path, required=True)
parser.add_argument("--merge", action="store_true", help="Preserve compatible previously captured floors in output")
parser.add_argument("--dimension-index", type=int, help="Use captured SavedValues defaults from this RuntimeFloor dimension-list index")
parser.add_argument("--omit-start-anchor", action="store_true", help="Keep per-entrance Anchor.json placements; retain the floor anchor only as floor_default_anchor for entries without StartAnchorID")
args = parser.parse_args()
floors, completed = {}, set()
placements = {}
placement_completed = set()
props_completed = set()
dimensions = {}
dimension_groups = {}
for line in args.log.read_text(encoding="utf-8", errors="replace").splitlines():
    if "[GAC_SCENE_DIMENSION] " in line:
        row = line.split("[GAC_SCENE_DIMENSION] ", 1)[1]
        groups = re.fullmatch(r"GROUPS (\d+) (\d+) (\d+) (.+)", row)
        if groups:
            floor, index, identity, raw = groups.groups()
            dimension_groups.setdefault(floor, {})[int(index)] = (int(identity), json.loads(raw))
        continue
    if "[GAC_SCENE_SAVED] " in line:
        row = line.split("[GAC_SCENE_SAVED] ", 1)[1]
        dimension = re.fullmatch(r"DIMENSION (\d+) (\d+) (.+)", row)
        if dimension:
            floor, index, raw = dimension.groups()
            dimensions.setdefault(floor, {})[int(index)] = json.loads(raw)
        continue
    if "[GAC_SCENE_PLACEMENT] " in line:
        row = line.split("[GAC_SCENE_PLACEMENT] ", 1)[1]
        end_placement = re.fullmatch(r"END (\d+)", row)
        if end_placement:
            placement_completed.add(end_placement[1])
        position = re.fullmatch(r"(ANCHOR|MONSTER) (\d+) (\d+)(?: (\d+))? pos=([^,]+),([^,]+),([^ ]+) rot=(.+)", row)
        if position:
            kind, floor, first_id, inst_id, x, y, z, rot = position.groups()
            # Unity metres/degrees -> protocol position/rotation units.
            value = {"pos": {"x": round(float(x)*1000), "y": round(float(y)*1000), "z": round(float(z)*1000)}, "rot": {"x": 0, "y": round(float(rot)*1000), "z": 0}}
            target = placements.setdefault(floor, {"monsters": {}})
            if kind == "ANCHOR":
                value["ID"] = int(first_id)
                target["start_anchor"] = value
            else:
                target["monsters"].setdefault(first_id, {})[inst_id] = value
        continue
    if "[GAC_SCENE_PROPS] " in line:
        row = line.split("[GAC_SCENE_PROPS] ", 1)[1]
        end_props = re.fullmatch(r"END (\d+)", row)
        prop = re.fullmatch(r"PROP (\d+) (\d+) (\d+) (\d+) (\d+) pos=([^,]+),([^,]+),([^ ]+) rot=(.+)", row)
        if end_props:
            props_completed.add(end_props[1])
        elif prop:
            floor, group_id, inst_id, prop_id, state, x, y, z, rot = prop.groups()
            if floor in floors:
                value = {
                    "pos": {"x": round(float(x) * 1000), "y": round(float(y) * 1000), "z": round(float(z) * 1000)},
                    "rot": {"x": 0, "y": round(float(rot) * 1000), "z": 0},
                    "groupId": int(group_id),
                    "instId": int(inst_id),
                    "propState": int(state),
                    "propId": int(prop_id),
                }
                floors[floor].setdefault("props", {}).setdefault(group_id, []).append(value)
        continue
    if "[GAC_SCENE_META]" not in line:
        continue
    row = line.split("[GAC_SCENE_META] ", 1)[1]
    start = re.fullmatch(r"FLOOR (\d+) (\d+) startGroup=(\d+) anchor=(\d+) count=(\d+)", row)
    group = re.fullmatch(r"GROUP (\d+) (\d+) side=(\w+): \d+ hoyo=(true|false) deleted=(true|false)", row)
    end = re.fullmatch(r"END (\d+)", row)
    if start:
        plane, floor, start_group, anchor, count = start.groups()
        floors[floor] = {"plane_id": int(plane), "group_count": int(count), "groups": {}, "props": {}}
        completed.discard(floor)
    elif group:
        floor, group_id, side, hoyo, deleted = group.groups()
        if floor in floors:
            floors[floor]["groups"][group_id] = {"load_side": side, "is_hoyo_group": hoyo == "true", "is_deleted": deleted == "true"}
    elif end:
        completed.add(end[1])
if not floors or any(key not in completed or key not in props_completed or len(value["groups"]) != value["group_count"] for key, value in floors.items()):
    raise SystemExit("Incomplete scene metadata capture; no output written")
for key, value in placements.items():
    if key in floors and key in placement_completed:
        floors[key].update(value)
for key, floor in floors.items():
    if args.omit_start_anchor:
        anchor = floor.pop("start_anchor", None)
        if anchor is not None:
            floor["floor_default_anchor"] = anchor
    if args.dimension_index is not None:
        rows = dimensions.get(key, {}).get(args.dimension_index)
        if not isinstance(rows, list):
            raise SystemExit(f"Missing SavedValues capture for floor {key}, dimension index {args.dimension_index}; no output written")
        defaults, rules = {}, {}
        for row in rows:
            if row.get("IsDelete"):
                continue
            name = row["Name"]
            value, minimum, maximum = row["DefaultValue"], row["MinValue"], row["MaxValue"]
            allowed = row.get("AllowedValues", [])
            if name in defaults or not name or not minimum <= value <= maximum or (allowed and value not in allowed):
                raise SystemExit(f"Invalid captured SavedValue {key}/{name}; no output written")
            defaults[name] = value
            rules[name] = {"DefaultValue": value, "MinValue": minimum, "MaxValue": maximum}
        floor["saved_values"], floor["saved_value_rules"] = defaults, rules
        selected = dimension_groups.get(key, {}).get(args.dimension_index)
        if selected is None:
            raise SystemExit(f"Missing dimension group membership for floor {key}; no output written")
        identity, ids = selected
        if not ids or any(str(group) not in floor["groups"] for group in ids):
            raise SystemExit(f"Invalid dimension group membership for floor {key}; no output written")
        floor["dimension_id"] = identity
        floor["dimension_groups"] = {str(identity): sorted(set(ids))}
result = {"client_version": args.client_version, "source": {
    "kind": "Client GameCoreConfigLoader.LoadRtLevelFloorInfo / GroupInstanceCommonMap",
    "assembly_sha256": hashlib.file_digest(args.assembly.open("rb"), "sha256").hexdigest(),
    "capture_sha256": hashlib.sha256(args.log.read_bytes()).hexdigest(),
}, "floors": floors}
if args.merge and args.output.exists():
    previous = json.loads(args.output.read_text(encoding="utf-8"))
    if previous["client_version"] != result["client_version"] or previous["source"]["assembly_sha256"] != result["source"]["assembly_sha256"]:
        raise SystemExit("Cannot merge metadata from different Clients")
    previous["floors"].update(floors)
    result["floors"] = previous["floors"]
    # Keep provenance and policies for untouched floors when adding a capture.
    result["source"] = {**previous["source"], **result["source"]}
    result["source"]["previous_capture_sha256"] = previous["source"].get("previous_capture_sha256", []) + [previous["source"]["capture_sha256"]]
args.output.write_text(json.dumps(result, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
print(f"Exported {len(floors)} floors to {args.output}")