arkparser 0.7.2__tar.gz → 0.7.4__tar.gz

{arkparser-0.7.2 → arkparser-0.7.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkparser
-Version: 0.7.2
+Version: 0.7.4
 Summary: Pure Python parser for ARK: Survival Evolved and ARK: Survival Ascended save files
 Author: Vertyco
 License-Expression: MIT

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/__init__.py

@@ -95,4 +95,4 @@ __all__ = [
     "ArkParseError",
 ]
 
-__version__ = "0.7.2"
+__version__ = "0.7.3"

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/export.py

@@ -2699,12 +2699,18 @@ def _allocate_profiles(
     profiles: list[Profile],
     file_tribe_ids: set[int],
     member_index: dict[int, int],
+    tribe_members: dict[int, set[int]],
 ) -> tuple[dict[int, int], dict[int, list[int]]]:
     """Allocate each profile to one tribe id (ContentContainer.cs:760-790).
 
-    Priority: explicit profile tribe id when it names an existing tribe ->
-    the tribe whose member list contains the player -> a solo tribe keyed on
-    the player id. Returns ``(player_id -> tribe_id, tribe_id -> [player_id])``.
+    Priority: explicit profile tribe id when it names an existing tribe AND
+    that tribe's member roster still lists the player -> the tribe whose
+    member list contains the player -> a solo tribe keyed on the player id.
+    The roster check diverges from live legacy on purpose: the game leaves
+    ``TribeID`` stale in the profile of a member removed while offline, and
+    honoring it unconditionally re-attaches ex-members (rosters then exceed
+    the server tribe cap). The ``.arktribe`` member list is authoritative.
+    Returns ``(player_id -> tribe_id, tribe_id -> [player_id])``.
     """
     profile_tribeid: dict[int, int] = {}
     players_by_tribe: dict[int, list[int]] = {}
@@ -2713,7 +2719,7 @@ def _allocate_profiles(
         if not pid:
             continue
         explicit = _int(prof.raw_tribe_id)
-        if explicit and explicit in file_tribe_ids:
+        if explicit and explicit in file_tribe_ids and pid in tribe_members.get(explicit, ()):
             target = explicit
         elif pid in member_index:
             target = member_index[pid]
@@ -2754,6 +2760,7 @@ def _assemble_tribes(save: t.Any) -> dict[str, t.Any]:
     logs: dict[int, list[str]] = {}
     member_index: dict[int, int] = {}
     member_name: dict[int, str] = {}
+    tribe_members: dict[int, set[int]] = {}
     order: list[int] = []
     for sid, sname in (
         (_UNCLAIMED_TRIBE_ID, "[ASV Unclaimed]"),
@@ -2773,12 +2780,15 @@ def _assemble_tribes(save: t.Any) -> dict[str, t.Any]:
         logs[tid] = log_entries
         for pid, mname in members:
             member_index.setdefault(pid, tid)
+            tribe_members.setdefault(tid, set()).add(pid)
             if mname:
                 member_name.setdefault(pid, mname)
     file_tribe_ids = set(rich) | {_UNCLAIMED_TRIBE_ID, _ABANDONED_TRIBE_ID}
     profiles = _profile_entries(save)
     profile_active = _profile_actives(profiles, save)
-    profile_tribeid, players_by_tribe = _allocate_profiles(profiles, file_tribe_ids, member_index)
+    profile_tribeid, players_by_tribe = _allocate_profiles(
+        profiles, file_tribe_ids, member_index, tribe_members
+    )
     for prof in profiles:
         pid = _int(prof.player_id)
         target = profile_tribeid.get(pid)

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/files/base.py

@@ -196,6 +196,12 @@ class ArkFile(ABC):
         - names_count (int32) - number of name strings to read
         - Names array (names_count strings)
         - Padding (20 bytes): 12 zeros + Int32 properties_offset + 4 zeros
+
+        Preconditions: reader is positioned at the first byte of this header.
+        Postconditions: reader is positioned at the first byte of the *next*
+        object's header; the returned object carries an absolute
+        ``properties_offset``. Nothing between headers is consumed, so headers
+        stay walkable regardless of GUID content.
         """
         if is_asa:
             obj = GameObject(id=obj_id)
@@ -220,18 +226,17 @@ class ArkFile(ABC):
 
             # Read padding/metadata after names:
             # Layout: 12 zero bytes + int32 properties_offset + 4 zero bytes = 20 bytes
-            # The stored properties_offset for v7 points to one byte before the actual
-            # properties start. That byte is a 0x00 terminator in most single-object
-            # files; in multi-object files, that byte is the first byte of the next
-            # object's GUID (non-zero). Either way, actual properties = stored + 1 for v7.
-            # For v6, the stored offset IS the exact properties start (no +1 needed).
+            # The header is fixed-width, so the next object's header begins exactly
+            # here and properties are reached by absolute seek (see
+            # GameObject.load_properties). Never sniff the upcoming byte to decide
+            # whether to advance: for every object but the last that byte is the first
+            # byte of the next object's GUID, which is 0x00 for ~1 in 256 profiles.
             reader.skip(12)  # 12 padding zeros
             stored_props_offset = reader.read_int32()  # stored absolute file offset
             reader.skip(4)  # 4 trailing zeros
-            # Consume the optional 0x00 terminator byte if present (most v7 single-obj files)
-            if version >= 7 and reader.remaining > 0 and reader.peek_bytes(1)[0] == 0x00:
-                reader.skip(1)
 
+            # v7 stores the offset of the 0x00 byte that terminates the header block;
+            # properties begin one byte later. v6 stores the exact properties start.
             obj.properties_offset = stored_props_offset + (1 if version >= 7 else 0)
 
             return obj

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/properties/__init__.py

@@ -16,7 +16,7 @@ Usage:
 """
 
 from .base import Property, PropertyHeader, read_property_header
-from .byte_property import ByteProperty
+from .byte_property import ByteProperty, EnumProperty
 from .compound import ArrayProperty, MapProperty, StructProperty
 from .primitives import (
     BoolProperty,
@@ -64,6 +64,7 @@ __all__ = [
     "SoftObjectProperty",
     # Byte
     "ByteProperty",
+    "EnumProperty",
     # Compound
     "ArrayProperty",
     "StructProperty",

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/properties/byte_property.py

@@ -13,6 +13,7 @@ from __future__ import annotations
 import typing as t
 from dataclasses import dataclass
 
+from ..common.exceptions import ArkParseError
 from .base import Property, PropertyHeader, read_name
 
 if t.TYPE_CHECKING:
@@ -265,3 +266,81 @@ class ByteProperty(Property):
                     enum_name=enum_name,
                     _enum_value=enum_value,
                 )
+
+
+class EnumProperty(ByteProperty):
+    """Unreal scoped-enum property, e.g. a Dragon Horn's ``LinkState``.
+
+    Unreal emits ByteProperty for the legacy ``TEnumAsByte`` and EnumProperty for
+    an ``enum class``. The body is an enum-form ByteProperty's plus a nested tag
+    naming the underlying storage type, so it needs its own reader.
+
+    Kept a distinct type rather than a registry alias because ``type_name`` is
+    load-bearing: GameObject forces ByteProperty groups to keep indexed dict form
+    for stat arrays, while a scoped enum is scalar and should collapse to a bare
+    value. ``__slots__ = ()`` keeps this free per instance.
+    """
+
+    __slots__ = ()
+
+    @property
+    def type_name(self) -> str:
+        return "EnumProperty"
+
+    @classmethod
+    def read(
+        cls,
+        reader: BinaryReader,
+        header: PropertyHeader,
+        is_asa: bool = False,
+        name_table: list[str] | None = None,  # noqa: ARG003 - unused; ASA enums are string-based
+        worldsave_format: bool = False,
+    ) -> EnumProperty:
+        """Read an EnumProperty body (ASA string-based files only).
+
+        Layout, verified byte-exactly against live ASA profiles::
+
+            enum_type_name  (header.position bytes, null-terminated)
+            extra1          (int32, observed 1)
+            blueprint_path  (string, e.g. "/Script/ShooterGame")
+            zeros           (int32)
+            underlying_type (string, e.g. "ByteProperty")   <- absent on ByteProperty
+            zeros2          (int32)                          <- absent on ByteProperty
+            data_size       (int32, counts the enum_value field)
+            flag            (uint8, bit 0 -> an array index follows)
+            enum_value      (string, e.g. "EDragonHornLinkState::Live")
+
+        Preconditions: ``header.position`` is the enum type name's length, which
+        is always > 1 because a scoped enum always names its type.
+        Postconditions: reader sits on the next property header.
+        Failure modes: raises ArkParseError on the ASE and worldsave paths, whose
+        layouts are unverified - EnumProperty appears in no ASE fixture and in no
+        ASA worldsave name table, so guessing a layout there would risk silently
+        desyncing the stream instead of failing loudly.
+        """
+        if worldsave_format or not is_asa:
+            raise ArkParseError(
+                f"EnumProperty {header.name!r} in an "
+                f"{'ASA worldsave' if worldsave_format else 'ASE'} stream: layout unverified. "
+                "Capture the save and confirm the byte layout before enabling this path."
+            )
+        assert header.position > 1, f"EnumProperty {header.name!r} must name its enum type"
+
+        enum_name = reader.read_bytes(header.position)[:-1].decode("latin-1")
+        _extra1 = reader.read_int32()
+        _blueprint_path = reader.read_string()
+        _zeros = reader.read_int32()
+        _underlying_type = reader.read_string()
+        _zeros2 = reader.read_int32()
+        _data_size = reader.read_int32()
+        flag = reader.read_uint8()
+        array_index = reader.read_int32() if flag & 0x01 else 0
+        enum_value = reader.read_string()
+
+        assert enum_value, f"EnumProperty {header.name!r} produced an empty value"
+        return cls(
+            name=header.name,
+            index=array_index,
+            enum_name=enum_name,
+            _enum_value=enum_value,
+        )

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser/properties/registry.py

@@ -15,7 +15,7 @@ from ..common.exceptions import ArkParseError, UnknownPropertyError
 from ..structs import registry as struct_registry
 from . import compound as compound_properties
 from .base import NameTable, Property, PropertyHeader, read_property_header
-from .byte_property import ByteProperty
+from .byte_property import ByteProperty, EnumProperty
 from .compound import ArrayProperty, MapProperty, StructProperty
 from .primitives import (
     BoolProperty,
@@ -51,6 +51,7 @@ _NEEDS_NAME_TABLE: frozenset[str] = frozenset({
     "StructProperty",
     "MapProperty",
     "ByteProperty",
+    "EnumProperty",
     "NameProperty",
     "ObjectProperty",
 })
@@ -78,6 +79,8 @@ PROPERTY_REGISTRY: dict[str, type[Property]] = {
     "SoftObjectProperty": SoftObjectProperty,
     # Byte (can be raw or enum)
     "ByteProperty": ByteProperty,
+    # Scoped enum - same wire layout as an enum-form ByteProperty
+    "EnumProperty": EnumProperty,
     # Compound types
     "ArrayProperty": ArrayProperty,
     "StructProperty": StructProperty,

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: arkparser
-Version: 0.7.2
+Version: 0.7.4
 Summary: Pure Python parser for ARK: Survival Evolved and ARK: Survival Ascended save files
 Author: Vertyco
 License-Expression: MIT

{arkparser-0.7.2 → arkparser-0.7.4}/arkparser.egg-info/SOURCES.txt

@@ -41,6 +41,7 @@ arkparser/structs/registry.py
 arkparser/structs/vectors.py
 tests/test_asa_header_position.py
 tests/test_asa_name_table.py
+tests/test_asa_profile_drop.py
 tests/test_ase_cluster_drift.py
 tests/test_binary_reader.py
 tests/test_binary_reader_layouts.py

{arkparser-0.7.2 → arkparser-0.7.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "arkparser"
-version = "0.7.2"
+version = "0.7.4"
 description = "Pure Python parser for ARK: Survival Evolved and ARK: Survival Ascended save files"
 readme = "README.md"
 license = "MIT"

arkparser-0.7.4/tests/test_asa_profile_drop.py

@@ -0,0 +1,186 @@
+"""Regression tests for the ASA profile drop (~12% of the live PvE playerbase).
+
+Two independent defects made ``Profile.load`` raise on real ASA profiles, and
+because every consumer skips a profile that fails to parse, the affected players
+survived only as tribe-file stubs with a blank EOS id, invisible to anything that
+matches on that id.
+
+1. A terminator-skip heuristic in the ASA object header sniffed the upcoming byte
+   and consumed it when it was 0x00. For every object but the last that byte is
+   the first byte of the *next* object's GUID, so ~1 in 256 profiles drifted one
+   byte and died on a garbage string length.
+2. ``EnumProperty`` (Unreal's scoped enum, carried by the Dragon Horn) had no
+   reader, so any player holding one lost their whole profile.
+
+Fixtures under ``asa_pve_rag_profiles`` are staged off a live server and are
+gitignored, so every test that needs them skips cleanly when they are absent.
+"""
+
+from __future__ import annotations
+
+import typing as t
+from pathlib import Path
+from uuid import UUID
+
+import pytest
+
+from arkparser import Profile
+from arkparser.common.binary_reader import BinaryReader
+from arkparser.common.exceptions import ArkParseError
+from arkparser.properties import ByteProperty, EnumProperty
+from arkparser.properties.base import PropertyHeader
+
+PROJECT_ROOT = Path(__file__).resolve().parent.parent
+ASA_DIR = PROJECT_ROOT / "references" / "local_saves" / "survival_ascended"
+PVE_RAG_DIR = PROJECT_ROOT / "references" / "local_saves" / "asa_pve_rag_profiles"
+
+# Reproduces defect 1 on its own: object 1's GUID starts with 0x00, so the old
+# heuristic ate it. Lives in the regular ASA snapshot, so this guard holds even
+# without the staged live fixtures.
+DRIFT_FIXTURE = ASA_DIR / "ragnarok_wp" / "00025da8dfe647d08e8cc55dc09228df.arkprofile"
+
+# Staged live fixtures, from the handoff. The 3 EnumProperty carriers, and the
+# control that always parsed.
+ENUM_FIXTURES = (
+    "0002b4979928425cbfb0751cf6ba615d",
+    "000233c5118447119bd608975a16e01b",
+    "000258582e594c1f89aa8584833b5d26",
+)
+DRIFT_FIXTURES = (
+    "00025da8dfe647d08e8cc55dc09228df",
+    "000274f09cae4b05a8b819082498d46f",
+    "00028651190a4946aaabe01f0172faca",
+    "00029cac9b9741f489776a58d0f53e19",
+)
+CONTROL_FIXTURE = "0002009b5c8a46a996e2b651c290bcff"
+ALL_FIXTURES = (*ENUM_FIXTURES, *DRIFT_FIXTURES, CONTROL_FIXTURE)
+
+MAX_PROFILES = 100_000
+
+
+def _staged(game_id: str) -> Path:
+    path = PVE_RAG_DIR / f"{game_id}.arkprofile"
+    if not path.is_file():
+        pytest.skip(f"live PvE fixture not staged: {path.name}")
+    return path
+
+
+def _enum_props(profile: Profile) -> list[EnumProperty]:
+    return [p for obj in profile.objects for p in obj.properties if isinstance(p, EnumProperty)]
+
+
+def test_next_object_guid_starting_with_zero_still_parses() -> None:
+    """A 0x00-leading GUID on a following object must not drift the reader.
+
+    This is the whole of defect 1: the header block is fixed-width, so the byte
+    after one header is the next header's first GUID byte, never a terminator to
+    be consumed.
+    """
+    if not DRIFT_FIXTURE.is_file():
+        pytest.skip(f"fixture not present: {DRIFT_FIXTURE}")
+
+    profile = Profile.load(DRIFT_FIXTURE)
+
+    assert len(profile.objects) == 2, "fixture should carry a player object plus a buff"
+    # GUIDs are read little-endian, so the on-disk first byte is bytes_le[0].
+    on_disk_first_byte = UUID(profile.objects[1].guid).bytes_le[0]
+    assert on_disk_first_byte == 0x00, "fixture must keep its 0x00-leading GUID to be a regression"
+    assert profile.unique_id, "player must resolve an EOS id, not a blank stub"
+
+
+def test_every_local_asa_profile_parses() -> None:
+    """No ASA profile in the local snapshot may fail to parse.
+
+    Guards the class of bug rather than one fixture: both defects presented as a
+    small, silent fraction of an otherwise healthy corpus.
+    """
+    if not ASA_DIR.is_dir():
+        pytest.skip(f"ASA snapshot not present: {ASA_DIR}")
+
+    files = sorted(ASA_DIR.glob("*/*.arkprofile"))
+    if not files:
+        pytest.skip("no ASA profiles on disk")
+    assert len(files) < MAX_PROFILES, "profile count exceeded bound"
+
+    failures: list[str] = []
+    for path in files:
+        try:
+            Profile.load(path)
+        except Exception as exc:  # noqa: BLE001 - collecting, so one bad file does not mask the rest
+            failures.append(f"{path.parent.name}/{path.name}: {type(exc).__name__}: {exc}")
+
+    assert not failures, f"{len(failures)} of {len(files)} ASA profiles failed to parse:\n  " + "\n  ".join(
+        failures[:10]
+    )
+
+
+@pytest.mark.parametrize("game_id", ALL_FIXTURES)
+def test_reported_fixture_loads_with_identity(game_id: str) -> None:
+    """Every fixture from the handoff loads and resolves its EOS id.
+
+    The blank id is what actually broke ``.mybase``: a stub carries no id, so the
+    player can never be matched back to their tribe.
+    """
+    profile = Profile.load(_staged(game_id))
+
+    assert profile.unique_id == game_id, f"EOS id should round-trip to the filename stem, got {profile.unique_id!r}"
+    assert profile.level > 0, "a real profile has a level; 0 means a stub"
+
+
+@pytest.mark.parametrize("game_id", ENUM_FIXTURES)
+def test_dragon_horn_enum_property(game_id: str) -> None:
+    """A Dragon Horn's LinkState decodes as a scoped enum.
+
+    Wire layout differs from an enum-form ByteProperty by a nested tag naming the
+    underlying storage type, so this pins the reader, not just the registration.
+    """
+    profile = Profile.load(_staged(game_id))
+
+    enums = _enum_props(profile)
+    assert enums, f"{game_id} should carry a Dragon Horn EnumProperty"
+
+    for prop in enums:
+        assert prop.type_name == "EnumProperty", "must not be reported as a ByteProperty"
+        assert prop.is_enum is True
+        assert prop.enum_name == "EDragonHornLinkState"
+        assert prop.enum_value is not None
+        assert prop.enum_value.startswith("EDragonHornLinkState::"), f"unexpected enum value {prop.enum_value!r}"
+        assert prop.byte_value is None, "a scoped enum carries a name, not a raw byte"
+
+
+def test_enum_property_is_a_byte_property_subclass() -> None:
+    """``isinstance(prop, ByteProperty)`` must keep holding for existing consumers."""
+    assert issubclass(EnumProperty, ByteProperty)
+    assert EnumProperty.__slots__ == (), (
+        "subclass must stay free per instance; ByteProperty is allocated in the millions"
+    )
+
+
+def test_enum_property_rejects_unverified_layouts() -> None:
+    """ASE and worldsave EnumProperty layouts are unverified, so they must fail loudly.
+
+    EnumProperty appears in no ASE fixture and in no ASA worldsave name table.
+    Guessing a layout there would desync the stream silently; raising keeps the
+    failure as loud as it was before the type existed.
+    """
+    header = PropertyHeader(name="LinkState", type_name="EnumProperty", data_size=2, index=0, position=21)
+    reader = BinaryReader.from_bytes(b"\x00" * 64)
+
+    with pytest.raises(ArkParseError, match="layout unverified"):
+        EnumProperty.read(reader, header, is_asa=True, worldsave_format=True)
+    with pytest.raises(ArkParseError, match="layout unverified"):
+        EnumProperty.read(reader, header, is_asa=False)
+
+
+def test_dragon_horn_stream_stays_aligned() -> None:
+    """Properties after the EnumProperty must still decode.
+
+    A wrong body length would not raise here, it would silently shift every later
+    property, so assert on a known field that sits after the Dragon Horn.
+    """
+    profile = Profile.load(_staged(ENUM_FIXTURES[0]))
+
+    assert _enum_props(profile), "fixture must actually carry the enum for this to prove alignment"
+    stats: dict[str, t.Any] = profile._persistent_stats
+    assert stats, "persistent character stats parse after the Dragon Horn, proving the reader stayed aligned"
+    assert profile.level == 105, "known level for this fixture"

{arkparser-0.7.2 → arkparser-0.7.4}/tests/test_export.py

@@ -3,6 +3,7 @@
 import datetime as dt
 import os
 import shutil
+import typing as t
 from pathlib import Path
 
 import pytest
@@ -157,6 +158,39 @@ def test_export_tribes_active_never_future() -> None:
     assert rec["active"] is None
 
 
+def test_allocate_profiles_ignores_stale_profile_tribe_id() -> None:
+    """A profile whose TribeID names a tribe that no longer lists the player
+    in its member roster must NOT be allocated to that tribe (the game leaves
+    TribeID stale in profiles of members removed while offline). It falls
+    through to a solo tribe keyed on the player id."""
+    from arkparser.export import _allocate_profiles
+
+    current = t.cast(Profile, type("P", (), {"player_id": 1, "raw_tribe_id": 555})())
+    stale = t.cast(Profile, type("P", (), {"player_id": 987, "raw_tribe_id": 555})())
+    member_index = {1: 555, 2: 555}
+    tribe_members = {555: {1, 2}}
+    profile_tribeid, players_by_tribe = _allocate_profiles(
+        [current, stale], {555}, member_index, tribe_members
+    )
+    assert profile_tribeid[1] == 555
+    assert profile_tribeid[987] == 987
+    assert players_by_tribe[555] == [1]
+    assert players_by_tribe[987] == [987]
+
+
+def test_allocate_profiles_member_list_still_wins_without_explicit_id() -> None:
+    """A profile with no usable TribeID but present in a tribe's roster is
+    still allocated to that tribe via the member-list fallback."""
+    from arkparser.export import _allocate_profiles
+
+    prof = t.cast(Profile, type("P", (), {"player_id": 2, "raw_tribe_id": 0})())
+    profile_tribeid, players_by_tribe = _allocate_profiles(
+        [prof], {555}, {2: 555}, {555: {2}}
+    )
+    assert profile_tribeid[2] == 555
+    assert players_by_tribe[555] == [2]
+
+
 def test_export_tribe_logs_uses_tribe_parser(ase_tribe_path: Path) -> None:
     tribe = Tribe.load(ase_tribe_path)
     exported = export_tribe_logs(type("Holder", (), {"tribes": [tribe]})())