Add Android Binary XML (ABX) parser

dissect/target/helpers/record.py

@@ -161,6 +161,17 @@ def DynamicDescriptor(types: Sequence[str]) -> RecordDescriptor:
     COMMON_UNIX_FIELDS,
 )
 
+AndroidUserRecord = TargetRecordDescriptor(
+    "android/user",
+    [
+        *COMMON_UNIX_FIELDS,
+        ("string", "usertype"),
+        ("datetime", "last_login"),
+        ("datetime", "last_foreground"),
+        ("varint", "flags"),
+    ],
+)
+
 EmptyRecord = RecordDescriptor(
     "empty",
     [],

dissect/target/plugins/os/unix/linux/android/_os.py

@@ -3,9 +3,10 @@
 from typing import TYPE_CHECKING
 
 from dissect.target.helpers import configutil
-from dissect.target.helpers.record import EmptyRecord
+from dissect.target.helpers.record import AndroidUserRecord
 from dissect.target.plugin import OperatingSystem, export
 from dissect.target.plugins.os.unix.linux._os import LinuxPlugin
+from dissect.target.plugins.os.unix.linux.android.util.abx import ABX, ABXSettingsFile
 
 if TYPE_CHECKING:
     from collections.abc import Iterator
@@ -59,6 +60,19 @@ def create(cls, target: Target, sysvol: Filesystem) -> Self:
 
     @export(property=True)
     def hostname(self) -> str | None:
+        """Return the likely hostname of this Android device."""
+        # Try the first user's ``device_name`` first
+        if (path := self.target.fs.path("/data/system/users/0/settings_global.xml")).is_file():
+            try:
+                return ABXSettingsFile(path).get("device_name")
+            except ValueError:
+                pass
+
+        # Try ``net.hostname``
+        if net_hostname := self.props.get("net.hostname"):
+            return net_hostname
+
+        # Fallback to ``ro.build.host`` property
         return self.props.get("ro.build.host")
 
     @export(property=True)
@@ -82,9 +96,43 @@ def version(self) -> str:
     def os(self) -> str:
         return OperatingSystem.ANDROID.value
 
-    @export(record=EmptyRecord)
-    def users(self) -> Iterator[EmptyRecord]:
-        yield from ()
+    @export(record=AndroidUserRecord)
+    def users(self) -> Iterator[AndroidUserRecord]:
+        """Yield all configured users of this Android system."""
+        if not (users_dir := self.target.fs.path("/data/system/users")).is_dir():
+            return
+
+        for path in users_dir.iterdir():
+            # Parse the /data/system/users/$id.xml ABX file for basic user information.
+            # Currently we do not parse /data/system/users/0/settings_(config|global|secure).xml
+            if not path.is_dir() or not (file := path.parent.joinpath(path.name + ".xml")).is_file():
+                continue
+            try:
+                abx = ABX(file)
+                user = abx.tree.find(".")
+            except ValueError as e:
+                self.target.log.warning("Unable to parse %s: %s", file, e)
+                continue
+
+            id = None
+            name = None
+            try:
+                id = user.attrib["id"]
+                name = user.find("./name").text
+            except Exception:
+                pass
+
+            yield AndroidUserRecord(
+                name=name,
+                uid=id,
+                home=f"/data/media/{id}",
+                last_login=user.attrib["lastLoggedIn"] / 1000,
+                last_foreground=user.attrib["lastEnteredForeground"] / 1000,
+                flags=user.attrib["flags"],
+                usertype=user.attrib["type"],
+                source=file,
+                _target=self.target,
+            )
 
 
 def find_build_props(fs: Filesystem) -> Iterator[Path]:

dissect/target/plugins/os/unix/linux/android/util/abx.py

@@ -0,0 +1,225 @@
+from __future__ import annotations
+
+import base64
+import struct
+from enum import IntEnum
+from pathlib import Path
+from typing import TYPE_CHECKING, Any, BinaryIO
+from xml.etree.ElementTree import Element, ElementTree, SubElement
+
+from dissect.cstruct import u16, u32, u64
+
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
+
+class XmlType(IntEnum):
+    START_DOCUMENT = 0
+    END_DOCUMENT = 1
+    START_TAG = 2
+    END_TAG = 3
+    TEXT = 4
+    CDSECT = 5
+    ENTITY_REF = 6
+    IGNORABLE_WHITESPACE = 7
+    PROCESSING_INSTRUCTION = 8
+    COMMENT = 9
+    DOCDECL = 10
+    ATTRIBUTE = 15
+
+
+class DataType(IntEnum):
+    NULL = 1 << 4
+    STRING = 2 << 4
+    STRING_INTERNED = 3 << 4
+    BYTES_HEX = 4 << 4
+    BYTES_BASE64 = 5 << 4
+    INT = 6 << 4
+    INT_HEX = 7 << 4
+    LONG = 8 << 4
+    LONG_HEX = 9 << 4
+    FLOAT = 10 << 4
+    DOUBLE = 11 << 4
+    BOOLEAN_TRUE = 12 << 4
+    BOOLEAN_FALSE = 13 << 4
+
+
+class ABX:
+    """Android binary XML (ABX) implementation.
+
+    References:
+        - https://www.cclsolutionsgroup.com/post/android-abx-binary-xml
+        - https://cs.android.com/android/platform/superproject/+/master:frameworks/base/core/java/com/android/internal/util/BinaryXmlSerializer.java
+        - https://cs.android.com/android/platform/superproject/+/master:frameworks/base/core/java/com/android/internal/util/BinaryXmlPullParser.java
+    """
+
+    def __init__(self, path: Path | BinaryIO, *, to_str: bool = False) -> None:
+        if isinstance(path, Path):
+            self.path = path
+            self.fh = path.open("rb")
+        elif hasattr(path, "read"):
+            self.path = None
+            self.fh = path
+        else:
+            raise ValueError("Expected Path or file-like object")
+
+        if (magic := self.fh.read(4)) != b"ABX\x00":
+            raise ValueError(f"Unexpected magic value {magic!r}")
+
+        self.READ_MAP: dict[DataType, Callable] = {
+            DataType.NULL: lambda: None,
+            DataType.BOOLEAN_TRUE: lambda: True,
+            DataType.BOOLEAN_FALSE: lambda: False,
+            DataType.INT: lambda: u32(self.fh.read(4), endian="big", sign=True),
+            DataType.INT_HEX: lambda: f"{u32(self.fh.read(4), endian='big', sign=True):x}",
+            DataType.LONG: lambda: u64(self.fh.read(8), endian="big", sign=True),
+            DataType.LONG_HEX: lambda: f"{u64(self.fh.read(8), endian='big', sign=True):x}",
+            DataType.FLOAT: lambda: struct.unpack(">f", self.fh.read(4))[0],
+            DataType.DOUBLE: lambda: struct.unpack(">d", self.fh.read(8))[0],
+            DataType.BYTES_HEX: self._read_bytes,
+            DataType.BYTES_BASE64: lambda: base64.b64encode(self._read_bytes()).decode().strip(),
+            DataType.STRING: self._read_string,
+            DataType.STRING_INTERNED: self._read_string_interned,
+        }
+
+        self.interned_strings = []
+        self.tree = self.read(to_str=to_str)
+
+    def __repr__(self) -> str:
+        return f"<AbxFile path={self.path.name if self.path else self.fh}>"
+
+    def _read_token(self) -> tuple[XmlType, DataType]:
+        """Reads a byte to determine :class:`XmlType` and :class:`DataType`."""
+        token = self.fh.read(1)
+        xml_type = XmlType(token[0] & 0x0F)  # lower nibble
+        data_type = DataType(token[0] & 0xF0)  # upper nibble
+        return xml_type, data_type
+
+    def _read_bytes(self) -> bytes:
+        len = u16(self.fh.read(2), endian="big", sign=False)
+        return self.fh.read(len)
+
+    def _read_string(self) -> str:
+        len = u16(self.fh.read(2), endian="big", sign=False)
+        return self.fh.read(len).decode()
+
+    def _read_string_interned(self) -> str:
+        ref = u16(self.fh.read(2), endian="big", sign=True)
+        if ref == -1:
+            value = self._read_string()
+            self.interned_strings.append(value)
+        else:
+            value = self.interned_strings[ref]
+        return value
+
+    def read(self, to_str: bool = False) -> ElementTree:
+        """Read the ABX file, returns XML :class:`ElementTree`."""
+        elements: list[Element] = []
+        document_open = False
+        root_closed = False
+
+        # Start by placing everything in a root element. If we later discover the document
+        # only has one element in this root, we replace the root element with that element.
+        root = Element("root")
+        elements.append(root)
+
+        while True:
+            xml_type, data_type = self._read_token()
+
+            if xml_type == XmlType.START_DOCUMENT:
+                if data_type != DataType.NULL:
+                    raise ValueError(f"XmlType.START_DOCUMENT should have DataType.NULL, got {data_type!r}")
+                document_open = True
+
+            elif xml_type == XmlType.END_DOCUMENT:
+                if data_type != DataType.NULL:
+                    raise ValueError(f"XmlType.END_DOCUMENT should have DataType.NULL, got {data_type!r}")
+                if len(elements) != 1:
+                    raise ValueError(f"XmlType.END_DOCUMENT with unclosed elements ({elements!r}) at {self.fh.tell()}")
+                if not document_open:
+                    raise ValueError(f"XmlType.END_DOCUMENT before XmlType.START_DOCUMENT at {self.fh.tell()}")
+                break
+
+            elif xml_type == XmlType.START_TAG:
+                if data_type != DataType.STRING_INTERNED:
+                    raise ValueError(f"XmlType.START_TAG should have DataType.STRING_INTERNED, got {data_type!r}")
+                if not document_open:
+                    raise ValueError(f"XmlType.START_TAG before XmlType.START_DOCUMENT at {self.fh.tell()}")
+                if root_closed:
+                    raise ValueError(f"XmlType.START_TAG after XmlType.END_TAG for root element at {self.fh.tell()}")
+
+                name = self._read_string_interned()
+                if len(elements) == 0:
+                    element = Element(name)
+                    root = element
+                else:
+                    element = SubElement(elements[-1], name)
+                elements.append(element)
+
+            elif xml_type == XmlType.END_TAG:
+                if data_type != DataType.STRING_INTERNED:
+                    raise ValueError(f"XmlType.END_TAG should have DataType.STRING_INTERNED, got {data_type!r}")
+                if len(elements) == 1:
+                    raise ValueError(f"XmlType.END_TAG without any elements in stack at {self.fh.tell()}")
+
+                name = self._read_string_interned()
+
+                if name != (other := elements[-1].tag):
+                    raise ValueError(f"XmlType.END_TAG for {name!r} encountered, expected for {other!r}")
+
+                last = elements.pop()
+                if len(elements) == 0:
+                    root_closed = True
+                    root = last
+
+            elif xml_type == XmlType.TEXT:
+                value = self._read_string()
+                if len(elements[-1]):
+                    if len(value.strip()) == 0:
+                        continue
+                    raise ValueError(f"XmlType.TEXT with mixed content encountered at {self.fh.tell()}")
+
+                if elements[-1].text is None:
+                    elements[-1].text = value
+                else:
+                    elements[-1].text += value
+
+            elif xml_type == XmlType.ATTRIBUTE:
+                if len(elements) == 1:
+                    raise ValueError(f"XmlType.ATTRIBUTE encountered outside any open element at {self.fh.tell()}")
+
+                name = self._read_string_interned()
+                if name in elements[-1].attrib:
+                    raise ValueError(
+                        f"Duplicate XmlType.ATTRIBUTE {name} encountered for element {elements[-1]} at {self.fh.tell()}"
+                    )
+
+                reader = self.READ_MAP.get(data_type)
+                if not reader:
+                    raise ValueError(f"Unsupported DataType {data_type!r}")
+
+                value = reader()
+                elements[-1].attrib[name] = str(value) if to_str else value
+
+        if not (root_closed or (len(elements) == 1 and elements[0] is root)):
+            raise ValueError("Document contains unclosed elements")
+
+        # If the root we created at the start only contains one child, we make that node the root.
+        if len(children := root.findall("./")) == 1:
+            root = children[0]
+
+        return ElementTree(root)
+
+
+class ABXSettingsFile(ABX):
+    """Android binary ABX settings file parser."""
+
+    def get(self, key: str, *, value_only: bool = True) -> Any:
+        """Return the value of the given setting name."""
+        if (node := self.tree.find(f"./settings/setting[@name='{key}']")) is not None:
+            return node.attrib["value"] if value_only else node
+
+        return None
+
+    def get_node(self, key: str) -> Any:
+        return self.get(key, value_only=False)

dissect/target/plugins/os/unix/linux/android/util/xml.py

@@ -0,0 +1,23 @@
+from __future__ import annotations
+
+from typing import BinaryIO
+from xml.etree.ElementTree import ElementTree
+
+from defusedxml import ElementTree as ET
+
+from dissect.target.plugins.os.unix.linux.android.util.abx import ABX
+
+
+def read_android_xml(fh: BinaryIO) -> ElementTree:
+    """Convert a readable stream of bytes which might be plaintext XML or ABX to an ElementTree."""
+    offset = fh.tell()
+    magic = fh.read(4)
+    fh.seek(offset)
+
+    if magic == b"ABX\x00":
+        return ABX(fh).tree
+
+    if magic == b"<?xm":
+        return ElementTree(ET.fromstring(fh.read().decode()))
+
+    raise ValueError(f"Unexpected magic {magic!r}")

dissect/target/tools/shell.py

@@ -23,9 +23,14 @@
 from datetime import datetime, timedelta, timezone
 from enum import IntEnum
 from pathlib import Path, PurePosixPath
+from pprint import pprint
 from tarfile import BLKTYPE, CHRTYPE, DIRTYPE, FIFOTYPE, LNKTYPE, REGTYPE, SYMTYPE
 from typing import TYPE_CHECKING, Any, BinaryIO, ClassVar, TextIO
 
+from defusedxml import ElementTree
+
+from dissect.target.plugins.os.unix.linux.android.util.abx import ABX
+
 try:
     # Allow disabling prompt_toolkit via environment variable
     if os.getenv("NO_PROMPT_TOOLKIT", "0") in {"1", "true", "True", "TRUE"}:
@@ -1504,6 +1509,22 @@ def cmd_hexdump(self, args: argparse.Namespace, stdout: TextIO) -> bool:
 
         return False
 
+    @arg("path", type=TargetPathArgument)
+    def cmd_abx(self, args: argparse.Namespace, stdout: TextIO) -> bool:
+        """Print the contents of an Android ABX file."""
+        paths = list(self.resolve_glob_path(args.path))
+        for path in paths:
+            if len(paths) > 1:
+                print(path)
+            try:
+                abx = ABX(path, to_str=True)
+                pprint(ElementTree.tostring(abx.tree.getroot()).decode(), stream=stdout)
+            except ValueError as e:
+                print(f"Failed to parse file {path}: {e}")
+                continue
+
+        return False
+
     @arg("path", type=TargetPathArgument)
     @alias("digest")
     @alias("shasum")