androscalpel/apk_frauder/src/file_header.rs
2025-06-19 10:54:23 +02:00

410 lines
15 KiB
Rust

use log::warn;
use std::io::{SeekFrom, Write};
use crate::compression::CompressionMethod;
use crate::error::Error;
use crate::extra_fields::{ExtraField, GenericExtraField, Zip64ExtraField};
use crate::{cp437, external_file_attributes, general_purpose_flags, Encoding, Signature};
use androscalpel_serializer::{ReadSeek, Result, Serializable};
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct FileHeader {
// signature: Signature(0x02014b50)
pub version_made_by: u16,
pub version_needed_to_extract: u16,
pub general_purpose_flags: u16,
pub compression_method: CompressionMethod,
pub last_mod_file_time: u16,
pub last_mod_file_data: u16,
pub crc_32: u32,
pub compressed_size: u32,
pub uncompressed_size: u32,
// file_name_length: u16,
// extra_field_length: u16,
// file_comment_length: u16,
pub disk_number_start: u16,
pub internal_file_attributes: u16,
pub external_file_attributes: u32,
pub offset_local_header: u32,
pub file_name: Vec<u8>,
pub extra_field: Vec<ExtraField>,
/// Remaining bytes in the extra_fields that could not be parsed as ExtraField
pub malformed_extra_field: Vec<u8>,
pub file_comment: Vec<u8>,
}
impl Serializable for FileHeader {
fn serialize(&self, output: &mut dyn Write) -> Result<()> {
Self::SIGNATURE.serialize(output)?;
self.version_made_by.serialize(output)?;
self.version_needed_to_extract.serialize(output)?;
self.general_purpose_flags.serialize(output)?;
self.compression_method.serialize(output)?;
self.last_mod_file_time.serialize(output)?;
self.last_mod_file_data.serialize(output)?;
self.crc_32.serialize(output)?;
self.compressed_size.serialize(output)?;
self.uncompressed_size.serialize(output)?;
(self.file_name.len() as u16).serialize(output)?;
(self
.extra_field
.iter()
.map(|f| f.size() as u16)
.sum::<u16>()
+ self.malformed_extra_field.len() as u16)
.serialize(output)?;
(self.file_comment.len() as u16).serialize(output)?;
self.disk_number_start.serialize(output)?;
self.internal_file_attributes.serialize(output)?;
self.external_file_attributes.serialize(output)?;
self.offset_local_header.serialize(output)?;
for c in &self.file_name {
c.serialize(output)?;
}
for c in &self.extra_field {
c.serialize(output)?;
}
for c in &self.malformed_extra_field {
c.serialize(output)?;
}
for c in &self.file_comment {
c.serialize(output)?;
}
Ok(())
}
fn deserialize(input: &mut dyn ReadSeek) -> Result<Self> {
let signature = Signature::deserialize(input)?;
assert_eq!(signature, Self::SIGNATURE); // TODO
let version_made_by = u16::deserialize(input)?;
let version_needed_to_extract = u16::deserialize(input)?;
let general_purpose_flags = u16::deserialize(input)?;
let compression_method = CompressionMethod::deserialize(input)?;
let last_mod_file_time = u16::deserialize(input)?;
let last_mod_file_data = u16::deserialize(input)?;
let crc_32 = u32::deserialize(input)?;
let compressed_size = u32::deserialize(input)?;
let uncompressed_size = u32::deserialize(input)?;
let file_name_length = u16::deserialize(input)?;
let extra_field_length = u16::deserialize(input)?;
let file_comment_length = u16::deserialize(input)?;
let disk_number_start = u16::deserialize(input)?;
let internal_file_attributes = u16::deserialize(input)?;
let external_file_attributes = u32::deserialize(input)?;
let offset_local_header = u32::deserialize(input)?;
let mut file_name = vec![];
for _ in 0..file_name_length {
file_name.push(u8::deserialize(input)?);
}
let mut header = Self {
version_made_by,
version_needed_to_extract,
general_purpose_flags,
compression_method,
last_mod_file_time,
last_mod_file_data,
crc_32,
compressed_size,
uncompressed_size,
disk_number_start,
internal_file_attributes,
external_file_attributes,
offset_local_header,
file_name,
extra_field: vec![],
malformed_extra_field: vec![],
file_comment: vec![],
};
//let end_of_extra_field = input.stream_position().unwrap() + extra_field_length as u64;
let extra_field_off = input.stream_position().unwrap();
let mut extra_size_read = 0;
while extra_size_read < extra_field_length as usize {
let field_off = input.stream_position().unwrap();
let field = ExtraField::deserialize(input);
if let Err(err) = field {
warn!(
"Failed to parsed extra field in {}: {err:?}",
header.get_name()
);
input.seek(SeekFrom::Start(field_off)).unwrap();
break;
} else {
let field = field.unwrap();
extra_size_read += field.size();
header.extra_field.push(field);
}
}
if extra_size_read > extra_field_length as usize {
let last_extra = header.extra_field.pop().unwrap();
let size = last_extra.size();
warn!(
"Failed to parse last extra field in {}, last field ({} bytes long) is {} bytes too long",
header.get_name(),
size,
extra_size_read - extra_field_length as usize
);
//warn!("Forgetting last extra field: {:?}", last_extra);
input.seek(SeekFrom::Current(-(size as i64))).unwrap();
}
let mut extra_size_read = input.stream_position().unwrap() - extra_field_off;
while extra_size_read < extra_field_length as u64 {
header.malformed_extra_field.push(u8::deserialize(input)?);
extra_size_read += 1;
}
//input.seek(SeekFrom::Start(end_of_extra_field)).unwrap();
for _ in 0..file_comment_length {
header.file_comment.push(u8::deserialize(input)?);
}
for field in &mut header.extra_field {
if let ExtraField::Generic(GenericExtraField {
id: Zip64ExtraField::ID,
data,
}) = field
{
let original_size = uncompressed_size == u32::MAX;
let compressed_size = compressed_size == u32::MAX;
let offset_header = offset_local_header == u32::MAX;
let disk_number = disk_number_start == u16::MAX;
let zip64_filed = Zip64ExtraField::from_generic(
&GenericExtraField {
id: Zip64ExtraField::ID,
data: data.clone(),
},
original_size,
compressed_size,
offset_header,
disk_number,
)
.unwrap();
*field = ExtraField::Zip64(zip64_filed);
}
}
Ok(header)
}
fn size(&self) -> usize {
Self::MIN_SIZE
+ self.file_name.len()
+ self.extra_field.iter().map(|f| f.size()).sum::<usize>()
+ self.malformed_extra_field.len()
+ self.file_comment.len()
}
}
impl FileHeader {
const SIGNATURE: Signature = Signature(0x02014b50);
const MIN_SIZE: usize = 4 + 6 * 2 + 4 * 3 + 5 * 2 + 4 * 2;
pub fn new_default(name: &str) -> Self {
let mut header = FileHeader {
version_made_by: 768,
version_needed_to_extract: 0,
general_purpose_flags: 0,
compression_method: CompressionMethod::Deflated,
last_mod_file_time: 0,
last_mod_file_data: 0,
crc_32: 0,
compressed_size: 0,
uncompressed_size: 0,
disk_number_start: 0,
internal_file_attributes: 0,
// TODO: why 0b10000000 ?
external_file_attributes: external_file_attributes::REGULAR_FILE
| external_file_attributes::PERM_UR
| external_file_attributes::PERM_UW
| external_file_attributes::PERM_GR
| external_file_attributes::PERM_OR, // TODO
offset_local_header: 0,
file_name: vec![],
extra_field: vec![],
malformed_extra_field: vec![],
file_comment: vec![],
};
header.set_name(name);
header
}
pub fn get_name_encoding(&self) -> Encoding {
if self.general_purpose_flags & general_purpose_flags::MASK_UTF8_FILENAME != 0 {
Encoding::UTF8
} else {
Encoding::CP437
}
}
pub fn get_name(&self) -> String {
match self.get_name_encoding() {
Encoding::UTF8 => std::str::from_utf8(&self.file_name).unwrap().into(),
Encoding::CP437 => cp437::cp437_to_string(&self.file_name),
}
}
pub fn external_file_attributes_set_flag(&mut self, flag: u32) {
self.external_file_attributes |= flag;
}
pub fn external_file_attributes_unset_flag(&mut self, flag: u32) {
self.external_file_attributes &= !flag;
}
pub fn set_file_type(&mut self, file_type: u32) {
self.external_file_attributes &= !external_file_attributes::MASK_FILE_TYPE;
self.external_file_attributes |= file_type;
}
pub fn set_name(&mut self, name: &str) {
let file_name = match cp437::string_to_cp437(name) {
Ok(name) => {
self.general_purpose_flags &= !general_purpose_flags::MASK_UTF8_FILENAME;
name
}
Err(Error::NotCp437) => {
self.general_purpose_flags |= general_purpose_flags::MASK_UTF8_FILENAME;
name.as_bytes().into()
}
};
self.file_name = file_name;
}
pub fn get_uncompressed_size(&self) -> u64 {
if self.uncompressed_size != u32::MAX {
self.uncompressed_size as u64
} else if let Some(ExtraField::Zip64(Zip64ExtraField {
original_size: Some(original_size),
..
})) = self
.extra_field
.iter()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*original_size
} else {
self.uncompressed_size as u64
}
}
pub fn set_uncompressed_size(&mut self, uncompressed_size: u64) {
if let Some(ExtraField::Zip64(Zip64ExtraField {
original_size: Some(original_size),
..
})) = self
.extra_field
.iter_mut()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*original_size = uncompressed_size;
self.uncompressed_size = u32::MAX;
} else if uncompressed_size > u32::MAX as u64 {
self.extra_field.push(ExtraField::Zip64(Zip64ExtraField {
original_size: Some(uncompressed_size),
compressed_size: Some(self.compressed_size as u64),
disk_number: None,
offset_header: None,
}));
self.uncompressed_size = u32::MAX;
self.compressed_size = u32::MAX;
} else {
self.uncompressed_size = uncompressed_size as u32;
}
}
pub fn get_compressed_size(&self) -> u64 {
if self.compressed_size != u32::MAX {
self.compressed_size as u64
} else if let Some(ExtraField::Zip64(Zip64ExtraField {
compressed_size: Some(compressed_size),
..
})) = self
.extra_field
.iter()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*compressed_size
} else {
self.compressed_size as u64
}
}
pub fn set_compressed_size(&mut self, compressed_size: u64) {
if let Some(ExtraField::Zip64(Zip64ExtraField {
compressed_size: Some(compressed_size_),
..
})) = self
.extra_field
.iter_mut()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*compressed_size_ = compressed_size;
self.compressed_size = u32::MAX;
} else if compressed_size > u32::MAX as u64 {
self.extra_field.push(ExtraField::Zip64(Zip64ExtraField {
original_size: Some(self.uncompressed_size as u64),
compressed_size: Some(compressed_size),
disk_number: None,
offset_header: None,
}));
self.uncompressed_size = u32::MAX;
self.compressed_size = u32::MAX;
} else {
self.compressed_size = compressed_size as u32;
}
}
pub fn get_offset_local_header(&self) -> u64 {
if self.offset_local_header != u32::MAX {
self.offset_local_header as u64
} else if let Some(ExtraField::Zip64(Zip64ExtraField {
offset_header: Some(offset_header),
..
})) = self
.extra_field
.iter()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*offset_header
} else {
self.offset_local_header as u64
}
}
pub fn set_offset_local_header(&mut self, offset: u64) {
if let Some(ExtraField::Zip64(Zip64ExtraField {
offset_header: Some(offset_header),
..
})) = self
.extra_field
.iter_mut()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*offset_header = offset;
self.offset_local_header = u32::MAX;
} else if offset > u32::MAX as u64 {
self.extra_field.push(ExtraField::Zip64(Zip64ExtraField {
original_size: Some(self.uncompressed_size as u64),
compressed_size: Some(self.compressed_size as u64),
disk_number: None,
offset_header: Some(offset),
}));
self.uncompressed_size = u32::MAX;
self.compressed_size = u32::MAX;
self.offset_local_header = u32::MAX;
} else {
self.offset_local_header = offset as u32;
}
}
pub fn get_disk_number_start(&self) -> u32 {
if self.disk_number_start != u16::MAX {
self.disk_number_start as u32
} else if let Some(ExtraField::Zip64(Zip64ExtraField {
disk_number: Some(disk_number),
..
})) = self
.extra_field
.iter()
.find(|f| matches!(f, ExtraField::Zip64(_)))
{
*disk_number
} else {
self.disk_number_start as u32
}
}
}