parse basic file header

This commit is contained in:
Jean-Marie 'Histausse' Mineau 2024-01-14 17:14:35 +01:00
parent 519d0748be
commit ea18640a78
Signed by: histausse
GPG key ID: B66AEEDA9B645AD2

View file

@ -7,6 +7,122 @@ use androscalpel_serializer::{ReadSeek, Result, Serializable};
//struct Signature(pub [u8; 4]);
struct Signature(pub u32);
#[derive(Debug, Clone, PartialEq, Eq)]
struct FileHeader {
// signature: Signature(0x02014b50)
version_made_by: u16,
version_needed_to_extract: u16,
general_purpose_flag: u16,
compression_method: u16,
last_mod_file_time: u16,
last_mod_file_data: u16,
crc_32: u32,
compressed_size: u32,
uncompressed_size: u32,
// file_name_length: u16,
// extra_field_length: u16,
// file_comment_length: u16,
disk_number_start: u16,
internal_file_attributes: u16,
external_file_attributes: u32,
offset_local_header: u32,
file_name: Vec<u8>,
extra_field: Vec<u8>, // TODO: zip64
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_flag.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.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.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_flag = u16::deserialize(input)?;
let compression_method = u16::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![];
let mut extra_field = vec![];
let mut file_comment = vec![];
for _ in 0..file_name_length {
file_name.push(u8::deserialize(input)?);
}
for _ in 0..extra_field_length {
extra_field.push(u8::deserialize(input)?);
}
for _ in 0..file_comment_length {
file_comment.push(u8::deserialize(input)?);
}
Ok(Self {
version_made_by,
version_needed_to_extract,
general_purpose_flag,
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,
file_comment,
})
}
fn size(&self) -> usize {
Self::MIN_SIZE + self.file_name.len() + self.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;
}
#[derive(Debug, Clone, PartialEq, Eq)]
struct Zip64EndCentralDirectory {
// signature: Signature
@ -164,7 +280,8 @@ impl Serializable for EndCentralDirectory {
struct ZipFile<T: Read + Seek> {
end_of_central_directory: EndCentralDirectory,
end_of_central_directory_off: u64,
zip64_end_of_central_directory: Option<Zip64EndCentralDirectory>,
files: Vec<FileHeader>,
data: T,
}
@ -183,7 +300,7 @@ impl<T: Read + Seek> ZipFile<T> {
))
.unwrap();
let zip64_ecd_locator = Zip64EndCentralDirectoryLocator::deserialize(&mut reader).ok();
if let Some(zip64_ecd_locator) = zip64_ecd_locator {
let zip64_end_of_central_directory = if let Some(zip64_ecd_locator) = zip64_ecd_locator {
assert_eq!(
zip64_ecd_locator.disk_number_of_zip64_end_central_directory_start,
0
@ -195,12 +312,93 @@ impl<T: Read + Seek> ZipFile<T> {
reader.seek(SeekFrom::Start(zip64_edc_record_off)).unwrap();
let zip64_edc_reccord = Zip64EndCentralDirectory::deserialize(&mut reader).ok();
println!("{zip64_edc_reccord:#?}");
}
zip64_edc_reccord
} else {
None
};
Self {
// At this point python's ziplib recompute the location of the central directory from the
// location of the end of central directory in case the zip was concanated after a file.
// We probably don't need that for now.
let mut zip_file = Self {
end_of_central_directory,
end_of_central_directory_off,
zip64_end_of_central_directory,
data: reader,
files: vec![],
};
zip_file
.data
.seek(SeekFrom::Start(zip_file.get_ed_offset()))
.unwrap();
let mut size_read = 0;
let cd_size = zip_file.get_ed_size();
while size_read < cd_size {
let file_header = FileHeader::deserialize(&mut zip_file.data).unwrap();
size_read += file_header.size() as u64;
zip_file.files.push(file_header);
}
assert_eq!(size_read, cd_size);
for f in &zip_file.files {
println!("{f:?}");
}
zip_file
}
/*
fn is_zip64(&self) -> bool {
self.zip64_end_of_central_directory.is_some()
}
*/
fn get_disk_num(&self) -> u32 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.number_of_this_disk
} else {
self.end_of_central_directory.disk_number as u32
}
}
fn get_disk_ed_start(&self) -> u32 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.disk_number_of_central_directory_start
} else {
self.end_of_central_directory
.disk_number_of_central_directory_start as u32
}
}
fn get_number_entries_on_disk(&self) -> u64 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.number_entry_in_central_directory_on_this_disk
} else {
self.end_of_central_directory
.number_of_entries_in_central_directory_on_disk as u64
}
}
fn get_number_entries(&self) -> u64 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.number_entry_in_central_directory
} else {
self.end_of_central_directory
.number_of_entries_in_central_directory as u64
}
}
fn get_ed_size(&self) -> u64 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.size_of_central_directory
} else {
self.end_of_central_directory.size_central_directory as u64
}
}
fn get_ed_offset(&self) -> u64 {
if let Some(zip64_end_of_central_directory) = &self.zip64_end_of_central_directory {
zip64_end_of_central_directory.offset_central_directory
} else {
self.end_of_central_directory.offset_central_directory as u64
}
}