Handle symlinks

[JMLX42]

Hello,

I have a ZIP archive created on linux using the --symlinks option.
So my archive contains symlinks.

Yet, when I extract the corresponding files, they end up being text files with the name of the symlink target as their content.

How can I detect/handle symlinks?

Regards,

[JMLX42]

After reading the ZIP specification, I think we need to add stuff here:

https://github.com/mvdnes/zip-rs/blob/master/src/read.rs#L408

I've started refactoring the code like this:

while (reader.position() as usize) < data.len()
    {
        let kind = try!(reader.read_u16::<LittleEndian>());
        let len = try!(reader.read_u16::<LittleEndian>());
        match kind
        {
            // Zip64 extended information extra field
            0x0001 => {
                file.uncompressed_size = try!(reader.read_u64::<LittleEndian>());
                file.compressed_size = try!(reader.read_u64::<LittleEndian>());
                try!(reader.read_u64::<LittleEndian>());  // relative header offset
                try!(reader.read_u32::<LittleEndian>());  // disk start number
            },
            // UNIX Extra Field
            0x000d => {
                let atime = try!(reader.read_u32::<LittleEndian>());
                let mtime = try!(reader.read_u32::<LittleEndian>());
                let uid = try!(reader.read_u16::<LittleEndian>());
                let gid = try!(reader.read_u16::<LittleEndian>());
            }
            _ => { try!(reader.seek(io::SeekFrom::Current(len as i64))); },
        };
    }

Now the problem is it never matches. It doesn't even match the original 0x0001 value.
In my case, I get a only two different values: 21589 or 30837. Which don't match any possible identifier in the specification.

So I'm guessing parse_extra_field() is not functional to begin with?

Regards,

[JMLX42]

Just so you know, I'm now using *.tar.gz files instead of *.zip files.

[cecton]

I would be interested in this library handling symlinks. I might make a PR at some point.

[David-OConnor]

I'm running into the same issue.

[Plecra]

Looks like symlinks are a non-standard, undocumented part of Info-ZIP, which we'd have to tease out of https://sourceforge.net/projects/infozip/files/. Yay.

Afaict, the only real option is to use unix-style symlinks inside the archives, and special case them on other platforms.

[markmmm]

I hacked up the following; throwing this out there in case someone wants to improve it (I don't know if I will get to it)..

The following code works to extract symlinks (for my example zip files) - I can't say whether it will work with other zips..

fn extract_from_zip_file(zip_path: &Path, extract_to_dir: &Path) -> Result<bool> {
    const S_IFLNK: u32 = 0o120000; // symbolic link
    let archive_file = std::fs::File::open(zip_path)?;
    let mut archive = zip::ZipArchive::new(archive_file)?;

    for file_number in 0..archive.len() {
        let mut next = archive.by_index(file_number).unwrap();
        #[allow(deprecated)]
        let sanitized_name = next.sanitized_name();

        if next.is_dir() {
            let extracted_folder_path = extract_to_dir.join(sanitized_name);
            std::fs::create_dir_all(extracted_folder_path).unwrap();
        } else if next.is_file() {
            let extracted_file_path = extract_to_dir.join(sanitized_name);
            // handle links - the link source is in the contents of the compressed file
            // it is not actually compressed
            if let Some(mode) = next.unix_mode() {
                if mode & S_IFLNK == S_IFLNK {
                    let mut contents = Vec::new();
                    next.read_to_end(&mut contents)?;

                    #[cfg(target_family = "unix")]
                    {
                        // Needed to be able to call `OsString::from_vec(Vec<u8>)`
                        use std::os::unix::ffi::OsStringExt as _;

                        let link_path =
                            std::path::PathBuf::from(std::ffi::OsString::from_vec(contents));
                        std::os::unix::fs::symlink(link_path, extracted_file_path)?;
                    }
                    #[cfg(target_family = "windows")]
                    {
                        // TODO: Support non-UTF-8 paths (currently only works for paths which are valid UTF-8)
                        let link_path = String::from_utf8(contents)?;
                        std::os::windows::fs::symlink_file(link_path, extracted_file_path)?;
                    }
                    continue;
                }
            }
            let mut buffer: Vec<u8> = Vec::new();
            let _bytes_read = next.read_to_end(&mut buffer)?;

            // Handle files where they are in a sub-directory but that directory hasn't
            // been created yet (found with __MACOSX/.filename)
            if let Some(parent) = extracted_file_path.parent() {
                if !parent.exists() {
                    std::fs::create_dir_all(parent)?;
                }
            }

            #[cfg(target_family = "unix")]
            {
                use std::fs::OpenOptions;
                use std::os::unix::fs::OpenOptionsExt;
                let mode = next.unix_mode().unwrap() & 0o7777;
                println!("{:?} {:?} ", extracted_file_path, mode);
                let mut file = OpenOptions::new()
                    .mode(mode)
                    .create(true)
                    .write(true)
                    .truncate(true)
                    .open(extracted_file_path)?;
                file.write_all(&buffer)?;
            }
            #[cfg(target_family = "windows")]
            {
                let mut file = std::fs::File::create(extracted_file_path)?;
                file.set_len(0)?;
                file.write_all(&buffer)?;
            }
        }
    }
    Ok(true)
}

[markmmm]

Hi, It would be great to have some eyes on my PR (it's much better than the mess I posted above)
Some things that would be helpful (from someone with better unix than me):

• Is the terminology of source/target correct in my pr
• Do you have instructions on how to create a zip file that has a link before the original file. (I encountered that situation in the files I was processing - which caused my first attempt to fail)

I have currently have a copy of this code running in my own project - it would be nice to switch back to zip-rs, anything I should be doing to help this get merged?

[Plecra]

Some lovely resources abt this linked in the @markmmm's implementation PR: #213 (comment)