transbeam/src/zip.rs

use std::io::Write;
use std::task::Waker;

use crc32fast::Hasher;
use log::debug;
use time::OffsetDateTime;

use crate::file::LiveWriter;
use crate::UploadedFile;

const SIGNATURE_SIZE: usize = 4;
const SHARED_FIELDS_SIZE: usize = 26;
const EXTRA_FIELD_SIZE: usize = 41;
const LOCAL_HEADER_SIZE_MINUS_FILENAME: usize =
    SIGNATURE_SIZE + SHARED_FIELDS_SIZE + EXTRA_FIELD_SIZE;
const DATA_DESCRIPTOR_SIZE: usize = 24;
const FILE_ENTRY_SIZE_MINUS_FILENAME_AND_FILE: usize =
    LOCAL_HEADER_SIZE_MINUS_FILENAME + DATA_DESCRIPTOR_SIZE;

const CENTRAL_DIRECTORY_HEADER_SIZE_MINUS_FILENAME: usize =
    SIGNATURE_SIZE + 2 + SHARED_FIELDS_SIZE + 14 + EXTRA_FIELD_SIZE;

const EOCD64_RECORD_SIZE: usize = 56;
const EOCD64_LOCATOR_SIZE: usize = 20;
const EOCD_RECORD_SIZE: usize = 22;
const EOCD_TOTAL_SIZE: usize = EOCD64_RECORD_SIZE + EOCD64_LOCATOR_SIZE + EOCD_RECORD_SIZE;

const EMPTY_STRING_CRC32: u32 = 0;

fn file_entry_size(file: &UploadedFile) -> usize {
    FILE_ENTRY_SIZE_MINUS_FILENAME_AND_FILE + file.name.len() + file.size
}

fn file_entries_size(files: &[UploadedFile]) -> usize {
    let mut total = 0;
    for file in files.iter() {
        total += file_entry_size(file)
    }
    total
}

fn central_directory_size(files: &[UploadedFile]) -> usize {
    let mut total = 0;
    for file in files.iter() {
        total += CENTRAL_DIRECTORY_HEADER_SIZE_MINUS_FILENAME + file.name.len();
    }
    total
}

fn zipfile_size(files: &[UploadedFile]) -> usize {
    file_entries_size(files) + central_directory_size(files) + EOCD_TOTAL_SIZE
}

fn fat_timestamp(time: OffsetDateTime) -> u32 {
    (((time.year() - 1980) as u32) << 25)
        | ((time.month() as u32) << 21)
        | ((time.day() as u32) << 16)
        | ((time.hour() as u32) << 11)
        | ((time.minute() as u32) << 5)
        | ((time.second() as u32) >> 1)
}

/// Append a value to a byte vector as little-endian bytes
fn append_value(data: &mut Vec<u8>, mut value: u64, len: usize) {
    data.resize_with(data.len() + len, || {
        let byte = value as u8;
        value >>= 8;
        byte
    });
}

fn append_repeated_byte(data: &mut Vec<u8>, byte: u8, count: usize) {
    data.resize(data.len() + count, byte);
}

fn append_0(data: &mut Vec<u8>, count: usize) {
    append_repeated_byte(data, 0, count);
}

fn append_ff(data: &mut Vec<u8>, count: usize) {
    append_repeated_byte(data, 0xff, count);
}

impl UploadedFile {
    /// Returns the fields shared by the ZIP local file header and
    /// central directory file header - "Version needed to extract"
    /// through "Extra field length".
    fn shared_header_fields(&self, hash: Option<u32>) -> Vec<u8> {
        let mut fields = vec![
            45, 0, // Minimum version required to extract: 4.5 for ZIP64 extensions
            0b00001000, 0, // General purpose bit flag: size and CRC-32 in data descriptor
            0, 0, // Compression method: none
        ];
        append_value(&mut fields, fat_timestamp(self.modtime) as u64, 4);
        // Use 0s as a placeholder if the CRC-32 hash isn't known yet
        append_value(&mut fields, hash.unwrap_or(0) as u64, 4);
        // Placeholders for compressed and uncompressed size in ZIP64 record, 4 bytes each
        append_ff(&mut fields, 8);
        append_value(&mut fields, self.name.len() as u64, 2);
        // Extra field length: 32 bytes for zip64, 9 bytes for timestamp
        fields.append(&mut vec![41, 0]);
        fields
    }

    fn extra_field(&self, local_header_offset: usize) -> Vec<u8> {
        let mut field = vec![
            0x01, 0x00, // Zip64 extended information
            28, 0, // 28 bytes of data
        ];
        // Original size and compressed size - if this is in the local
        // header, we're supposed to leave these blank and point to
        // the data descriptor, but I'm assuming it won't hurt to fill
        // them in regardless
        append_value(&mut field, self.size as u64, 8);
        append_value(&mut field, self.size as u64, 8);
        append_value(&mut field, local_header_offset as u64, 8);
        append_0(&mut field, 4); // File starts on disk 0, there's no other disk

        field.append(&mut vec![
            0x55, 0x54, // Extended timestamp
            5, 0,          // 5 bytes of data
            0b00000001, // Flags: Only modification time is present
        ]);
        append_value(&mut field, self.modtime.unix_timestamp() as u64, 4);

        field
    }

    fn local_header(&self, local_header_offset: usize) -> Vec<u8> {
        let mut header = vec![0x50, 0x4b, 0x03, 0x04]; // Local file header signature
        header.append(&mut self.shared_header_fields(None));
        header.append(&mut self.name.clone().into_bytes());
        header.append(&mut self.extra_field(local_header_offset));
        header
    }

    fn central_directory_header(&self, local_header_offset: usize, hash: u32) -> Vec<u8> {
        let mut header = vec![
            // Central directory file header signature
            0x50, 0x4b, 0x01, 0x02,
            // Made by a "DOS" system supporting version 4.5 - if we
            // say it's made by a Unix system, then unzip will expect
            // it to have embedded Unix permission information, and
            // will set all the files to mode 000 when that's not
            // present :/
            45, 0,
        ];
        header.append(&mut self.shared_header_fields(Some(hash)));
        header.append(&mut vec![
            0, 0, // File comment length: 0
            0, 0, // Disk number where file starts: 0
            0, 0, // Internal file attributes: nothing
            0, 0, 0, 0, // External file attributes: nothing
            0xff, 0xff, 0xff,
            0xff, // Relative offset of local file header: placeholder, see ZIP64 data
        ]);
        header.append(&mut self.name.clone().into_bytes());
        header.append(&mut self.extra_field(local_header_offset));
        header
    }

    fn data_descriptor(&self, hash: u32) -> Vec<u8> {
        let mut descriptor = vec![0x50, 0x4b, 0x07, 0x08]; // Data descriptor signature
        append_value(&mut descriptor, hash as u64, 4);
        // Compressed and uncompressed sizes
        append_value(&mut descriptor, self.size as u64, 8);
        append_value(&mut descriptor, self.size as u64, 8);
        descriptor
    }
}

fn end_of_central_directory(files: &[UploadedFile]) -> Vec<u8> {
    let entries_size = file_entries_size(files) as u64;
    let directory_size = central_directory_size(files) as u64;

    let mut eocd = vec![
        0x50, 0x4b, 0x06, 0x06, // EOCD64 record signature
        44,   // Size of remaining EOCD64 record
    ];
    append_0(&mut eocd, 7); // pad out the rest of the size field
    eocd.append(&mut vec![
        45, 0, // Made by a "DOS" system supporting version 4.5, see above
        45, 0, // Minimum version 4.5 to extract
    ]);
    append_0(&mut eocd, 8); // Two 4-byte disk numbers, both 0
                            // Number of central directory records, on this disk and in total
    append_value(&mut eocd, files.len() as u64, 8);
    append_value(&mut eocd, files.len() as u64, 8);
    append_value(&mut eocd, directory_size, 8);
    append_value(&mut eocd, entries_size, 8); // Offset of start of central directory

    eocd.append(&mut vec![0x50, 0x4b, 0x06, 0x07]); // EOCD64 locator signature
    append_0(&mut eocd, 4); // disk number
    append_value(&mut eocd, entries_size + directory_size, 8); // EOCD64 record offset
    append_0(&mut eocd, 4); // total number of disks;

    eocd.append(&mut vec![0x50, 0x4b, 0x05, 0x06]); // EOCD record signature
    append_ff(&mut eocd, 16); // Zip64 placeholders for disk numbers, record counts, and offsets
    append_0(&mut eocd, 2); // Comment length: 0

    eocd
}

pub struct ZipGenerator<'a> {
    files: Vec<UploadedFile>,
    file_index: usize,
    byte_index: usize,
    pending_metadata: Vec<u8>,
    hasher: Hasher,
    hashes: Vec<u32>,
    output: Box<dyn LiveWriter + 'a>,
}

impl<'a> ZipGenerator<'a> {
    pub fn new(files: Vec<UploadedFile>, output: Box<dyn LiveWriter + 'a>) -> Self {
        let mut result = Self {
            files,
            file_index: 0,
            byte_index: 0,
            pending_metadata: vec![],
            hasher: Hasher::new(),
            hashes: vec![],
            output,
        };
        result.start_new_file();
        result
    }

    pub fn total_size(&self) -> usize {
        zipfile_size(&self.files)
    }

    fn finish_file(&mut self) {
        let hash = std::mem::replace(&mut self.hasher, Hasher::new()).finalize();
        self.hashes.push(hash);
        self.pending_metadata
            .append(&mut self.files[self.file_index].data_descriptor(hash));
        debug!(
            "Finishing file entry in zipfile: {}, hash {}",
            self.files[self.file_index].name, hash
        );
        self.file_index += 1;
        self.start_new_file();
    }

    fn start_new_file(&mut self) {
        let mut offset = file_entries_size(&self.files[..self.file_index]);
        while self.file_index < self.files.len() && self.files[self.file_index].size == 0 {
            debug!(
                "Empty file entry in zipfile: {}",
                self.files[self.file_index].name
            );
            self.hashes.push(EMPTY_STRING_CRC32);
            let mut local_header = self.files[self.file_index].local_header(offset);
            let mut data_descriptor =
                self.files[self.file_index].data_descriptor(EMPTY_STRING_CRC32);
            offset += local_header.len() + data_descriptor.len();
            self.file_index += 1;
            self.pending_metadata.append(&mut local_header);
            self.pending_metadata.append(&mut data_descriptor);
        }
        if self.file_index < self.files.len() {
            debug!(
                "Starting file entry in zipfile: {}",
                self.files[self.file_index].name
            );
            self.byte_index = 0;
            self.pending_metadata
                .append(&mut self.files[self.file_index].local_header(offset));
        } else {
            self.finish_zipfile();
        }
    }

    fn finish_zipfile(&mut self) {
        debug!("Writing zipfile central directory");
        let mut offset = 0;
        for (i, file) in self.files.iter().enumerate() {
            debug!(
                "Writing central directory entry: {}, hash {}",
                file.name, self.hashes[i]
            );
            self.pending_metadata
                .append(&mut file.central_directory_header(offset, self.hashes[i]));
            offset += file_entry_size(file);
        }
        debug!("Writing end of central directory");
        self.pending_metadata
            .append(&mut end_of_central_directory(&self.files));
    }
}

impl<'a> LiveWriter for ZipGenerator<'a> {
    fn add_waker(&mut self, waker: Waker) {
        self.output.add_waker(waker);
    }
}

impl<'a> Write for ZipGenerator<'a> {
    fn write(&mut self, mut buf: &[u8]) -> std::io::Result<usize> {
        while !self.pending_metadata.is_empty() {
            let result = self.output.write(self.pending_metadata.as_slice());
            match result {
                Ok(0) | Err(_) => {
                    return result;
                }
                Ok(n) => {
                    self.pending_metadata.drain(..n);
                }
            }
        }
        if self.file_index >= self.files.len() {
            return Ok(0);
        }
        let bytes_remaining = self.files[self.file_index].size - self.byte_index;
        if bytes_remaining < buf.len() {
            buf = &buf[..bytes_remaining];
        }
        let result = self.output.write(buf);
        match result {
            Ok(0) | Err(_) => (),
            Ok(n) => {
                self.hasher.update(&buf[..n]);
                self.byte_index += n;
                if n == bytes_remaining {
                    self.finish_file();
                }
            }
        }
        result
    }

    fn flush(&mut self) -> std::io::Result<()> {
        debug!("Flushing zipfile writer");
        if !self.pending_metadata.is_empty() {
            self.output.write_all(self.pending_metadata.as_slice())?;
            self.pending_metadata.clear();
        }
        self.output.flush()
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    fn test_no_files() {
        let mut output: Vec<u8> = vec![];
        {
            let mut zipgen = ZipGenerator::new(vec![], Box::new(std::io::Cursor::new(&mut output)));
            zipgen.write_all(&[]).unwrap();
            zipgen.flush().unwrap();
        }

        eprintln!("{:?}", &output);
        {
            let mut reader = std::io::BufReader::new(output.as_slice());
            let zipfile = zip::read::read_zipfile_from_stream(&mut reader).unwrap();
            assert!(zipfile.is_none());
        }
        let archive = zip::ZipArchive::new(std::io::Cursor::new(output)).unwrap();
        assert!(archive.is_empty());
    }
}