zipstream-ng

A modern and easy to use streamable zip file generator. It can package and stream many files and folders into a zip on the fly without needing temporary files or excessive memory. It can also calculate the final size of the zip file before streaming it.

Features:

• Generates zip data on the fly as it's requested.
• Can calculate the total size of the resulting zip file before generation even begins.
• Low memory usage: Since the zip is generated as it's requested, very little has to be kept in memory (peak usage of less than 20MB is typical, even for TBs of files).
• Performant: On-par or faster than using the standard library to create non-streamed zip files.
• Flexible API: Typical use cases are simple, complicated ones are possible.
• Supports zipping data from files, bytes, strings, and any other iterable objects.
• Keeps track of the date of the most recently modified file added to the zip file.
• Threadsafe: Won't mangle data if multiple threads concurrently add data to the same stream.
• Includes a clone of Python's http.server module with zip support added. Try python -m zipstream.server.
• Automatically uses Zip64 extensions, but only if they are required.
• No external dependencies.

Ideal for web backends:

• Generating zip data on the fly requires very little memory, no disk usage, and starts producing data with less latency than creating the entire zip up-front. This means faster responses, no temporary files, and very low memory usage.
• The ability to calculate the total size of the stream before any data is actually generated (provided no compression is used) means web backends can provide a Content-Length header in their responses. This allows clients to show a progress bar as the stream is transferred.
• By keeping track of the date of the most recently modified file added to the zip, web backends can provide a Last-Modified header. This allows clients to check if they have the most up-to-date version of the zip with just a HEAD request instead of having to download the entire thing.

Installation

pip install zipstream-ng

Examples

Create a local zip file (simple example)

Make an archive named files.zip in the current directory that contains all files under /path/to/files.

from zipstream import ZipStream

zs = ZipStream.from_path("/path/to/files/")

with open("files.zip", "wb") as f:
    f.writelines(zs)

Create a local zip file (demos more of the API)

from zipstream import ZipStream, ZIP_DEFLATED

# Create a ZipStream that uses the maximum level of Deflate compression.
zs = ZipStream(compress_type=ZIP_DEFLATED, compress_level=9)

# Set the zip file's comment.
zs.comment = "Contains compressed important files"

# Add all the files under a path.
# Will add all files under a top-level folder called "files" in the zip.
zs.add_path("/path/to/files/")

# Add another file (will be added as "data.txt" in the zip file).
zs.add_path("/path/to/file.txt", "data.txt")

# Add some random data from an iterable.
# This generator will only be run when the stream is generated.
def random_data():
    import random
    for _ in range(10):
        yield random.randbytes(1024)

zs.add(random_data(), "random.bin")

# Add a file containing some static text.
# Will automatically be encoded to bytes before being added (uses utf-8).
zs.add("This is some text", "README.txt")

# Write out the zip file as it's being generated.
# At this point the data in the files will be read in and the generator
# will be iterated over.
with open("files.zip", "wb") as f:
    f.writelines(zs)

zipserver (included)

A fully-functional and useful example can be found in the included zipstream.server module. It's a clone of Python's built in http.server with the added ability to serve multiple files and folders as a single zip file. Try it out by installing the package and running zipserver --help or python -m zipstream.server --help.

Integration with a Flask webapp

A very basic Flask-based file server that streams all the files under the requested path to the client as a zip file. It provides the total size of the stream in the Content-Length header so the client can show a progress bar as the stream is downloaded. It also provides a Last-Modified header so the client can check if it already has the most recent copy of the zipped data with a HEAD request instead of having to download the file and check.

Note that while this example works, it's not a good idea to deploy it as-is due to the lack of input validation and other checks.

import os.path
from flask import Flask, Response
from zipstream import ZipStream

app = Flask(__name__)

@app.route("/", defaults={"path": "."})
@app.route("/<path:path>")
def stream_zip(path):
    name = os.path.basename(os.path.abspath(path))
    zs = ZipStream.from_path(path)
    return Response(
        zs,
        mimetype="application/zip",
        headers={
            "Content-Disposition": f"attachment; filename={name}.zip",
            "Content-Length": len(zs),
            "Last-Modified": zs.last_modified,
        }
    )

if __name__ == "__main__":
    app.run(host="0.0.0.0", port=5000)

Partial generation and last-minute file additions

It's possible to generate the zip stream, but stop before finalizing it. This enables adding something like a file manifest or compression log after all the files have been added.

ZipStream provides a info_list method that returns information on all the files added to the stream. In this example, all that information will be added to the zip in a file named "manifest.json" before finalizing it.

from zipstream import ZipStream
import json

def gen_zipfile():
    zs = ZipStream.from_path("/path/to/files")
    yield from zs.all_files()
    zs.add(
        json.dumps(
            zs.info_list(),
            indent=2
        ),
        "manifest.json"
    )
    yield from zs.finalize()

Comparison to stdlib

Since Python 3.6 it has actually been possible to generate zip files as a stream using just the standard library, it just hasn't been very ergonomic or efficient. Consider the typical use case of zipping up a directory of files while streaming it over a network connection:

(note that the size of the stream is not pre-calculated in this case as this would make the stdlib example way too long).

Using ZipStream:

from zipstream import ZipStream

send_stream(
    ZipStream.from_path("/path/to/files/")
)

import os
import io
from zipfile import ZipFile, ZipInfo

class Stream(io.RawIOBase):
    """An unseekable stream for the ZipFile to write to"""

    def __init__(self):
        self._buffer = bytearray()
        self._closed = False

    def close(self):
        self._closed = True

    def write(self, b):
        if self._closed:
            raise ValueError("Can't write to a closed stream")
        self._buffer += b
        return len(b)

    def readall(self):
        chunk = bytes(self._buffer)
        self._buffer.clear()
        return chunk

def iter_files(path):
    for dirpath, _, files in os.walk(path, followlinks=True):
        if not files:
            yield dirpath  # Preserve empty directories
        for f in files:
            yield os.path.join(dirpath, f)

def read_file(path):
    with open(path, "rb") as fp:
        while True:
            buf = fp.read(1024 * 64)
            if not buf:
                break
            yield buf

def generate_zipstream(path):
    stream = Stream()
    with ZipFile(stream, mode="w") as zf:
        toplevel = os.path.basename(os.path.normpath(path))
        for f in iter_files(path):
            # Use the basename of the path to set the arcname
            arcname = os.path.join(toplevel, os.path.relpath(f, path))
            zinfo = ZipInfo.from_file(f, arcname)

            # Write data to the zip file then yield the stream content
            with zf.open(zinfo, mode="w") as fp:
                if zinfo.is_dir():
                    continue
                for buf in read_file(f):
                    fp.write(buf)
                    yield stream.readall()
    yield stream.readall()

send_stream(
    generate_zipstream("/path/to/files/")
)

Tests

This package contains extensive tests. To run them, install pytest (pip install pytest) and run py.test in the project directory.

License

Licensed under the GNU LGPLv3.