Server-side chunk uploader.
This package provides PHP the ability to upload possibly ultra-big file in smaller chunks.
But why? What's wrong with standard HTTP form?
Some primary considerations:
pros:
- No worry about RAM and other server intrinsic limitations.
- If configured properly, no worry about
upload_max_filesize
and consequentlymax_execution_time
directives. Filesize limit may still come from disk space or filesystem limitation. - Each chunk can be processed for, e.g. fingerprinting, so there's a way to fail early when file is corrupt in transit.
- Easily configurable to work with your other web app routers.
- Works on plain PHP. No web server tweaks necessary.
cons:
- Network overhead explodes because multiple requests must be made to upload even a single file.
- A special client must be crafted. Standard HTTP upload form will not work.
$ composer require bfitech/zapchupload
$ vim index.php
Quick index.php
setup:
<?php
require_once __DIR__ . '/vendor/autoload.php';
use BFITech\ZapCore\Router;
use BFITech\ZapCore\Logger;
use BFITech\ZapChupload\ChunkUpload;
// create a logging service
$log = new Logger;
// create a router
$core = (new Router)->config('logger', $log);
// instantiate the chunk uploader class
$chup = new ChunkUpload(
$core, '/tmp/tempdir', '/tmp/destdir',
null, null, null, $log);
// uploader route
$core->route('/upload', [$chup, 'upload'], 'POST');
// downloader route for testing
$core->route('/', function($args) use($core) {
$file = $args['get']['file'] ?? null;
if ($file)
$core->static_file('/tmp/destdir/' . $file);
$core::halt('HELLO WORLD');
});
// that's it
You can run it on your local machine with builtin server as follows:
$ php -S 0.0.0.0:9999 &
Here's a simple client written in Python:
#!/usr/bin/env python3
# chupload-client.py
import os
import sys
import stat
import hashlib
# use pip3 to install this
import requests
# from your toy service
UL_URL = 'http://localhost:9999/upload'
# from your toy service
DL_URL = 'http://localhost:9999/?file='
# ChunkUpload default prefix
PREFIX = "__chupload_"
# ChunkUpload default chunk size
CHUNK_SIZE = 1024 * 100
def upload(path):
try:
fst = os.stat(path)
except FileNotFoundError:
sys.exit(2)
if stat.S_ISDIR(fst.st_mode):
sys.exit(3)
size = fst.st_size
base = os.path.basename(path)
# chupload must have this as _POST data
data = {
PREFIX + 'index': 0,
PREFIX + 'size': size,
PREFIX + 'name': base,
}
# calculate max number of chunks for test
chunk_max = divmod(size, CHUNK_SIZE)[0]
index = 0
with open(path, 'rb') as fhn:
while True:
# chupload must have this as uploaded chunk, where the
# filename doesn't matter
files = {
PREFIX + 'blob': ('noop', fhn.read(CHUNK_SIZE)),
}
# make request on each chunk
resp = requests.post(UL_URL, files=files, data=data).json()
assert(resp['errno'] == 0)
rdata = resp['data']
# upload complete
if rdata['done'] == True:
assert(index == chunk_max == rdata['index'])
print("UPLOAD: OK")
return "%s%s" % (DL_URL, rdata['path'])
# increment index for next chunk
index += 1
data[PREFIX + 'index'] = index
raise Exception("Oops! Something went wrong.")
def compare(path, url):
resp = requests.get(url)
assert(resp.status_code == 200)
print("DOWNLOAD: OK")
# compare download with local file
rhash = hashlib.sha256(resp.content)
lhash = hashlib.sha256(open(path, 'rb').read())
assert(rhash.hexdigest() == lhash.hexdigest())
print("COMPARE: OK")
if __name__ == '__main__':
try:
path = sys.argv[1]
except IndexError:
sys.exit(1)
compare(path, upload(path))
that you can run from the CLI as follows:
$ python3 chupload-client.py ~/some-file.dat || echo FAIL
To see how it works on the browser, run the demo:
$ php -S 0.0.0.0:9998 -t ./demo &
$ x-www-browser localhost:9998
Complete documentation is available with:
$ doxygen
$ x-www-browser docs/html/index.html