Merge pull request #92 from lbryio/file-uploads

Add support for file uploads (for Publish page)
This commit is contained in:
Jack Robison 2016-07-20 22:37:08 -04:00 committed by GitHub
commit 276a7f9d2a
3 changed files with 219 additions and 2 deletions

View file

@ -175,6 +175,7 @@ class LBRYDaemon(jsonrpc.JSONRPC):
self.streams = {}
self.known_dht_nodes = KNOWN_DHT_NODES
self.first_run_after_update = False
self.uploaded_temp_files = []
if os.name == "nt":
from lbrynet.winhelpers.knownpaths import get_path, FOLDERID, UserHandle
@ -713,6 +714,13 @@ class LBRYDaemon(jsonrpc.JSONRPC):
else:
return defer.succeed(None)
def _clean_up_temp_files(self):
for path in self.uploaded_temp_files:
try:
os.remove(path)
except OSError:
pass
def _shutdown(self):
log.info("Closing lbrynet session")
log.info("Status at time of shutdown: " + self.startup_status[0])
@ -725,6 +733,8 @@ class LBRYDaemon(jsonrpc.JSONRPC):
if self.lbry_ui_manager.update_checker.running:
self.lbry_ui_manager.update_checker.stop()
self._clean_up_temp_files()
d = self._upload_log(log_type="close", exclude_previous=False if self.first_run else True)
d.addCallback(lambda _: self._stop_server())
d.addErrback(lambda err: True)

View file

@ -12,7 +12,7 @@ from twisted.web import server
from twisted.internet import reactor, defer
from jsonrpc.proxy import JSONRPCProxy
from lbrynet.lbrynet_daemon.LBRYDaemonServer import LBRYDaemonServer
from lbrynet.lbrynet_daemon.LBRYDaemonServer import LBRYDaemonServer, LBRYDaemonRequest
from lbrynet.conf import API_CONNECTION_STRING, API_INTERFACE, API_ADDRESS, API_PORT, \
DEFAULT_WALLET, UI_ADDRESS, DEFAULT_UI_BRANCH, LOG_FILE_NAME
@ -121,7 +121,9 @@ def start():
if args.launchui:
d.addCallback(lambda _: webbrowser.open(UI_ADDRESS))
reactor.listenTCP(API_PORT, server.Site(lbry.root), interface=API_INTERFACE)
lbrynet_server = server.Site(lbry.root)
lbrynet_server.requestFactory = LBRYDaemonRequest
reactor.listenTCP(API_PORT, lbrynet_server, interface=API_INTERFACE)
reactor.run()
if not args.logtoconsole and not args.quiet:

View file

@ -4,6 +4,10 @@ import shutil
import json
import sys
import mimetypes
import mimetools
import tempfile
import time
import cgi
from datetime import datetime
from appdirs import user_data_dir
@ -29,6 +33,184 @@ handler = logging.handlers.RotatingFileHandler(lbrynet_log, maxBytes=2097152, ba
log.addHandler(handler)
log.setLevel(logging.INFO)
class LBRYDaemonRequest(server.Request):
"""
For LBRY specific request functionality. Currently just provides
handling for large multipart POST requests, taken from here:
http://sammitch.ca/2013/07/handling-large-requests-in-twisted/
For multipart POST requests, this populates self.args with temp
file objects instead of strings. Note that these files don't auto-delete
on close because we want to be able to move and rename them.
"""
# max amount of memory to allow any ~single~ request argument [ie: POSTed file]
# note: this value seems to be taken with a grain of salt, memory usage may spike
# FAR above this value in some cases.
# eg: set the memory limit to 5 MB, write 2 blocks of 4MB, mem usage will
# have spiked to 8MB before the data is rolled to disk after the
# second write completes.
memorylimit = 1024*1024*100
# enable/disable debug logging
do_log = True
# re-defined only for debug/logging purposes
def gotLength(self, length):
if self.do_log:
print '%f Headers received, Content-Length: %d' % (time.time(), length)
server.Request.gotLength(self, length)
# re-definition of twisted.web.server.Request.requestreceived, the only difference
# is that self.parse_multipart() is used rather than cgi.parse_multipart()
def requestReceived(self, command, path, version):
from twisted.web.http import parse_qs
if self.do_log:
print '%f Request Received' % time.time()
print self.content
self.content.seek(0,0)
self.args = {}
self.stack = []
self.method, self.uri = command, path
self.clientproto = version
x = self.uri.split(b'?', 1)
if len(x) == 1:
self.path = self.uri
else:
self.path, argstring = x
self.args = parse_qs(argstring, 1)
# cache the client and server information, we'll need this later to be
# serialized and sent with the request so CGIs will work remotely
self.client = self.channel.transport.getPeer()
self.host = self.channel.transport.getHost()
# Argument processing
args = self.args
ctype = self.requestHeaders.getRawHeaders(b'content-type')
if ctype is not None:
ctype = ctype[0]
if self.method == b"POST" and ctype:
mfd = b'multipart/form-data'
key, pdict = cgi.parse_header(ctype)
if key == b'application/x-www-form-urlencoded':
args.update(parse_qs(self.content.read(), 1))
elif key == mfd:
try:
self.content.seek(0,0)
args.update(self.parse_multipart(self.content, pdict))
#args.update(cgi.parse_multipart(self.content, pdict))
except KeyError as e:
if e.args[0] == b'content-disposition':
# Parse_multipart can't cope with missing
# content-dispostion headers in multipart/form-data
# parts, so we catch the exception and tell the client
# it was a bad request.
self.channel.transport.write(
b"HTTP/1.1 400 Bad Request\r\n\r\n")
self.channel.transport.loseConnection()
return
raise
self.content.seek(0, 0)
self.process()
# re-definition of cgi.parse_multipart that uses a single temporary file to store
# data rather than storing 2 to 3 copies in various lists.
def parse_multipart(self, fp, pdict):
if self.do_log:
print '%f Parsing Multipart data: ' % time.time()
rewind = fp.tell() #save cursor
fp.seek(0,0) #reset cursor
boundary = ""
if 'boundary' in pdict:
boundary = pdict['boundary']
if not cgi.valid_boundary(boundary):
raise ValueError, ('Invalid boundary in multipart form: %r'
% (boundary,))
nextpart = "--" + boundary
lastpart = "--" + boundary + "--"
partdict = {}
terminator = ""
while terminator != lastpart:
c_bytes = -1
data = tempfile.NamedTemporaryFile(delete=False)
if terminator:
# At start of next part. Read headers first.
headers = mimetools.Message(fp)
clength = headers.getheader('content-length')
if clength:
try:
c_bytes = int(clength)
except ValueError:
pass
if c_bytes > 0:
data.write(fp.read(c_bytes))
# Read lines until end of part.
while 1:
line = fp.readline()
if not line:
terminator = lastpart # End outer loop
break
if line[:2] == "--":
terminator = line.strip()
if terminator in (nextpart, lastpart):
break
data.write(line)
# Done with part.
if data.tell() == 0:
continue
if c_bytes < 0:
# if a Content-Length header was not supplied with the MIME part
# then the trailing line break must be removed.
# we have data, read the last 2 bytes
rewind = min(2, data.tell())
data.seek(-rewind, os.SEEK_END)
line = data.read(2)
if line[-2:] == "\r\n":
data.seek(-2, os.SEEK_END)
data.truncate()
elif line[-1:] == "\n":
data.seek(-1, os.SEEK_END)
data.truncate()
line = headers['content-disposition']
if not line:
continue
key, params = cgi.parse_header(line)
if key != 'form-data':
continue
if 'name' in params:
name = params['name']
# kludge in the filename
if 'filename' in params:
fname_index = name + '_filename'
if fname_index in partdict:
partdict[fname_index].append(params['filename'])
else:
partdict[fname_index] = [params['filename']]
else:
# Unnamed parts are not returned at all.
continue
data.seek(0,0)
if name in partdict:
partdict[name].append(data)
else:
partdict[name] = [data]
fp.seek(rewind) # Restore cursor
return partdict
class LBRYindex(resource.Resource):
def __init__(self, ui_dir):
@ -179,12 +361,35 @@ class HostedLBRYFile(resource.Resource):
# call.addErrback(lambda err: log.info("Error: " + str(err)))
# call.cancel()
class LBRYFileUpload(resource.Resource):
"""
Accepts a file sent via the file upload widget in the web UI, saves
it into a temporary dir, and responds with a JSON string containing
the path of the newly created file.
"""
def __init__(self, api):
self._api = api
def render_POST(self, request):
origfilename = request.args['file_filename'][0]
uploaded_file = request.args['file'][0] # Temp file created by request
# Move to a new temporary dir and restore the original file name
newdirpath = tempfile.mkdtemp()
newpath = os.path.join(newdirpath, origfilename)
shutil.move(uploaded_file.name, newpath)
self._api.uploaded_temp_files.append(newpath)
return json.dumps(newpath)
class LBRYDaemonServer(object):
def _setup_server(self, wallet):
self.root = LBRYindex(os.path.join(os.path.join(data_dir, "lbry-ui"), "active"))
self._api = LBRYDaemon(self.root, wallet_type=wallet)
self.root.putChild("view", HostedLBRYFile(self._api))
self.root.putChild("upload", LBRYFileUpload(self._api))
self.root.putChild(API_ADDRESS, self._api)
return defer.succeed(True)