michael@0: #!/usr/bin/env python
michael@0: #
michael@0: # Copyright 2007 Google Inc.
michael@0: #
michael@0: # Licensed under the Apache License, Version 2.0 (the "License");
michael@0: # you may not use this file except in compliance with the License.
michael@0: # You may obtain a copy of the License at
michael@0: #
michael@0: #     http://www.apache.org/licenses/LICENSE-2.0
michael@0: #
michael@0: # Unless required by applicable law or agreed to in writing, software
michael@0: # distributed under the License is distributed on an "AS IS" BASIS,
michael@0: # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
michael@0: # See the License for the specific language governing permissions and
michael@0: # limitations under the License.
michael@0: 
michael@0: """Tool for uploading diffs from a version control system to the codereview app.
michael@0: 
michael@0: Usage summary: upload.py [options] [-- diff_options]
michael@0: 
michael@0: Diff options are passed to the diff command of the underlying system.
michael@0: 
michael@0: Supported version control systems:
michael@0:   Git
michael@0:   Mercurial
michael@0:   Subversion
michael@0: 
michael@0: It is important for Git/Mercurial users to specify a tree/node/branch to diff
michael@0: against by using the '--rev' option.
michael@0: """
michael@0: # This code is derived from appcfg.py in the App Engine SDK (open source),
michael@0: # and from ASPN recipe #146306.
michael@0: 
michael@0: import cookielib
michael@0: import getpass
michael@0: import logging
michael@0: import md5
michael@0: import mimetypes
michael@0: import optparse
michael@0: import os
michael@0: import re
michael@0: import socket
michael@0: import subprocess
michael@0: import sys
michael@0: import urllib
michael@0: import urllib2
michael@0: import urlparse
michael@0: 
michael@0: try:
michael@0:   import readline
michael@0: except ImportError:
michael@0:   pass
michael@0: 
michael@0: # The logging verbosity:
michael@0: #  0: Errors only.
michael@0: #  1: Status messages.
michael@0: #  2: Info logs.
michael@0: #  3: Debug logs.
michael@0: verbosity = 1
michael@0: 
michael@0: # Max size of patch or base file.
michael@0: MAX_UPLOAD_SIZE = 900 * 1024
michael@0: 
michael@0: 
michael@0: def GetEmail(prompt):
michael@0:   """Prompts the user for their email address and returns it.
michael@0: 
michael@0:   The last used email address is saved to a file and offered up as a suggestion
michael@0:   to the user. If the user presses enter without typing in anything the last
michael@0:   used email address is used. If the user enters a new address, it is saved
michael@0:   for next time we prompt.
michael@0: 
michael@0:   """
michael@0:   last_email_file_name = os.path.expanduser("~/.last_codereview_email_address")
michael@0:   last_email = ""
michael@0:   if os.path.exists(last_email_file_name):
michael@0:     try:
michael@0:       last_email_file = open(last_email_file_name, "r")
michael@0:       last_email = last_email_file.readline().strip("\n")
michael@0:       last_email_file.close()
michael@0:       prompt += " [%s]" % last_email
michael@0:     except IOError, e:
michael@0:       pass
michael@0:   email = raw_input(prompt + ": ").strip()
michael@0:   if email:
michael@0:     try:
michael@0:       last_email_file = open(last_email_file_name, "w")
michael@0:       last_email_file.write(email)
michael@0:       last_email_file.close()
michael@0:     except IOError, e:
michael@0:       pass
michael@0:   else:
michael@0:     email = last_email
michael@0:   return email
michael@0: 
michael@0: 
michael@0: def StatusUpdate(msg):
michael@0:   """Print a status message to stdout.
michael@0: 
michael@0:   If 'verbosity' is greater than 0, print the message.
michael@0: 
michael@0:   Args:
michael@0:     msg: The string to print.
michael@0:   """
michael@0:   if verbosity > 0:
michael@0:     print msg
michael@0: 
michael@0: 
michael@0: def ErrorExit(msg):
michael@0:   """Print an error message to stderr and exit."""
michael@0:   print >>sys.stderr, msg
michael@0:   sys.exit(1)
michael@0: 
michael@0: 
michael@0: class ClientLoginError(urllib2.HTTPError):
michael@0:   """Raised to indicate there was an error authenticating with ClientLogin."""
michael@0: 
michael@0:   def __init__(self, url, code, msg, headers, args):
michael@0:     urllib2.HTTPError.__init__(self, url, code, msg, headers, None)
michael@0:     self.args = args
michael@0:     self.reason = args["Error"]
michael@0: 
michael@0: 
michael@0: class AbstractRpcServer(object):
michael@0:   """Provides a common interface for a simple RPC server."""
michael@0: 
michael@0:   def __init__(self, host, auth_function, host_override=None, extra_headers={},
michael@0:                save_cookies=False):
michael@0:     """Creates a new HttpRpcServer.
michael@0: 
michael@0:     Args:
michael@0:       host: The host to send requests to.
michael@0:       auth_function: A function that takes no arguments and returns an
michael@0:         (email, password) tuple when called. Will be called if authentication
michael@0:         is required.
michael@0:       host_override: The host header to send to the server (defaults to host).
michael@0:       extra_headers: A dict of extra headers to append to every request.
michael@0:       save_cookies: If True, save the authentication cookies to local disk.
michael@0:         If False, use an in-memory cookiejar instead.  Subclasses must
michael@0:         implement this functionality.  Defaults to False.
michael@0:     """
michael@0:     self.host = host
michael@0:     self.host_override = host_override
michael@0:     self.auth_function = auth_function
michael@0:     self.authenticated = False
michael@0:     self.extra_headers = extra_headers
michael@0:     self.save_cookies = save_cookies
michael@0:     self.opener = self._GetOpener()
michael@0:     if self.host_override:
michael@0:       logging.info("Server: %s; Host: %s", self.host, self.host_override)
michael@0:     else:
michael@0:       logging.info("Server: %s", self.host)
michael@0: 
michael@0:   def _GetOpener(self):
michael@0:     """Returns an OpenerDirector for making HTTP requests.
michael@0: 
michael@0:     Returns:
michael@0:       A urllib2.OpenerDirector object.
michael@0:     """
michael@0:     raise NotImplementedError()
michael@0: 
michael@0:   def _CreateRequest(self, url, data=None):
michael@0:     """Creates a new urllib request."""
michael@0:     logging.debug("Creating request for: '%s' with payload:\n%s", url, data)
michael@0:     req = urllib2.Request(url, data=data)
michael@0:     if self.host_override:
michael@0:       req.add_header("Host", self.host_override)
michael@0:     for key, value in self.extra_headers.iteritems():
michael@0:       req.add_header(key, value)
michael@0:     return req
michael@0: 
michael@0:   def _GetAuthToken(self, email, password):
michael@0:     """Uses ClientLogin to authenticate the user, returning an auth token.
michael@0: 
michael@0:     Args:
michael@0:       email:    The user's email address
michael@0:       password: The user's password
michael@0: 
michael@0:     Raises:
michael@0:       ClientLoginError: If there was an error authenticating with ClientLogin.
michael@0:       HTTPError: If there was some other form of HTTP error.
michael@0: 
michael@0:     Returns:
michael@0:       The authentication token returned by ClientLogin.
michael@0:     """
michael@0:     account_type = "GOOGLE"
michael@0:     if self.host.endswith(".google.com"):
michael@0:       # Needed for use inside Google.
michael@0:       account_type = "HOSTED"
michael@0:     req = self._CreateRequest(
michael@0:         url="https://www.google.com/accounts/ClientLogin",
michael@0:         data=urllib.urlencode({
michael@0:             "Email": email,
michael@0:             "Passwd": password,
michael@0:             "service": "ah",
michael@0:             "source": "rietveld-codereview-upload",
michael@0:             "accountType": account_type,
michael@0:         }),
michael@0:     )
michael@0:     try:
michael@0:       response = self.opener.open(req)
michael@0:       response_body = response.read()
michael@0:       response_dict = dict(x.split("=")
michael@0:                            for x in response_body.split("\n") if x)
michael@0:       return response_dict["Auth"]
michael@0:     except urllib2.HTTPError, e:
michael@0:       if e.code == 403:
michael@0:         body = e.read()
michael@0:         response_dict = dict(x.split("=", 1) for x in body.split("\n") if x)
michael@0:         raise ClientLoginError(req.get_full_url(), e.code, e.msg,
michael@0:                                e.headers, response_dict)
michael@0:       else:
michael@0:         raise
michael@0: 
michael@0:   def _GetAuthCookie(self, auth_token):
michael@0:     """Fetches authentication cookies for an authentication token.
michael@0: 
michael@0:     Args:
michael@0:       auth_token: The authentication token returned by ClientLogin.
michael@0: 
michael@0:     Raises:
michael@0:       HTTPError: If there was an error fetching the authentication cookies.
michael@0:     """
michael@0:     # This is a dummy value to allow us to identify when we're successful.
michael@0:     continue_location = "http://localhost/"
michael@0:     args = {"continue": continue_location, "auth": auth_token}
michael@0:     req = self._CreateRequest("http://%s/_ah/login?%s" %
michael@0:                               (self.host, urllib.urlencode(args)))
michael@0:     try:
michael@0:       response = self.opener.open(req)
michael@0:     except urllib2.HTTPError, e:
michael@0:       response = e
michael@0:     if (response.code != 302 or
michael@0:         response.info()["location"] != continue_location):
michael@0:       raise urllib2.HTTPError(req.get_full_url(), response.code, response.msg,
michael@0:                               response.headers, response.fp)
michael@0:     self.authenticated = True
michael@0: 
michael@0:   def _Authenticate(self):
michael@0:     """Authenticates the user.
michael@0: 
michael@0:     The authentication process works as follows:
michael@0:      1) We get a username and password from the user
michael@0:      2) We use ClientLogin to obtain an AUTH token for the user
michael@0:         (see http://code.google.com/apis/accounts/AuthForInstalledApps.html).
michael@0:      3) We pass the auth token to /_ah/login on the server to obtain an
michael@0:         authentication cookie. If login was successful, it tries to redirect
michael@0:         us to the URL we provided.
michael@0: 
michael@0:     If we attempt to access the upload API without first obtaining an
michael@0:     authentication cookie, it returns a 401 response and directs us to
michael@0:     authenticate ourselves with ClientLogin.
michael@0:     """
michael@0:     for i in range(3):
michael@0:       credentials = self.auth_function()
michael@0:       try:
michael@0:         auth_token = self._GetAuthToken(credentials[0], credentials[1])
michael@0:       except ClientLoginError, e:
michael@0:         if e.reason == "BadAuthentication":
michael@0:           print >>sys.stderr, "Invalid username or password."
michael@0:           continue
michael@0:         if e.reason == "CaptchaRequired":
michael@0:           print >>sys.stderr, (
michael@0:               "Please go to\n"
michael@0:               "https://www.google.com/accounts/DisplayUnlockCaptcha\n"
michael@0:               "and verify you are a human.  Then try again.")
michael@0:           break
michael@0:         if e.reason == "NotVerified":
michael@0:           print >>sys.stderr, "Account not verified."
michael@0:           break
michael@0:         if e.reason == "TermsNotAgreed":
michael@0:           print >>sys.stderr, "User has not agreed to TOS."
michael@0:           break
michael@0:         if e.reason == "AccountDeleted":
michael@0:           print >>sys.stderr, "The user account has been deleted."
michael@0:           break
michael@0:         if e.reason == "AccountDisabled":
michael@0:           print >>sys.stderr, "The user account has been disabled."
michael@0:           break
michael@0:         if e.reason == "ServiceDisabled":
michael@0:           print >>sys.stderr, ("The user's access to the service has been "
michael@0:                                "disabled.")
michael@0:           break
michael@0:         if e.reason == "ServiceUnavailable":
michael@0:           print >>sys.stderr, "The service is not available; try again later."
michael@0:           break
michael@0:         raise
michael@0:       self._GetAuthCookie(auth_token)
michael@0:       return
michael@0: 
michael@0:   def Send(self, request_path, payload=None,
michael@0:            content_type="application/octet-stream",
michael@0:            timeout=None,
michael@0:            **kwargs):
michael@0:     """Sends an RPC and returns the response.
michael@0: 
michael@0:     Args:
michael@0:       request_path: The path to send the request to, eg /api/appversion/create.
michael@0:       payload: The body of the request, or None to send an empty request.
michael@0:       content_type: The Content-Type header to use.
michael@0:       timeout: timeout in seconds; default None i.e. no timeout.
michael@0:         (Note: for large requests on OS X, the timeout doesn't work right.)
michael@0:       kwargs: Any keyword arguments are converted into query string parameters.
michael@0: 
michael@0:     Returns:
michael@0:       The response body, as a string.
michael@0:     """
michael@0:     # TODO: Don't require authentication.  Let the server say
michael@0:     # whether it is necessary.
michael@0:     if not self.authenticated:
michael@0:       self._Authenticate()
michael@0: 
michael@0:     old_timeout = socket.getdefaulttimeout()
michael@0:     socket.setdefaulttimeout(timeout)
michael@0:     try:
michael@0:       tries = 0
michael@0:       while True:
michael@0:         tries += 1
michael@0:         args = dict(kwargs)
michael@0:         url = "http://%s%s" % (self.host, request_path)
michael@0:         if args:
michael@0:           url += "?" + urllib.urlencode(args)
michael@0:         req = self._CreateRequest(url=url, data=payload)
michael@0:         req.add_header("Content-Type", content_type)
michael@0:         try:
michael@0:           f = self.opener.open(req)
michael@0:           response = f.read()
michael@0:           f.close()
michael@0:           return response
michael@0:         except urllib2.HTTPError, e:
michael@0:           if tries > 3:
michael@0:             raise
michael@0:           elif e.code == 401:
michael@0:             self._Authenticate()
michael@0: ##           elif e.code >= 500 and e.code < 600:
michael@0: ##             # Server Error - try again.
michael@0: ##             continue
michael@0:           else:
michael@0:             raise
michael@0:     finally:
michael@0:       socket.setdefaulttimeout(old_timeout)
michael@0: 
michael@0: 
michael@0: class HttpRpcServer(AbstractRpcServer):
michael@0:   """Provides a simplified RPC-style interface for HTTP requests."""
michael@0: 
michael@0:   def _Authenticate(self):
michael@0:     """Save the cookie jar after authentication."""
michael@0:     super(HttpRpcServer, self)._Authenticate()
michael@0:     if self.save_cookies:
michael@0:       StatusUpdate("Saving authentication cookies to %s" % self.cookie_file)
michael@0:       self.cookie_jar.save()
michael@0: 
michael@0:   def _GetOpener(self):
michael@0:     """Returns an OpenerDirector that supports cookies and ignores redirects.
michael@0: 
michael@0:     Returns:
michael@0:       A urllib2.OpenerDirector object.
michael@0:     """
michael@0:     opener = urllib2.OpenerDirector()
michael@0:     opener.add_handler(urllib2.ProxyHandler())
michael@0:     opener.add_handler(urllib2.UnknownHandler())
michael@0:     opener.add_handler(urllib2.HTTPHandler())
michael@0:     opener.add_handler(urllib2.HTTPDefaultErrorHandler())
michael@0:     opener.add_handler(urllib2.HTTPSHandler())
michael@0:     opener.add_handler(urllib2.HTTPErrorProcessor())
michael@0:     if self.save_cookies:
michael@0:       self.cookie_file = os.path.expanduser("~/.codereview_upload_cookies")
michael@0:       self.cookie_jar = cookielib.MozillaCookieJar(self.cookie_file)
michael@0:       if os.path.exists(self.cookie_file):
michael@0:         try:
michael@0:           self.cookie_jar.load()
michael@0:           self.authenticated = True
michael@0:           StatusUpdate("Loaded authentication cookies from %s" %
michael@0:                        self.cookie_file)
michael@0:         except (cookielib.LoadError, IOError):
michael@0:           # Failed to load cookies - just ignore them.
michael@0:           pass
michael@0:       else:
michael@0:         # Create an empty cookie file with mode 600
michael@0:         fd = os.open(self.cookie_file, os.O_CREAT, 0600)
michael@0:         os.close(fd)
michael@0:       # Always chmod the cookie file
michael@0:       os.chmod(self.cookie_file, 0600)
michael@0:     else:
michael@0:       # Don't save cookies across runs of update.py.
michael@0:       self.cookie_jar = cookielib.CookieJar()
michael@0:     opener.add_handler(urllib2.HTTPCookieProcessor(self.cookie_jar))
michael@0:     return opener
michael@0: 
michael@0: 
michael@0: parser = optparse.OptionParser(usage="%prog [options] [-- diff_options]")
michael@0: parser.add_option("-y", "--assume_yes", action="store_true",
michael@0:                   dest="assume_yes", default=False,
michael@0:                   help="Assume that the answer to yes/no questions is 'yes'.")
michael@0: # Logging
michael@0: group = parser.add_option_group("Logging options")
michael@0: group.add_option("-q", "--quiet", action="store_const", const=0,
michael@0:                  dest="verbose", help="Print errors only.")
michael@0: group.add_option("-v", "--verbose", action="store_const", const=2,
michael@0:                  dest="verbose", default=1,
michael@0:                  help="Print info level logs (default).")
michael@0: group.add_option("--noisy", action="store_const", const=3,
michael@0:                  dest="verbose", help="Print all logs.")
michael@0: # Review server
michael@0: group = parser.add_option_group("Review server options")
michael@0: group.add_option("-s", "--server", action="store", dest="server",
michael@0:                  default="codereview.appspot.com",
michael@0:                  metavar="SERVER",
michael@0:                  help=("The server to upload to. The format is host[:port]. "
michael@0:                        "Defaults to 'codereview.appspot.com'."))
michael@0: group.add_option("-e", "--email", action="store", dest="email",
michael@0:                  metavar="EMAIL", default=None,
michael@0:                  help="The username to use. Will prompt if omitted.")
michael@0: group.add_option("-H", "--host", action="store", dest="host",
michael@0:                  metavar="HOST", default=None,
michael@0:                  help="Overrides the Host header sent with all RPCs.")
michael@0: group.add_option("--no_cookies", action="store_false",
michael@0:                  dest="save_cookies", default=True,
michael@0:                  help="Do not save authentication cookies to local disk.")
michael@0: # Issue
michael@0: group = parser.add_option_group("Issue options")
michael@0: group.add_option("-d", "--description", action="store", dest="description",
michael@0:                  metavar="DESCRIPTION", default=None,
michael@0:                  help="Optional description when creating an issue.")
michael@0: group.add_option("-f", "--description_file", action="store",
michael@0:                  dest="description_file", metavar="DESCRIPTION_FILE",
michael@0:                  default=None,
michael@0:                  help="Optional path of a file that contains "
michael@0:                       "the description when creating an issue.")
michael@0: group.add_option("-r", "--reviewers", action="store", dest="reviewers",
michael@0:                  metavar="REVIEWERS", default=None,
michael@0:                  help="Add reviewers (comma separated email addresses).")
michael@0: group.add_option("--cc", action="store", dest="cc",
michael@0:                  metavar="CC", default=None,
michael@0:                  help="Add CC (comma separated email addresses).")
michael@0: # Upload options
michael@0: group = parser.add_option_group("Patch options")
michael@0: group.add_option("-m", "--message", action="store", dest="message",
michael@0:                  metavar="MESSAGE", default=None,
michael@0:                  help="A message to identify the patch. "
michael@0:                       "Will prompt if omitted.")
michael@0: group.add_option("-i", "--issue", type="int", action="store",
michael@0:                  metavar="ISSUE", default=None,
michael@0:                  help="Issue number to which to add. Defaults to new issue.")
michael@0: group.add_option("--download_base", action="store_true",
michael@0:                  dest="download_base", default=False,
michael@0:                  help="Base files will be downloaded by the server "
michael@0:                  "(side-by-side diffs may not work on files with CRs).")
michael@0: group.add_option("--rev", action="store", dest="revision",
michael@0:                  metavar="REV", default=None,
michael@0:                  help="Branch/tree/revision to diff against (used by DVCS).")
michael@0: group.add_option("--send_mail", action="store_true",
michael@0:                  dest="send_mail", default=False,
michael@0:                  help="Send notification email to reviewers.")
michael@0: 
michael@0: 
michael@0: def GetRpcServer(options):
michael@0:   """Returns an instance of an AbstractRpcServer.
michael@0: 
michael@0:   Returns:
michael@0:     A new AbstractRpcServer, on which RPC calls can be made.
michael@0:   """
michael@0: 
michael@0:   rpc_server_class = HttpRpcServer
michael@0: 
michael@0:   def GetUserCredentials():
michael@0:     """Prompts the user for a username and password."""
michael@0:     email = options.email
michael@0:     if email is None:
michael@0:       email = GetEmail("Email (login for uploading to %s)" % options.server)
michael@0:     password = getpass.getpass("Password for %s: " % email)
michael@0:     return (email, password)
michael@0: 
michael@0:   # If this is the dev_appserver, use fake authentication.
michael@0:   host = (options.host or options.server).lower()
michael@0:   if host == "localhost" or host.startswith("localhost:"):
michael@0:     email = options.email
michael@0:     if email is None:
michael@0:       email = "test@example.com"
michael@0:       logging.info("Using debug user %s.  Override with --email" % email)
michael@0:     server = rpc_server_class(
michael@0:         options.server,
michael@0:         lambda: (email, "password"),
michael@0:         host_override=options.host,
michael@0:         extra_headers={"Cookie":
michael@0:                        'dev_appserver_login="%s:False"' % email},
michael@0:         save_cookies=options.save_cookies)
michael@0:     # Don't try to talk to ClientLogin.
michael@0:     server.authenticated = True
michael@0:     return server
michael@0: 
michael@0:   return rpc_server_class(options.server, GetUserCredentials,
michael@0:                           host_override=options.host,
michael@0:                           save_cookies=options.save_cookies)
michael@0: 
michael@0: 
michael@0: def EncodeMultipartFormData(fields, files):
michael@0:   """Encode form fields for multipart/form-data.
michael@0: 
michael@0:   Args:
michael@0:     fields: A sequence of (name, value) elements for regular form fields.
michael@0:     files: A sequence of (name, filename, value) elements for data to be
michael@0:            uploaded as files.
michael@0:   Returns:
michael@0:     (content_type, body) ready for httplib.HTTP instance.
michael@0: 
michael@0:   Source:
michael@0:     http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/146306
michael@0:   """
michael@0:   BOUNDARY = '-M-A-G-I-C---B-O-U-N-D-A-R-Y-'
michael@0:   CRLF = '\r\n'
michael@0:   lines = []
michael@0:   for (key, value) in fields:
michael@0:     lines.append('--' + BOUNDARY)
michael@0:     lines.append('Content-Disposition: form-data; name="%s"' % key)
michael@0:     lines.append('')
michael@0:     lines.append(value)
michael@0:   for (key, filename, value) in files:
michael@0:     lines.append('--' + BOUNDARY)
michael@0:     lines.append('Content-Disposition: form-data; name="%s"; filename="%s"' %
michael@0:              (key, filename))
michael@0:     lines.append('Content-Type: %s' % GetContentType(filename))
michael@0:     lines.append('')
michael@0:     lines.append(value)
michael@0:   lines.append('--' + BOUNDARY + '--')
michael@0:   lines.append('')
michael@0:   body = CRLF.join(lines)
michael@0:   content_type = 'multipart/form-data; boundary=%s' % BOUNDARY
michael@0:   return content_type, body
michael@0: 
michael@0: 
michael@0: def GetContentType(filename):
michael@0:   """Helper to guess the content-type from the filename."""
michael@0:   return mimetypes.guess_type(filename)[0] or 'application/octet-stream'
michael@0: 
michael@0: 
michael@0: # Use a shell for subcommands on Windows to get a PATH search.
michael@0: use_shell = sys.platform.startswith("win")
michael@0: 
michael@0: def RunShellWithReturnCode(command, print_output=False,
michael@0:                            universal_newlines=True):
michael@0:   """Executes a command and returns the output from stdout and the return code.
michael@0: 
michael@0:   Args:
michael@0:     command: Command to execute.
michael@0:     print_output: If True, the output is printed to stdout.
michael@0:                   If False, both stdout and stderr are ignored.
michael@0:     universal_newlines: Use universal_newlines flag (default: True).
michael@0: 
michael@0:   Returns:
michael@0:     Tuple (output, return code)
michael@0:   """
michael@0:   logging.info("Running %s", command)
michael@0:   p = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
michael@0:                        shell=use_shell, universal_newlines=universal_newlines)
michael@0:   if print_output:
michael@0:     output_array = []
michael@0:     while True:
michael@0:       line = p.stdout.readline()
michael@0:       if not line:
michael@0:         break
michael@0:       print line.strip("\n")
michael@0:       output_array.append(line)
michael@0:     output = "".join(output_array)
michael@0:   else:
michael@0:     output = p.stdout.read()
michael@0:   p.wait()
michael@0:   errout = p.stderr.read()
michael@0:   if print_output and errout:
michael@0:     print >>sys.stderr, errout
michael@0:   p.stdout.close()
michael@0:   p.stderr.close()
michael@0:   return output, p.returncode
michael@0: 
michael@0: 
michael@0: def RunShell(command, silent_ok=False, universal_newlines=True,
michael@0:              print_output=False):
michael@0:   data, retcode = RunShellWithReturnCode(command, print_output,
michael@0:                                          universal_newlines)
michael@0:   if retcode:
michael@0:     ErrorExit("Got error status from %s:\n%s" % (command, data))
michael@0:   if not silent_ok and not data:
michael@0:     ErrorExit("No output from %s" % command)
michael@0:   return data
michael@0: 
michael@0: 
michael@0: class VersionControlSystem(object):
michael@0:   """Abstract base class providing an interface to the VCS."""
michael@0: 
michael@0:   def __init__(self, options):
michael@0:     """Constructor.
michael@0: 
michael@0:     Args:
michael@0:       options: Command line options.
michael@0:     """
michael@0:     self.options = options
michael@0: 
michael@0:   def GenerateDiff(self, args):
michael@0:     """Return the current diff as a string.
michael@0: 
michael@0:     Args:
michael@0:       args: Extra arguments to pass to the diff command.
michael@0:     """
michael@0:     raise NotImplementedError(
michael@0:         "abstract method -- subclass %s must override" % self.__class__)
michael@0: 
michael@0:   def GetUnknownFiles(self):
michael@0:     """Return a list of files unknown to the VCS."""
michael@0:     raise NotImplementedError(
michael@0:         "abstract method -- subclass %s must override" % self.__class__)
michael@0: 
michael@0:   def CheckForUnknownFiles(self):
michael@0:     """Show an "are you sure?" prompt if there are unknown files."""
michael@0:     unknown_files = self.GetUnknownFiles()
michael@0:     if unknown_files:
michael@0:       print "The following files are not added to version control:"
michael@0:       for line in unknown_files:
michael@0:         print line
michael@0:       prompt = "Are you sure to continue?(y/N) "
michael@0:       answer = raw_input(prompt).strip()
michael@0:       if answer != "y":
michael@0:         ErrorExit("User aborted")
michael@0: 
michael@0:   def GetBaseFile(self, filename):
michael@0:     """Get the content of the upstream version of a file.
michael@0: 
michael@0:     Returns:
michael@0:       A tuple (base_content, new_content, is_binary, status)
michael@0:         base_content: The contents of the base file.
michael@0:         new_content: For text files, this is empty.  For binary files, this is
michael@0:           the contents of the new file, since the diff output won't contain
michael@0:           information to reconstruct the current file.
michael@0:         is_binary: True iff the file is binary.
michael@0:         status: The status of the file.
michael@0:     """
michael@0: 
michael@0:     raise NotImplementedError(
michael@0:         "abstract method -- subclass %s must override" % self.__class__)
michael@0: 
michael@0: 
michael@0:   def GetBaseFiles(self, diff):
michael@0:     """Helper that calls GetBase file for each file in the patch.
michael@0: 
michael@0:     Returns:
michael@0:       A dictionary that maps from filename to GetBaseFile's tuple.  Filenames
michael@0:       are retrieved based on lines that start with "Index:" or
michael@0:       "Property changes on:".
michael@0:     """
michael@0:     files = {}
michael@0:     for line in diff.splitlines(True):
michael@0:       if line.startswith('Index:') or line.startswith('Property changes on:'):
michael@0:         unused, filename = line.split(':', 1)
michael@0:         # On Windows if a file has property changes its filename uses '\'
michael@0:         # instead of '/'.
michael@0:         filename = filename.strip().replace('\\', '/')
michael@0:         files[filename] = self.GetBaseFile(filename)
michael@0:     return files
michael@0: 
michael@0: 
michael@0:   def UploadBaseFiles(self, issue, rpc_server, patch_list, patchset, options,
michael@0:                       files):
michael@0:     """Uploads the base files (and if necessary, the current ones as well)."""
michael@0: 
michael@0:     def UploadFile(filename, file_id, content, is_binary, status, is_base):
michael@0:       """Uploads a file to the server."""
michael@0:       file_too_large = False
michael@0:       if is_base:
michael@0:         type = "base"
michael@0:       else:
michael@0:         type = "current"
michael@0:       if len(content) > MAX_UPLOAD_SIZE:
michael@0:         print ("Not uploading the %s file for %s because it's too large." %
michael@0:                (type, filename))
michael@0:         file_too_large = True
michael@0:         content = ""
michael@0:       checksum = md5.new(content).hexdigest()
michael@0:       if options.verbose > 0 and not file_too_large:
michael@0:         print "Uploading %s file for %s" % (type, filename)
michael@0:       url = "/%d/upload_content/%d/%d" % (int(issue), int(patchset), file_id)
michael@0:       form_fields = [("filename", filename),
michael@0:                      ("status", status),
michael@0:                      ("checksum", checksum),
michael@0:                      ("is_binary", str(is_binary)),
michael@0:                      ("is_current", str(not is_base)),
michael@0:                     ]
michael@0:       if file_too_large:
michael@0:         form_fields.append(("file_too_large", "1"))
michael@0:       if options.email:
michael@0:         form_fields.append(("user", options.email))
michael@0:       ctype, body = EncodeMultipartFormData(form_fields,
michael@0:                                             [("data", filename, content)])
michael@0:       response_body = rpc_server.Send(url, body,
michael@0:                                       content_type=ctype)
michael@0:       if not response_body.startswith("OK"):
michael@0:         StatusUpdate("  --> %s" % response_body)
michael@0:         sys.exit(1)
michael@0: 
michael@0:     patches = dict()
michael@0:     [patches.setdefault(v, k) for k, v in patch_list]
michael@0:     for filename in patches.keys():
michael@0:       base_content, new_content, is_binary, status = files[filename]
michael@0:       file_id_str = patches.get(filename)
michael@0:       if file_id_str.find("nobase") != -1:
michael@0:         base_content = None
michael@0:         file_id_str = file_id_str[file_id_str.rfind("_") + 1:]
michael@0:       file_id = int(file_id_str)
michael@0:       if base_content != None:
michael@0:         UploadFile(filename, file_id, base_content, is_binary, status, True)
michael@0:       if new_content != None:
michael@0:         UploadFile(filename, file_id, new_content, is_binary, status, False)
michael@0: 
michael@0:   def IsImage(self, filename):
michael@0:     """Returns true if the filename has an image extension."""
michael@0:     mimetype =  mimetypes.guess_type(filename)[0]
michael@0:     if not mimetype:
michael@0:       return False
michael@0:     return mimetype.startswith("image/")
michael@0: 
michael@0: 
michael@0: class SubversionVCS(VersionControlSystem):
michael@0:   """Implementation of the VersionControlSystem interface for Subversion."""
michael@0: 
michael@0:   def __init__(self, options):
michael@0:     super(SubversionVCS, self).__init__(options)
michael@0:     if self.options.revision:
michael@0:       match = re.match(r"(\d+)(:(\d+))?", self.options.revision)
michael@0:       if not match:
michael@0:         ErrorExit("Invalid Subversion revision %s." % self.options.revision)
michael@0:       self.rev_start = match.group(1)
michael@0:       self.rev_end = match.group(3)
michael@0:     else:
michael@0:       self.rev_start = self.rev_end = None
michael@0:     # Cache output from "svn list -r REVNO dirname".
michael@0:     # Keys: dirname, Values: 2-tuple (ouput for start rev and end rev).
michael@0:     self.svnls_cache = {}
michael@0:     # SVN base URL is required to fetch files deleted in an older revision.
michael@0:     # Result is cached to not guess it over and over again in GetBaseFile().
michael@0:     required = self.options.download_base or self.options.revision is not None
michael@0:     self.svn_base = self._GuessBase(required)
michael@0: 
michael@0:   def GuessBase(self, required):
michael@0:     """Wrapper for _GuessBase."""
michael@0:     return self.svn_base
michael@0: 
michael@0:   def _GuessBase(self, required):
michael@0:     """Returns the SVN base URL.
michael@0: 
michael@0:     Args:
michael@0:       required: If true, exits if the url can't be guessed, otherwise None is
michael@0:         returned.
michael@0:     """
michael@0:     info = RunShell(["svn", "info"])
michael@0:     for line in info.splitlines():
michael@0:       words = line.split()
michael@0:       if len(words) == 2 and words[0] == "URL:":
michael@0:         url = words[1]
michael@0:         scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
michael@0:         username, netloc = urllib.splituser(netloc)
michael@0:         if username:
michael@0:           logging.info("Removed username from base URL")
michael@0:         if netloc.endswith("svn.python.org"):
michael@0:           if netloc == "svn.python.org":
michael@0:             if path.startswith("/projects/"):
michael@0:               path = path[9:]
michael@0:           elif netloc != "pythondev@svn.python.org":
michael@0:             ErrorExit("Unrecognized Python URL: %s" % url)
michael@0:           base = "http://svn.python.org/view/*checkout*%s/" % path
michael@0:           logging.info("Guessed Python base = %s", base)
michael@0:         elif netloc.endswith("svn.collab.net"):
michael@0:           if path.startswith("/repos/"):
michael@0:             path = path[6:]
michael@0:           base = "http://svn.collab.net/viewvc/*checkout*%s/" % path
michael@0:           logging.info("Guessed CollabNet base = %s", base)
michael@0:         elif netloc.endswith(".googlecode.com"):
michael@0:           path = path + "/"
michael@0:           base = urlparse.urlunparse(("http", netloc, path, params,
michael@0:                                       query, fragment))
michael@0:           logging.info("Guessed Google Code base = %s", base)
michael@0:         else:
michael@0:           path = path + "/"
michael@0:           base = urlparse.urlunparse((scheme, netloc, path, params,
michael@0:                                       query, fragment))
michael@0:           logging.info("Guessed base = %s", base)
michael@0:         return base
michael@0:     if required:
michael@0:       ErrorExit("Can't find URL in output from svn info")
michael@0:     return None
michael@0: 
michael@0:   def GenerateDiff(self, args):
michael@0:     cmd = ["svn", "diff"]
michael@0:     if self.options.revision:
michael@0:       cmd += ["-r", self.options.revision]
michael@0:     cmd.extend(args)
michael@0:     data = RunShell(cmd)
michael@0:     count = 0
michael@0:     for line in data.splitlines():
michael@0:       if line.startswith("Index:") or line.startswith("Property changes on:"):
michael@0:         count += 1
michael@0:         logging.info(line)
michael@0:     if not count:
michael@0:       ErrorExit("No valid patches found in output from svn diff")
michael@0:     return data
michael@0: 
michael@0:   def _CollapseKeywords(self, content, keyword_str):
michael@0:     """Collapses SVN keywords."""
michael@0:     # svn cat translates keywords but svn diff doesn't. As a result of this
michael@0:     # behavior patching.PatchChunks() fails with a chunk mismatch error.
michael@0:     # This part was originally written by the Review Board development team
michael@0:     # who had the same problem (http://reviews.review-board.org/r/276/).
michael@0:     # Mapping of keywords to known aliases
michael@0:     svn_keywords = {
michael@0:       # Standard keywords
michael@0:       'Date':                ['Date', 'LastChangedDate'],
michael@0:       'Revision':            ['Revision', 'LastChangedRevision', 'Rev'],
michael@0:       'Author':              ['Author', 'LastChangedBy'],
michael@0:       'HeadURL':             ['HeadURL', 'URL'],
michael@0:       'Id':                  ['Id'],
michael@0: 
michael@0:       # Aliases
michael@0:       'LastChangedDate':     ['LastChangedDate', 'Date'],
michael@0:       'LastChangedRevision': ['LastChangedRevision', 'Rev', 'Revision'],
michael@0:       'LastChangedBy':       ['LastChangedBy', 'Author'],
michael@0:       'URL':                 ['URL', 'HeadURL'],
michael@0:     }
michael@0: 
michael@0:     def repl(m):
michael@0:        if m.group(2):
michael@0:          return "$%s::%s$" % (m.group(1), " " * len(m.group(3)))
michael@0:        return "$%s$" % m.group(1)
michael@0:     keywords = [keyword
michael@0:                 for name in keyword_str.split(" ")
michael@0:                 for keyword in svn_keywords.get(name, [])]
michael@0:     return re.sub(r"\$(%s):(:?)([^\$]+)\$" % '|'.join(keywords), repl, content)
michael@0: 
michael@0:   def GetUnknownFiles(self):
michael@0:     status = RunShell(["svn", "status", "--ignore-externals"], silent_ok=True)
michael@0:     unknown_files = []
michael@0:     for line in status.split("\n"):
michael@0:       if line and line[0] == "?":
michael@0:         unknown_files.append(line)
michael@0:     return unknown_files
michael@0: 
michael@0:   def ReadFile(self, filename):
michael@0:     """Returns the contents of a file."""
michael@0:     file = open(filename, 'rb')
michael@0:     result = ""
michael@0:     try:
michael@0:       result = file.read()
michael@0:     finally:
michael@0:       file.close()
michael@0:     return result
michael@0: 
michael@0:   def GetStatus(self, filename):
michael@0:     """Returns the status of a file."""
michael@0:     if not self.options.revision:
michael@0:       status = RunShell(["svn", "status", "--ignore-externals", filename])
michael@0:       if not status:
michael@0:         ErrorExit("svn status returned no output for %s" % filename)
michael@0:       status_lines = status.splitlines()
michael@0:       # If file is in a cl, the output will begin with
michael@0:       # "\n--- Changelist 'cl_name':\n".  See
michael@0:       # http://svn.collab.net/repos/svn/trunk/notes/changelist-design.txt
michael@0:       if (len(status_lines) == 3 and
michael@0:           not status_lines[0] and
michael@0:           status_lines[1].startswith("--- Changelist")):
michael@0:         status = status_lines[2]
michael@0:       else:
michael@0:         status = status_lines[0]
michael@0:     # If we have a revision to diff against we need to run "svn list"
michael@0:     # for the old and the new revision and compare the results to get
michael@0:     # the correct status for a file.
michael@0:     else:
michael@0:       dirname, relfilename = os.path.split(filename)
michael@0:       if dirname not in self.svnls_cache:
michael@0:         cmd = ["svn", "list", "-r", self.rev_start, dirname or "."]
michael@0:         out, returncode = RunShellWithReturnCode(cmd)
michael@0:         if returncode:
michael@0:           ErrorExit("Failed to get status for %s." % filename)
michael@0:         old_files = out.splitlines()
michael@0:         args = ["svn", "list"]
michael@0:         if self.rev_end:
michael@0:           args += ["-r", self.rev_end]
michael@0:         cmd = args + [dirname or "."]
michael@0:         out, returncode = RunShellWithReturnCode(cmd)
michael@0:         if returncode:
michael@0:           ErrorExit("Failed to run command %s" % cmd)
michael@0:         self.svnls_cache[dirname] = (old_files, out.splitlines())
michael@0:       old_files, new_files = self.svnls_cache[dirname]
michael@0:       if relfilename in old_files and relfilename not in new_files:
michael@0:         status = "D   "
michael@0:       elif relfilename in old_files and relfilename in new_files:
michael@0:         status = "M   "
michael@0:       else:
michael@0:         status = "A   "
michael@0:     return status
michael@0: 
michael@0:   def GetBaseFile(self, filename):
michael@0:     status = self.GetStatus(filename)
michael@0:     base_content = None
michael@0:     new_content = None
michael@0: 
michael@0:     # If a file is copied its status will be "A  +", which signifies
michael@0:     # "addition-with-history".  See "svn st" for more information.  We need to
michael@0:     # upload the original file or else diff parsing will fail if the file was
michael@0:     # edited.
michael@0:     if status[0] == "A" and status[3] != "+":
michael@0:       # We'll need to upload the new content if we're adding a binary file
michael@0:       # since diff's output won't contain it.
michael@0:       mimetype = RunShell(["svn", "propget", "svn:mime-type", filename],
michael@0:                           silent_ok=True)
michael@0:       base_content = ""
michael@0:       is_binary = mimetype and not mimetype.startswith("text/")
michael@0:       if is_binary and self.IsImage(filename):
michael@0:         new_content = self.ReadFile(filename)
michael@0:     elif (status[0] in ("M", "D", "R") or
michael@0:           (status[0] == "A" and status[3] == "+") or  # Copied file.
michael@0:           (status[0] == " " and status[1] == "M")):  # Property change.
michael@0:       args = []
michael@0:       if self.options.revision:
michael@0:         url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
michael@0:       else:
michael@0:         # Don't change filename, it's needed later.
michael@0:         url = filename
michael@0:         args += ["-r", "BASE"]
michael@0:       cmd = ["svn"] + args + ["propget", "svn:mime-type", url]
michael@0:       mimetype, returncode = RunShellWithReturnCode(cmd)
michael@0:       if returncode:
michael@0:         # File does not exist in the requested revision.
michael@0:         # Reset mimetype, it contains an error message.
michael@0:         mimetype = ""
michael@0:       get_base = False
michael@0:       is_binary = mimetype and not mimetype.startswith("text/")
michael@0:       if status[0] == " ":
michael@0:         # Empty base content just to force an upload.
michael@0:         base_content = ""
michael@0:       elif is_binary:
michael@0:         if self.IsImage(filename):
michael@0:           get_base = True
michael@0:           if status[0] == "M":
michael@0:             if not self.rev_end:
michael@0:               new_content = self.ReadFile(filename)
michael@0:             else:
michael@0:               url = "%s/%s@%s" % (self.svn_base, filename, self.rev_end)
michael@0:               new_content = RunShell(["svn", "cat", url],
michael@0:                                      universal_newlines=True, silent_ok=True)
michael@0:         else:
michael@0:           base_content = ""
michael@0:       else:
michael@0:         get_base = True
michael@0: 
michael@0:       if get_base:
michael@0:         if is_binary:
michael@0:           universal_newlines = False
michael@0:         else:
michael@0:           universal_newlines = True
michael@0:         if self.rev_start:
michael@0:           # "svn cat -r REV delete_file.txt" doesn't work. cat requires
michael@0:           # the full URL with "@REV" appended instead of using "-r" option.
michael@0:           url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
michael@0:           base_content = RunShell(["svn", "cat", url],
michael@0:                                   universal_newlines=universal_newlines,
michael@0:                                   silent_ok=True)
michael@0:         else:
michael@0:           base_content = RunShell(["svn", "cat", filename],
michael@0:                                   universal_newlines=universal_newlines,
michael@0:                                   silent_ok=True)
michael@0:         if not is_binary:
michael@0:           args = []
michael@0:           if self.rev_start:
michael@0:             url = "%s/%s@%s" % (self.svn_base, filename, self.rev_start)
michael@0:           else:
michael@0:             url = filename
michael@0:             args += ["-r", "BASE"]
michael@0:           cmd = ["svn"] + args + ["propget", "svn:keywords", url]
michael@0:           keywords, returncode = RunShellWithReturnCode(cmd)
michael@0:           if keywords and not returncode:
michael@0:             base_content = self._CollapseKeywords(base_content, keywords)
michael@0:     else:
michael@0:       StatusUpdate("svn status returned unexpected output: %s" % status)
michael@0:       sys.exit(1)
michael@0:     return base_content, new_content, is_binary, status[0:5]
michael@0: 
michael@0: 
michael@0: class GitVCS(VersionControlSystem):
michael@0:   """Implementation of the VersionControlSystem interface for Git."""
michael@0: 
michael@0:   def __init__(self, options):
michael@0:     super(GitVCS, self).__init__(options)
michael@0:     # Map of filename -> hash of base file.
michael@0:     self.base_hashes = {}
michael@0: 
michael@0:   def GenerateDiff(self, extra_args):
michael@0:     # This is more complicated than svn's GenerateDiff because we must convert
michael@0:     # the diff output to include an svn-style "Index:" line as well as record
michael@0:     # the hashes of the base files, so we can upload them along with our diff.
michael@0:     if self.options.revision:
michael@0:       extra_args = [self.options.revision] + extra_args
michael@0:     gitdiff = RunShell(["git", "diff", "--full-index"] + extra_args)
michael@0:     svndiff = []
michael@0:     filecount = 0
michael@0:     filename = None
michael@0:     for line in gitdiff.splitlines():
michael@0:       match = re.match(r"diff --git a/(.*) b/.*$", line)
michael@0:       if match:
michael@0:         filecount += 1
michael@0:         filename = match.group(1)
michael@0:         svndiff.append("Index: %s\n" % filename)
michael@0:       else:
michael@0:         # The "index" line in a git diff looks like this (long hashes elided):
michael@0:         #   index 82c0d44..b2cee3f 100755
michael@0:         # We want to save the left hash, as that identifies the base file.
michael@0:         match = re.match(r"index (\w+)\.\.", line)
michael@0:         if match:
michael@0:           self.base_hashes[filename] = match.group(1)
michael@0:       svndiff.append(line + "\n")
michael@0:     if not filecount:
michael@0:       ErrorExit("No valid patches found in output from git diff")
michael@0:     return "".join(svndiff)
michael@0: 
michael@0:   def GetUnknownFiles(self):
michael@0:     status = RunShell(["git", "ls-files", "--exclude-standard", "--others"],
michael@0:                       silent_ok=True)
michael@0:     return status.splitlines()
michael@0: 
michael@0:   def GetBaseFile(self, filename):
michael@0:     hash = self.base_hashes[filename]
michael@0:     base_content = None
michael@0:     new_content = None
michael@0:     is_binary = False
michael@0:     if hash == "0" * 40:  # All-zero hash indicates no base file.
michael@0:       status = "A"
michael@0:       base_content = ""
michael@0:     else:
michael@0:       status = "M"
michael@0:       base_content, returncode = RunShellWithReturnCode(["git", "show", hash])
michael@0:       if returncode:
michael@0:         ErrorExit("Got error status from 'git show %s'" % hash)
michael@0:     return (base_content, new_content, is_binary, status)
michael@0: 
michael@0: 
michael@0: class MercurialVCS(VersionControlSystem):
michael@0:   """Implementation of the VersionControlSystem interface for Mercurial."""
michael@0: 
michael@0:   def __init__(self, options, repo_dir):
michael@0:     super(MercurialVCS, self).__init__(options)
michael@0:     # Absolute path to repository (we can be in a subdir)
michael@0:     self.repo_dir = os.path.normpath(repo_dir)
michael@0:     # Compute the subdir
michael@0:     cwd = os.path.normpath(os.getcwd())
michael@0:     assert cwd.startswith(self.repo_dir)
michael@0:     self.subdir = cwd[len(self.repo_dir):].lstrip(r"\/")
michael@0:     if self.options.revision:
michael@0:       self.base_rev = self.options.revision
michael@0:     else:
michael@0:       self.base_rev = RunShell(["hg", "parent", "-q"]).split(':')[1].strip()
michael@0: 
michael@0:   def _GetRelPath(self, filename):
michael@0:     """Get relative path of a file according to the current directory,
michael@0:     given its logical path in the repo."""
michael@0:     assert filename.startswith(self.subdir), filename
michael@0:     return filename[len(self.subdir):].lstrip(r"\/")
michael@0: 
michael@0:   def GenerateDiff(self, extra_args):
michael@0:     # If no file specified, restrict to the current subdir
michael@0:     extra_args = extra_args or ["."]
michael@0:     cmd = ["hg", "diff", "--git", "-r", self.base_rev] + extra_args
michael@0:     data = RunShell(cmd, silent_ok=True)
michael@0:     svndiff = []
michael@0:     filecount = 0
michael@0:     for line in data.splitlines():
michael@0:       m = re.match("diff --git a/(\S+) b/(\S+)", line)
michael@0:       if m:
michael@0:         # Modify line to make it look like as it comes from svn diff.
michael@0:         # With this modification no changes on the server side are required
michael@0:         # to make upload.py work with Mercurial repos.
michael@0:         # NOTE: for proper handling of moved/copied files, we have to use
michael@0:         # the second filename.
michael@0:         filename = m.group(2)
michael@0:         svndiff.append("Index: %s" % filename)
michael@0:         svndiff.append("=" * 67)
michael@0:         filecount += 1
michael@0:         logging.info(line)
michael@0:       else:
michael@0:         svndiff.append(line)
michael@0:     if not filecount:
michael@0:       ErrorExit("No valid patches found in output from hg diff")
michael@0:     return "\n".join(svndiff) + "\n"
michael@0: 
michael@0:   def GetUnknownFiles(self):
michael@0:     """Return a list of files unknown to the VCS."""
michael@0:     args = []
michael@0:     status = RunShell(["hg", "status", "--rev", self.base_rev, "-u", "."],
michael@0:         silent_ok=True)
michael@0:     unknown_files = []
michael@0:     for line in status.splitlines():
michael@0:       st, fn = line.split(" ", 1)
michael@0:       if st == "?":
michael@0:         unknown_files.append(fn)
michael@0:     return unknown_files
michael@0: 
michael@0:   def GetBaseFile(self, filename):
michael@0:     # "hg status" and "hg cat" both take a path relative to the current subdir
michael@0:     # rather than to the repo root, but "hg diff" has given us the full path
michael@0:     # to the repo root.
michael@0:     base_content = ""
michael@0:     new_content = None
michael@0:     is_binary = False
michael@0:     oldrelpath = relpath = self._GetRelPath(filename)
michael@0:     # "hg status -C" returns two lines for moved/copied files, one otherwise
michael@0:     out = RunShell(["hg", "status", "-C", "--rev", self.base_rev, relpath])
michael@0:     out = out.splitlines()
michael@0:     # HACK: strip error message about missing file/directory if it isn't in
michael@0:     # the working copy
michael@0:     if out[0].startswith('%s: ' % relpath):
michael@0:       out = out[1:]
michael@0:     if len(out) > 1:
michael@0:       # Moved/copied => considered as modified, use old filename to
michael@0:       # retrieve base contents
michael@0:       oldrelpath = out[1].strip()
michael@0:       status = "M"
michael@0:     else:
michael@0:       status, _ = out[0].split(' ', 1)
michael@0:     if status != "A":
michael@0:       base_content = RunShell(["hg", "cat", "-r", self.base_rev, oldrelpath],
michael@0:         silent_ok=True)
michael@0:       is_binary = "\0" in base_content  # Mercurial's heuristic
michael@0:     if status != "R":
michael@0:       new_content = open(relpath, "rb").read()
michael@0:       is_binary = is_binary or "\0" in new_content
michael@0:     if is_binary and base_content:
michael@0:       # Fetch again without converting newlines
michael@0:       base_content = RunShell(["hg", "cat", "-r", self.base_rev, oldrelpath],
michael@0:         silent_ok=True, universal_newlines=False)
michael@0:     if not is_binary or not self.IsImage(relpath):
michael@0:       new_content = None
michael@0:     return base_content, new_content, is_binary, status
michael@0: 
michael@0: 
michael@0: # NOTE: The SplitPatch function is duplicated in engine.py, keep them in sync.
michael@0: def SplitPatch(data):
michael@0:   """Splits a patch into separate pieces for each file.
michael@0: 
michael@0:   Args:
michael@0:     data: A string containing the output of svn diff.
michael@0: 
michael@0:   Returns:
michael@0:     A list of 2-tuple (filename, text) where text is the svn diff output
michael@0:       pertaining to filename.
michael@0:   """
michael@0:   patches = []
michael@0:   filename = None
michael@0:   diff = []
michael@0:   for line in data.splitlines(True):
michael@0:     new_filename = None
michael@0:     if line.startswith('Index:'):
michael@0:       unused, new_filename = line.split(':', 1)
michael@0:       new_filename = new_filename.strip()
michael@0:     elif line.startswith('Property changes on:'):
michael@0:       unused, temp_filename = line.split(':', 1)
michael@0:       # When a file is modified, paths use '/' between directories, however
michael@0:       # when a property is modified '\' is used on Windows.  Make them the same
michael@0:       # otherwise the file shows up twice.
michael@0:       temp_filename = temp_filename.strip().replace('\\', '/')
michael@0:       if temp_filename != filename:
michael@0:         # File has property changes but no modifications, create a new diff.
michael@0:         new_filename = temp_filename
michael@0:     if new_filename:
michael@0:       if filename and diff:
michael@0:         patches.append((filename, ''.join(diff)))
michael@0:       filename = new_filename
michael@0:       diff = [line]
michael@0:       continue
michael@0:     if diff is not None:
michael@0:       diff.append(line)
michael@0:   if filename and diff:
michael@0:     patches.append((filename, ''.join(diff)))
michael@0:   return patches
michael@0: 
michael@0: 
michael@0: def UploadSeparatePatches(issue, rpc_server, patchset, data, options):
michael@0:   """Uploads a separate patch for each file in the diff output.
michael@0: 
michael@0:   Returns a list of [patch_key, filename] for each file.
michael@0:   """
michael@0:   patches = SplitPatch(data)
michael@0:   rv = []
michael@0:   for patch in patches:
michael@0:     if len(patch[1]) > MAX_UPLOAD_SIZE:
michael@0:       print ("Not uploading the patch for " + patch[0] +
michael@0:              " because the file is too large.")
michael@0:       continue
michael@0:     form_fields = [("filename", patch[0])]
michael@0:     if not options.download_base:
michael@0:       form_fields.append(("content_upload", "1"))
michael@0:     files = [("data", "data.diff", patch[1])]
michael@0:     ctype, body = EncodeMultipartFormData(form_fields, files)
michael@0:     url = "/%d/upload_patch/%d" % (int(issue), int(patchset))
michael@0:     print "Uploading patch for " + patch[0]
michael@0:     response_body = rpc_server.Send(url, body, content_type=ctype)
michael@0:     lines = response_body.splitlines()
michael@0:     if not lines or lines[0] != "OK":
michael@0:       StatusUpdate("  --> %s" % response_body)
michael@0:       sys.exit(1)
michael@0:     rv.append([lines[1], patch[0]])
michael@0:   return rv
michael@0: 
michael@0: 
michael@0: def GuessVCS(options):
michael@0:   """Helper to guess the version control system.
michael@0: 
michael@0:   This examines the current directory, guesses which VersionControlSystem
michael@0:   we're using, and returns an instance of the appropriate class.  Exit with an
michael@0:   error if we can't figure it out.
michael@0: 
michael@0:   Returns:
michael@0:     A VersionControlSystem instance. Exits if the VCS can't be guessed.
michael@0:   """
michael@0:   # Mercurial has a command to get the base directory of a repository
michael@0:   # Try running it, but don't die if we don't have hg installed.
michael@0:   # NOTE: we try Mercurial first as it can sit on top of an SVN working copy.
michael@0:   try:
michael@0:     out, returncode = RunShellWithReturnCode(["hg", "root"])
michael@0:     if returncode == 0:
michael@0:       return MercurialVCS(options, out.strip())
michael@0:   except OSError, (errno, message):
michael@0:     if errno != 2:  # ENOENT -- they don't have hg installed.
michael@0:       raise
michael@0: 
michael@0:   # Subversion has a .svn in all working directories.
michael@0:   if os.path.isdir('.svn'):
michael@0:     logging.info("Guessed VCS = Subversion")
michael@0:     return SubversionVCS(options)
michael@0: 
michael@0:   # Git has a command to test if you're in a git tree.
michael@0:   # Try running it, but don't die if we don't have git installed.
michael@0:   try:
michael@0:     out, returncode = RunShellWithReturnCode(["git", "rev-parse",
michael@0:                                               "--is-inside-work-tree"])
michael@0:     if returncode == 0:
michael@0:       return GitVCS(options)
michael@0:   except OSError, (errno, message):
michael@0:     if errno != 2:  # ENOENT -- they don't have git installed.
michael@0:       raise
michael@0: 
michael@0:   ErrorExit(("Could not guess version control system. "
michael@0:              "Are you in a working copy directory?"))
michael@0: 
michael@0: 
michael@0: def RealMain(argv, data=None):
michael@0:   """The real main function.
michael@0: 
michael@0:   Args:
michael@0:     argv: Command line arguments.
michael@0:     data: Diff contents. If None (default) the diff is generated by
michael@0:       the VersionControlSystem implementation returned by GuessVCS().
michael@0: 
michael@0:   Returns:
michael@0:     A 2-tuple (issue id, patchset id).
michael@0:     The patchset id is None if the base files are not uploaded by this
michael@0:     script (applies only to SVN checkouts).
michael@0:   """
michael@0:   logging.basicConfig(format=("%(asctime).19s %(levelname)s %(filename)s:"
michael@0:                               "%(lineno)s %(message)s "))
michael@0:   os.environ['LC_ALL'] = 'C'
michael@0:   options, args = parser.parse_args(argv[1:])
michael@0:   global verbosity
michael@0:   verbosity = options.verbose
michael@0:   if verbosity >= 3:
michael@0:     logging.getLogger().setLevel(logging.DEBUG)
michael@0:   elif verbosity >= 2:
michael@0:     logging.getLogger().setLevel(logging.INFO)
michael@0:   vcs = GuessVCS(options)
michael@0:   if isinstance(vcs, SubversionVCS):
michael@0:     # base field is only allowed for Subversion.
michael@0:     # Note: Fetching base files may become deprecated in future releases.
michael@0:     base = vcs.GuessBase(options.download_base)
michael@0:   else:
michael@0:     base = None
michael@0:   if not base and options.download_base:
michael@0:     options.download_base = True
michael@0:     logging.info("Enabled upload of base file")
michael@0:   if not options.assume_yes:
michael@0:     vcs.CheckForUnknownFiles()
michael@0:   if data is None:
michael@0:     data = vcs.GenerateDiff(args)
michael@0:   files = vcs.GetBaseFiles(data)
michael@0:   if verbosity >= 1:
michael@0:     print "Upload server:", options.server, "(change with -s/--server)"
michael@0:   if options.issue:
michael@0:     prompt = "Message describing this patch set: "
michael@0:   else:
michael@0:     prompt = "New issue subject: "
michael@0:   message = options.message or raw_input(prompt).strip()
michael@0:   if not message:
michael@0:     ErrorExit("A non-empty message is required")
michael@0:   rpc_server = GetRpcServer(options)
michael@0:   form_fields = [("subject", message)]
michael@0:   if base:
michael@0:     form_fields.append(("base", base))
michael@0:   if options.issue:
michael@0:     form_fields.append(("issue", str(options.issue)))
michael@0:   if options.email:
michael@0:     form_fields.append(("user", options.email))
michael@0:   if options.reviewers:
michael@0:     for reviewer in options.reviewers.split(','):
michael@0:       if "@" in reviewer and not reviewer.split("@")[1].count(".") == 1:
michael@0:         ErrorExit("Invalid email address: %s" % reviewer)
michael@0:     form_fields.append(("reviewers", options.reviewers))
michael@0:   if options.cc:
michael@0:     for cc in options.cc.split(','):
michael@0:       if "@" in cc and not cc.split("@")[1].count(".") == 1:
michael@0:         ErrorExit("Invalid email address: %s" % cc)
michael@0:     form_fields.append(("cc", options.cc))
michael@0:   description = options.description
michael@0:   if options.description_file:
michael@0:     if options.description:
michael@0:       ErrorExit("Can't specify description and description_file")
michael@0:     file = open(options.description_file, 'r')
michael@0:     description = file.read()
michael@0:     file.close()
michael@0:   if description:
michael@0:     form_fields.append(("description", description))
michael@0:   # Send a hash of all the base file so the server can determine if a copy
michael@0:   # already exists in an earlier patchset.
michael@0:   base_hashes = ""
michael@0:   for file, info in files.iteritems():
michael@0:     if not info[0] is None:
michael@0:       checksum = md5.new(info[0]).hexdigest()
michael@0:       if base_hashes:
michael@0:         base_hashes += "|"
michael@0:       base_hashes += checksum + ":" + file
michael@0:   form_fields.append(("base_hashes", base_hashes))
michael@0:   # If we're uploading base files, don't send the email before the uploads, so
michael@0:   # that it contains the file status.
michael@0:   if options.send_mail and options.download_base:
michael@0:     form_fields.append(("send_mail", "1"))
michael@0:   if not options.download_base:
michael@0:     form_fields.append(("content_upload", "1"))
michael@0:   if len(data) > MAX_UPLOAD_SIZE:
michael@0:     print "Patch is large, so uploading file patches separately."
michael@0:     uploaded_diff_file = []
michael@0:     form_fields.append(("separate_patches", "1"))
michael@0:   else:
michael@0:     uploaded_diff_file = [("data", "data.diff", data)]
michael@0:   ctype, body = EncodeMultipartFormData(form_fields, uploaded_diff_file)
michael@0:   response_body = rpc_server.Send("/upload", body, content_type=ctype)
michael@0:   patchset = None
michael@0:   if not options.download_base or not uploaded_diff_file:
michael@0:     lines = response_body.splitlines()
michael@0:     if len(lines) >= 2:
michael@0:       msg = lines[0]
michael@0:       patchset = lines[1].strip()
michael@0:       patches = [x.split(" ", 1) for x in lines[2:]]
michael@0:     else:
michael@0:       msg = response_body
michael@0:   else:
michael@0:     msg = response_body
michael@0:   StatusUpdate(msg)
michael@0:   if not response_body.startswith("Issue created.") and \
michael@0:   not response_body.startswith("Issue updated."):
michael@0:     sys.exit(0)
michael@0:   issue = msg[msg.rfind("/")+1:]
michael@0: 
michael@0:   if not uploaded_diff_file:
michael@0:     result = UploadSeparatePatches(issue, rpc_server, patchset, data, options)
michael@0:     if not options.download_base:
michael@0:       patches = result
michael@0: 
michael@0:   if not options.download_base:
michael@0:     vcs.UploadBaseFiles(issue, rpc_server, patches, patchset, options, files)
michael@0:     if options.send_mail:
michael@0:       rpc_server.Send("/" + issue + "/mail", payload="")
michael@0:   return issue, patchset
michael@0: 
michael@0: 
michael@0: def main():
michael@0:   try:
michael@0:     RealMain(sys.argv)
michael@0:   except KeyboardInterrupt:
michael@0:     print
michael@0:     StatusUpdate("Interrupted.")
michael@0:     sys.exit(1)
michael@0: 
michael@0: 
michael@0: if __name__ == "__main__":
michael@0:   main()