~mterry/duplicity/require-2.6 : revision 972

32

import getpass

33

import gettext

34

import urllib

35

import urlparse

35

36

37

from duplicity import dup_temp

37

from duplicity import dup_threading

38

from duplicity import file_naming

39

from duplicity import globals

40

from duplicity import log

41

from duplicity import urlparse_2_5 as urlparser

42

41

from duplicity import progress

43

42

44

43

from duplicity.util import exception_traceback

58

57

_forced_backend = None

59

58

_backends = {}

60

59

60

# These URL schemes have a backend with a notion of an RFC "network location".

61

# The 'file' and 's3+http' schemes should not be in this list.

62

# 'http' and 'https' are not actually used for duplicity backend urls, but are needed

63

# in order to properly support urls returned from some webdav servers. adding them here

64

# is a hack. we should instead not stomp on the url parsing module to begin with.

65

#

66

# This looks similar to urlparse's 'uses_netloc' list, but urlparse doesn't use

67

# that list for parsing, only creating urls. And doesn't include our custom

68

# schemes anyway. So we keep our own here for our own use.

69

uses_netloc = ['ftp',

70

'ftps',

71

'hsi',

72

'rsync',

73

's3',

74

'u1',

75

'scp', 'ssh', 'sftp',

76

'webdav', 'webdavs',

77

'gdocs',

78

'http', 'https',

79

'imap', 'imaps',

80

'mega']

81

61

82

62

83

def import_backends():

63

84

"""

165

186

raise BackendException(_("Could not initialize backend: %s") % str(sys.exc_info()[1]))

166

187

167

188

168

_urlparser_initialized = False

169

_urlparser_initialized_lock = dup_threading.threading_module().Lock()

170

171

def _ensure_urlparser_initialized():

172

"""

173

Ensure that the appropriate clobbering of variables in the

174

urlparser module has been done. In the future, the need for this

175

clobbering to begin with should preferably be eliminated.

176

"""

177

def init():

178

global _urlparser_initialized

179

180

if not _urlparser_initialized:

181

# These URL schemes have a backend with a notion of an RFC "network location".

182

# The 'file' and 's3+http' schemes should not be in this list.

183

# 'http' and 'https' are not actually used for duplicity backend urls, but are needed

184

# in order to properly support urls returned from some webdav servers. adding them here

185

# is a hack. we should instead not stomp on the url parsing module to begin with.

186

#

187

# todo: eliminate the need for backend specific hacking here completely.

188

urlparser.uses_netloc = ['ftp',

189

'ftps',

190

'hsi',

191

'rsync',

192

's3',

193

'u1',

194

'scp', 'ssh', 'sftp',

195

'webdav', 'webdavs',

196

'gdocs',

197

'http', 'https',

198

'imap', 'imaps',

199

'mega']

200

201

# Do not transform or otherwise parse the URL path component.

202

urlparser.uses_query = []

203

urlparser.uses_fragm = []

204

205

_urlparser_initialized = True

206

207

dup_threading.with_lock(_urlparser_initialized_lock, init)

208

209

189

class ParsedUrl:

210

190

"""

211

191

Parse the given URL as a duplicity backend URL.

219

199

"""

220

200

def __init__(self, url_string):

221

201

self.url_string = url_string

222

_ensure_urlparser_initialized()

223

202

224

203

# While useful in some cases, the fact is that the urlparser makes

225

204

# all the properties in the URL deferred or lazy. This means that

227

206

# problems here, so they will be caught early.

228

207

229

208

try:

230

pu = urlparser.urlparse(url_string)

209

pu = urlparse.urlparse(url_string)

231

210

except Exception:

232

211

raise InvalidBackendURL("Syntax error in: %s" % url_string)

233

212

273

252

self.port = None

274

253

try:

275

254

self.port = pu.port

276

except Exception:

255

except Exception: # not raised in python2.7+, just returns None

277

256

# old style rsync://host::[/]dest, are still valid, though they contain no port

278

257

if not ( self.scheme in ['rsync'] and re.search('::[^:]*$', self.url_string)):

279

258

raise InvalidBackendURL("Syntax error (port) in: %s A%s B%s C%s" % (url_string, (self.scheme in ['rsync']), re.search('::[^:]+$', self.netloc), self.netloc ) )

280

259

260

# Our URL system uses two slashes more than urlparse's does when using

261

# non-netloc URLs. And we want to make sure that if urlparse assuming

262

# a netloc where we don't want one, that we correct it.

263

if self.scheme not in uses_netloc:

264

if self.netloc:

265

self.path = '//' + self.netloc + self.path

266

self.netloc = ''

267

self.hostname = None

268

elif self.path.startswith('/'):

269

self.path = '//' + self.path

270

281

271

# This happens for implicit local paths.

282

if not pu.scheme:

272

if not self.scheme:

283

273

return

284

274

285

275

# Our backends do not handle implicit hosts.

286

if pu.scheme in urlparser.uses_netloc and not pu.hostname:

276

if self.scheme in uses_netloc and not self.hostname:

287

277

raise InvalidBackendURL("Missing hostname in a backend URL which "

288

278

"requires an explicit hostname: %s"

289

279

"" % (url_string))

290

280

291

281

# Our backends do not handle implicit relative paths.

292

if pu.scheme not in urlparser.uses_netloc and not pu.path.startswith('//'):

282

if self.scheme not in uses_netloc and not self.path.startswith('//'):

293

283

raise InvalidBackendURL("missing // - relative paths not supported "

294

284

"for scheme %s: %s"

295

"" % (pu.scheme, url_string))

285

"" % (self.scheme, url_string))

296

286

297

287

def geturl(self):

298

288

return self.url_string