root / snf-pithos-app / pithos / api / util.py @ f4fbb0fa
History | View | Annotate | Download (36.6 kB)
1 |
# Copyright 2011-2012 GRNET S.A. All rights reserved.
|
---|---|
2 |
#
|
3 |
# Redistribution and use in source and binary forms, with or
|
4 |
# without modification, are permitted provided that the following
|
5 |
# conditions are met:
|
6 |
#
|
7 |
# 1. Redistributions of source code must retain the above
|
8 |
# copyright notice, this list of conditions and the following
|
9 |
# disclaimer.
|
10 |
#
|
11 |
# 2. Redistributions in binary form must reproduce the above
|
12 |
# copyright notice, this list of conditions and the following
|
13 |
# disclaimer in the documentation and/or other materials
|
14 |
# provided with the distribution.
|
15 |
#
|
16 |
# THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS
|
17 |
# OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
|
18 |
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
19 |
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR
|
20 |
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
21 |
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
22 |
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
|
23 |
# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
|
24 |
# AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
25 |
# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
|
26 |
# ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
27 |
# POSSIBILITY OF SUCH DAMAGE.
|
28 |
#
|
29 |
# The views and conclusions contained in the software and
|
30 |
# documentation are those of the authors and should not be
|
31 |
# interpreted as representing official policies, either expressed
|
32 |
# or implied, of GRNET S.A.
|
33 |
|
34 |
from functools import wraps |
35 |
from time import time |
36 |
from traceback import format_exc |
37 |
from wsgiref.handlers import format_date_time |
38 |
from binascii import hexlify, unhexlify |
39 |
from datetime import datetime, tzinfo, timedelta |
40 |
from urllib import quote, unquote |
41 |
|
42 |
from django.conf import settings |
43 |
from django.http import HttpResponse |
44 |
from django.template.loader import render_to_string |
45 |
from django.utils import simplejson as json |
46 |
from django.utils.http import http_date, parse_etags |
47 |
from django.utils.encoding import smart_unicode, smart_str |
48 |
from django.core.files.uploadhandler import FileUploadHandler |
49 |
from django.core.files.uploadedfile import UploadedFile |
50 |
|
51 |
from synnefo.lib.parsedate import parse_http_date_safe, parse_http_date |
52 |
from synnefo.lib.astakos import get_user |
53 |
|
54 |
from pithos.api.faults import ( |
55 |
Fault, NotModified, BadRequest, Unauthorized, Forbidden, ItemNotFound, |
56 |
Conflict, LengthRequired, PreconditionFailed, RequestEntityTooLarge, |
57 |
RangeNotSatisfiable, InternalServerError, NotImplemented)
|
58 |
from pithos.api.short_url import encode_url |
59 |
from pithos.api.settings import (BACKEND_DB_MODULE, BACKEND_DB_CONNECTION, |
60 |
BACKEND_BLOCK_MODULE, BACKEND_BLOCK_PATH, |
61 |
BACKEND_BLOCK_UMASK, |
62 |
BACKEND_QUEUE_MODULE, BACKEND_QUEUE_HOSTS, |
63 |
BACKEND_QUEUE_EXCHANGE, |
64 |
BACKEND_QUOTA, BACKEND_VERSIONING, |
65 |
AUTHENTICATION_URL, AUTHENTICATION_USERS, |
66 |
SERVICE_TOKEN, COOKIE_NAME) |
67 |
|
68 |
from pithos.backends import connect_backend |
69 |
from pithos.backends.base import NotAllowedError, QuotaError, ItemNotExists, VersionNotExists |
70 |
|
71 |
import logging |
72 |
import re |
73 |
import hashlib |
74 |
import uuid |
75 |
import decimal |
76 |
|
77 |
|
78 |
logger = logging.getLogger(__name__) |
79 |
|
80 |
|
81 |
class UTC(tzinfo): |
82 |
def utcoffset(self, dt): |
83 |
return timedelta(0) |
84 |
|
85 |
def tzname(self, dt): |
86 |
return 'UTC' |
87 |
|
88 |
def dst(self, dt): |
89 |
return timedelta(0) |
90 |
|
91 |
|
92 |
def json_encode_decimal(obj): |
93 |
if isinstance(obj, decimal.Decimal): |
94 |
return str(obj) |
95 |
raise TypeError(repr(obj) + " is not JSON serializable") |
96 |
|
97 |
|
98 |
def isoformat(d): |
99 |
"""Return an ISO8601 date string that includes a timezone."""
|
100 |
|
101 |
return d.replace(tzinfo=UTC()).isoformat()
|
102 |
|
103 |
|
104 |
def rename_meta_key(d, old, new): |
105 |
if old not in d: |
106 |
return
|
107 |
d[new] = d[old] |
108 |
del(d[old])
|
109 |
|
110 |
|
111 |
def printable_header_dict(d): |
112 |
"""Format a meta dictionary for printing out json/xml.
|
113 |
|
114 |
Convert all keys to lower case and replace dashes with underscores.
|
115 |
Format 'last_modified' timestamp.
|
116 |
"""
|
117 |
|
118 |
if 'last_modified' in d and d['last_modified']: |
119 |
d['last_modified'] = isoformat(
|
120 |
datetime.fromtimestamp(d['last_modified']))
|
121 |
return dict([(k.lower().replace('-', '_'), v) for k, v in d.iteritems()]) |
122 |
|
123 |
|
124 |
def format_header_key(k): |
125 |
"""Convert underscores to dashes and capitalize intra-dash strings."""
|
126 |
return '-'.join([x.capitalize() for x in k.replace('_', '-').split('-')]) |
127 |
|
128 |
|
129 |
def get_header_prefix(request, prefix): |
130 |
"""Get all prefix-* request headers in a dict. Reformat keys with format_header_key()."""
|
131 |
|
132 |
prefix = 'HTTP_' + prefix.upper().replace('-', '_') |
133 |
# TODO: Document or remove '~' replacing.
|
134 |
return dict([(format_header_key(k[5:]), v.replace('~', '')) for k, v in request.META.iteritems() if k.startswith(prefix) and len(k) > len(prefix)]) |
135 |
|
136 |
|
137 |
def check_meta_headers(meta): |
138 |
if len(meta) > 90: |
139 |
raise BadRequest('Too many headers.') |
140 |
for k, v in meta.iteritems(): |
141 |
if len(k) > 128: |
142 |
raise BadRequest('Header name too large.') |
143 |
if len(v) > 256: |
144 |
raise BadRequest('Header value too large.') |
145 |
|
146 |
|
147 |
def get_account_headers(request): |
148 |
meta = get_header_prefix(request, 'X-Account-Meta-')
|
149 |
check_meta_headers(meta) |
150 |
groups = {} |
151 |
for k, v in get_header_prefix(request, 'X-Account-Group-').iteritems(): |
152 |
n = k[16:].lower()
|
153 |
if '-' in n or '_' in n: |
154 |
raise BadRequest('Bad characters in group name') |
155 |
groups[n] = v.replace(' ', '').split(',') |
156 |
while '' in groups[n]: |
157 |
groups[n].remove('')
|
158 |
return meta, groups
|
159 |
|
160 |
|
161 |
def put_account_headers(response, meta, groups, policy): |
162 |
if 'count' in meta: |
163 |
response['X-Account-Container-Count'] = meta['count'] |
164 |
if 'bytes' in meta: |
165 |
response['X-Account-Bytes-Used'] = meta['bytes'] |
166 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
167 |
for k in [x for x in meta.keys() if x.startswith('X-Account-Meta-')]: |
168 |
response[smart_str( |
169 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
170 |
if 'until_timestamp' in meta: |
171 |
response['X-Account-Until-Timestamp'] = http_date(
|
172 |
int(meta['until_timestamp'])) |
173 |
for k, v in groups.iteritems(): |
174 |
k = smart_str(k, strings_only=True)
|
175 |
k = format_header_key('X-Account-Group-' + k)
|
176 |
v = smart_str(','.join(v), strings_only=True) |
177 |
response[k] = v |
178 |
for k, v in policy.iteritems(): |
179 |
response[smart_str(format_header_key('X-Account-Policy-' + k), strings_only=True)] = smart_str(v, strings_only=True) |
180 |
|
181 |
|
182 |
def get_container_headers(request): |
183 |
meta = get_header_prefix(request, 'X-Container-Meta-')
|
184 |
check_meta_headers(meta) |
185 |
policy = dict([(k[19:].lower(), v.replace(' ', '')) for k, v in get_header_prefix(request, 'X-Container-Policy-').iteritems()]) |
186 |
return meta, policy
|
187 |
|
188 |
|
189 |
def put_container_headers(request, response, meta, policy): |
190 |
if 'count' in meta: |
191 |
response['X-Container-Object-Count'] = meta['count'] |
192 |
if 'bytes' in meta: |
193 |
response['X-Container-Bytes-Used'] = meta['bytes'] |
194 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
195 |
for k in [x for x in meta.keys() if x.startswith('X-Container-Meta-')]: |
196 |
response[smart_str( |
197 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
198 |
l = [smart_str(x, strings_only=True) for x in meta['object_meta'] |
199 |
if x.startswith('X-Object-Meta-')] |
200 |
response['X-Container-Object-Meta'] = ','.join([x[14:] for x in l]) |
201 |
response['X-Container-Block-Size'] = request.backend.block_size
|
202 |
response['X-Container-Block-Hash'] = request.backend.hash_algorithm
|
203 |
if 'until_timestamp' in meta: |
204 |
response['X-Container-Until-Timestamp'] = http_date(
|
205 |
int(meta['until_timestamp'])) |
206 |
for k, v in policy.iteritems(): |
207 |
response[smart_str(format_header_key('X-Container-Policy-' + k), strings_only=True)] = smart_str(v, strings_only=True) |
208 |
|
209 |
|
210 |
def get_object_headers(request): |
211 |
content_type = request.META.get('CONTENT_TYPE', None) |
212 |
meta = get_header_prefix(request, 'X-Object-Meta-')
|
213 |
check_meta_headers(meta) |
214 |
if request.META.get('HTTP_CONTENT_ENCODING'): |
215 |
meta['Content-Encoding'] = request.META['HTTP_CONTENT_ENCODING'] |
216 |
if request.META.get('HTTP_CONTENT_DISPOSITION'): |
217 |
meta['Content-Disposition'] = request.META['HTTP_CONTENT_DISPOSITION'] |
218 |
if request.META.get('HTTP_X_OBJECT_MANIFEST'): |
219 |
meta['X-Object-Manifest'] = request.META['HTTP_X_OBJECT_MANIFEST'] |
220 |
return content_type, meta, get_sharing(request), get_public(request)
|
221 |
|
222 |
|
223 |
def put_object_headers(response, meta, restricted=False): |
224 |
response['ETag'] = meta['checksum'] |
225 |
response['Content-Length'] = meta['bytes'] |
226 |
response['Content-Type'] = meta.get('type', 'application/octet-stream') |
227 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
228 |
if not restricted: |
229 |
response['X-Object-Hash'] = meta['hash'] |
230 |
response['X-Object-UUID'] = meta['uuid'] |
231 |
response['X-Object-Modified-By'] = smart_str(
|
232 |
meta['modified_by'], strings_only=True) |
233 |
response['X-Object-Version'] = meta['version'] |
234 |
response['X-Object-Version-Timestamp'] = http_date(
|
235 |
int(meta['version_timestamp'])) |
236 |
for k in [x for x in meta.keys() if x.startswith('X-Object-Meta-')]: |
237 |
response[smart_str( |
238 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
239 |
for k in ( |
240 |
'Content-Encoding', 'Content-Disposition', 'X-Object-Manifest', |
241 |
'X-Object-Sharing', 'X-Object-Shared-By', 'X-Object-Allowed-To', |
242 |
'X-Object-Public'):
|
243 |
if k in meta: |
244 |
response[k] = smart_str(meta[k], strings_only=True)
|
245 |
else:
|
246 |
for k in ('Content-Encoding', 'Content-Disposition'): |
247 |
if k in meta: |
248 |
response[k] = smart_str(meta[k], strings_only=True)
|
249 |
|
250 |
|
251 |
def update_manifest_meta(request, v_account, meta): |
252 |
"""Update metadata if the object has an X-Object-Manifest."""
|
253 |
|
254 |
if 'X-Object-Manifest' in meta: |
255 |
etag = ''
|
256 |
bytes = 0
|
257 |
try:
|
258 |
src_container, src_name = split_container_object_string( |
259 |
'/' + meta['X-Object-Manifest']) |
260 |
objects = request.backend.list_objects( |
261 |
request.user_uniq, v_account, |
262 |
src_container, prefix=src_name, virtual=False)
|
263 |
for x in objects: |
264 |
src_meta = request.backend.get_object_meta(request.user_uniq, |
265 |
v_account, src_container, x[0], 'pithos', x[1]) |
266 |
etag += src_meta['checksum']
|
267 |
bytes += src_meta['bytes'] |
268 |
except:
|
269 |
# Ignore errors.
|
270 |
return
|
271 |
meta['bytes'] = bytes |
272 |
md5 = hashlib.md5() |
273 |
md5.update(etag) |
274 |
meta['checksum'] = md5.hexdigest().lower()
|
275 |
|
276 |
|
277 |
def update_sharing_meta(request, permissions, v_account, v_container, v_object, meta): |
278 |
if permissions is None: |
279 |
return
|
280 |
allowed, perm_path, perms = permissions |
281 |
if len(perms) == 0: |
282 |
return
|
283 |
ret = [] |
284 |
r = ','.join(perms.get('read', [])) |
285 |
if r:
|
286 |
ret.append('read=' + r)
|
287 |
w = ','.join(perms.get('write', [])) |
288 |
if w:
|
289 |
ret.append('write=' + w)
|
290 |
meta['X-Object-Sharing'] = '; '.join(ret) |
291 |
if '/'.join((v_account, v_container, v_object)) != perm_path: |
292 |
meta['X-Object-Shared-By'] = perm_path
|
293 |
if request.user_uniq != v_account:
|
294 |
meta['X-Object-Allowed-To'] = allowed
|
295 |
|
296 |
|
297 |
def update_public_meta(public, meta): |
298 |
if not public: |
299 |
return
|
300 |
meta['X-Object-Public'] = '/public/' + encode_url(public) |
301 |
|
302 |
|
303 |
def validate_modification_preconditions(request, meta): |
304 |
"""Check that the modified timestamp conforms with the preconditions set."""
|
305 |
|
306 |
if 'modified' not in meta: |
307 |
return # TODO: Always return? |
308 |
|
309 |
if_modified_since = request.META.get('HTTP_IF_MODIFIED_SINCE')
|
310 |
if if_modified_since is not None: |
311 |
if_modified_since = parse_http_date_safe(if_modified_since) |
312 |
if if_modified_since is not None and int(meta['modified']) <= if_modified_since: |
313 |
raise NotModified('Resource has not been modified') |
314 |
|
315 |
if_unmodified_since = request.META.get('HTTP_IF_UNMODIFIED_SINCE')
|
316 |
if if_unmodified_since is not None: |
317 |
if_unmodified_since = parse_http_date_safe(if_unmodified_since) |
318 |
if if_unmodified_since is not None and int(meta['modified']) > if_unmodified_since: |
319 |
raise PreconditionFailed('Resource has been modified') |
320 |
|
321 |
|
322 |
def validate_matching_preconditions(request, meta): |
323 |
"""Check that the ETag conforms with the preconditions set."""
|
324 |
|
325 |
etag = meta['checksum']
|
326 |
if not etag: |
327 |
etag = None
|
328 |
|
329 |
if_match = request.META.get('HTTP_IF_MATCH')
|
330 |
if if_match is not None: |
331 |
if etag is None: |
332 |
raise PreconditionFailed('Resource does not exist') |
333 |
if if_match != '*' and etag not in [x.lower() for x in parse_etags(if_match)]: |
334 |
raise PreconditionFailed('Resource ETag does not match') |
335 |
|
336 |
if_none_match = request.META.get('HTTP_IF_NONE_MATCH')
|
337 |
if if_none_match is not None: |
338 |
# TODO: If this passes, must ignore If-Modified-Since header.
|
339 |
if etag is not None: |
340 |
if if_none_match == '*' or etag in [x.lower() for x in parse_etags(if_none_match)]: |
341 |
# TODO: Continue if an If-Modified-Since header is present.
|
342 |
if request.method in ('HEAD', 'GET'): |
343 |
raise NotModified('Resource ETag matches') |
344 |
raise PreconditionFailed('Resource exists or ETag matches') |
345 |
|
346 |
|
347 |
def split_container_object_string(s): |
348 |
if not len(s) > 0 or s[0] != '/': |
349 |
raise ValueError |
350 |
s = s[1:]
|
351 |
pos = s.find('/')
|
352 |
if pos == -1 or pos == len(s) - 1: |
353 |
raise ValueError |
354 |
return s[:pos], s[(pos + 1):] |
355 |
|
356 |
|
357 |
def copy_or_move_object(request, src_account, src_container, src_name, dest_account, dest_container, dest_name, move=False, delimiter=None): |
358 |
"""Copy or move an object."""
|
359 |
|
360 |
if 'ignore_content_type' in request.GET and 'CONTENT_TYPE' in request.META: |
361 |
del(request.META['CONTENT_TYPE']) |
362 |
content_type, meta, permissions, public = get_object_headers(request) |
363 |
src_version = request.META.get('HTTP_X_SOURCE_VERSION')
|
364 |
try:
|
365 |
if move:
|
366 |
version_id = request.backend.move_object( |
367 |
request.user_uniq, src_account, src_container, src_name, |
368 |
dest_account, dest_container, dest_name, |
369 |
content_type, 'pithos', meta, False, permissions, delimiter) |
370 |
else:
|
371 |
version_id = request.backend.copy_object( |
372 |
request.user_uniq, src_account, src_container, src_name, |
373 |
dest_account, dest_container, dest_name, |
374 |
content_type, 'pithos', meta, False, permissions, src_version, delimiter) |
375 |
except NotAllowedError:
|
376 |
raise Forbidden('Not allowed') |
377 |
except (ItemNotExists, VersionNotExists):
|
378 |
raise ItemNotFound('Container or object does not exist') |
379 |
except ValueError: |
380 |
raise BadRequest('Invalid sharing header') |
381 |
except QuotaError:
|
382 |
raise RequestEntityTooLarge('Quota exceeded') |
383 |
if public is not None: |
384 |
try:
|
385 |
request.backend.update_object_public(request.user_uniq, dest_account, dest_container, dest_name, public) |
386 |
except NotAllowedError:
|
387 |
raise Forbidden('Not allowed') |
388 |
except ItemNotExists:
|
389 |
raise ItemNotFound('Object does not exist') |
390 |
return version_id
|
391 |
|
392 |
|
393 |
def get_int_parameter(p): |
394 |
if p is not None: |
395 |
try:
|
396 |
p = int(p)
|
397 |
except ValueError: |
398 |
return None |
399 |
if p < 0: |
400 |
return None |
401 |
return p
|
402 |
|
403 |
|
404 |
def get_content_length(request): |
405 |
content_length = get_int_parameter(request.META.get('CONTENT_LENGTH'))
|
406 |
if content_length is None: |
407 |
raise LengthRequired('Missing or invalid Content-Length header') |
408 |
return content_length
|
409 |
|
410 |
|
411 |
def get_range(request, size): |
412 |
"""Parse a Range header from the request.
|
413 |
|
414 |
Either returns None, when the header is not existent or should be ignored,
|
415 |
or a list of (offset, length) tuples - should be further checked.
|
416 |
"""
|
417 |
|
418 |
ranges = request.META.get('HTTP_RANGE', '').replace(' ', '') |
419 |
if not ranges.startswith('bytes='): |
420 |
return None |
421 |
|
422 |
ret = [] |
423 |
for r in (x.strip() for x in ranges[6:].split(',')): |
424 |
p = re.compile('^(?P<offset>\d*)-(?P<upto>\d*)$')
|
425 |
m = p.match(r) |
426 |
if not m: |
427 |
return None |
428 |
offset = m.group('offset')
|
429 |
upto = m.group('upto')
|
430 |
if offset == '' and upto == '': |
431 |
return None |
432 |
|
433 |
if offset != '': |
434 |
offset = int(offset)
|
435 |
if upto != '': |
436 |
upto = int(upto)
|
437 |
if offset > upto:
|
438 |
return None |
439 |
ret.append((offset, upto - offset + 1))
|
440 |
else:
|
441 |
ret.append((offset, size - offset)) |
442 |
else:
|
443 |
length = int(upto)
|
444 |
ret.append((size - length, length)) |
445 |
|
446 |
return ret
|
447 |
|
448 |
|
449 |
def get_content_range(request): |
450 |
"""Parse a Content-Range header from the request.
|
451 |
|
452 |
Either returns None, when the header is not existent or should be ignored,
|
453 |
or an (offset, length, total) tuple - check as length, total may be None.
|
454 |
Returns (None, None, None) if the provided range is '*/*'.
|
455 |
"""
|
456 |
|
457 |
ranges = request.META.get('HTTP_CONTENT_RANGE', '') |
458 |
if not ranges: |
459 |
return None |
460 |
|
461 |
p = re.compile('^bytes (?P<offset>\d+)-(?P<upto>\d*)/(?P<total>(\d+|\*))$')
|
462 |
m = p.match(ranges) |
463 |
if not m: |
464 |
if ranges == 'bytes */*': |
465 |
return (None, None, None) |
466 |
return None |
467 |
offset = int(m.group('offset')) |
468 |
upto = m.group('upto')
|
469 |
total = m.group('total')
|
470 |
if upto != '': |
471 |
upto = int(upto)
|
472 |
else:
|
473 |
upto = None
|
474 |
if total != '*': |
475 |
total = int(total)
|
476 |
else:
|
477 |
total = None
|
478 |
if (upto is not None and offset > upto) or \ |
479 |
(total is not None and offset >= total) or \ |
480 |
(total is not None and upto is not None and upto >= total): |
481 |
return None |
482 |
|
483 |
if upto is None: |
484 |
length = None
|
485 |
else:
|
486 |
length = upto - offset + 1
|
487 |
return (offset, length, total)
|
488 |
|
489 |
|
490 |
def get_sharing(request): |
491 |
"""Parse an X-Object-Sharing header from the request.
|
492 |
|
493 |
Raises BadRequest on error.
|
494 |
"""
|
495 |
|
496 |
permissions = request.META.get('HTTP_X_OBJECT_SHARING')
|
497 |
if permissions is None: |
498 |
return None |
499 |
|
500 |
# TODO: Document or remove '~' replacing.
|
501 |
permissions = permissions.replace('~', '') |
502 |
|
503 |
ret = {} |
504 |
permissions = permissions.replace(' ', '') |
505 |
if permissions == '': |
506 |
return ret
|
507 |
for perm in (x for x in permissions.split(';')): |
508 |
if perm.startswith('read='): |
509 |
ret['read'] = list(set( |
510 |
[v.replace(' ', '').lower() for v in perm[5:].split(',')])) |
511 |
if '' in ret['read']: |
512 |
ret['read'].remove('') |
513 |
if '*' in ret['read']: |
514 |
ret['read'] = ['*'] |
515 |
if len(ret['read']) == 0: |
516 |
raise BadRequest('Bad X-Object-Sharing header value') |
517 |
elif perm.startswith('write='): |
518 |
ret['write'] = list(set( |
519 |
[v.replace(' ', '').lower() for v in perm[6:].split(',')])) |
520 |
if '' in ret['write']: |
521 |
ret['write'].remove('') |
522 |
if '*' in ret['write']: |
523 |
ret['write'] = ['*'] |
524 |
if len(ret['write']) == 0: |
525 |
raise BadRequest('Bad X-Object-Sharing header value') |
526 |
else:
|
527 |
raise BadRequest('Bad X-Object-Sharing header value') |
528 |
|
529 |
# Keep duplicates only in write list.
|
530 |
dups = [x for x in ret.get( |
531 |
'read', []) if x in ret.get('write', []) and x != '*'] |
532 |
if dups:
|
533 |
for x in dups: |
534 |
ret['read'].remove(x)
|
535 |
if len(ret['read']) == 0: |
536 |
del(ret['read']) |
537 |
|
538 |
return ret
|
539 |
|
540 |
|
541 |
def get_public(request): |
542 |
"""Parse an X-Object-Public header from the request.
|
543 |
|
544 |
Raises BadRequest on error.
|
545 |
"""
|
546 |
|
547 |
public = request.META.get('HTTP_X_OBJECT_PUBLIC')
|
548 |
if public is None: |
549 |
return None |
550 |
|
551 |
public = public.replace(' ', '').lower() |
552 |
if public == 'true': |
553 |
return True |
554 |
elif public == 'false' or public == '': |
555 |
return False |
556 |
raise BadRequest('Bad X-Object-Public header value') |
557 |
|
558 |
|
559 |
def raw_input_socket(request): |
560 |
"""Return the socket for reading the rest of the request."""
|
561 |
|
562 |
server_software = request.META.get('SERVER_SOFTWARE')
|
563 |
if server_software and server_software.startswith('mod_python'): |
564 |
return request._req
|
565 |
if 'wsgi.input' in request.environ: |
566 |
return request.environ['wsgi.input'] |
567 |
raise NotImplemented('Unknown server software') |
568 |
|
569 |
MAX_UPLOAD_SIZE = 5 * (1024 * 1024 * 1024) # 5GB |
570 |
|
571 |
|
572 |
def socket_read_iterator(request, length=0, blocksize=4096): |
573 |
"""Return a maximum of blocksize data read from the socket in each iteration.
|
574 |
|
575 |
Read up to 'length'. If 'length' is negative, will attempt a chunked read.
|
576 |
The maximum ammount of data read is controlled by MAX_UPLOAD_SIZE.
|
577 |
"""
|
578 |
|
579 |
sock = raw_input_socket(request) |
580 |
if length < 0: # Chunked transfers |
581 |
# Small version (server does the dechunking).
|
582 |
if request.environ.get('mod_wsgi.input_chunked', None) or request.META['SERVER_SOFTWARE'].startswith('gunicorn'): |
583 |
while length < MAX_UPLOAD_SIZE:
|
584 |
data = sock.read(blocksize) |
585 |
if data == '': |
586 |
return
|
587 |
yield data
|
588 |
raise BadRequest('Maximum size is reached') |
589 |
|
590 |
# Long version (do the dechunking).
|
591 |
data = ''
|
592 |
while length < MAX_UPLOAD_SIZE:
|
593 |
# Get chunk size.
|
594 |
if hasattr(sock, 'readline'): |
595 |
chunk_length = sock.readline() |
596 |
else:
|
597 |
chunk_length = ''
|
598 |
while chunk_length[-1:] != '\n': |
599 |
chunk_length += sock.read(1)
|
600 |
chunk_length.strip() |
601 |
pos = chunk_length.find(';')
|
602 |
if pos >= 0: |
603 |
chunk_length = chunk_length[:pos] |
604 |
try:
|
605 |
chunk_length = int(chunk_length, 16) |
606 |
except Exception, e: |
607 |
raise BadRequest('Bad chunk size') |
608 |
# TODO: Change to something more appropriate.
|
609 |
# Check if done.
|
610 |
if chunk_length == 0: |
611 |
if len(data) > 0: |
612 |
yield data
|
613 |
return
|
614 |
# Get the actual data.
|
615 |
while chunk_length > 0: |
616 |
chunk = sock.read(min(chunk_length, blocksize))
|
617 |
chunk_length -= len(chunk)
|
618 |
if length > 0: |
619 |
length += len(chunk)
|
620 |
data += chunk |
621 |
if len(data) >= blocksize: |
622 |
ret = data[:blocksize] |
623 |
data = data[blocksize:] |
624 |
yield ret
|
625 |
sock.read(2) # CRLF |
626 |
raise BadRequest('Maximum size is reached') |
627 |
else:
|
628 |
if length > MAX_UPLOAD_SIZE:
|
629 |
raise BadRequest('Maximum size is reached') |
630 |
while length > 0: |
631 |
data = sock.read(min(length, blocksize))
|
632 |
if not data: |
633 |
raise BadRequest()
|
634 |
length -= len(data)
|
635 |
yield data
|
636 |
|
637 |
|
638 |
class SaveToBackendHandler(FileUploadHandler): |
639 |
"""Handle a file from an HTML form the django way."""
|
640 |
|
641 |
def __init__(self, request=None): |
642 |
super(SaveToBackendHandler, self).__init__(request) |
643 |
self.backend = request.backend
|
644 |
|
645 |
def put_data(self, length): |
646 |
if len(self.data) >= length: |
647 |
block = self.data[:length]
|
648 |
self.file.hashmap.append(self.backend.put_block(block)) |
649 |
self.md5.update(block)
|
650 |
self.data = self.data[length:] |
651 |
|
652 |
def new_file(self, field_name, file_name, content_type, content_length, charset=None): |
653 |
self.md5 = hashlib.md5()
|
654 |
self.data = '' |
655 |
self.file = UploadedFile(
|
656 |
name=file_name, content_type=content_type, charset=charset) |
657 |
self.file.size = 0 |
658 |
self.file.hashmap = []
|
659 |
|
660 |
def receive_data_chunk(self, raw_data, start): |
661 |
self.data += raw_data
|
662 |
self.file.size += len(raw_data) |
663 |
self.put_data(self.request.backend.block_size) |
664 |
return None |
665 |
|
666 |
def file_complete(self, file_size): |
667 |
l = len(self.data) |
668 |
if l > 0: |
669 |
self.put_data(l)
|
670 |
self.file.etag = self.md5.hexdigest().lower() |
671 |
return self.file |
672 |
|
673 |
|
674 |
class ObjectWrapper(object): |
675 |
"""Return the object's data block-per-block in each iteration.
|
676 |
|
677 |
Read from the object using the offset and length provided in each entry of the range list.
|
678 |
"""
|
679 |
|
680 |
def __init__(self, backend, ranges, sizes, hashmaps, boundary): |
681 |
self.backend = backend
|
682 |
self.ranges = ranges
|
683 |
self.sizes = sizes
|
684 |
self.hashmaps = hashmaps
|
685 |
self.boundary = boundary
|
686 |
self.size = sum(self.sizes) |
687 |
|
688 |
self.file_index = 0 |
689 |
self.block_index = 0 |
690 |
self.block_hash = -1 |
691 |
self.block = '' |
692 |
|
693 |
self.range_index = -1 |
694 |
self.offset, self.length = self.ranges[0] |
695 |
|
696 |
def __iter__(self): |
697 |
return self |
698 |
|
699 |
def part_iterator(self): |
700 |
if self.length > 0: |
701 |
# Get the file for the current offset.
|
702 |
file_size = self.sizes[self.file_index] |
703 |
while self.offset >= file_size: |
704 |
self.offset -= file_size
|
705 |
self.file_index += 1 |
706 |
file_size = self.sizes[self.file_index] |
707 |
|
708 |
# Get the block for the current position.
|
709 |
self.block_index = int(self.offset / self.backend.block_size) |
710 |
if self.block_hash != self.hashmaps[self.file_index][self.block_index]: |
711 |
self.block_hash = self.hashmaps[ |
712 |
self.file_index][self.block_index] |
713 |
try:
|
714 |
self.block = self.backend.get_block(self.block_hash) |
715 |
except ItemNotExists:
|
716 |
raise ItemNotFound('Block does not exist') |
717 |
|
718 |
# Get the data from the block.
|
719 |
bo = self.offset % self.backend.block_size |
720 |
bs = self.backend.block_size
|
721 |
if (self.block_index == len(self.hashmaps[self.file_index]) - 1 and |
722 |
self.sizes[self.file_index] % self.backend.block_size): |
723 |
bs = self.sizes[self.file_index] % self.backend.block_size |
724 |
bl = min(self.length, bs - bo) |
725 |
data = self.block[bo:bo + bl]
|
726 |
self.offset += bl
|
727 |
self.length -= bl
|
728 |
return data
|
729 |
else:
|
730 |
raise StopIteration |
731 |
|
732 |
def next(self): |
733 |
if len(self.ranges) == 1: |
734 |
return self.part_iterator() |
735 |
if self.range_index == len(self.ranges): |
736 |
raise StopIteration |
737 |
try:
|
738 |
if self.range_index == -1: |
739 |
raise StopIteration |
740 |
return self.part_iterator() |
741 |
except StopIteration: |
742 |
self.range_index += 1 |
743 |
out = [] |
744 |
if self.range_index < len(self.ranges): |
745 |
# Part header.
|
746 |
self.offset, self.length = self.ranges[self.range_index] |
747 |
self.file_index = 0 |
748 |
if self.range_index > 0: |
749 |
out.append('')
|
750 |
out.append('--' + self.boundary) |
751 |
out.append('Content-Range: bytes %d-%d/%d' % (
|
752 |
self.offset, self.offset + self.length - 1, self.size)) |
753 |
out.append('Content-Transfer-Encoding: binary')
|
754 |
out.append('')
|
755 |
out.append('')
|
756 |
return '\r\n'.join(out) |
757 |
else:
|
758 |
# Footer.
|
759 |
out.append('')
|
760 |
out.append('--' + self.boundary + '--') |
761 |
out.append('')
|
762 |
return '\r\n'.join(out) |
763 |
|
764 |
|
765 |
def object_data_response(request, sizes, hashmaps, meta, public=False): |
766 |
"""Get the HttpResponse object for replying with the object's data."""
|
767 |
|
768 |
# Range handling.
|
769 |
size = sum(sizes)
|
770 |
ranges = get_range(request, size) |
771 |
if ranges is None: |
772 |
ranges = [(0, size)]
|
773 |
ret = 200
|
774 |
else:
|
775 |
check = [True for offset, length in ranges if |
776 |
length <= 0 or length > size or |
777 |
offset < 0 or offset >= size or |
778 |
offset + length > size] |
779 |
if len(check) > 0: |
780 |
raise RangeNotSatisfiable('Requested range exceeds object limits') |
781 |
ret = 206
|
782 |
if_range = request.META.get('HTTP_IF_RANGE')
|
783 |
if if_range:
|
784 |
try:
|
785 |
# Modification time has passed instead.
|
786 |
last_modified = parse_http_date(if_range) |
787 |
if last_modified != meta['modified']: |
788 |
ranges = [(0, size)]
|
789 |
ret = 200
|
790 |
except ValueError: |
791 |
if if_range != meta['checksum']: |
792 |
ranges = [(0, size)]
|
793 |
ret = 200
|
794 |
|
795 |
if ret == 206 and len(ranges) > 1: |
796 |
boundary = uuid.uuid4().hex |
797 |
else:
|
798 |
boundary = ''
|
799 |
wrapper = ObjectWrapper(request.backend, ranges, sizes, hashmaps, boundary) |
800 |
response = HttpResponse(wrapper, status=ret) |
801 |
put_object_headers(response, meta, public) |
802 |
if ret == 206: |
803 |
if len(ranges) == 1: |
804 |
offset, length = ranges[0]
|
805 |
response[ |
806 |
'Content-Length'] = length # Update with the correct length. |
807 |
response['Content-Range'] = 'bytes %d-%d/%d' % ( |
808 |
offset, offset + length - 1, size)
|
809 |
else:
|
810 |
del(response['Content-Length']) |
811 |
response['Content-Type'] = 'multipart/byteranges; boundary=%s' % ( |
812 |
boundary,) |
813 |
return response
|
814 |
|
815 |
|
816 |
def put_object_block(request, hashmap, data, offset): |
817 |
"""Put one block of data at the given offset."""
|
818 |
|
819 |
bi = int(offset / request.backend.block_size)
|
820 |
bo = offset % request.backend.block_size |
821 |
bl = min(len(data), request.backend.block_size - bo) |
822 |
if bi < len(hashmap): |
823 |
hashmap[bi] = request.backend.update_block(hashmap[bi], data[:bl], bo) |
824 |
else:
|
825 |
hashmap.append(request.backend.put_block(('\x00' * bo) + data[:bl]))
|
826 |
return bl # Return ammount of data written. |
827 |
|
828 |
|
829 |
def hashmap_md5(backend, hashmap, size): |
830 |
"""Produce the MD5 sum from the data in the hashmap."""
|
831 |
|
832 |
# TODO: Search backend for the MD5 of another object with the same hashmap and size...
|
833 |
md5 = hashlib.md5() |
834 |
bs = backend.block_size |
835 |
for bi, hash in enumerate(hashmap): |
836 |
data = backend.get_block(hash) # Blocks come in padded. |
837 |
if bi == len(hashmap) - 1: |
838 |
data = data[:size % bs] |
839 |
md5.update(data) |
840 |
return md5.hexdigest().lower()
|
841 |
|
842 |
|
843 |
def simple_list_response(request, l): |
844 |
if request.serialization == 'text': |
845 |
return '\n'.join(l) + '\n' |
846 |
if request.serialization == 'xml': |
847 |
return render_to_string('items.xml', {'items': l}) |
848 |
if request.serialization == 'json': |
849 |
return json.dumps(l)
|
850 |
|
851 |
|
852 |
|
853 |
def _get_backend(): |
854 |
backend = connect_backend(db_module=BACKEND_DB_MODULE, |
855 |
db_connection=BACKEND_DB_CONNECTION, |
856 |
block_module=BACKEND_BLOCK_MODULE, |
857 |
block_path=BACKEND_BLOCK_PATH, |
858 |
block_umask=BACKEND_BLOCK_UMASK, |
859 |
queue_module=BACKEND_QUEUE_MODULE, |
860 |
queue_hosts=BACKEND_QUEUE_HOSTS, |
861 |
queue_exchange=BACKEND_QUEUE_EXCHANGE) |
862 |
backend.default_policy['quota'] = BACKEND_QUOTA
|
863 |
backend.default_policy['versioning'] = BACKEND_VERSIONING
|
864 |
return backend
|
865 |
|
866 |
|
867 |
def _pooled_backend_close(backend): |
868 |
backend._pool.pool_put(backend) |
869 |
|
870 |
|
871 |
from synnefo.lib.pool import ObjectPool |
872 |
from new import instancemethod |
873 |
|
874 |
USAGE_LIMIT = 500
|
875 |
POOL_SIZE = 5
|
876 |
|
877 |
class PithosBackendPool(ObjectPool): |
878 |
def _pool_create(self): |
879 |
backend = _get_backend() |
880 |
backend._real_close = backend.close |
881 |
backend.close = instancemethod(_pooled_backend_close, backend, |
882 |
type(backend))
|
883 |
backend._pool = self
|
884 |
backend._use_count = USAGE_LIMIT |
885 |
return backend
|
886 |
|
887 |
def _pool_cleanup(self, backend): |
888 |
c = backend._use_count - 1
|
889 |
if c < 0: |
890 |
backend._real_close() |
891 |
return True |
892 |
|
893 |
backend._use_count = c |
894 |
return False |
895 |
|
896 |
_pithos_backend_pool = PithosBackendPool(size=POOL_SIZE) |
897 |
|
898 |
|
899 |
def get_backend(): |
900 |
return _pithos_backend_pool.pool_get()
|
901 |
|
902 |
|
903 |
def update_request_headers(request): |
904 |
# Handle URL-encoded keys and values.
|
905 |
meta = dict([(
|
906 |
k, v) for k, v in request.META.iteritems() if k.startswith('HTTP_')]) |
907 |
for k, v in meta.iteritems(): |
908 |
try:
|
909 |
k.decode('ascii')
|
910 |
v.decode('ascii')
|
911 |
except UnicodeDecodeError: |
912 |
raise BadRequest('Bad character in headers.') |
913 |
if '%' in k or '%' in v: |
914 |
del(request.META[k])
|
915 |
request.META[unquote(k)] = smart_unicode(unquote( |
916 |
v), strings_only=True)
|
917 |
|
918 |
|
919 |
def update_response_headers(request, response): |
920 |
if request.serialization == 'xml': |
921 |
response['Content-Type'] = 'application/xml; charset=UTF-8' |
922 |
elif request.serialization == 'json': |
923 |
response['Content-Type'] = 'application/json; charset=UTF-8' |
924 |
elif not response['Content-Type']: |
925 |
response['Content-Type'] = 'text/plain; charset=UTF-8' |
926 |
|
927 |
if (not response.has_header('Content-Length') and |
928 |
not (response.has_header('Content-Type') and |
929 |
response['Content-Type'].startswith('multipart/byteranges'))): |
930 |
response['Content-Length'] = len(response.content) |
931 |
|
932 |
# URL-encode unicode in headers.
|
933 |
meta = response.items() |
934 |
for k, v in meta: |
935 |
if (k.startswith('X-Account-') or k.startswith('X-Container-') or |
936 |
k.startswith('X-Object-') or k.startswith('Content-')): |
937 |
del(response[k])
|
938 |
response[quote(k)] = quote(v, safe='/=,:@; ')
|
939 |
|
940 |
|
941 |
def render_fault(request, fault): |
942 |
if isinstance(fault, InternalServerError) and settings.DEBUG: |
943 |
fault.details = format_exc(fault) |
944 |
|
945 |
request.serialization = 'text'
|
946 |
data = fault.message + '\n'
|
947 |
if fault.details:
|
948 |
data += '\n' + fault.details
|
949 |
response = HttpResponse(data, status=fault.code) |
950 |
update_response_headers(request, response) |
951 |
return response
|
952 |
|
953 |
|
954 |
def request_serialization(request, format_allowed=False): |
955 |
"""Return the serialization format requested.
|
956 |
|
957 |
Valid formats are 'text' and 'json', 'xml' if 'format_allowed' is True.
|
958 |
"""
|
959 |
|
960 |
if not format_allowed: |
961 |
return 'text' |
962 |
|
963 |
format = request.GET.get('format')
|
964 |
if format == 'json': |
965 |
return 'json' |
966 |
elif format == 'xml': |
967 |
return 'xml' |
968 |
|
969 |
for item in request.META.get('HTTP_ACCEPT', '').split(','): |
970 |
accept, sep, rest = item.strip().partition(';')
|
971 |
if accept == 'application/json': |
972 |
return 'json' |
973 |
elif accept == 'application/xml' or accept == 'text/xml': |
974 |
return 'xml' |
975 |
|
976 |
return 'text' |
977 |
|
978 |
|
979 |
def api_method(http_method=None, format_allowed=False, user_required=True): |
980 |
"""Decorator function for views that implement an API method."""
|
981 |
|
982 |
def decorator(func): |
983 |
@wraps(func)
|
984 |
def wrapper(request, *args, **kwargs): |
985 |
try:
|
986 |
if http_method and request.method != http_method: |
987 |
raise BadRequest('Method not allowed.') |
988 |
|
989 |
if user_required:
|
990 |
token = None
|
991 |
if request.method in ('HEAD', 'GET') and COOKIE_NAME in request.COOKIES: |
992 |
cookie_value = unquote( |
993 |
request.COOKIES.get(COOKIE_NAME, ''))
|
994 |
if cookie_value and '|' in cookie_value: |
995 |
token = cookie_value.split('|', 1)[1] |
996 |
get_user(request, |
997 |
AUTHENTICATION_URL, AUTHENTICATION_USERS, token) |
998 |
if getattr(request, 'user', None) is None: |
999 |
raise Unauthorized('Access denied') |
1000 |
|
1001 |
# The args variable may contain up to (account, container, object).
|
1002 |
if len(args) > 1 and len(args[1]) > 256: |
1003 |
raise BadRequest('Container name too large.') |
1004 |
if len(args) > 2 and len(args[2]) > 1024: |
1005 |
raise BadRequest('Object name too large.') |
1006 |
|
1007 |
# Format and check headers.
|
1008 |
update_request_headers(request) |
1009 |
|
1010 |
# Fill in custom request variables.
|
1011 |
request.serialization = request_serialization( |
1012 |
request, format_allowed) |
1013 |
request.backend = get_backend() |
1014 |
|
1015 |
response = func(request, *args, **kwargs) |
1016 |
update_response_headers(request, response) |
1017 |
return response
|
1018 |
except Fault, fault:
|
1019 |
return render_fault(request, fault)
|
1020 |
except BaseException, e: |
1021 |
logger.exception('Unexpected error: %s' % e)
|
1022 |
fault = InternalServerError('Unexpected error: %s' % e)
|
1023 |
return render_fault(request, fault)
|
1024 |
finally:
|
1025 |
if getattr(request, 'backend', None) is not None: |
1026 |
request.backend.close() |
1027 |
return wrapper
|
1028 |
return decorator
|