root / snf-pithos-app / pithos / api / util.py @ 06e7d2f0
History | View | Annotate | Download (39 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 datetime import datetime |
36 |
from urllib import quote, unquote |
37 |
|
38 |
from django.http import HttpResponse |
39 |
from django.template.loader import render_to_string |
40 |
from django.utils import simplejson as json |
41 |
from django.utils.http import http_date, parse_etags |
42 |
from django.utils.encoding import smart_unicode, smart_str |
43 |
from django.core.files.uploadhandler import FileUploadHandler |
44 |
from django.core.files.uploadedfile import UploadedFile |
45 |
|
46 |
from snf_django.lib.api.parsedate import parse_http_date_safe, parse_http_date |
47 |
from snf_django.lib import api |
48 |
from snf_django.lib.api import faults, utils |
49 |
|
50 |
from pithos.api.settings import (BACKEND_DB_MODULE, BACKEND_DB_CONNECTION, |
51 |
BACKEND_BLOCK_MODULE, BACKEND_BLOCK_PATH, |
52 |
BACKEND_BLOCK_UMASK, |
53 |
BACKEND_QUEUE_MODULE, BACKEND_QUEUE_HOSTS, |
54 |
BACKEND_QUEUE_EXCHANGE, |
55 |
ASTAKOSCLIENT_POOLSIZE, |
56 |
SERVICE_TOKEN, |
57 |
ASTAKOS_BASE_URL, |
58 |
BACKEND_ACCOUNT_QUOTA, BACKEND_CONTAINER_QUOTA, |
59 |
BACKEND_VERSIONING, |
60 |
BACKEND_FREE_VERSIONING, BACKEND_POOL_SIZE, |
61 |
RADOS_STORAGE, RADOS_POOL_BLOCKS, |
62 |
RADOS_POOL_MAPS, TRANSLATE_UUIDS, |
63 |
PUBLIC_URL_SECURITY, |
64 |
PUBLIC_URL_ALPHABET) |
65 |
from pithos.api.resources import resources |
66 |
from pithos.backends.base import (NotAllowedError, QuotaError, ItemNotExists, |
67 |
VersionNotExists) |
68 |
from astakosclient import AstakosClient |
69 |
from astakosclient.errors import NoUserName, NoUUID |
70 |
|
71 |
import logging |
72 |
import re |
73 |
import hashlib |
74 |
import uuid |
75 |
import decimal |
76 |
|
77 |
logger = logging.getLogger(__name__) |
78 |
|
79 |
|
80 |
def json_encode_decimal(obj): |
81 |
if isinstance(obj, decimal.Decimal): |
82 |
return str(obj) |
83 |
raise TypeError(repr(obj) + " is not JSON serializable") |
84 |
|
85 |
|
86 |
def rename_meta_key(d, old, new): |
87 |
if old not in d: |
88 |
return
|
89 |
d[new] = d[old] |
90 |
del(d[old])
|
91 |
|
92 |
|
93 |
def printable_header_dict(d): |
94 |
"""Format a meta dictionary for printing out json/xml.
|
95 |
|
96 |
Convert all keys to lower case and replace dashes with underscores.
|
97 |
Format 'last_modified' timestamp.
|
98 |
"""
|
99 |
|
100 |
if 'last_modified' in d and d['last_modified']: |
101 |
d['last_modified'] = utils.isoformat(
|
102 |
datetime.fromtimestamp(d['last_modified']))
|
103 |
return dict([(k.lower().replace('-', '_'), v) for k, v in d.iteritems()]) |
104 |
|
105 |
|
106 |
def format_header_key(k): |
107 |
"""Convert underscores to dashes and capitalize intra-dash strings."""
|
108 |
return '-'.join([x.capitalize() for x in k.replace('_', '-').split('-')]) |
109 |
|
110 |
|
111 |
def get_header_prefix(request, prefix): |
112 |
"""Get all prefix-* request headers in a dict.
|
113 |
Reformat keys with format_header_key()."""
|
114 |
|
115 |
prefix = 'HTTP_' + prefix.upper().replace('-', '_') |
116 |
# TODO: Document or remove '~' replacing.
|
117 |
return dict([(format_header_key(k[5:]), v.replace('~', '')) |
118 |
for k, v in request.META.iteritems() |
119 |
if k.startswith(prefix) and len(k) > len(prefix)]) |
120 |
|
121 |
|
122 |
def check_meta_headers(meta): |
123 |
if len(meta) > 90: |
124 |
raise faults.BadRequest('Too many headers.') |
125 |
for k, v in meta.iteritems(): |
126 |
if len(k) > 128: |
127 |
raise faults.BadRequest('Header name too large.') |
128 |
if len(v) > 256: |
129 |
raise faults.BadRequest('Header value too large.') |
130 |
|
131 |
|
132 |
def get_account_headers(request): |
133 |
meta = get_header_prefix(request, 'X-Account-Meta-')
|
134 |
check_meta_headers(meta) |
135 |
groups = {} |
136 |
for k, v in get_header_prefix(request, 'X-Account-Group-').iteritems(): |
137 |
n = k[16:].lower()
|
138 |
if '-' in n or '_' in n: |
139 |
raise faults.BadRequest('Bad characters in group name') |
140 |
groups[n] = v.replace(' ', '').split(',') |
141 |
while '' in groups[n]: |
142 |
groups[n].remove('')
|
143 |
return meta, groups
|
144 |
|
145 |
|
146 |
def put_account_headers(response, meta, groups, policy): |
147 |
if 'count' in meta: |
148 |
response['X-Account-Container-Count'] = meta['count'] |
149 |
if 'bytes' in meta: |
150 |
response['X-Account-Bytes-Used'] = meta['bytes'] |
151 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
152 |
for k in [x for x in meta.keys() if x.startswith('X-Account-Meta-')]: |
153 |
response[smart_str( |
154 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
155 |
if 'until_timestamp' in meta: |
156 |
response['X-Account-Until-Timestamp'] = http_date(
|
157 |
int(meta['until_timestamp'])) |
158 |
for k, v in groups.iteritems(): |
159 |
k = smart_str(k, strings_only=True)
|
160 |
k = format_header_key('X-Account-Group-' + k)
|
161 |
v = smart_str(','.join(v), strings_only=True) |
162 |
response[k] = v |
163 |
for k, v in policy.iteritems(): |
164 |
response[smart_str(format_header_key('X-Account-Policy-' + k),
|
165 |
strings_only=True)] = smart_str(v, strings_only=True) |
166 |
|
167 |
|
168 |
def get_container_headers(request): |
169 |
meta = get_header_prefix(request, 'X-Container-Meta-')
|
170 |
check_meta_headers(meta) |
171 |
policy = dict([(k[19:].lower(), v.replace(' ', '')) for k, v in |
172 |
get_header_prefix(request, |
173 |
'X-Container-Policy-').iteritems()])
|
174 |
return meta, policy
|
175 |
|
176 |
|
177 |
def put_container_headers(request, response, meta, policy): |
178 |
if 'count' in meta: |
179 |
response['X-Container-Object-Count'] = meta['count'] |
180 |
if 'bytes' in meta: |
181 |
response['X-Container-Bytes-Used'] = meta['bytes'] |
182 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
183 |
for k in [x for x in meta.keys() if x.startswith('X-Container-Meta-')]: |
184 |
response[smart_str( |
185 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
186 |
l = [smart_str(x, strings_only=True) for x in meta['object_meta'] |
187 |
if x.startswith('X-Object-Meta-')] |
188 |
response['X-Container-Object-Meta'] = ','.join([x[14:] for x in l]) |
189 |
response['X-Container-Block-Size'] = request.backend.block_size
|
190 |
response['X-Container-Block-Hash'] = request.backend.hash_algorithm
|
191 |
if 'until_timestamp' in meta: |
192 |
response['X-Container-Until-Timestamp'] = http_date(
|
193 |
int(meta['until_timestamp'])) |
194 |
for k, v in policy.iteritems(): |
195 |
response[smart_str(format_header_key('X-Container-Policy-' + k),
|
196 |
strings_only=True)] = smart_str(v, strings_only=True) |
197 |
|
198 |
|
199 |
def get_object_headers(request): |
200 |
content_type = request.META.get('CONTENT_TYPE', None) |
201 |
meta = get_header_prefix(request, 'X-Object-Meta-')
|
202 |
check_meta_headers(meta) |
203 |
if request.META.get('HTTP_CONTENT_ENCODING'): |
204 |
meta['Content-Encoding'] = request.META['HTTP_CONTENT_ENCODING'] |
205 |
if request.META.get('HTTP_CONTENT_DISPOSITION'): |
206 |
meta['Content-Disposition'] = request.META['HTTP_CONTENT_DISPOSITION'] |
207 |
if request.META.get('HTTP_X_OBJECT_MANIFEST'): |
208 |
meta['X-Object-Manifest'] = request.META['HTTP_X_OBJECT_MANIFEST'] |
209 |
return content_type, meta, get_sharing(request), get_public(request)
|
210 |
|
211 |
|
212 |
def put_object_headers(response, meta, restricted=False, token=None): |
213 |
response['ETag'] = meta['checksum'] |
214 |
response['Content-Length'] = meta['bytes'] |
215 |
response['Content-Type'] = meta.get('type', 'application/octet-stream') |
216 |
response['Last-Modified'] = http_date(int(meta['modified'])) |
217 |
if not restricted: |
218 |
response['X-Object-Hash'] = meta['hash'] |
219 |
response['X-Object-UUID'] = meta['uuid'] |
220 |
if TRANSLATE_UUIDS:
|
221 |
meta['modified_by'] = \
|
222 |
retrieve_displayname(token, meta['modified_by'])
|
223 |
response['X-Object-Modified-By'] = smart_str(
|
224 |
meta['modified_by'], strings_only=True) |
225 |
response['X-Object-Version'] = meta['version'] |
226 |
response['X-Object-Version-Timestamp'] = http_date(
|
227 |
int(meta['version_timestamp'])) |
228 |
for k in [x for x in meta.keys() if x.startswith('X-Object-Meta-')]: |
229 |
response[smart_str( |
230 |
k, strings_only=True)] = smart_str(meta[k], strings_only=True) |
231 |
for k in ( |
232 |
'Content-Encoding', 'Content-Disposition', 'X-Object-Manifest', |
233 |
'X-Object-Sharing', 'X-Object-Shared-By', 'X-Object-Allowed-To', |
234 |
'X-Object-Public'):
|
235 |
if k in meta: |
236 |
response[k] = smart_str(meta[k], strings_only=True)
|
237 |
else:
|
238 |
for k in ('Content-Encoding', 'Content-Disposition'): |
239 |
if k in meta: |
240 |
response[k] = smart_str(meta[k], strings_only=True)
|
241 |
|
242 |
|
243 |
def update_manifest_meta(request, v_account, meta): |
244 |
"""Update metadata if the object has an X-Object-Manifest."""
|
245 |
|
246 |
if 'X-Object-Manifest' in meta: |
247 |
etag = ''
|
248 |
bytes = 0
|
249 |
try:
|
250 |
src_container, src_name = split_container_object_string( |
251 |
'/' + meta['X-Object-Manifest']) |
252 |
objects = request.backend.list_objects( |
253 |
request.user_uniq, v_account, |
254 |
src_container, prefix=src_name, virtual=False)
|
255 |
for x in objects: |
256 |
src_meta = request.backend.get_object_meta(request.user_uniq, |
257 |
v_account, |
258 |
src_container, |
259 |
x[0], 'pithos', x[1]) |
260 |
etag += src_meta['checksum']
|
261 |
bytes += src_meta['bytes'] |
262 |
except:
|
263 |
# Ignore errors.
|
264 |
return
|
265 |
meta['bytes'] = bytes |
266 |
md5 = hashlib.md5() |
267 |
md5.update(etag) |
268 |
meta['checksum'] = md5.hexdigest().lower()
|
269 |
|
270 |
|
271 |
def is_uuid(str): |
272 |
if str is None: |
273 |
return False |
274 |
try:
|
275 |
uuid.UUID(str)
|
276 |
except ValueError: |
277 |
return False |
278 |
else:
|
279 |
return True |
280 |
|
281 |
|
282 |
##########################
|
283 |
# USER CATALOG utilities #
|
284 |
##########################
|
285 |
|
286 |
def retrieve_displayname(token, uuid, fail_silently=True): |
287 |
astakos = AstakosClient(ASTAKOS_BASE_URL, retry=2, use_pool=True, |
288 |
logger=logger) |
289 |
try:
|
290 |
displayname = astakos.get_username(token, uuid) |
291 |
except NoUserName:
|
292 |
if not fail_silently: |
293 |
raise ItemNotExists(uuid)
|
294 |
else:
|
295 |
# just return the uuid
|
296 |
return uuid
|
297 |
return displayname
|
298 |
|
299 |
|
300 |
def retrieve_displaynames(token, uuids, return_dict=False, fail_silently=True): |
301 |
astakos = AstakosClient(ASTAKOS_BASE_URL, retry=2, use_pool=True, |
302 |
logger=logger) |
303 |
catalog = astakos.get_usernames(token, uuids) or {}
|
304 |
missing = list(set(uuids) - set(catalog)) |
305 |
if missing and not fail_silently: |
306 |
raise ItemNotExists('Unknown displaynames: %s' % ', '.join(missing)) |
307 |
return catalog if return_dict else [catalog.get(i) for i in uuids] |
308 |
|
309 |
|
310 |
def retrieve_uuid(token, displayname): |
311 |
if is_uuid(displayname):
|
312 |
return displayname
|
313 |
|
314 |
astakos = AstakosClient(ASTAKOS_BASE_URL, retry=2, use_pool=True, |
315 |
logger=logger) |
316 |
try:
|
317 |
uuid = astakos.get_uuid(token, displayname) |
318 |
except NoUUID:
|
319 |
raise ItemNotExists(displayname)
|
320 |
return uuid
|
321 |
|
322 |
|
323 |
def retrieve_uuids(token, displaynames, return_dict=False, fail_silently=True): |
324 |
astakos = AstakosClient(ASTAKOS_BASE_URL, retry=2, use_pool=True, |
325 |
logger=logger) |
326 |
catalog = astakos.get_uuids(token, displaynames) or {}
|
327 |
missing = list(set(displaynames) - set(catalog)) |
328 |
if missing and not fail_silently: |
329 |
raise ItemNotExists('Unknown uuids: %s' % ', '.join(missing)) |
330 |
return catalog if return_dict else [catalog.get(i) for i in displaynames] |
331 |
|
332 |
|
333 |
def replace_permissions_displayname(token, holder): |
334 |
if holder == '*': |
335 |
return holder
|
336 |
try:
|
337 |
# check first for a group permission
|
338 |
account, group = holder.split(':', 1) |
339 |
except ValueError: |
340 |
return retrieve_uuid(token, holder)
|
341 |
else:
|
342 |
return ':'.join([retrieve_uuid(token, account), group]) |
343 |
|
344 |
|
345 |
def replace_permissions_uuid(token, holder): |
346 |
if holder == '*': |
347 |
return holder
|
348 |
try:
|
349 |
# check first for a group permission
|
350 |
account, group = holder.split(':', 1) |
351 |
except ValueError: |
352 |
return retrieve_displayname(token, holder)
|
353 |
else:
|
354 |
return ':'.join([retrieve_displayname(token, account), group]) |
355 |
|
356 |
|
357 |
def update_sharing_meta(request, permissions, v_account, |
358 |
v_container, v_object, meta): |
359 |
if permissions is None: |
360 |
return
|
361 |
allowed, perm_path, perms = permissions |
362 |
if len(perms) == 0: |
363 |
return
|
364 |
|
365 |
# replace uuid with displayname
|
366 |
if TRANSLATE_UUIDS:
|
367 |
perms['read'] = [replace_permissions_uuid(
|
368 |
getattr(request, 'token', None), x) |
369 |
for x in perms.get('read', [])] |
370 |
perms['write'] = [replace_permissions_uuid(
|
371 |
getattr(request, 'token', None), x) |
372 |
for x in perms.get('write', [])] |
373 |
|
374 |
ret = [] |
375 |
|
376 |
r = ','.join(perms.get('read', [])) |
377 |
if r:
|
378 |
ret.append('read=' + r)
|
379 |
w = ','.join(perms.get('write', [])) |
380 |
if w:
|
381 |
ret.append('write=' + w)
|
382 |
meta['X-Object-Sharing'] = '; '.join(ret) |
383 |
if '/'.join((v_account, v_container, v_object)) != perm_path: |
384 |
meta['X-Object-Shared-By'] = perm_path
|
385 |
if request.user_uniq != v_account:
|
386 |
meta['X-Object-Allowed-To'] = allowed
|
387 |
|
388 |
|
389 |
def update_public_meta(public, meta): |
390 |
if not public: |
391 |
return
|
392 |
meta['X-Object-Public'] = '/public/' + public |
393 |
|
394 |
|
395 |
def validate_modification_preconditions(request, meta): |
396 |
"""Check that the modified timestamp conforms with the preconditions set."""
|
397 |
|
398 |
if 'modified' not in meta: |
399 |
return # TODO: Always return? |
400 |
|
401 |
if_modified_since = request.META.get('HTTP_IF_MODIFIED_SINCE')
|
402 |
if if_modified_since is not None: |
403 |
if_modified_since = parse_http_date_safe(if_modified_since) |
404 |
if (if_modified_since is not None |
405 |
and int(meta['modified']) <= if_modified_since): |
406 |
raise faults.NotModified('Resource has not been modified') |
407 |
|
408 |
if_unmodified_since = request.META.get('HTTP_IF_UNMODIFIED_SINCE')
|
409 |
if if_unmodified_since is not None: |
410 |
if_unmodified_since = parse_http_date_safe(if_unmodified_since) |
411 |
if (if_unmodified_since is not None |
412 |
and int(meta['modified']) > if_unmodified_since): |
413 |
raise faults.PreconditionFailed('Resource has been modified') |
414 |
|
415 |
|
416 |
def validate_matching_preconditions(request, meta): |
417 |
"""Check that the ETag conforms with the preconditions set."""
|
418 |
|
419 |
etag = meta['checksum']
|
420 |
if not etag: |
421 |
etag = None
|
422 |
|
423 |
if_match = request.META.get('HTTP_IF_MATCH')
|
424 |
if if_match is not None: |
425 |
if etag is None: |
426 |
raise faults.PreconditionFailed('Resource does not exist') |
427 |
if (if_match != '*' |
428 |
and etag not in [x.lower() for x in parse_etags(if_match)]): |
429 |
raise faults.PreconditionFailed('Resource ETag does not match') |
430 |
|
431 |
if_none_match = request.META.get('HTTP_IF_NONE_MATCH')
|
432 |
if if_none_match is not None: |
433 |
# TODO: If this passes, must ignore If-Modified-Since header.
|
434 |
if etag is not None: |
435 |
if (if_none_match == '*' |
436 |
or etag in [x.lower() for x in parse_etags(if_none_match)]): |
437 |
# TODO: Continue if an If-Modified-Since header is present.
|
438 |
if request.method in ('HEAD', 'GET'): |
439 |
raise faults.NotModified('Resource ETag matches') |
440 |
raise faults.PreconditionFailed(
|
441 |
'Resource exists or ETag matches')
|
442 |
|
443 |
|
444 |
def split_container_object_string(s): |
445 |
if not len(s) > 0 or s[0] != '/': |
446 |
raise ValueError |
447 |
s = s[1:]
|
448 |
pos = s.find('/')
|
449 |
if pos == -1 or pos == len(s) - 1: |
450 |
raise ValueError |
451 |
return s[:pos], s[(pos + 1):] |
452 |
|
453 |
|
454 |
def copy_or_move_object(request, src_account, src_container, src_name, |
455 |
dest_account, dest_container, dest_name, |
456 |
move=False, delimiter=None): |
457 |
"""Copy or move an object."""
|
458 |
|
459 |
if 'ignore_content_type' in request.GET and 'CONTENT_TYPE' in request.META: |
460 |
del(request.META['CONTENT_TYPE']) |
461 |
content_type, meta, permissions, public = get_object_headers(request) |
462 |
src_version = request.META.get('HTTP_X_SOURCE_VERSION')
|
463 |
try:
|
464 |
if move:
|
465 |
version_id = request.backend.move_object( |
466 |
request.user_uniq, src_account, src_container, src_name, |
467 |
dest_account, dest_container, dest_name, |
468 |
content_type, 'pithos', meta, False, permissions, delimiter) |
469 |
else:
|
470 |
version_id = request.backend.copy_object( |
471 |
request.user_uniq, src_account, src_container, src_name, |
472 |
dest_account, dest_container, dest_name, |
473 |
content_type, 'pithos', meta, False, permissions, |
474 |
src_version, delimiter) |
475 |
except NotAllowedError:
|
476 |
raise faults.Forbidden('Not allowed') |
477 |
except (ItemNotExists, VersionNotExists):
|
478 |
raise faults.ItemNotFound('Container or object does not exist') |
479 |
except ValueError: |
480 |
raise faults.BadRequest('Invalid sharing header') |
481 |
except QuotaError, e:
|
482 |
raise faults.RequestEntityTooLarge('Quota error: %s' % e) |
483 |
if public is not None: |
484 |
try:
|
485 |
request.backend.update_object_public( |
486 |
request.user_uniq, dest_account, |
487 |
dest_container, dest_name, public) |
488 |
except NotAllowedError:
|
489 |
raise faults.Forbidden('Not allowed') |
490 |
except ItemNotExists:
|
491 |
raise faults.ItemNotFound('Object does not exist') |
492 |
return version_id
|
493 |
|
494 |
|
495 |
def get_int_parameter(p): |
496 |
if p is not None: |
497 |
try:
|
498 |
p = int(p)
|
499 |
except ValueError: |
500 |
return None |
501 |
if p < 0: |
502 |
return None |
503 |
return p
|
504 |
|
505 |
|
506 |
def get_content_length(request): |
507 |
content_length = get_int_parameter(request.META.get('CONTENT_LENGTH'))
|
508 |
if content_length is None: |
509 |
raise faults.LengthRequired('Missing or invalid Content-Length header') |
510 |
return content_length
|
511 |
|
512 |
|
513 |
def get_range(request, size): |
514 |
"""Parse a Range header from the request.
|
515 |
|
516 |
Either returns None, when the header is not existent or should be ignored,
|
517 |
or a list of (offset, length) tuples - should be further checked.
|
518 |
"""
|
519 |
|
520 |
ranges = request.META.get('HTTP_RANGE', '').replace(' ', '') |
521 |
if not ranges.startswith('bytes='): |
522 |
return None |
523 |
|
524 |
ret = [] |
525 |
for r in (x.strip() for x in ranges[6:].split(',')): |
526 |
p = re.compile('^(?P<offset>\d*)-(?P<upto>\d*)$')
|
527 |
m = p.match(r) |
528 |
if not m: |
529 |
return None |
530 |
offset = m.group('offset')
|
531 |
upto = m.group('upto')
|
532 |
if offset == '' and upto == '': |
533 |
return None |
534 |
|
535 |
if offset != '': |
536 |
offset = int(offset)
|
537 |
if upto != '': |
538 |
upto = int(upto)
|
539 |
if offset > upto:
|
540 |
return None |
541 |
ret.append((offset, upto - offset + 1))
|
542 |
else:
|
543 |
ret.append((offset, size - offset)) |
544 |
else:
|
545 |
length = int(upto)
|
546 |
ret.append((size - length, length)) |
547 |
|
548 |
return ret
|
549 |
|
550 |
|
551 |
def get_content_range(request): |
552 |
"""Parse a Content-Range header from the request.
|
553 |
|
554 |
Either returns None, when the header is not existent or should be ignored,
|
555 |
or an (offset, length, total) tuple - check as length, total may be None.
|
556 |
Returns (None, None, None) if the provided range is '*/*'.
|
557 |
"""
|
558 |
|
559 |
ranges = request.META.get('HTTP_CONTENT_RANGE', '') |
560 |
if not ranges: |
561 |
return None |
562 |
|
563 |
p = re.compile('^bytes (?P<offset>\d+)-(?P<upto>\d*)/(?P<total>(\d+|\*))$')
|
564 |
m = p.match(ranges) |
565 |
if not m: |
566 |
if ranges == 'bytes */*': |
567 |
return (None, None, None) |
568 |
return None |
569 |
offset = int(m.group('offset')) |
570 |
upto = m.group('upto')
|
571 |
total = m.group('total')
|
572 |
if upto != '': |
573 |
upto = int(upto)
|
574 |
else:
|
575 |
upto = None
|
576 |
if total != '*': |
577 |
total = int(total)
|
578 |
else:
|
579 |
total = None
|
580 |
if (upto is not None and offset > upto) or \ |
581 |
(total is not None and offset >= total) or \ |
582 |
(total is not None and upto is not None and upto >= total): |
583 |
return None |
584 |
|
585 |
if upto is None: |
586 |
length = None
|
587 |
else:
|
588 |
length = upto - offset + 1
|
589 |
return (offset, length, total)
|
590 |
|
591 |
|
592 |
def get_sharing(request): |
593 |
"""Parse an X-Object-Sharing header from the request.
|
594 |
|
595 |
Raises BadRequest on error.
|
596 |
"""
|
597 |
|
598 |
permissions = request.META.get('HTTP_X_OBJECT_SHARING')
|
599 |
if permissions is None: |
600 |
return None |
601 |
|
602 |
# TODO: Document or remove '~' replacing.
|
603 |
permissions = permissions.replace('~', '') |
604 |
|
605 |
ret = {} |
606 |
permissions = permissions.replace(' ', '') |
607 |
if permissions == '': |
608 |
return ret
|
609 |
for perm in (x for x in permissions.split(';')): |
610 |
if perm.startswith('read='): |
611 |
ret['read'] = list(set( |
612 |
[v.replace(' ', '').lower() for v in perm[5:].split(',')])) |
613 |
if '' in ret['read']: |
614 |
ret['read'].remove('') |
615 |
if '*' in ret['read']: |
616 |
ret['read'] = ['*'] |
617 |
if len(ret['read']) == 0: |
618 |
raise faults.BadRequest(
|
619 |
'Bad X-Object-Sharing header value: invalid length')
|
620 |
elif perm.startswith('write='): |
621 |
ret['write'] = list(set( |
622 |
[v.replace(' ', '').lower() for v in perm[6:].split(',')])) |
623 |
if '' in ret['write']: |
624 |
ret['write'].remove('') |
625 |
if '*' in ret['write']: |
626 |
ret['write'] = ['*'] |
627 |
if len(ret['write']) == 0: |
628 |
raise faults.BadRequest(
|
629 |
'Bad X-Object-Sharing header value: invalid length')
|
630 |
else:
|
631 |
raise faults.BadRequest(
|
632 |
'Bad X-Object-Sharing header value: missing prefix')
|
633 |
|
634 |
# replace displayname with uuid
|
635 |
if TRANSLATE_UUIDS:
|
636 |
try:
|
637 |
ret['read'] = [replace_permissions_displayname(
|
638 |
getattr(request, 'token', None), x) |
639 |
for x in ret.get('read', [])] |
640 |
ret['write'] = [replace_permissions_displayname(
|
641 |
getattr(request, 'token', None), x) |
642 |
for x in ret.get('write', [])] |
643 |
except ItemNotExists, e:
|
644 |
raise faults.BadRequest(
|
645 |
'Bad X-Object-Sharing header value: unknown account: %s' % e)
|
646 |
|
647 |
# Keep duplicates only in write list.
|
648 |
dups = [x for x in ret.get( |
649 |
'read', []) if x in ret.get('write', []) and x != '*'] |
650 |
if dups:
|
651 |
for x in dups: |
652 |
ret['read'].remove(x)
|
653 |
if len(ret['read']) == 0: |
654 |
del(ret['read']) |
655 |
|
656 |
return ret
|
657 |
|
658 |
|
659 |
def get_public(request): |
660 |
"""Parse an X-Object-Public header from the request.
|
661 |
|
662 |
Raises BadRequest on error.
|
663 |
"""
|
664 |
|
665 |
public = request.META.get('HTTP_X_OBJECT_PUBLIC')
|
666 |
if public is None: |
667 |
return None |
668 |
|
669 |
public = public.replace(' ', '').lower() |
670 |
if public == 'true': |
671 |
return True |
672 |
elif public == 'false' or public == '': |
673 |
return False |
674 |
raise faults.BadRequest('Bad X-Object-Public header value') |
675 |
|
676 |
|
677 |
def raw_input_socket(request): |
678 |
"""Return the socket for reading the rest of the request."""
|
679 |
|
680 |
server_software = request.META.get('SERVER_SOFTWARE')
|
681 |
if server_software and server_software.startswith('mod_python'): |
682 |
return request._req
|
683 |
if 'wsgi.input' in request.environ: |
684 |
return request.environ['wsgi.input'] |
685 |
raise NotImplemented('Unknown server software') |
686 |
|
687 |
MAX_UPLOAD_SIZE = 5 * (1024 * 1024 * 1024) # 5GB |
688 |
|
689 |
|
690 |
def socket_read_iterator(request, length=0, blocksize=4096): |
691 |
"""Return a maximum of blocksize data read from the socket in each iteration
|
692 |
|
693 |
Read up to 'length'. If 'length' is negative, will attempt a chunked read.
|
694 |
The maximum ammount of data read is controlled by MAX_UPLOAD_SIZE.
|
695 |
"""
|
696 |
|
697 |
sock = raw_input_socket(request) |
698 |
if length < 0: # Chunked transfers |
699 |
# Small version (server does the dechunking).
|
700 |
if (request.environ.get('mod_wsgi.input_chunked', None) |
701 |
or request.META['SERVER_SOFTWARE'].startswith('gunicorn')): |
702 |
while length < MAX_UPLOAD_SIZE:
|
703 |
data = sock.read(blocksize) |
704 |
if data == '': |
705 |
return
|
706 |
yield data
|
707 |
raise faults.BadRequest('Maximum size is reached') |
708 |
|
709 |
# Long version (do the dechunking).
|
710 |
data = ''
|
711 |
while length < MAX_UPLOAD_SIZE:
|
712 |
# Get chunk size.
|
713 |
if hasattr(sock, 'readline'): |
714 |
chunk_length = sock.readline() |
715 |
else:
|
716 |
chunk_length = ''
|
717 |
while chunk_length[-1:] != '\n': |
718 |
chunk_length += sock.read(1)
|
719 |
chunk_length.strip() |
720 |
pos = chunk_length.find(';')
|
721 |
if pos >= 0: |
722 |
chunk_length = chunk_length[:pos] |
723 |
try:
|
724 |
chunk_length = int(chunk_length, 16) |
725 |
except Exception: |
726 |
raise faults.BadRequest('Bad chunk size') |
727 |
# TODO: Change to something more appropriate.
|
728 |
# Check if done.
|
729 |
if chunk_length == 0: |
730 |
if len(data) > 0: |
731 |
yield data
|
732 |
return
|
733 |
# Get the actual data.
|
734 |
while chunk_length > 0: |
735 |
chunk = sock.read(min(chunk_length, blocksize))
|
736 |
chunk_length -= len(chunk)
|
737 |
if length > 0: |
738 |
length += len(chunk)
|
739 |
data += chunk |
740 |
if len(data) >= blocksize: |
741 |
ret = data[:blocksize] |
742 |
data = data[blocksize:] |
743 |
yield ret
|
744 |
sock.read(2) # CRLF |
745 |
raise faults.BadRequest('Maximum size is reached') |
746 |
else:
|
747 |
if length > MAX_UPLOAD_SIZE:
|
748 |
raise faults.BadRequest('Maximum size is reached') |
749 |
while length > 0: |
750 |
data = sock.read(min(length, blocksize))
|
751 |
if not data: |
752 |
raise faults.BadRequest()
|
753 |
length -= len(data)
|
754 |
yield data
|
755 |
|
756 |
|
757 |
class SaveToBackendHandler(FileUploadHandler): |
758 |
"""Handle a file from an HTML form the django way."""
|
759 |
|
760 |
def __init__(self, request=None): |
761 |
super(SaveToBackendHandler, self).__init__(request) |
762 |
self.backend = request.backend
|
763 |
|
764 |
def put_data(self, length): |
765 |
if len(self.data) >= length: |
766 |
block = self.data[:length]
|
767 |
self.file.hashmap.append(self.backend.put_block(block)) |
768 |
self.md5.update(block)
|
769 |
self.data = self.data[length:] |
770 |
|
771 |
def new_file(self, field_name, file_name, content_type, |
772 |
content_length, charset=None):
|
773 |
self.md5 = hashlib.md5()
|
774 |
self.data = '' |
775 |
self.file = UploadedFile(
|
776 |
name=file_name, content_type=content_type, charset=charset) |
777 |
self.file.size = 0 |
778 |
self.file.hashmap = []
|
779 |
|
780 |
def receive_data_chunk(self, raw_data, start): |
781 |
self.data += raw_data
|
782 |
self.file.size += len(raw_data) |
783 |
self.put_data(self.request.backend.block_size) |
784 |
return None |
785 |
|
786 |
def file_complete(self, file_size): |
787 |
l = len(self.data) |
788 |
if l > 0: |
789 |
self.put_data(l)
|
790 |
self.file.etag = self.md5.hexdigest().lower() |
791 |
return self.file |
792 |
|
793 |
|
794 |
class ObjectWrapper(object): |
795 |
"""Return the object's data block-per-block in each iteration.
|
796 |
|
797 |
Read from the object using the offset and length provided
|
798 |
in each entry of the range list.
|
799 |
"""
|
800 |
|
801 |
def __init__(self, backend, ranges, sizes, hashmaps, boundary): |
802 |
self.backend = backend
|
803 |
self.ranges = ranges
|
804 |
self.sizes = sizes
|
805 |
self.hashmaps = hashmaps
|
806 |
self.boundary = boundary
|
807 |
self.size = sum(self.sizes) |
808 |
|
809 |
self.file_index = 0 |
810 |
self.block_index = 0 |
811 |
self.block_hash = -1 |
812 |
self.block = '' |
813 |
|
814 |
self.range_index = -1 |
815 |
self.offset, self.length = self.ranges[0] |
816 |
|
817 |
def __iter__(self): |
818 |
return self |
819 |
|
820 |
def part_iterator(self): |
821 |
if self.length > 0: |
822 |
# Get the file for the current offset.
|
823 |
file_size = self.sizes[self.file_index] |
824 |
while self.offset >= file_size: |
825 |
self.offset -= file_size
|
826 |
self.file_index += 1 |
827 |
file_size = self.sizes[self.file_index] |
828 |
|
829 |
# Get the block for the current position.
|
830 |
self.block_index = int(self.offset / self.backend.block_size) |
831 |
if self.block_hash != \ |
832 |
self.hashmaps[self.file_index][self.block_index]: |
833 |
self.block_hash = self.hashmaps[ |
834 |
self.file_index][self.block_index] |
835 |
try:
|
836 |
self.block = self.backend.get_block(self.block_hash) |
837 |
except ItemNotExists:
|
838 |
raise faults.ItemNotFound('Block does not exist') |
839 |
|
840 |
# Get the data from the block.
|
841 |
bo = self.offset % self.backend.block_size |
842 |
bs = self.backend.block_size
|
843 |
if (self.block_index == len(self.hashmaps[self.file_index]) - 1 and |
844 |
self.sizes[self.file_index] % self.backend.block_size): |
845 |
bs = self.sizes[self.file_index] % self.backend.block_size |
846 |
bl = min(self.length, bs - bo) |
847 |
data = self.block[bo:bo + bl]
|
848 |
self.offset += bl
|
849 |
self.length -= bl
|
850 |
return data
|
851 |
else:
|
852 |
raise StopIteration |
853 |
|
854 |
def next(self): |
855 |
if len(self.ranges) == 1: |
856 |
return self.part_iterator() |
857 |
if self.range_index == len(self.ranges): |
858 |
raise StopIteration |
859 |
try:
|
860 |
if self.range_index == -1: |
861 |
raise StopIteration |
862 |
return self.part_iterator() |
863 |
except StopIteration: |
864 |
self.range_index += 1 |
865 |
out = [] |
866 |
if self.range_index < len(self.ranges): |
867 |
# Part header.
|
868 |
self.offset, self.length = self.ranges[self.range_index] |
869 |
self.file_index = 0 |
870 |
if self.range_index > 0: |
871 |
out.append('')
|
872 |
out.append('--' + self.boundary) |
873 |
out.append('Content-Range: bytes %d-%d/%d' % (
|
874 |
self.offset, self.offset + self.length - 1, self.size)) |
875 |
out.append('Content-Transfer-Encoding: binary')
|
876 |
out.append('')
|
877 |
out.append('')
|
878 |
return '\r\n'.join(out) |
879 |
else:
|
880 |
# Footer.
|
881 |
out.append('')
|
882 |
out.append('--' + self.boundary + '--') |
883 |
out.append('')
|
884 |
return '\r\n'.join(out) |
885 |
|
886 |
|
887 |
def object_data_response(request, sizes, hashmaps, meta, public=False): |
888 |
"""Get the HttpResponse object for replying with the object's data."""
|
889 |
|
890 |
# Range handling.
|
891 |
size = sum(sizes)
|
892 |
ranges = get_range(request, size) |
893 |
if ranges is None: |
894 |
ranges = [(0, size)]
|
895 |
ret = 200
|
896 |
else:
|
897 |
check = [True for offset, length in ranges if |
898 |
length <= 0 or length > size or |
899 |
offset < 0 or offset >= size or |
900 |
offset + length > size] |
901 |
if len(check) > 0: |
902 |
raise faults.RangeNotSatisfiable(
|
903 |
'Requested range exceeds object limits')
|
904 |
ret = 206
|
905 |
if_range = request.META.get('HTTP_IF_RANGE')
|
906 |
if if_range:
|
907 |
try:
|
908 |
# Modification time has passed instead.
|
909 |
last_modified = parse_http_date(if_range) |
910 |
if last_modified != meta['modified']: |
911 |
ranges = [(0, size)]
|
912 |
ret = 200
|
913 |
except ValueError: |
914 |
if if_range != meta['checksum']: |
915 |
ranges = [(0, size)]
|
916 |
ret = 200
|
917 |
|
918 |
if ret == 206 and len(ranges) > 1: |
919 |
boundary = uuid.uuid4().hex |
920 |
else:
|
921 |
boundary = ''
|
922 |
wrapper = ObjectWrapper(request.backend, ranges, sizes, hashmaps, boundary) |
923 |
response = HttpResponse(wrapper, status=ret) |
924 |
response.override_serialization = True
|
925 |
put_object_headers( |
926 |
response, meta, restricted=public, |
927 |
token=getattr(request, 'token', None)) |
928 |
if ret == 206: |
929 |
if len(ranges) == 1: |
930 |
offset, length = ranges[0]
|
931 |
response[ |
932 |
'Content-Length'] = length # Update with the correct length. |
933 |
response['Content-Range'] = 'bytes %d-%d/%d' % ( |
934 |
offset, offset + length - 1, size)
|
935 |
else:
|
936 |
del(response['Content-Length']) |
937 |
response['Content-Type'] = 'multipart/byteranges; boundary=%s' % ( |
938 |
boundary,) |
939 |
return response
|
940 |
|
941 |
|
942 |
def put_object_block(request, hashmap, data, offset): |
943 |
"""Put one block of data at the given offset."""
|
944 |
|
945 |
bi = int(offset / request.backend.block_size)
|
946 |
bo = offset % request.backend.block_size |
947 |
bl = min(len(data), request.backend.block_size - bo) |
948 |
if bi < len(hashmap): |
949 |
hashmap[bi] = request.backend.update_block(hashmap[bi], data[:bl], bo) |
950 |
else:
|
951 |
hashmap.append(request.backend.put_block(('\x00' * bo) + data[:bl]))
|
952 |
return bl # Return ammount of data written. |
953 |
|
954 |
|
955 |
def hashmap_md5(backend, hashmap, size): |
956 |
"""Produce the MD5 sum from the data in the hashmap."""
|
957 |
|
958 |
# TODO: Search backend for the MD5 of another object
|
959 |
# with the same hashmap and size...
|
960 |
md5 = hashlib.md5() |
961 |
bs = backend.block_size |
962 |
for bi, hash in enumerate(hashmap): |
963 |
data = backend.get_block(hash) # Blocks come in padded. |
964 |
if bi == len(hashmap) - 1: |
965 |
data = data[:size % bs] |
966 |
md5.update(data) |
967 |
return md5.hexdigest().lower()
|
968 |
|
969 |
|
970 |
def simple_list_response(request, l): |
971 |
if request.serialization == 'text': |
972 |
return '\n'.join(l) + '\n' |
973 |
if request.serialization == 'xml': |
974 |
return render_to_string('items.xml', {'items': l}) |
975 |
if request.serialization == 'json': |
976 |
return json.dumps(l)
|
977 |
|
978 |
|
979 |
from pithos.backends.util import PithosBackendPool |
980 |
|
981 |
if RADOS_STORAGE:
|
982 |
BLOCK_PARAMS = {'mappool': RADOS_POOL_MAPS,
|
983 |
'blockpool': RADOS_POOL_BLOCKS, }
|
984 |
else:
|
985 |
BLOCK_PARAMS = {'mappool': None, |
986 |
'blockpool': None, } |
987 |
|
988 |
|
989 |
_pithos_backend_pool = PithosBackendPool( |
990 |
size=BACKEND_POOL_SIZE, |
991 |
db_module=BACKEND_DB_MODULE, |
992 |
db_connection=BACKEND_DB_CONNECTION, |
993 |
block_module=BACKEND_BLOCK_MODULE, |
994 |
block_path=BACKEND_BLOCK_PATH, |
995 |
block_umask=BACKEND_BLOCK_UMASK, |
996 |
queue_module=BACKEND_QUEUE_MODULE, |
997 |
queue_hosts=BACKEND_QUEUE_HOSTS, |
998 |
queue_exchange=BACKEND_QUEUE_EXCHANGE, |
999 |
astakos_url=ASTAKOS_BASE_URL, |
1000 |
service_token=SERVICE_TOKEN, |
1001 |
astakosclient_poolsize=ASTAKOSCLIENT_POOLSIZE, |
1002 |
free_versioning=BACKEND_FREE_VERSIONING, |
1003 |
block_params=BLOCK_PARAMS, |
1004 |
public_url_security=PUBLIC_URL_SECURITY, |
1005 |
public_url_alphabet=PUBLIC_URL_ALPHABET, |
1006 |
account_quota_policy=BACKEND_ACCOUNT_QUOTA, |
1007 |
container_quota_policy=BACKEND_CONTAINER_QUOTA, |
1008 |
container_versioning_policy=BACKEND_VERSIONING) |
1009 |
|
1010 |
|
1011 |
def get_backend(): |
1012 |
backend = _pithos_backend_pool.pool_get() |
1013 |
backend.messages = [] |
1014 |
return backend
|
1015 |
|
1016 |
|
1017 |
def update_request_headers(request): |
1018 |
# Handle URL-encoded keys and values.
|
1019 |
meta = dict([(
|
1020 |
k, v) for k, v in request.META.iteritems() if k.startswith('HTTP_')]) |
1021 |
for k, v in meta.iteritems(): |
1022 |
try:
|
1023 |
k.decode('ascii')
|
1024 |
v.decode('ascii')
|
1025 |
except UnicodeDecodeError: |
1026 |
raise faults.BadRequest('Bad character in headers.') |
1027 |
if '%' in k or '%' in v: |
1028 |
del(request.META[k])
|
1029 |
request.META[unquote(k)] = smart_unicode(unquote( |
1030 |
v), strings_only=True)
|
1031 |
|
1032 |
|
1033 |
def update_response_headers(request, response): |
1034 |
if (not response.has_header('Content-Length') and |
1035 |
not (response.has_header('Content-Type') and |
1036 |
response['Content-Type'].startswith('multipart/byteranges'))): |
1037 |
response['Content-Length'] = len(response.content) |
1038 |
|
1039 |
# URL-encode unicode in headers.
|
1040 |
meta = response.items() |
1041 |
for k, v in meta: |
1042 |
if (k.startswith('X-Account-') or k.startswith('X-Container-') or |
1043 |
k.startswith('X-Object-') or k.startswith('Content-')): |
1044 |
del(response[k])
|
1045 |
response[quote(k)] = quote(v, safe='/=,:@; ')
|
1046 |
|
1047 |
|
1048 |
def get_pithos_usage(token): |
1049 |
"""Get Pithos Usage from astakos."""
|
1050 |
astakos = AstakosClient(ASTAKOS_BASE_URL, retry=2, use_pool=True, |
1051 |
logger=logger) |
1052 |
quotas = astakos.get_quotas(token)['system']
|
1053 |
pithos_resources = [r['name'] for r in resources] |
1054 |
map(quotas.pop, filter(lambda k: k not in pithos_resources, quotas.keys())) |
1055 |
return quotas.popitem()[-1] # assume only one resource |
1056 |
|
1057 |
|
1058 |
def api_method(http_method=None, user_required=True, logger=None, |
1059 |
format_allowed=False, default_serialization="json"): |
1060 |
def decorator(func): |
1061 |
@api.api_method(http_method=http_method, user_required=user_required,
|
1062 |
logger=logger, format_allowed=format_allowed, |
1063 |
astakos_url=ASTAKOS_BASE_URL, |
1064 |
default_serialization=default_serialization) |
1065 |
@wraps(func)
|
1066 |
def wrapper(request, *args, **kwargs): |
1067 |
# The args variable may contain up to (account, container, object).
|
1068 |
if len(args) > 1 and len(args[1]) > 256: |
1069 |
raise faults.BadRequest("Container name too large") |
1070 |
if len(args) > 2 and len(args[2]) > 1024: |
1071 |
raise faults.BadRequest('Object name too large.') |
1072 |
|
1073 |
try:
|
1074 |
# Add a PithosBackend as attribute of the request object
|
1075 |
request.backend = get_backend() |
1076 |
# Many API method expect thet X-Auth-Token in request,token
|
1077 |
request.token = request.x_auth_token |
1078 |
update_request_headers(request) |
1079 |
response = func(request, *args, **kwargs) |
1080 |
update_response_headers(request, response) |
1081 |
return response
|
1082 |
finally:
|
1083 |
# Always close PithosBackend connection
|
1084 |
if getattr(request, "backend", None) is not None: |
1085 |
request.backend.close() |
1086 |
return wrapper
|
1087 |
return decorator
|