Synnefo » Pithos

# Copyright 2011 GRNET S.A. All rights reserved.

# 

# Redistribution and use in source and binary forms, with or

# without modification, are permitted provided that the following

# conditions are met:

# 

#   1. Redistributions of source code must retain the above

#      copyright notice, this list of conditions and the following

#      disclaimer.

# 

#   2. Redistributions in binary form must reproduce the above

#      copyright notice, this list of conditions and the following

#      disclaimer in the documentation and/or other materials

#      provided with the distribution.

# 

# THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS

# OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED

# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR

# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR

# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,

# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT

# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF

# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED

# AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT

# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN

# ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE

# POSSIBILITY OF SUCH DAMAGE.

# 

# The views and conclusions contained in the software and

# documentation are those of the authors and should not be

# interpreted as representing official policies, either expressed

# or implied, of GRNET S.A.

from functools import wraps

from time import time

from traceback import format_exc

from wsgiref.handlers import format_date_time

from binascii import hexlify

from django.conf import settings

from django.http import HttpResponse

from django.utils.http import http_date, parse_etags

from pithos.api.compat import parse_http_date_safe, parse_http_date

from pithos.api.faults import (Fault, NotModified, BadRequest, Unauthorized, ItemNotFound,

                                LengthRequired, PreconditionFailed, RangeNotSatisfiable,

                                ServiceUnavailable)

from pithos.backends import backend

from pithos.backends.base import NotAllowedError

import datetime

import logging

import re

import hashlib

import uuid

logger = logging.getLogger(__name__)

def printable_header_dict(d):

    """Format a meta dictionary for printing out json/xml.

    Convert all keys to lower case and replace dashes to underscores.

    Change 'modified' key from backend to 'last_modified' and format date.

    """

    if 'modified' in d:

        d['last_modified'] = datetime.datetime.fromtimestamp(int(d['modified'])).isoformat()

        del(d['modified'])

    return dict([(k.lower().replace('-', '_'), v) for k, v in d.iteritems()])

def format_header_key(k):

    """Convert underscores to dashes and capitalize intra-dash strings."""

    return '-'.join([x.capitalize() for x in k.replace('_', '-').split('-')])

def get_header_prefix(request, prefix):

    """Get all prefix-* request headers in a dict. Reformat keys with format_header_key()."""

    prefix = 'HTTP_' + prefix.upper().replace('-', '_')

    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)])

def get_account_headers(request):

    meta = get_header_prefix(request, 'X-Account-Meta-')

    groups = {}

    for k, v in get_header_prefix(request, 'X-Account-Group-').iteritems():

        n = k[16:].lower()

        if '-' in n or '_' in n:

            raise BadRequest('Bad characters in group name')

        groups[n] = v.replace(' ', '').split(',')

        if '' in groups[n]:

            groups[n].remove('')

    return meta, groups

def put_account_headers(response, meta, groups):

    response['X-Account-Container-Count'] = meta['count']

    response['X-Account-Bytes-Used'] = meta['bytes']

    if 'modified' in meta:

        response['Last-Modified'] = http_date(int(meta['modified']))

    for k in [x for x in meta.keys() if x.startswith('X-Account-Meta-')]:

        response[k.encode('utf-8')] = meta[k].encode('utf-8')

    if 'until_timestamp' in meta:

        response['X-Account-Until-Timestamp'] = http_date(int(meta['until_timestamp']))

    for k, v in groups.iteritems():

        response[format_header_key('X-Account-Group-' + k).encode('utf-8')] = (','.join(v)).encode('utf-8')

def get_container_headers(request):

    meta = get_header_prefix(request, 'X-Container-Meta-')

    return meta

def put_container_headers(response, meta):

    response['X-Container-Object-Count'] = meta['count']

    response['X-Container-Bytes-Used'] = meta['bytes']

    response['Last-Modified'] = http_date(int(meta['modified']))

    for k in [x for x in meta.keys() if x.startswith('X-Container-Meta-')]:

        response[k.encode('utf-8')] = meta[k].encode('utf-8')

    response['X-Container-Object-Meta'] = [x[14:] for x in meta['object_meta'] if x.startswith('X-Object-Meta-')]

    response['X-Container-Block-Size'] = backend.block_size

    response['X-Container-Block-Hash'] = backend.hash_algorithm

    if 'until_timestamp' in meta:

        response['X-Container-Until-Timestamp'] = http_date(int(meta['until_timestamp']))

def get_object_headers(request):

    meta = get_header_prefix(request, 'X-Object-Meta-')

    if request.META.get('CONTENT_TYPE'):

        meta['Content-Type'] = request.META['CONTENT_TYPE']

    if request.META.get('HTTP_CONTENT_ENCODING'):

        meta['Content-Encoding'] = request.META['HTTP_CONTENT_ENCODING']

    if request.META.get('HTTP_CONTENT_DISPOSITION'):

        meta['Content-Disposition'] = request.META['HTTP_CONTENT_DISPOSITION']

    if request.META.get('HTTP_X_OBJECT_MANIFEST'):

        meta['X-Object-Manifest'] = request.META['HTTP_X_OBJECT_MANIFEST']

    return meta

def put_object_headers(response, meta, public=False):

    response['ETag'] = meta['hash']

    response['Content-Length'] = meta['bytes']

    response['Content-Type'] = meta.get('Content-Type', 'application/octet-stream')

    response['Last-Modified'] = http_date(int(meta['modified']))

    if not public:

        response['X-Object-Modified-By'] = meta['modified_by']

        response['X-Object-Version'] = meta['version']

        response['X-Object-Version-Timestamp'] = http_date(int(meta['version_timestamp']))

        for k in [x for x in meta.keys() if x.startswith('X-Object-Meta-')]:

            response[k.encode('utf-8')] = meta[k].encode('utf-8')

        for k in ('Content-Encoding', 'Content-Disposition', 'X-Object-Manifest', 'X-Object-Sharing', 'X-Object-Shared-By'):

            if k in meta:

                response[k] = meta[k]

    else:

        for k in ('Content-Encoding', 'Content-Disposition'):

            if k in meta:

                response[k] = meta[k]

def update_manifest_meta(request, v_account, meta):

    """Update metadata if the object has an X-Object-Manifest."""

    if 'X-Object-Manifest' in meta:

        hash = ''

        bytes = 0

        try:

            src_container, src_name = split_container_object_string('/' + meta['X-Object-Manifest'])

            objects = backend.list_objects(request.user, v_account, src_container, prefix=src_name, virtual=False)

            for x in objects:

                src_meta = backend.get_object_meta(request.user, v_account, src_container, x[0], x[1])

                hash += src_meta['hash']

                bytes += src_meta['bytes']

        except:

            # Ignore errors.

            return

        meta['bytes'] = bytes

        md5 = hashlib.md5()

        md5.update(hash)

        meta['hash'] = md5.hexdigest().lower()

def update_sharing_meta(permissions, v_account, v_container, v_object, meta):

    if permissions is None:

        return

    perm_path, perms = permissions

    if len(perms) == 0:

        return

    ret = []

    r = ','.join(perms.get('read', []))

    if r:

        ret.append('read=' + r)

    w = ','.join(perms.get('write', []))

    if w:

        ret.append('write=' + w)

    meta['X-Object-Sharing'] = '; '.join(ret)

    if '/'.join((v_account, v_container, v_object)) != perm_path:

        meta['X-Object-Shared-By'] = perm_path

def validate_modification_preconditions(request, meta):

    """Check that the modified timestamp conforms with the preconditions set."""

    if 'modified' not in meta:

        return # TODO: Always return?

    if_modified_since = request.META.get('HTTP_IF_MODIFIED_SINCE')

    if if_modified_since is not None:

        if_modified_since = parse_http_date_safe(if_modified_since)

    if if_modified_since is not None and int(meta['modified']) <= if_modified_since:

        raise NotModified('Resource has not been modified')

    if_unmodified_since = request.META.get('HTTP_IF_UNMODIFIED_SINCE')

    if if_unmodified_since is not None:

        if_unmodified_since = parse_http_date_safe(if_unmodified_since)

    if if_unmodified_since is not None and int(meta['modified']) > if_unmodified_since:

        raise PreconditionFailed('Resource has been modified')

def validate_matching_preconditions(request, meta):

    """Check that the ETag conforms with the preconditions set."""

    if 'hash' not in meta:

        return # TODO: Always return?

    if_match = request.META.get('HTTP_IF_MATCH')

    if if_match is not None and if_match != '*':

        if meta['hash'] not in [x.lower() for x in parse_etags(if_match)]:

            raise PreconditionFailed('Resource Etag does not match')

    if_none_match = request.META.get('HTTP_IF_NONE_MATCH')

    if if_none_match is not None:

        if if_none_match == '*' or meta['hash'] in [x.lower() for x in parse_etags(if_none_match)]:

            raise NotModified('Resource Etag matches')

def split_container_object_string(s):

    if not len(s) > 0 or s[0] != '/':

        raise ValueError

    s = s[1:]

    pos = s.find('/')

    if pos == -1:

        raise ValueError

    return s[:pos], s[(pos + 1):]

def copy_or_move_object(request, v_account, src_container, src_name, dest_container, dest_name, move=False):

    """Copy or move an object."""

    meta = get_object_headers(request)

    permissions = get_sharing(request)

    src_version = request.META.get('HTTP_X_SOURCE_VERSION')    

    try:

        if move:

            backend.move_object(request.user, v_account, src_container, src_name, dest_container, dest_name, meta, False, permissions)

        else:

            backend.copy_object(request.user, v_account, src_container, src_name, dest_container, dest_name, meta, False, permissions, src_version)

    except NotAllowedError:

        raise Unauthorized('Access denied')

    except NameError, IndexError:

        raise ItemNotFound('Container or object does not exist')

    except ValueError:

        raise BadRequest('Invalid sharing header')

    except AttributeError:

        raise Conflict('Sharing already set above or below this path in the hierarchy')

def get_int_parameter(request, name):

    p = request.GET.get(name)

    if p is not None:

        try:

            p = int(p)

        except ValueError:

            return None

        if p < 0:

            return None

    return p

def get_content_length(request):

    content_length = request.META.get('CONTENT_LENGTH')

    if not content_length:

        raise LengthRequired('Missing Content-Length header')

    try:

        content_length = int(content_length)

        if content_length < 0:

            raise ValueError

    except ValueError:

        raise BadRequest('Invalid Content-Length header')

    return content_length

def get_range(request, size):

    """Parse a Range header from the request.

    Either returns None, when the header is not existent or should be ignored,

    or a list of (offset, length) tuples - should be further checked.

    """

    ranges = request.META.get('HTTP_RANGE', '').replace(' ', '')

    if not ranges.startswith('bytes='):

        return None

    ret = []

    for r in (x.strip() for x in ranges[6:].split(',')):

        p = re.compile('^(?P<offset>\d*)-(?P<upto>\d*)$')

        m = p.match(r)

        if not m:

            return None

        offset = m.group('offset')

        upto = m.group('upto')

        if offset == '' and upto == '':

            return None

        if offset != '':

            offset = int(offset)

            if upto != '':

                upto = int(upto)

                if offset > upto:

                    return None

                ret.append((offset, upto - offset + 1))

            else:

                ret.append((offset, size - offset))

        else:

            length = int(upto)

            ret.append((size - length, length))

    return ret

def get_content_range(request):

    """Parse a Content-Range header from the request.

    Either returns None, when the header is not existent or should be ignored,

    or an (offset, length, total) tuple - check as length, total may be None.

    Returns (None, None, None) if the provided range is '*/*'.

    """

    ranges = request.META.get('HTTP_CONTENT_RANGE', '')

    if not ranges:

        return None

    p = re.compile('^bytes (?P<offset>\d+)-(?P<upto>\d*)/(?P<total>(\d+|\*))$')

    m = p.match(ranges)

    if not m:

        if ranges == 'bytes */*':

            return (None, None, None)

        return None

    offset = int(m.group('offset'))

    upto = m.group('upto')

    total = m.group('total')

    if upto != '':

        upto = int(upto)

    else:

        upto = None

    if total != '*':

        total = int(total)

    else:

        total = None

    if (upto is not None and offset > upto) or \

        (total is not None and offset >= total) or \

        (total is not None and upto is not None and upto >= total):

        return None

    if upto is None:

        length = None

    else:

        length = upto - offset + 1

    return (offset, length, total)

def get_sharing(request):

    """Parse an X-Object-Sharing header from the request.

    Raises BadRequest on error.

    """

    permissions = request.META.get('HTTP_X_OBJECT_SHARING')

    if permissions is None:

        return None

    ret = {}

    permissions = permissions.replace(' ', '')

    if permissions == '':

        return ret

    for perm in (x for x in permissions.split(';')):

        if perm.startswith('read='):

            ret['read'] = [v.replace(' ','').lower() for v in perm[5:].split(',')]

            if '' in ret['read']:

                ret['read'].remove('')

            if '*' in ret['read']:

                ret['read'] = ['*']

            if len(ret['read']) == 0:

                raise BadRequest('Bad X-Object-Sharing header value')

        elif perm.startswith('write='):

            ret['write'] = [v.replace(' ','').lower() for v in perm[6:].split(',')]

            if '' in ret['write']:

                ret['write'].remove('')

            if '*' in ret['write']:

                ret['write'] = ['*']

            if len(ret['write']) == 0:

                raise BadRequest('Bad X-Object-Sharing header value')

        else:

            raise BadRequest('Bad X-Object-Sharing header value')

    return ret

def raw_input_socket(request):

    """Return the socket for reading the rest of the request."""

    server_software = request.META.get('SERVER_SOFTWARE')

    if not server_software:

        if 'wsgi.input' in request.environ:

            return request.environ['wsgi.input']

        raise ServiceUnavailable('Unknown server software')

    if server_software.startswith('WSGIServer'):

        return request.environ['wsgi.input']

    elif server_software.startswith('mod_python'):

        return request._req

    raise ServiceUnavailable('Unknown server software')

MAX_UPLOAD_SIZE = 10 * (1024 * 1024) # 10MB

def socket_read_iterator(sock, length=0, blocksize=4096):

    """Return a maximum of blocksize data read from the socket in each iteration.

    Read up to 'length'. If 'length' is negative, will attempt a chunked read.

    The maximum ammount of data read is controlled by MAX_UPLOAD_SIZE.

    """

    if length < 0: # Chunked transfers

        data = ''

        while length < MAX_UPLOAD_SIZE:

            # Get chunk size.

            if hasattr(sock, 'readline'):

                chunk_length = sock.readline()

            else:

                chunk_length = ''

                while chunk_length[-1:] != '\n':

                    chunk_length += sock.read(1)

                chunk_length.strip()

            pos = chunk_length.find(';')

            if pos >= 0:

                chunk_length = chunk_length[:pos]

            try:

                chunk_length = int(chunk_length, 16)

            except Exception, e:

                raise BadRequest('Bad chunk size') # TODO: Change to something more appropriate.

            # Check if done.

            if chunk_length == 0:

                if len(data) > 0:

                    yield data

                return

            # Get the actual data.

            while chunk_length > 0:

                chunk = sock.read(min(chunk_length, blocksize))

                chunk_length -= len(chunk)

                if length > 0:

                    length += len(chunk)

                data += chunk

                if len(data) >= blocksize:

                    ret = data[:blocksize]

                    data = data[blocksize:]

                    yield ret

            sock.read(2) # CRLF

        # TODO: Raise something to note that maximum size is reached.

    else:

        if length > MAX_UPLOAD_SIZE:

            # TODO: Raise something to note that maximum size is reached.

            pass

        while length > 0:

            data = sock.read(min(length, blocksize))

            length -= len(data)

            yield data

class ObjectWrapper(object):

    """Return the object's data block-per-block in each iteration.

    Read from the object using the offset and length provided in each entry of the range list.

    """

    def __init__(self, ranges, sizes, hashmaps, boundary):

        self.ranges = ranges

        self.sizes = sizes

        self.hashmaps = hashmaps

        self.boundary = boundary

        self.size = sum(self.sizes)

        self.file_index = 0

        self.block_index = 0

        self.block_hash = -1

        self.block = ''

        self.range_index = -1

        self.offset, self.length = self.ranges[0]

    def __iter__(self):

        return self

    def part_iterator(self):

        if self.length > 0:

            # Get the file for the current offset.

            file_size = self.sizes[self.file_index]

            while self.offset >= file_size:

                self.offset -= file_size

                self.file_index += 1

                file_size = self.sizes[self.file_index]

            # Get the block for the current position.

            self.block_index = int(self.offset / backend.block_size)

            if self.block_hash != self.hashmaps[self.file_index][self.block_index]:

                self.block_hash = self.hashmaps[self.file_index][self.block_index]

                try:

                    self.block = backend.get_block(self.block_hash)

                except NameError:

                    raise ItemNotFound('Block does not exist')

            # Get the data from the block.

            bo = self.offset % backend.block_size

            bl = min(self.length, len(self.block) - bo)

            data = self.block[bo:bo + bl]

            self.offset += bl

            self.length -= bl

            return data

        else:

            raise StopIteration

    def next(self):

        if len(self.ranges) == 1:

            return self.part_iterator()

        if self.range_index == len(self.ranges):

            raise StopIteration

        try:

            if self.range_index == -1:

                raise StopIteration

            return self.part_iterator()

        except StopIteration:

            self.range_index += 1

            out = []

            if self.range_index < len(self.ranges):

                # Part header.

                self.offset, self.length = self.ranges[self.range_index]

                self.file_index = 0

                if self.range_index > 0:

                    out.append('')

                out.append('--' + self.boundary)

                out.append('Content-Range: bytes %d-%d/%d' % (self.offset, self.offset + self.length - 1, self.size))

                out.append('Content-Transfer-Encoding: binary')

                out.append('')

                out.append('')

                return '\r\n'.join(out)

            else:

                # Footer.

                out.append('')

                out.append('--' + self.boundary + '--')

                out.append('')

                return '\r\n'.join(out)

def object_data_response(request, sizes, hashmaps, meta, public=False):

    """Get the HttpResponse object for replying with the object's data."""

    # Range handling.

    size = sum(sizes)

    ranges = get_range(request, size)

    if ranges is None:

        ranges = [(0, size)]

        ret = 200

    else:

        check = [True for offset, length in ranges if

                    length <= 0 or length > size or

                    offset < 0 or offset >= size or

                    offset + length > size]

        if len(check) > 0:

            raise RangeNotSatisfiable('Requested range exceeds object limits')

        ret = 206

        if_range = request.META.get('HTTP_IF_RANGE', '')

        if if_range and if_range.startswith('If-Range:'):

            if_range = if_range.split('If-Range:')[1]

            try:

                # modification time has passed instead

                last_modified = parse_http_date(if_range)

                if last_modified != meta['modified']:

                    ranges = [(0, size)]

                    ret = 200

            except ValueError:

                if if_range != meta['hash']:

                    ranges = [(0, size)]

                    ret = 200

    if ret == 206 and len(ranges) > 1:

        boundary = uuid.uuid4().hex

    else:

        boundary = ''

    wrapper = ObjectWrapper(ranges, sizes, hashmaps, boundary)

    response = HttpResponse(wrapper, status=ret)

    put_object_headers(response, meta, public)

    if ret == 206:

        if len(ranges) == 1:

            offset, length = ranges[0]

            response['Content-Length'] = length # Update with the correct length.

            response['Content-Range'] = 'bytes %d-%d/%d' % (offset, offset + length - 1, size)

        else:

            del(response['Content-Length'])

            response['Content-Type'] = 'multipart/byteranges; boundary=%s' % (boundary,)

    return response

def put_object_block(hashmap, data, offset):

    """Put one block of data at the given offset."""

    bi = int(offset / backend.block_size)

    bo = offset % backend.block_size

    bl = min(len(data), backend.block_size - bo)

    if bi < len(hashmap):

        hashmap[bi] = backend.update_block(hashmap[bi], data[:bl], bo)

    else:

        hashmap.append(backend.put_block(('\x00' * bo) + data[:bl]))

    return bl # Return ammount of data written.

def hashmap_hash(hashmap):

    """Produce the root hash, treating the hashmap as a Merkle-like tree."""

    def subhash(d):

        h = hashlib.new(backend.hash_algorithm)

        h.update(d)

        return h.digest()

    if len(hashmap) == 0:

        return hexlify(subhash(''))

    if len(hashmap) == 1:

        return hexlify(subhash(hashmap[0]))

    s = 2

    while s < len(hashmap):

        s = s * 2

    h = hashmap + ([('\x00' * len(hashmap[0]))] * (s - len(hashmap)))

    h = [subhash(h[x] + (h[x + 1] if x + 1 < len(h) else '')) for x in range(0, len(h), 2)]

    while len(h) > 1:

        h = [subhash(h[x] + (h[x + 1] if x + 1 < len(h) else '')) for x in range(0, len(h), 2)]

    return hexlify(h[0])

def update_response_headers(request, response):

    if request.serialization == 'xml':

        response['Content-Type'] = 'application/xml; charset=UTF-8'

    elif request.serialization == 'json':

        response['Content-Type'] = 'application/json; charset=UTF-8'

    elif not response['Content-Type']:

        response['Content-Type'] = 'text/plain; charset=UTF-8'

    if settings.TEST:

        response['Date'] = format_date_time(time())

def render_fault(request, fault):

    if settings.DEBUG or settings.TEST:

        fault.details = format_exc(fault)

    request.serialization = 'text'

    data = '\n'.join((fault.message, fault.details)) + '\n'

    response = HttpResponse(data, status=fault.code)

    update_response_headers(request, response)

    return response

def request_serialization(request, format_allowed=False):

    """Return the serialization format requested.

    Valid formats are 'text' and 'json', 'xml' if 'format_allowed' is True.

    """

    if not format_allowed:

        return 'text'

    format = request.GET.get('format')

    if format == 'json':

        return 'json'

    elif format == 'xml':

        return 'xml'

    for item in request.META.get('HTTP_ACCEPT', '').split(','):

        accept, sep, rest = item.strip().partition(';')

        if accept == 'application/json':

            return 'json'

        elif accept == 'application/xml' or accept == 'text/xml':

            return 'xml'

    return 'text'

def api_method(http_method=None, format_allowed=False):

    """Decorator function for views that implement an API method."""

    def decorator(func):

        @wraps(func)

        def wrapper(request, *args, **kwargs):

            try:

                if http_method and request.method != http_method:

                    raise BadRequest('Method not allowed.')

                # The args variable may contain up to (account, container, object).

                if len(args) > 1 and len(args[1]) > 256:

                    raise BadRequest('Container name too large.')

                if len(args) > 2 and len(args[2]) > 1024:

                    raise BadRequest('Object name too large.')

                # Fill in custom request variables.

                request.serialization = request_serialization(request, format_allowed)

                response = func(request, *args, **kwargs)

                update_response_headers(request, response)

                return response

            except Fault, fault:

                return render_fault(request, fault)

            except BaseException, e:

                logger.exception('Unexpected error: %s' % e)

                fault = ServiceUnavailable('Unexpected error')

                return render_fault(request, fault)

        return wrapper

    return decorator