code.grnet.gr Git - pithos/blob - tools/migration

   1 #!/usr/bin/env python
   2
   3 # Copyright 2011 GRNET S.A. All rights reserved.
   4 #
   5 # Redistribution and use in source and binary forms, with or
   6 # without modification, are permitted provided that the following
   7 # conditions are met:
   8 #
   9 #   1. Redistributions of source code must retain the above
  10 #      copyright notice, this list of conditions and the following
  11 #      disclaimer.
  12 #
  13 #   2. Redistributions in binary form must reproduce the above
  14 #      copyright notice, this list of conditions and the following
  15 #      disclaimer in the documentation and/or other materials
  16 #      provided with the distribution.
  17 #
  18 # THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS
  19 # OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  20 # WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  21 # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR
  22 # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  23 # SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  24 # LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
  25 # USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
  26 # AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  27 # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
  28 # ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  29 # POSSIBILITY OF SUCH DAMAGE.
  30 #
  31 # The views and conclusions contained in the software and
  32 # documentation are those of the authors and should not be
  33 # interpreted as representing official policies, either expressed
  34 # or implied, of GRNET S.A.
  35
  36 from sqlalchemy import create_engine
  37 from sqlalchemy import Table, MetaData
  38 from sqlalchemy.sql import select
  39
  40 from pithos.api.util import hashmap_hash, get_container_headers
  41 from pithos.backends.lib.hashfiler import Blocker, Mapper
  42 from pithos.aai.models import PithosUser
  43
  44 from django.conf import settings
  45
  46 from pithos.backends.modular import ModularBackend
  47
  48 import json
  49 import base64
  50 import os
  51
  52 class Migration(object):
  53     def __init__(self, db):
  54         self.engine = create_engine(db)
  55         self.metadata = MetaData(self.engine)
  56         #self.engine.echo = True
  57         self.conn = self.engine.connect()
  58
  59     def execute(self):
  60         pass
  61
  62 class UserMigration(Migration):
  63     def __init__(self, db):
  64         Migration.__init__(self, db)
  65         self.gss_users = Table('gss_user', self.metadata, autoload=True)
  66
  67     def execute(self):
  68         s = self.gss_users.select()
  69         users = self.conn.execute(s).fetchall()
  70         l = []
  71         for u in users:
  72             user = PithosUser()
  73             user.pk = u['id']
  74             user.uniq = u['username']
  75             user.realname = u['name']
  76             user.is_admin = False
  77             user.affiliation = u['homeorganization'] if u['homeorganization'] else ''
  78             user.auth_token = base64.b64encode(u['authtoken'])
  79             user.auth_token_created = u['creationdate']
  80             user.auth_token_expires = u['authtokenexpirydate']
  81             user.created = u['creationdate']
  82             user.updated = u['modificationdate']
  83             print '#', user
  84             user.save(update_timestamps=False)
  85
  86 class DataMigration(Migration):
  87     def __init__(self, db, path, block_size, hash_algorithm):
  88         Migration.__init__(self, db)
  89         params = {'blocksize': block_size,
  90                   'blockpath': os.path.join(path + '/blocks'),
  91                   'hashtype': hash_algorithm}
  92         self.blocker = Blocker(**params)
  93
  94         params = {'mappath': os.path.join(path + '/maps'),
  95                   'namelen': self.blocker.hashlen}
  96         self.mapper = Mapper(**params)
  97
  98     def execute(self):
  99         filebody = Table('filebody', self.metadata, autoload=True)
 100         s = select([filebody.c.id, filebody.c.storedfilepath])
 101         rp = self.conn.execute(s)
 102
 103         while True:
 104             t = rp.fetchone()
 105             if not t:
 106                 break
 107             id, path = t
 108             print '#', id, path
 109             hashlist = self.blocker.block_stor_file(open(path))[1]
 110             self.mapper.map_stor(id, hashlist)
 111         rp.close()
 112
 113 class ObjectMigration(DataMigration):
 114     def __init__(self, db, path, block_size, hash_algorithm):
 115         DataMigration.__init__(self, db, path, block_size, hash_algorithm)
 116         options = getattr(settings, 'BACKEND', None)[1]
 117         self.backend = ModularBackend(*options)
 118
 119     def create_default_containers(self):
 120         users = PithosUser.objects.all()
 121         for u in users:
 122             try:
 123                 self.backend.put_container(u.uniq, u.uniq, 'pithos', {})
 124                 self.backend.put_container(u.uniq, u.uniq, 'trash', {})
 125             except NameError, e:
 126                 pass
 127
 128     def create_directory_markers(self, parent_id=None, path=None):
 129         folderTable = Table('folder', self.metadata, autoload=True)
 130         userTable = Table('gss_user', self.metadata, autoload=True)
 131         s = select([folderTable.c.id, folderTable.c.name, userTable.c.username])
 132         s = s.where(folderTable.c.parent_id == parent_id)
 133         s = s.where(folderTable.c.owner_id == userTable.c.id)
 134         rp = self.conn.execute(s)
 135         while True:
 136             t = rp.fetchone()
 137             if not t:
 138                 path = None
 139                 break
 140             id, name, uuniq = t[0], t[1], t[2]
 141             #print id, name, uuniq
 142             if parent_id:
 143                 obj = '%s/%s' %(path, name) if path else name
 144                 print '#', obj
 145                 self.backend.update_object_hashmap(uuniq, uuniq, 'pithos', obj, 0, [])
 146             else:
 147                 obj = ''
 148             self.create_directory_markers(id, path=obj)
 149         rp.close()
 150         path = None
 151
 152     def execute(self):
 153         filebody = Table('filebody', self.metadata, autoload=True)
 154         s = select([filebody.c.id])
 155         rp = self.conn.execute(s)
 156         while True:
 157             id = rp.fetchone()
 158             if not id:
 159                 break
 160             meta = {}
 161             hashlist = self.mapper.map_retr(id)
 162             #hashmap = d['hashes']
 163             #size = int(d['bytes'])
 164             #meta.update({'hash': hashmap_hash(request, hashmap)})
 165             #version_id = backend.update_object_hashmap(request.user, v_account,
 166             #                                           v_container, v_object,
 167             #                                           size, hashmap)
 168         rp.close()
 169
 170 if __name__ == "__main__":
 171     db = ''
 172     t = UserMigration(db)
 173     t.execute()
 174
 175     basepath = options = getattr(settings, 'PROJECT_PATH', None)
 176     params = {'db':db,
 177               'path':os.path.join(basepath, 'data/pithos/'),
 178               'block_size':(4 * 1024 * 1024),
 179               'hash_algorithm':'sha256'}
 180     dt = DataMigration(**params)
 181     dt.execute()
 182
 183     ot = ObjectMigration(**params)
 184     ot.create_default_containers()
 185     ot.create_directory_markers()
 186