Revision 5fdaf331 tools/migrate-db

b/tools/migrate-db
40 40

  
41 41
from pithos.backends.lib.hashfiler import Blocker
42 42
from pithos.backends.lib.sqlalchemy import Node
43
from pithos.aai.models import PithosUser
44 43

  
45 44
from django.conf import settings
46 45

  
47 46
from pithos.backends.modular import CLUSTER_NORMAL, CLUSTER_HISTORY, CLUSTER_DELETED
48
from pithos.backends.lib.sqlalchemy.node import Node
47
from pithos.backends.lib.sqlalchemy.node import Node, ROOTNODE
49 48

  
50 49
from lib.transfer import upload
51 50
from lib.hashmap import HashMap, file_read_iterator
52 51
from lib.client import Fault
53 52
from lib.migrate import Migration, Cache
54 53
from calendar import timegm
54
from decimal import Decimal
55
from collections import defaultdict
55 56

  
56 57
import json
57 58
import os
58 59
import sys
59 60
import hashlib
60 61
import mimetypes
62
import time
63
import datetime
64

  
65
(ID, CREATIONDATE, MODIFICATIONDATE, DELETED, ICON, NAME, VERSION, CREATEDBY_ID, MODIFIEDBY_ID, OWNER_ID, PARENT_ID, READFORALL, SHARED, USER) = range(14)
61 66

  
62 67
class ObjectMigration(Migration):
63 68
    def __init__(self, old_db, db, f):
64 69
        Migration.__init__(self, old_db)
65 70
        self.cache = Cache(db)
66 71
    
67
    def create_node(self, username, container, object, filepath, mimetype):
68
        obj = ''
69
        path = '/'.join(object.split('/')[:-1])
70
        name =  object.split('/')[-1]
71
        #create directory markers
72
        for f in path.split('/'):
73
            obj = '%s/%s' %(obj, f) if obj else f
74
            try:
75
                md5 = hashlib.md5()
76
                meta = {'Content-Type':'application/directory',
77
                        'hash':  md5.hexdigest().lower()}
78
                self.backend.update_object_hashmap(username, username, container, obj, 0, [], meta) 
79
            except NameError, e:
80
                pass
81
        
82
        parent_path = '%s/%s' %(username, container)
83
        parent_node = self.backend.node.node_lookup(parent_path)
84
        path = '%s/%s' %(parent_path, object)
85
        nodeid = self.backend.node.node_create(parent_node, path)
86
        return nodeid
72
    def create_node(self, username, container, object):
73
        node = self.backend.node.node_lookup(object)
74
        if not node:
75
            parent_path = '%s/%s' %(username, container)
76
            parent_node = self.backend.node.node_lookup(parent_path)
77
            if not parent_node:
78
                raise Exception('Missing node')
79
            node = self.backend.node.node_create(parent_node, object)
80
        return node
87 81
    
88 82
    def create_history(self, header_id, node_id, deleted=False):
89 83
        i = 0
90 84
        map = HashMap(self.backend.block_size, self.backend.hash_algorithm)
85
        v = []
86
        stored_versions = self.backend.node.node_get_versions(node_id, ['mtime'])
87
        stored_versions_mtime = [datetime.datetime.utcfromtimestamp(elem[0]) for elem in stored_versions]
91 88
        for t, rowcount  in self.retrieve_node_versions(header_id):
92
            size, modyfied_by, filepath, mimetype, modificationdate = t
89
            size, modyfied_by, filepath, mimetype, mdate = t
90
            if mdate in stored_versions_mtime:
91
                continue
93 92
            cluster = CLUSTER_HISTORY if i < rowcount - 1 else CLUSTER_NORMAL
94 93
            cluster = cluster if not deleted else CLUSTER_DELETED
95 94
            hash = self.cache.get(filepath)
96 95
            if hash == None:
97
                raise Exception("Missing hash") 
98
            args = (node_id, hash, size, None, modyfied_by, cluster)
99
            serial = self.backend.node.version_create(*args)[0]
100
            meta = {'hash':hash,
101
                    'content-type':mimetype}
102
            self.backend.node.attribute_set(serial, ((k, v) for k, v in meta.iteritems()))
103
            timestamp = timegm(modificationdate.timetuple())
104
            microseconds = modificationdate.time().microsecond
105
            f.write('update versions set mtime=\'%10d.%6d\' where serial=%s;' %(timestamp, microseconds, serial))
96
                raise Exception("Missing hash")
97
            args = node_id, hash, size, modyfied_by, cluster, mimetype, mdate
98
            v.append(self.create_version(*args))
106 99
            i += 1
100
        return v
101
    
102
    def create_version(self, node_id, hash, size, modyfied_by, cluster, mimetype, mdate):
103
        args = (node_id, hash, size, None, modyfied_by, cluster)
104
        serial = self.backend.node.version_create(*args)[0]
105
        meta = {'hash':hash,
106
                'content-type':mimetype}
107
        self.backend.node.attribute_set(serial, ((k, v) for k, v in meta.iteritems()))
108
        timestamp = timegm(mdate.timetuple())
109
        microseconds = mdate.time().microsecond
110
        values = timestamp, microseconds, serial
111
        f.write('update versions set mtime=\'%10d.%6d\' where serial=%s;' %values)
112
        return serial
113
    
114
    def create_tags(self, header_id, node_id, vserials):
115
        tags = self.retrieve_tags(header_id)
116
        if not tags:
117
            return
118
        for v in vserials:
119
            self.backend.node.attribute_set(v, (('tag', tags),))
107 120
    
108
    def create_metadata(self, header_id, node_id):
109
        for t in self.retrieve_metadata(header_id):
110
            pass
121
    def create_permissions(self, fid, path, owner):
122
        fpath, permissions = self.backend.permissions.access_inherit(path)
123
        if not permissions:
124
            permissions = self.retrieve_permissions(fid)
125
            keys = ('read', 'write')
126
            for k in keys:
127
                if owner in permissions[k]:
128
                    permissions[k].remove(owner)
129
            self.backend.permissions.access_set(path, permissions)
111 130
    
112 131
    def create_objects(self):
113
        for username, headerid, folderid, filename, deleted, filepath, mimetype in self.retrieve_current_nodes():
114
            path = self.retrieve_path(folderid)[1:]
115
            container = 'pithos' if not deleted else 'trash'
132
        for t in self.retrieve_current_nodes():
133
            username, headerid, folderid, filename, deleted, filepath, mimetype, public, owner_id = t
134
            containers = ['pithos', 'trash']
135
            
136
            for c in containers:
137
                #create container if it does not exist
138
                try:
139
                    self.backend._lookup_container(username, c)
140
                except NameError, e:
141
                    self.backend.put_container(username, username, c) 
116 142
            
117
            #create container if it does not exist
118
            try:
119
                self.backend._lookup_container(username, container)
120
            except NameError:
121
                self.backend.put_container(username, username, container) 
143
            container = 'pithos' if not deleted else 'trash'
144
            path = self.build_path(folderid)
122 145
            
123 146
            #create node
124
            object = '%s/%s' %(path, filename)
125
            nodeid = self.create_node(username, container, object, filepath, mimetype)
147
            object = '%s/%s/%s/%s' %(username, container, path, filename)
148
            args = username, container, object, filepath, mimetype
149
            nodeid = self.create_node(*args)
126 150
            
127
            #create node history
128
            self.create_history(headerid, nodeid, deleted)
151
            #create node history 
152
            vserials = self.create_history(headerid, nodeid, deleted)
129 153
            
130
            self.create_metadata(headerid, nodeid)
131
            #self.set_public()
132
            #self.statistics()
133
            #self.set_permissions()
154
            #set object tags
155
            self.create_tags(headerid, nodeid, vserials)
156
            
157
            #set object's publicity
158
            if public:
159
                self.backend.permissions.public_set(object)
160
            
161
            #set object's permissions
162
            self.create_permissions(headerid, object, username)
134 163
    
135
    def retrieve_path(self, child_id):
136
        folderTable = Table('folder', self.metadata, autoload=True)
137
        s = select([folderTable.c.parent_id, folderTable.c.name])
138
        s = s.where(folderTable.c.id == child_id)
164
    def build_path(self, child_id):
165
        folder = Table('folder', self.metadata, autoload=True)
166
        user = Table('gss_user', self.metadata, autoload=True)
167
        j = folder.join(user, folder.c.owner_id == user.c.id)
168
        s = select([folder, user.c.username], from_obj=j)
169
        s = s.where(folder.c.id == child_id)
170
        s.order_by(folder.c.modificationdate)
139 171
        rp = self.conn.execute(s)
140
        parent_id, foldername = rp.fetchone()
141
        if not parent_id:
172
        t = rp.fetchone()
173
        md5 = hashlib.md5()
174
        hash = md5.hexdigest().lower()
175
        size = 0
176
        if not t[PARENT_ID]:
142 177
            return ''
143 178
        else:
144
            return '%s/%s' %(self.retrieve_path(parent_id), foldername)
179
            container_path = t[USER]
180
            container_path += '/trash' if t[DELETED] else '/pithos'
181
            parent_node = self.backend.node.node_lookup(container_path)
182
            if not parent_node:
183
                raise Exception('Missing node:', container_path)
184
            parent_path = self.build_path(t[PARENT_ID])
185
            path = '%s/%s/%s' %(container_path, parent_path, t[NAME]) if parent_path else '%s/%s' %(container_path, t[NAME])
186
            node = self.backend.node.node_lookup(path)
187
            if not node:
188
                node = self.backend.node.node_create(parent_node, path)
189
                if not node:
190
                    raise Exception('Unable to create node:', path)
191
                
192
                #create versions
193
                v = self.create_version(node, hash, size, t[USER], CLUSTER_NORMAL, 'application/directory', t[CREATIONDATE])
194
                if t[CREATIONDATE] != t[MODIFICATIONDATE]:
195
                    self.backend.node.version_recluster(v, CLUSTER_HISTORY)
196
                    self.create_version(node, hash, size, t[USER], CLUSTER_NORMAL, 'application/directory', t[MODIFICATIONDATE])
197
                
198
                #set permissions
199
                self.create_permissions(t[ID], path, t[USER])
200
            return '%s/%s' %(parent_path, t[NAME]) if parent_path else t[NAME]
145 201
    
146 202
    def retrieve_current_nodes(self):
147 203
        fileheader = Table('fileheader', self.metadata, autoload=True)
......
152 208
        j = j.join(folder, fileheader.c.folder_id == folder.c.id)
153 209
        j = j.join(gss_user, fileheader.c.owner_id == gss_user.c.id)
154 210
        s = select([gss_user.c.username,  fileheader.c.id, fileheader.c.folder_id,
155
                    fileheader.c.name,  fileheader.c.deleted, filebody.c.storedfilepath,
156
                    filebody.c.mimetype], from_obj=j)
157
        s = s.limit(1)
211
                    fileheader.c.name,  fileheader.c.deleted,
212
                    filebody.c.storedfilepath, filebody.c.mimetype,
213
                    fileheader.c.readforall, fileheader.c.owner_id], from_obj=j)
158 214
        rp = self.conn.execute(s)
159 215
        object = rp.fetchone()
160 216
        while object:
......
178 234
            version = rp.fetchone()
179 235
        rp.close()
180 236
    
181
    def retrieve_metadata(self, header_id):
237
    def retrieve_tags(self, header_id):
182 238
        filetag = Table('filetag', self.metadata, autoload=True)
183
        s = filetag.select(filetag.c.fileid == header_id)
239
        s = select([filetag.c.tag], filetag.c.fileid == header_id)
184 240
        rp = self.conn.execute(s)
185
        tag = rp.fetchone()
186
        while tag:
187
            yield tag
188
            tag = tp.fetchone()
241
        tags = rp.fetchall() if rp.returns_rows else []
242
        tags = [elem[0] for elem in tags]
189 243
        rp.close()
244
        return ','.join(tags) if tags else ''
190 245
    
191
    def handle_deleted(self):
192
        pass
246
    def retrieve_permissions(self, id, is_folder=True):
247
        permissions = {}
248
        if is_folder:
249
            ftable = Table('folder_permission', self.metadata, autoload=True)
250
        else:
251
            ftable = Table('fileheader_permission', self.metadata, autoload=True)
252
        permission = Table('permission', self.metadata, autoload=True)
253
        group = Table('gss_group', self.metadata, autoload=True)
254
        user = Table('gss_user', self.metadata, autoload=True)
255
        j = ftable.join(permission, ftable.c.permissions_id == permission.c.id)
256
        j1 = j.join(group, group.c.id == permission.c.group_id)
257
        j2 = j.join(user, user.c.id == permission.c.user_id)
258
        
259
        permissions = defaultdict(list)
260
        
261
        #get folder read groups
262
        s = select([group.c.name], from_obj=j1)
263
        if is_folder:
264
            s = s.where(ftable.c.folder_id == id)
265
        else:
266
            s = s.where(ftable.c.fileheader_id == id)
267
        s = s.where(permission.c.read == True)
268
        s = s.where(permission.c.group_id != None)
269
        rp = self.conn.execute(s)
270
        permissions['read'].extend([e[0] for e in rp.fetchall()])
193 271

  
272
        #get folder read users
273
        s = select([user.c.username], from_obj=j2)
274
        if is_folder:
275
            s = s.where(ftable.c.folder_id == id)
276
        else:
277
            s = s.where(ftable.c.fileheader_id == id)
278
        s = s.where(permission.c.read == True)
279
        s = s.where(permission.c.user_id != None)
280
        rp = self.conn.execute(s)
281
        permissions['read'].extend([e[0] for e in rp.fetchall()])
282
        
283
        #get folder write groups
284
        s = select([group.c.name], from_obj=j1)
285
        if is_folder:
286
            s = s.where(ftable.c.folder_id == id)
287
        else:
288
            s = s.where(ftable.c.fileheader_id == id)
289
        s = s.where(permission.c.write == True)
290
        s = s.where(permission.c.group_id != None)
291
        rp = self.conn.execute(s)
292
        permissions['write'].extend([e[0] for e in rp.fetchall()])
293
        
294
        #get folder write groups
295
        s = select([user.c.username], from_obj=j2)
296
        if is_folder:
297
            s = s.where(ftable.c.folder_id == id)
298
        else:
299
            s = s.where(ftable.c.fileheader_id == id)
300
        s = s.where(permission.c.write == True)
301
        s = s.where(permission.c.user_id != None)
302
        rp = self.conn.execute(s)
303
        permissions['write'].extend([e[0] for e in rp.fetchall()])
304
        
305
        rp.close()
306
        return permissions
307
    
194 308
if __name__ == "__main__":
195 309
    old_db = ''
196 310
    db = ''

Also available in: Unified diff