1 # -*- encoding: utf-8 -*-
2 ##############################################################################
4 # OpenERP, Open Source Management Solution
6 # Copyright (C) P. Christeas, 2009, all rights reserved
8 # This program is free software: you can redistribute it and/or modify
9 # it under the terms of the GNU General Public License as published by
10 # the Free Software Foundation, either version 3 of the License, or
11 # (at your option) any later version.
13 # This program is distributed in the hope that it will be useful,
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 # GNU General Public License for more details.
18 # You should have received a copy of the GNU General Public License
19 # along with this program. If not, see <http://www.gnu.org/licenses/>.
21 ##############################################################################
23 from osv import osv, fields
27 from tools.misc import ustr
28 from tools.translate import _
30 from osv.orm import except_orm
35 from content_index import cntIndex
37 DMS_ROOT_PATH = tools.config.get('document_path', os.path.join(tools.config.get('root_path'), 'filestore'))
40 """ The algorithm of data storage
42 We have to consider 3 cases of data /retrieval/:
43 Given (context,path) we need to access the file (aka. node).
44 given (directory, context), we need one of its children (for listings, views)
45 given (ir.attachment, context), we needs its data and metadata (node).
47 For data /storage/ we have the cases:
48 Have (ir.attachment, context), we modify the file (save, update, rename etc).
49 Have (directory, context), we create a file.
50 Have (path, context), we create or modify a file.
52 Note that in all above cases, we don't explicitly choose the storage media,
53 but always require a context to be present.
55 Note that a node will not always have a corresponding ir.attachment. Dynamic
56 nodes, for once, won't. Their metadata will be computed by the parent storage
59 The algorithm says that in any of the above cases, our first goal is to locate
60 the node for any combination of search criteria. It would be wise NOT to
61 represent each node in the path (like node[/] + node[/dir1] + node[/dir1/dir2])
62 but directly jump to the end node (like node[/dir1/dir2]) whenever possible.
64 We also contain all the parenting loop code in one function. This is intentional,
65 because one day this will be optimized in the db (Pg 8.4).
72 d = [random.choice(string.ascii_letters) for x in xrange(10) ]
76 INVALID_CHARS = {'*':str(hash('*')), '|':str(hash('|')) , "\\":str(hash("\\")), '/':'__', ':':str(hash(':')), '"':str(hash('"')), '<':str(hash('<')) , '>':str(hash('>')) , '?':str(hash('?'))}
79 def create_directory(path):
80 dir_name = random_name()
81 path = os.path.join(path, dir_name)
86 class document_storage(osv.osv):
87 """ The primary object for data storage.
88 Each instance of this object is a storage media, in which our application
89 can store contents. The object here controls the behaviour of the storage
91 The referring document.directory-ies will control the placement of data
94 It is a bad idea to have multiple document.storage objects pointing to
95 the same tree of filesystem storage.
97 _name = 'document.storage'
98 _description = 'Storage Media'
102 'name': fields.char('Name', size=64, required=True, select=1),
103 'write_date': fields.datetime('Date Modified', readonly=True),
104 'write_uid': fields.many2one('res.users', 'Last Modification User', readonly=True),
105 'create_date': fields.datetime('Date Created', readonly=True),
106 'create_uid': fields.many2one('res.users', 'Creator', readonly=True),
107 'user_id': fields.many2one('res.users', 'Owner'),
108 'group_ids': fields.many2many('res.groups', 'document_storage_group_rel', 'item_id', 'group_id', 'Groups'),
109 'dir_ids': fields.one2many('document.directory', 'parent_id', 'Directories'),
110 'type': fields.selection([('db', 'Database'), ('filestore', 'Internal File storage'),
111 ('realstore', 'External file storage'), ('virtual', 'Virtual storage')], 'Type', required=True),
112 'path': fields.char('Path', size=250, select=1, help="For file storage, the root path of the storage"),
113 'online': fields.boolean('Online', help="If not checked, media is currently offline and its contents not available", required=True),
114 'readonly': fields.boolean('Read Only', help="If set, media is for reading only"),
117 def _get_rootpath(self, cr, uid, context=None):
118 return os.path.join(DMS_ROOT_PATH, cr.dbname)
121 'user_id': lambda self, cr, uid, ctx: uid,
122 'online': lambda *args: True,
123 'readonly': lambda *args: False,
124 # Note: the defaults below should only be used ONCE for the default
125 # storage media. All other times, we should create different paths at least.
126 'type': lambda *args: 'filestore',
127 'path': _get_rootpath,
130 # SQL note: a path = NULL doesn't have to be unique.
131 ('path_uniq', 'UNIQUE(type,path)', "The storage path must be unique!")
134 def get_data(self, cr, uid, id, file_node, context=None, fil_obj=None):
135 """ retrieve the contents of some file_node having storage_id = id
136 optionally, fil_obj could point to the browse object of the file
141 boo = self.browse(cr, uid, id, context)
145 ira = self.pool.get('ir.attachment').browse(cr, uid, file_node.file_id, context=context)
146 return self.__get_data_3(cr, uid, boo, ira, context)
148 def __get_data_3(self, cr, uid, boo, ira, context):
150 raise RuntimeError('media offline')
151 if boo.type == 'filestore':
152 if not ira.store_fname:
153 # On a migrated db, some files may have the wrong storage type
154 # try to fix their directory.
156 netsvc.Logger().notifyChannel('document', netsvc.LOG_WARNING, "ir.attachment #%d does not have a filename, but is at filestore, fix it!" % ira.id)
158 fpath = os.path.join(boo.path, ira.store_fname)
159 return file(fpath, 'rb').read()
160 elif boo.type == 'db':
161 # TODO: we need a better api for large files
163 out = base64.decodestring(ira.db_datas)
167 elif boo.type == 'realstore':
168 if not ira.store_fname:
169 # On a migrated db, some files may have the wrong storage type
170 # try to fix their directory.
172 netsvc.Logger().notifyChannel('document',netsvc.LOG_WARNING,"ir.attachment #%d does not have a filename, trying the name." %ira.id)
174 fpath = os.path.join(boo.path,ira.store_fname or ira.name)
175 if os.path.exists(fpath):
176 return file(fpath,'rb').read()
177 elif not ira.store_fname:
180 raise IOError("File not found: %s" % fpath)
182 raise TypeError("No %s storage" % boo.type)
184 def set_data(self, cr, uid, id, file_node, data, context=None, fil_obj=None):
186 This function MUST be used from an ir.attachment. It wouldn't make sense
187 to store things persistently for other types (dynamic).
191 boo = self.browse(cr, uid, id, context)
192 logger = netsvc.Logger()
196 ira = self.pool.get('ir.attachment').browse(cr, uid, file_node.file_id, context=context)
199 raise RuntimeError('media offline')
200 logger.notifyChannel('document', netsvc.LOG_DEBUG, "Store data for ir.attachment #%d" % ira.id)
203 if boo.type == 'filestore':
207 # This can be improved
208 if os.path.isdir(path):
209 for dirs in os.listdir(path):
210 if os.path.isdir(os.path.join(path, dirs)) and len(os.listdir(os.path.join(path, dirs))) < 4000:
213 flag = flag or create_directory(path)
214 filename = random_name()
215 fname = os.path.join(path, flag, filename)
216 fp = file(fname, 'wb')
219 logger.notifyChannel('document', netsvc.LOG_DEBUG, "Saved data to %s" % fname)
220 filesize = len(data) # os.stat(fname).st_size
221 store_fname = os.path.join(flag, filename)
223 # TODO Here, an old file would be left hanging.
225 except Exception, e :
226 netsvc.Logger().notifyChannel('document', netsvc.LOG_WARNING, "Couldn't save data: %s" % str(e))
227 raise except_orm(_('Error!'), str(e))
228 elif boo.type == 'db':
230 # will that work for huge data? TODO
231 out = base64.encodestring(data)
232 cr.execute('UPDATE ir_attachment SET db_datas = %s WHERE id = %s',
233 (out, file_node.file_id))
234 elif boo.type == 'realstore':
236 file_node.fix_ppath(cr, ira)
237 npath = file_node.full_path() or []
238 # npath may contain empty elements, for root directory etc.
239 for i, n in enumerate(npath):
243 for ch in ('*', '|', "\\", '/', ':', '"', '<', '>', '?', '..'):
245 raise ValueError("Invalid char %s in path %s" %(ch, n))
248 path = os.path.join(*dpath)
249 if not os.path.isdir(path):
251 fname = os.path.join(path, npath[-1])
252 fp = file(fname,'wb')
255 logger.notifyChannel('document',netsvc.LOG_DEBUG,"Saved data to %s" % fname)
256 filesize = len(data) # os.stat(fname).st_size
257 store_fname = os.path.join(*npath)
258 # TODO Here, an old file would be left hanging.
261 traceback.print_exc()
262 netsvc.Logger().notifyChannel('document',netsvc.LOG_WARNING,"Couldn't save data: %s" % e)
263 raise except_orm(_('Error!'), str(e))
265 raise TypeError("No %s storage" % boo.type)
267 # 2nd phase: store the metadata
274 mime, icont = cntIndex.doIndex(data, ira.datas_fname,
275 ira.file_type or None, fname)
277 logger.notifyChannel('document', netsvc.LOG_DEBUG, 'Cannot index file: %s' % str(e))
280 # a hack: /assume/ that the calling write operation will not try
281 # to write the fname and size, and update them in the db concurrently.
282 # We cannot use a write() here, because we are already in one.
283 cr.execute('UPDATE ir_attachment SET store_fname = %s, file_size = %s, index_content = %s, file_type = %s WHERE id = %s',
284 (store_fname, filesize, ustr(icont), mime, file_node.file_id))
285 file_node.content_length = filesize
286 file_node.content_type = mime
288 except Exception, e :
289 netsvc.Logger().notifyChannel('document', netsvc.LOG_WARNING, "Couldn't save data: %s" % str(e))
290 # should we really rollback once we have written the actual data?
291 # at the db case (only), that rollback would be safe
292 raise except_orm(_('Error at doc write!'), str(e))
294 def prepare_unlink(self, cr, uid, storage_bo, fil_bo):
295 """ Before we unlink a file (fil_boo), prepare the list of real
296 files that have to be removed, too. """
298 if not storage_bo.online:
299 raise RuntimeError('media offline')
301 if storage_bo.type == 'filestore':
302 fname = fil_bo.store_fname
305 path = storage_bo.path
306 return (storage_bo.id, 'file', os.path.join(path, fname))
307 elif storage_bo.type == 'db':
309 elif storage_bo.type == 'realstore':
310 fname = fil_bo.store_fname
313 path = storage_bo.path
314 return ( storage_bo.id, 'file', os.path.join(path,fname))
316 raise TypeError("No %s storage" % boo.type)
318 def do_unlink(self, cr, uid, unres):
319 for id, ktype, fname in unres:
324 netsvc.Logger().notifyChannel('document', netsvc.LOG_WARNING, "Could not remove file %s, please remove manually." % fname)
326 netsvc.Logger().notifyChannel('document', netsvc.LOG_WARNING, "Unknown unlink key %s" % ktype)