In GDrive, filenames aren't unique, and directories are just
special files - so you can have multiple files with the same
name, multiple directories with the same name, and even files
with the same names as directories.
OC doesn't handle this at all, though, and just wants to act
as if file and directory names *are* unique. So when renaming,
we must check if there's an existing object with the same
file or directory name before we commit the rename, and
explicitly delete it if the rename is successful. (Other
providers like dropbox do the same for files, but intentionally
don't do it for directories; we really need to do it for
directories too.)
A good way to observe this is to run the storage unit tests
and look at the state of the Drive afterwards. Without this
commit, there will be several copies of all the test files
and directories. After this commit, there's just one of each.
We can't just say "hey, Drive lets us do this, what's the
problem?" because we don't handle multiple-objects, same-name
cases - getDriveFile() just bails and prints an error if it
searches for the file or directory with a given name and gets
multiple results.
Sometimes there are bugs that cause setupFS() to be called for
non-existing users. Instead of failing hard and breaking the instance,
this fix simply logs a warning.
opendir() trims the path passed then calls getDriveFile() -
which immediately does the same trim operation. This breaks
opendir() on the root directory, which causes the failure of
the testStat() test when it checks the ctime of the root dir.
ownCloud passes us a Unix time integer, but the GDrive API wants
an RFC3339-formatted date. Actually it wants a single particular
RFC3339 format, not just anything that complies will do - it
requires the fractions to be specified, though RFC3339 doesn't.
This resolves issue #11267 (and was also noted by PVince81 in
reviewing PR #6989).
This is a slightly hacky workaround for
https://github.com/google/google-api-php-client/issues/59 .
There's a bug in the Google library which makes it go nuts on
file uploads and transfer *way* too much data if compression is
enabled and it's using its own IO handler (not curl). Upstream
'fixed' this (by disabling compression) for one upload
mechanism, but not for the one we use. The bug doesn't seem to
happen if the google lib detects that curl is available and
decides to use it instead of its own handler. So, let's disable
compression, but only if it looks like the Google lib's check
for curl is going to fail.
Allow specifying a protocol in the host field when mounting another
ownCloud instance. Note that this was already possible with the WebDAV
config but this bug made it inconsistent.
folder size and mtime is always unknown in s3
more s3 fixes
make rescanDelay of root dir configurable, add on the fly update of legacy storage ids, !isset -> empty when checking strings
reduce number of http calls on remove and rmdir
fix typo
maintain deprecated \OC::$session when getting or setting the session via the server container or UserSession
restore order os OC::$session and OC::$CLI
remove unneded initialization of dummy session
write back session when $useCustomSession is true
log warning when deprecated app is used
As constants not defined within a class cannot be automatically found by the
autoloader moving those constants into a class makes them accessible to
code which uses them.
Signed-off-by: Stephan Peijnik <speijnik@anexia-it.com>
For some reason the aws-sdk-php package does not caclulate the
signiture correctly when accessing an object in a bucket with a name of
'.'.
When we are at the top of a S3 bucket there is a need(?) to have a directory
name. Per standard Unix the name picked was '.' (dot or period). This
choice exercises the aws-sdk bug.
This fix is to add a field to the method to store the name to use instead of
'.' which at this point is hard coded to '<root>'. We also add a private
function 'cleanKey()' which will test for the '.' name and replace it with
the variable. Finally all calls to manipulate objects where the path is
not obviously not '.' are processed through cleanKey().
An example where we don't process through clean key would be
'Key' => $path.'/',
Use correct relationship operator
Per feed back use === instead of ==
use '/' instead of '<root>'
Now the external storage correctly returns the mount points visible only
for the current user by using the method getAbsoluteMountPoints() which
is already filtered.
Since that call was missing the backend name which is important for the
UI, this one was added as well.
Each storage backend has a default priority, assigned to any system mounts
created in ownCloud. mount.json can be manually modified to change these
priorities.
The priority order is as follows:
* Personal
* User
* Group
* Global
Within each mount type, the mount with the highest priority is active.
The storage backend defaults were chosen to be the following:
* Local - 150
* Remote storage - 100
* SMB / CIFS with OC login - 90
An issue existed where `readData` used `OCP\User::getUser()` to get the user
for personal mount points, which worked in all situations apart from when a
personal mount point was used for sharing, so the return from `getUser()` is
not the user that owns the share. As such, any personal mount points would not
work correctly when shared.
`readData` and `writeData` have been changed from using a `$isPersonal`
boolean to using a `$user` string|null. `$isPersonal = false` can now be
written as `$user = NULL` (or left out in the case of `readData`), and
`$isPersonal = true` can be written as `$user = OCP\User::getUser()`.
Dependency messages now appear below the configuration options instead of above
Reworked dependency check method to support consolidated messages for multiple backends
Conflicts:
apps/files_external/lib/google.php
apps/files_external/lib/swift.php
apps/files_external/templates/settings.php
Added check for backend's checkDependencies method to OC_Mount_Config::getBackends() when backend is configured to have some instead of blindly calling it and crashing
Conflicts:
apps/files_external/lib/config.php
apps/files_external/lib/google.php
Moved dependency check to the individual backends instead of the config
Conflicts:
apps/files_external/appinfo/app.php
apps/files_external/lib/config.php
The touch() operation now uses "UpdateMetadata()" instead of "Update()"
which doesn't clear the object's contents.
This fixes syncing, as the sync client needs to use touch to update the
object's mtime.
To check for shares, the code attempts to connect anonymously to the share.
In most cases this will fail with NT_STATUS_ACCESS_DENIED, so the regex array
used for parsing the output of smbclient in smb4php has been overridden to
treat such output as success.
The 'test' method for storage classes can now take a single parameter,
$isPersonal, which allows the storage to adjust the tests performed based on
if they are being configured as personal shares or as system shares.
Shares authenticated with user credentials (aka not hard coded) cannot be
accessed by other users, breaking sharing. This change completely revokes
sharing for such shares
getDisplayName would return the display name of the user, not great if it is a
canonical string. The uid passed back from 'login' is the UUID of the user, so
also not suitable. The login name from the session is the username the user
used to log in to ownCloud in the first place, which is what is needed.
Selecting 'SMB/CIFS Auto' in the mounts configuration allows an SMB/CIFS
mount to be configured that uses the credentials of the user logging in to
authenticate to the server.
Optionally, the username can be used as the share name, permitting home shares
to be dynamically mounted.
Merging of configurations is whenever the same config is available for
multiple users/groups, in which case the config is considered as a
single one by the UI, and shows multiple users/groups selected.
Fixed merging logic to make sure that class, mount point and options are
the same before merging them.
Fixed merging to work correctly when the same mount point path is used
for separate users and configs. These are now correctly shows in the UI
as separate entries.
The old password field is now preserved in the JSON structure to make
sure that the order is preserved.
This is a quick fix until the UI is fixed to not rely on the PHP array
key order.
- test config was written correctly (global and local)
- test that personal mount points can be added for allowed backends
- test that personal mount points cannot be added for disallowed backends
- added $skipTest flag to make it possible add mount points without
doing an actual check/connection (note: this is necessary because the
static class OC_Mount_Config cannot be mocked)
- Some WebDAV servers like lighttpd need paths in URLs to be properly
encoded
- Added error log output when curl connection failed
- Added check for 'resourcetype' in case the WebDAV server doesn't
support/return it
- Fixed touch() to return false if the server doesn't implement
PROPPATCH
- Added optional delay in WebDAV unit tests to use when testing against
lighttpd's WebDAV
Added private var $certPath to store the user root cert
Move logic to determine the $certPath path to the constructor and modify to get the path from OC_User::getHome()
Add curl options to use the certificate to avoid certificate errors with self-signed certicates in the downdload/upload method so we don't get blank files
Some storages need to use different calls for deleting files or folders,
usually unlink() and rmdir().
Fixes#4532 (SMB dir deletion)
Fixes#5941 (FTP dir deletion)
Note that the extra is_dir() should be fast because it's read from the
stat cache.
- fixed touch function to return true on success
- fixed local metadata cache to remove deleted files/directories
- fixed getMetaData() to ignore files reported as deleted by
Dropbox
- fixed "file not found" case to not log as exception
- fixed "overwrite on rename" case
- fixed unit tests to exclude unsupported cases
- added unit test for touch return value
- unit tests for Dropbox should all run correctly now
When uploading files through WebDAV, a part file is created and a rename
operation is performed with the expectation that the part file
overwrites an existing file, if any.
This fix makes the SMB external storage delete the target file before
renaming, as smbclient doesn't support overwrite on move/rename.
Fixes#5348
When an ext storage doesn't implement isReadable(), always returning
true made the file scanner believe that the file exists and creates a
cache entry with the size zero.
This fix makes the default impl of isReadable() use file_exists().
Fixes#5940
