In an attempt to finally address the longstanding SQLite locking issues, I'm
introducing a way to explicitly, lexically scope transactions. The Transaction
class is a context manager that always fully fetches after SELECTs and
automatically commits on exit. No direct access to the library is allowed, so
all changes will eventually be committed and all queries will be completed. This
will also provide a debugging mechanism to show where concurrent transactions
are beginning and ending.
To support composition (transaction reentrancy), an internal, per-Library stack
of transactions is maintained. Commits only happen when the outermost
transaction exits. This means that, while it's possible to introduce atomicity
bugs by invoking Library methods outside of a transaction, you can conveniently
call them *without* a currently-active transaction to get a single atomic
action.
Note that this "transaction stack" concepts assumes a single Library object per
thread. Because we need to duplicate Library objects for concurrent access due
to sqlite3 limitation already, this is fine for now. Later, the interface should
provide one transaction stack per thread for shared Library objects.
Instead of parsing the template at each call to destination(), it's now possible
to parse them *once*, a priori, and re-use the resulting template object. This
is analogous to the re module's compiled expressions.
For a less cumbersome uniquifying string, only a single field value is now used
instead of a prefix of a list of fields. The old semantics had two problems that
made it both unnecessary and insufficient:
- In the vast majority of cases, a single field suffices (year OR label OR
catalog number, for example) and forcing the string to include many identical
fields is unnecessary.
- If the albums are very similar, a prefix may be insufficient; a better
solution may be found with an arbitrary subset. (Of course, we can't afford to
search the whole power set.)
So we're going with a single field for now. This should cause far less
confusion.
- Copying and moving are mutually exclusive. Moving overrides copying so the
user only has to add one line ("import_move: true") to disable copying and
enable moving in its place.
- Deleting is only possible when copying.
- Deprecating the "delete" option (moving is almost always better).
- Removed command-line switch for moving. It's somewhat "unsafe", so this
removes some potential for accidental irreversible changes.
- Changelog & thanks.
- Update docs to refer to import_move instead of import_delete as the
correct solution for ending up with only one copy of the file.
The new fields are:
ALBUM: mb_releasegroupid asin catalognum script language country albumstatus
media albumdisambig
TRACK: disctitle encoder
These are not yet parsed from MusicBrainz responses (just added to MediaFile
and the database).
There's no longer a distinction between Unix and Windows substitutions. Enough
users reported problems with Windows-forbidden characters on Samba shares that
it seems appropriate to make all filenames Windows-safe, even on Unix. Users who
really want those additional characters (<>:"?*|\) can re-enable them via the
"replace" option. Nobody has complained about beets being *too* conservative.
This also adds sanitization of control characters, which is an all-around good
idea, and the substitution now runs in the Unicode (rather than byte) domain.
Previously, there was just an "artist sort name" field -- now there's a
corresponding sort name for both track artists and album artists. I also made
the names shorter (artist_sort and albumartist_sort).
Generates disambiguating strings to distinguish albums from one another. To be
used as the basis for a simpler path field, $unique, as a default disambiguator.
Based on the "remove_duplicates" flag on ImportTask, the apply_choices coroutine
now looks for duplicates (using an extended version of the _duplicate_check
functions) and removes items from the library. It also *deletes* files
associated with those items when they are located inside the beets library
directory. Files outside of the directory are left on disk (but their DB entry
is still removed). This should "do the right thing" in most cases -- again, this
is something we can add a config option for if it comes up.
For the recently-added samplerate, bitdepth, and channels properties on
MediaFile, a few things were fixed:
- tests in test_mediafile_basic
- never return None (zero when unavailable)
- make channels work with MP3 files (by looking at the codec "mode")
Also added some docstrings on all of the properties.
With import_delete enabled and performing a re-import (which moves files), the
former location of the file would be "deleted". This would lead to a "file does
not exist" error.
When a partial match is found, its first item (task.items[0]) may be None, and
_infer_album_fields would crash in this case. This solution walks through the
items list and finds the first non-None item.
Previously, an empty argument was treated as "not an argument at all". Now,
every function call always has at least one argument -- i.e., %foo{} is a
function call whose only argument is "" -- and %foo{,bar} is valid syntax.
This was causing a problem with situation where }} would have semantic meaning
other than escaping a }. Specifically, %func{%func{arg}} contains a }} but
should not escape the }. $} seems to cover this situation. However, ${ is not
permitted as an escape sequence because it looks like the beginning of a symbol
(variable reference) like ${foo}. This is OK because { can be used anywhere as a
literal.
- Plugins are sent the unadulterated, None-ridden ordered items lists. Changed
the lastid plugin to accommodate this.
- Make colorization optional in partial album warnings.
- Fix some tests.
In the function order_items, instead of automatically reject the
canonical candidate if it has more tracks, the function still tries to
find matches for the tracks amongst the items, and otherwise uses None
to fill the void in order to keep the information about the track
numbers
If the user has some songs from a specific album, but not all of them,
the real solution is immediately discarded. This commit is the first of a
series that will implement support for these incomplete albums.
The point of this patch is to make sure missing commits are taken into
account when calculating the distance between an album and its canonical
data.
Note that in order not to break API compatibility, the album_distance
call for the plugins receives a purged version of both the items and the
album info, resulting in some potential accuracy if the plugin bases
itself on the index of a track in album_info.tracks.
The interface no longer specifies the type of the image embedded in the file; it
just returns a bytestring blob. When a type must be stored, it is inferred using
the imghdr module, which shoudl reduce the potential for weird bugs when the
formats don't correspond.
"beet import -i" now tags items instead of albums. There are many loose ends to
tie up (marked with TODOs in the source):
- What to do about applying non-track metadata to matched tracks? Currently it's
just left in place.
- Plugin autotag candidates for tracks.
- No user querying yet.
- Non-autotagged -i import are unimplemented.
And, on top of those:
- Need to remove the action.TRACKS workflow and replace it with an option that
lets you jump over to the individual-track interface from the album tagger.
