For #899, we need to change MediaFile's behavior (pre-write) based on whether
we're doing ID3v2.3 or not. So we need a field on the object, not a parameter
to `save()`.
Due to the new exception nesting stuff, we were catching and emitting
exceptions where none was necessary: specifically, when the file was non-music
(which is expected, especially when there are images).
There were a number of problems with the changes to the util melange:
- It used print rather than logging, and its string formatting was probably
not Unicode-ready.
- The shell-command-like print lines were not quite compatible, which makes
their general usefulness questionable.
- Used an unsafe/leaky global variable for mkdirall.
- Used deprecated sets.Set.
Seemed better just to add this to the plugin where we need it so it's easier
to see where this goes.
It also seems unnecessary to me to print `mkdir -p` commands. They just
clutter up the output for me when I really just want to see the transcoding
commands.
This option allows the user to specify the format on the command line
instead of editing the configuration.
The commit also includes some refactoring. In particular adding
arguments to functions to avoid dependence on global state.
Doc and Changelog in next commit
Partially resolves#877 showing:
- Directory creation
- Copies
- Deletes
- Moves
- Encodings
Information about tagging and plugins on _after_convert_ is not
currently shown. That requires changing the plugins to support the
pretend option, so a lot of work may be needed and it doesn't seem to be
helpful enough for me.
Progress is a `defaultdict` and we assume presence of any key. This
means we have to assign keys the default value instead of deleting
them. Fixes#873.
* Control flow and implementation of help command is now
similar to the other commands.
* Simplifies and flattens some code and removes unused method.
* Makes SubcommandOptionParser agnostic of Subcommand.parser.
Instead of recording only the most recently imported path and assuming
that all previous path have also been imported, we record all imported
paths and skip recorded paths when resuming the import.
This fixes an issue where parallelism would screw up the import order
and reimport some files. It also allows us to record singleton items.
We can apply the same strategy can be applied for incremental imports.
