Restructure importers

[tobiasdiez]

The aim of this PR is to reduce the overlap between the Importer package and some code in GUI (in particular, the class OpenDatabaseAction).
Moreover, #1080 and #1153 are fixed.

In detail, the following things were changed:

• The abstract ImportFormat class is almost completely rewritten. Please have a look at this class. Almost all other changes are a direct consequence of the edits made to the ImportFormat class. The biggest change is that the import method returns a ParserResult instead of a list of entries (in this way, metadata can also be imported and error messages are communicated more effectively instead of throwing IOExceptions). Moreover, the import method operates on a reader instead of an input stream in order to reduce some common code in all importers.
• Add method importDatabase(Path file, Charset encoding) to the importer interface which opens the file and passes the stream to the other import method.
• The BibTexImporter overwrites this new import method and uses the logic from OpenDatabaseAction.loadDatabase to determine the encoding of the bib file.
• Change OpenDatabaseAction to operate against the BibTex importer.
• Remove get/set isCustomImporter: was not really used (just for sorting - now everything is alphabetical)
• Empty entries are no longer imported (thus a few test had to be changed)

Remark: The methods getExtension and getDescription are not implemented by almost any importer. This should be done. Maybe this is something for the stupro?

(Will add the changelog entry and add a few tests later)

• Change in CHANGELOG.md described
• Tests created for changes
• Screenshots added (for bigger UI changes)

[tobiasdiez]

Any comments or can I rebase and merge?

[oscargus]

Why return null here (and in similar places)?

[tobiasdiez]

Since I have no idea for which extensions these importers support.

See above:

Remark: The methods getExtension and getDescription are not implemented by almost any importer. This should be done. Maybe this is something for the stupro?

[oscargus]

Ah, OK. Tired... Somehow, the questions was also, why not an empty list?

[oscargus]

I do not fully understand why the getCLIId method is removed? How is the importer selected from CLI now?

Also, since there is a method setCliId, maybe that should change name as well (as it appears as getCLIId is sometimes replaces with getId, somehow, it seems like getCLIId is a better name, but no strong opinions).

Many changes are related to imports and replacing extension strings with a list of extension strings, right?

[oscargus]

Now I think I understand why. However, I think that getCLIId and getFormatName in general should return different things...

[oscargus]

Well, maybe I do not understand the code fully (again, tired), but what I was worried about with getId doesn't seem to the case, so to me this looks good!

[tobiasdiez]

There is a method getId (previously getCLIID) in ImportFormat which constructs the id based on the name of the formatter. Thus some of the hand-coded ids in the specific importers could be removed.

I couldn't find the method setCLIID. Where does it exists?

[oscargus]

I saw "REPEC New Economic Papers (NEP)" and then that getId() used that string which caused one comment, but then I saw the tests and realized that one wasn't supposed to type something like that. :-) Now, I also know why.

👍

[simonharrer]

Waiting for the last #547 to be merged - then, this PR can clean up all the importer code uniformly.

[simonharrer]

This is unblocked now. Go 4 it. :)