Unify OpenDatabaseAction and Importer

[tobiasdiez]

I had a look at how to fix #1080 and #1153 and recognized that there is quite some overlap between the Importer package and some code in GUI (in particular, the class OpenDatabaseAction).

I propose to unify the code used to open or import a database.

• Change List<BibEntry> importEntries(InputStream in, OutputPrinter status) in ImportFormat to a method ParserResult import(InputStream in) which might also return some metadata. The returned ParserResult contains potential error messages (which in my opinion is cleaner than letting JabRefFrame implement OutputPrinter).
• Add default method import(Path file, Charset defaultEncoding) to the importer interface which opens the file and passes the stream to the other import method.
• The BibTexImporter can overwrite import(Path file, Charset defaultEncoding) and use the logic from OpenDatabaseAction.loadDatabase to determine the correct encoding.
• Change OpenDatabaseAction to operate against the (BibTex)importer.
• At this point the importFromFile methods in ImportFormatReader can be removed.
• Remove isRecognizedFormat: this method is only used directly before the import, so ImportFormat.import can just return a corresponding warning message.
• Remove get/set isCustomImporter: is not really used (just for sorting and I think a completely alphabetic display is more logical)
• getExtension and getDescription is not implemented for a lot of importer?!
• Changed: Empty entries are no longer imported (thus a few test had to be changed)

What do you think about the proposed changes?

[lenhard]

Sounds fine to me and I think you can go ahead with a PR. There is no reason to duplicate the BibTeX file reading logic.

Just be prepared to run into unforseen complications that thwart your plan (as always in JabRef) :)

[oscargus]

Also, unifying append and import to existing would be nice, see #262

[oscargus]

Regarding isRecognizedFormat, I think it makes sense to keep. The whole point is to be able to guess format without doing an actual import. Given that it is implemented correctly, there should be a significant complexity reduction compared to actually trying to import in each and every format.

[tobiasdiez]

Thanks for the feedback. The rest can be discussed in the PR #1207