Parse non-tabular grammar information

[radomirbosak]

>>> python3 -c "import duden; print(duden.get('Kragen').grammar_raw)"
[]

duden.de, however has this text supplied in the grammar section:

der Kragen; Genitiv: des Kragens, Plural: die Kragen, süddeutsch, österreichisch, schweizerisch: Krägen

The script shouldn't omit this. However it is not clear in which form should it present the information in this rare case of non-table-like format.

[radomirbosak]

The word Mönch uses a similar format for grammar.

der Mönch; Genitiv: des Mönch[e]s, Plural: die Mönche

[radomirbosak]

Verbs like schwindeln also use a non-tabular grammar format.

schwaches Verb; Perfektbildung mit »hat«

We should first decide what should the function return for words like these.

[radomirbosak]

This bug is still valid in 0.10.0 .

[radomirbosak]

the word Bereich has the same problem - the section doesn't contain any tables, just the text der, selten: das Bereich; Genitiv: des Bereich[e]s, Plural: die Bereiche.

[radomirbosak]

The word Meme is also a bit special since it has multiple spellings. This will complicate grammar parsing.

[radomirbosak]

A related attribute grammar_overview was added in #168 . However, it does not parse the text data there.

[radomirbosak]

The above new attribute should be enough and I won't be implementing more detailed parsing of a highly variable text attribute. I will close this as out of scope.