Can we generate verb-lists from lexicon

[dhowe]

Can we generate verb-lists from lexicon, specifically VB_ENDS_IN_E and VB_ENDS_IN_DOUBLE ??

[dhowe]

also, IRREG_VERBS_NOLEX is missing in java

[Real-John-Cheung]

Can we generate verb-lists from lexicon, specifically VB_ENDS_IN_E and VB_ENDS_IN_DOUBLE ??

Yes we can, in fact now the lists are generated from the lexicon and then hard-coded into the file (may be we should do this generation when initializing conjugator? but it might slow down the process) this is also related to

IRREG_VERBS_NOLEX is missing in java

coz the reason that it is missing is that if we also hard-coded it in, the static initializer of Conjugator will exceed the 65535 bytes limit (another solution is we keep the lists in another file?)

[Real-John-Cheung]

I implant the method in the above PRs It did slow down the initialisation but I think not a lot

[dhowe]

I've done some refactors to unconjugate and its tests:

1. lazy-load verb lists only when calling unconjugate()
2. remove unneeded regular expressions (these are slower than simple string methods)
3. added tests for a few missing cases

TODO:

• [x] add tests for any missing cases (I marked one as a TODO) -- each branch should have at least one test
• [x] sync with java and re-PR