MorphGNT

morphgnt_lexicon.yaml -- stem lexicon with all the verbs in MorphGNT
morphgnt_generate.py -- script that goes through (using py-sblgnt) all the verb forms in given NT books and validates that the code + dataset generates the correct form
generate_morphgnt_lexicon.py -- similar to morphgnt_generate.py but instead of just validating and showing unexplainable forms, builds the starting point for a lexicon file to explain the forms
make_morphgnt_test.py -- script for generating a YAML test file (like those in test_data/) based on verbs in John (or change line 10)
make_morphgnt_nominal_test.py -- like make-morphgnt_test.py but for nouns and adjectives.
morphgnt_utils.py -- common code used by the above scripts

Using `morphgnt_generate.py`

To test John's gospel:

$ ./morphgnt_generate.py 4

To test the Johannine Epistles:

$ ./morphgnt_generate.py 23 24 25

morphgnt_generate.py also takes --lexicon and --stemming arguments to change the stem lexicon and stemming rule files respectively.

run generate_morphgnt_lexicon.py on the book(s) you want to add coverage for: ./generate_morphgnt_lexicon.py BOOK1 BOOK2 ... > tmp1
cat morphgnt_lexicon.yaml tmp1 > tmp2
./sort_lexicon.py tmp2 > morphgnt_lexicon.yaml
review all lines in morphgnt_lexicon.yaml that have # @ (you can review about 10 a minute once you get good at it)

@m means multiple possible stems, for example:

ἀγνοέω:
    stems:
        1-: {'ἀγνοου{athematic}', 'ἀγνοε', 'ἀγνοο'}  # @m

which should be manually corrected to:

ἀγνοέω:
    stems:
        1-: ἀγνοε

@1 means a single possible stem. Verify and make sure if the lemma already exists that the new stem is moved in with the others.

For example:

ἀποκόπτω:
    stems:
        3+: ἀπεκοψ
    stems:
        2-: ἀποκοψ  # @1

needs to be changed to

ἀποκόπτω:
    stems:
        2-: ἀποκοψ
        3+: ἀπεκοψ
    stems:

@0 means no stem could be guessed. This normally means a missing stemming.yaml rule.

Run ./morphgnt_generate.py with appropriate arguments to test your new lexicon.

Failures could just be mistakes you made in step 5 or could be missing stemming.yaml rules.