def test_parsing():
assert rule_engine.parse(RULES1) == [
(['NOUN'], ['_S']),
(['COMP', 'Cmp2'], ['_A', '_comp2']),
(['|'], ['_='])
]
def _transform_tag(tag, **kwargs):
rules = rule_engine.parse(RULES2)
return ','.join(
rule_engine.apply_rules(rules, tag.split(','), **kwargs)
)
def test_parsing():
assert rule_engine.parse(RULES1) == [(['NOUN'], ['_S']),
(['COMP', 'Cmp2'], ['_A', '_comp2']),
(['|'], ['_='])]
FROM_OPENCORPORA = rule_engine.parse("""
# part of speech
NOUN => S
ADJF,Apro,Subx,Anph => A-PRO
ADJF,Apro,Subx => S-PRO
ADJF,Apro => A-PRO
ADJF,Anum => ANUM
ADJF => A,plen
ADJS => A,brev
COMP,Cmp2 => A,comp2
COMP,Supr => A,supr
COMP => A,comp
# INFN => V,inf,med
# INFN => V,inf,act
INFN => V,inf
PRTF,inan => V,partcp,plen
PRTF,anim => V,partcp,plen
PRTF => V,partcp,plen
PRTS,intr => V,intr,partcp,plen
PRTS => V,tran,partcp,brev
GRND => V,ger
VERB,impr,incl,sing => V,1p,imper,pl
VERB,impr,incl,plur => V,1p,imper2,pl
VERB,impr,excl => V,2p,imper
VERB,impr,incl => V,1p,imper
VERB => V
NUMR,inan => NUM
NUMR => NUM
PRED,pres => PRAEDIC
PRED => PRAEDIC
ADVB,Ques => ADV-PRO
ADVB => ADV
NPRO => S-PRO
PREP => PR
PRCL => PART
INTJ => INTJ
CONJ,Prnt => PARENTH
CONJ => CONJ
# animacy
anim => anim
inan => inan
# gender
femn,Ms-f => m-f
masc,Ms-f => m-f
neut,Ms-f => m-f
Ms-f => m-f
masc => m
femn => f
neut => n
# number
sing => sg
plur => pl
Fixd => 0
# case
nomn => nom
gent => gen
datv => dat
accs => acc
ablt => ins
loct => loc
voct => voc
gen1 => gen
gen2 => gen2
acc2 => acc2
loc1 => loc
loc2 => loc2
# Abbr doesn't mean the same, but convert it nevertheless
Abbr => abbr
Name => persn
Surn => famn
Patr => patrn
Dist => distort
Arch => anom
# ??
Litr => anom
# Infr => anom
# V-oy => anom
V-be => distort
Supr => supr
# --
# ??
Coun => adnum
# aspect
perf => pf
impf => ipf
# transitivity
tran => tran
intr => intr
# voice
actv => act
pssv => pass
# person
1per => 1p
2per => 2p
3per => 3p
# tense
pres => praes
past => praet
futr => fut
# mood
indc => indic
impr => imper
# extra grammemes
LATN => NONLEX
ROMN => ANUM,ciph
# non-standard & pymorphy2-specific
# (RNC doesn't tag punctuation marks)
PNCT => PNCT
NUMB => NUM,ciph
# hack to preserve whitespace info:
| => =
""")
def _transform_tag(tag, **kwargs):
rules = rule_engine.parse(RULES2)
return ','.join(rule_engine.apply_rules(rules, tag.split(','), **kwargs))
