Skip to content

Inflection

Every analysis knows its lexeme - all forms of the word with the same lemma. From an analysis you can get the list of forms, the form with given UD features or the form that agrees with a number. The lexeme is defined by the lemma, so it follows the UD lemma conventions: the lexeme of «Ивановой» is the feminine forms of the surname «Иванова», the lexeme of «важнейшего» is the superlative forms.

The word forms are spelled as in the dictionary: lowercase and with «ё» where the dictionary has it; the forms of guessed words are described in Unknown words.

Methods

Forms of the lexeme

forms · list[Analysis]

forms()

The forms of the lexeme in dictionary order - analyses like those of analyze, with probability 1. Form variants differ in labels: «кошкою» has V-oy, the second locative «в лесу» has loc2 (its Case=Loc feature is the same as in «о лесе»): «лес» with inflect(Case="Loc") gives «лесе», and «лесу» is chosen from forms() by the loc2 label.

from rumorph import Analyzer

morph = Analyzer()
cat = morph.analyze("кошка")[0]
[form.word for form in cat.forms()]
form = cat.forms()[8]
form.word, form.feats

['кошка',
 'кошки',
 'кошке',
 'кошку',
 'кошкой',
 'кошкою',
 'кошке',
 'кошки',
 'кошек',
 'кошкам',
 'кошек',
 'кошками',
 'кошках']
('кошек', {'Animacy': 'Anim', 'Case': 'Gen', 'Gender': 'Fem', 'Number': 'Plur'})

Form by features

inflect · str

inflect(**feats)

The form of the lexeme with the given UD features. Features that are not given are kept where possible: «кошка» in the genitive is «кошки», not «кошек». The short form, the animate accusative («нового») and the inclusive imperative («сделаем») are chosen only when asked for: Variant="Short", Animacy="Anim", Person="1". The future of «быть» in SynTagRus is Tense=Pres with the futr label: «будет» in the plural is «будут», while inflect(Tense="Fut") of «быть» raises FormNotFoundError - the future is taken from forms() by the futr label.

Parameter Type Default Description
**feats str - UD features and their values: Case="Gen", Number="Plur"

Exceptions:

  • FormNotFoundError - the lexeme has no form with these features;
  • ValueError - the feature or the value is not in the dictionary (a typo like Case="Genitive"), the message lists the valid ones;
  • TypeError - a feature value that is not a string.
cat.inflect(Case="Gen"), cat.inflect(Case="Gen", Number="Plur")
verb = morph.analyze("читал")[0]
verb.inflect(Gender="Fem"), verb.inflect(VerbForm="Inf")
verb.inflect(Tense="Pres", Person="3", Number="Plur")
morph.analyze("новый")[0].inflect(Degree="Cmp")

('кошки', 'кошек')
('читала', 'читать')
'читают'
'новее'

Agreement with a number

agree · str

agree(number)

The form of a noun, an adjective, a pronominal adjective or a participle that agrees with a number. The case of the analysis is kept: «рублях» with 2 is «рублях» («о 2 рублях»), the accusative of animate nouns gives «вижу двух котов». Adjectives and nouns declined like adjectives agree by gender: «две новые», «два новых», «два учёных», «две столовые». After a number the counting form is taken: «5 вольт», «5 лет», «5 человек». Words without a singular take the plural: «1 сутки», «2 суток». Feminine surnames in -ова, -ина agree like adjectives: «две Ивановы», «пять Ивановых». A negative number agrees as its absolute value, a fractional one («1,5», Decimal("2.5"), Fraction(3, 2)) takes the genitive singular in any case: «1,5 рубля», «о 2,5 рубля».

Parameter Type Default Description
number int / float / Decimal / Fraction - The number; an integer is any number with __index__, and also a Decimal or Fraction without a fractional part

Exceptions:

  • FormNotFoundError - the word does not agree with numbers (a verb, an adverb) or the lexeme has no needed form («5 молока»);
  • TypeError - a value that is not a number;
  • ValueError - infinity or NaN.
ruble = morph.analyze("рубль")[0]
[f"{n} {ruble.agree(n)}" for n in (1, 2, 5, 21)]
morph.analyze("сутки")[0].agree(2)

['1 рубль', '2 рубля', '5 рублей', '21 рубль']
'суток'