def corr_gem_cutting(text, search_result):
"""
>>> corr_gem_cutting("(бриолетовый восковые опалы)")
'(бриолетовые восковые опалы)'
>>> corr_gem_cutting("большой шерлы")
'большие шерлы'
"""
words = search_result.group(1).split()
if words[-1] in body_parts:
return corr_item_body_parts(text)
gender = get_gender(words[-1], {"NOUN", "nomn"})
new_list = []
for word in words[:-1]:
if word in make_adjective:
adj = make_adjective[word]
word = inflect_adjective(adj, gender)
elif is_adjective(word):
word = inflect_adjective(word, gender)
new_list.append(word)
new_list.append(words[-1])
return text.replace(search_result.group(0), " ".join(new_list))
def corr_settlement(_, search_result):
"""
>>> corr_settlement(" человеческий крепость Belrokalle")
'Человеческая крепость Belrokalle'
>>> corr_settlement(" эльфийский лесное убежище Etathuatha")
'Эльфийское лесное убежище Etathuatha'
>>> corr_settlement(" дварфийский горный город КилрудОстач")
'Дварфийский горный город Килрудостач'
>>> corr_settlement(" лесное убежище Cinilidisa")
'Лесное убежище Cinilidisa'
"""
adjective = search_result.group(1).strip()
settlement = search_result.group(2)
name = search_result.group(3)
if len(adjective) == 0:
return "{} {}".format(settlement.capitalize(), name.capitalize())
if adjective in {"Покинуть", "Разрушить"}:
return
gender = get_main_word_gender(settlement)
if " " not in adjective:
adjective_2 = inflect_adjective(adjective, gender)
else:
adjective_2 = " ".join(inflect_adjective(word, gender) for word in adjective.split(" "))
if adjective_2 is None:
adjective_2 = adjective
return "{} {} {}".format(adjective_2.capitalize(), settlement, name.capitalize())
def corr_adjective_relief(text, search_result):
"""
>>> corr_adjective_relief("Заснеженный Густой овсяница")
'Заснеженная густая овсяница'
>>> corr_adjective_relief("Густой мюленбергия")
'Густая мюленбергия'
>>> corr_adjective_relief("Густой морошка")
'Густая морошка'
>>> corr_adjective_relief("Заснеженный Густой куропаточья трава")
'Заснеженная густая куропаточья трава'
"""
adjective = search_result.group(1)
obj = search_result.group(2)
if " " in obj:
words = obj.split(" ")
if is_adjective(words[0]):
gender = get_gender(words[-1])
new_word = inflect_adjective(words[0], gender, "nomn")
text = text.replace(words[0], new_word)
new_word = inflect_adjective(adjective, gender, "nomn")
text = text.replace(adjective, new_word)
else:
gender = get_gender(obj)
new_word = inflect_adjective(adjective, gender, "nomn")
if new_word:
text = "{} {}".format(new_word, obj)
return text.capitalize()
def corr_jewelers_shop(_, search_result):
"""
>>> corr_jewelers_shop("Огранить из необработанного адамантина")
'Огранить необработанный адамантин'
>>> corr_jewelers_shop("Инкрустировать Предметы обстановки с из необработанного адамантина")
'Инкрустировать предметы обстановки необработанным адамантином'
>>> corr_jewelers_shop("Огранить из фарфора")
'Огранить фарфор'
"""
first_part = search_result.group(1)
words = search_result.group(2).split()
if first_part == "Огранить":
# accusative case
tags = None
if words[0] == "из":
words = words[1:]
tags = {"gent"}
item = words[-1]
gender = get_gender(item, known_tags=tags)
words = [inflect_adjective(word, gender, "accs", animated=False) for word in words[:-1]]
parse = list(filter(lambda x: {gender, "inan"} in x.tag, custom_parse(item)))
if item == "адамантина":
item = "адамантин"
else:
item = parse[0].inflect({"accs"}).word
words.append(item)
else:
# instrumental/ablative case ('incrust with smth')
words = [custom_parse(word)[0].inflect({"ablt"}).word for word in words if word != "из"]
if first_part.endswith(" с"):
first_part = first_part[:-2]
text = first_part + " " + " ".join(words)
return text.capitalize()
def corr_craft_glass(text, search_result): # TODO: Combine into single crafting-related function
"""
>>> corr_craft_glass("Делать грубый зелёное стекло")
'Варить грубое зелёное стекло'
>>> corr_craft_glass("Делать гигантский хрусталь лезвие топора")
'Делать гигантское лезвие топора из хрусталя'
"""
material = search_result.group(3)
material_gender = get_gender(material)
words = search_result.group(2).split()
product = search_result.group(4).split()
verb = search_result.group(1)
if not product:
verb = "Варить"
adjectives = (inflect_adjective(adj, material_gender, "accs", animated=False) for adj in words)
result = "{} {} {}".format(verb, " ".join(adjectives), material)
else:
index = next(
(i for i, item in enumerate(words) if item in {"грубое", "зелёное", "прозрачное", "грубый"}), len(words)
)
product_adjectives = words[:index]
if any_in_tag({"NOUN", "nomn"}, custom_parse(product[0])):
product_gender = get_gender(product[0])
product[0] = inflect_noun(product[0], case="accs")
else:
product_gender = get_gender(product[-1])
product_adjectives += product[:-1]
product = [inflect_noun(product[-1], case="accs")]
product_adjectives = [
inflect_adjective(adj, product_gender, case="accs", animated=False) for adj in product_adjectives
]
def corr_relief(_, search_result):
"""
>>> corr_relief("Мёртвый клён деревце")
'Мёртвое деревце (клён)'
>>> corr_relief("Глинистый суглинок Стена")
'Стена из глинистого суглинка'
>>> corr_relief("кремень подъем")
'Подъем из кремня'
"""
group1 = search_result.group(1)
obj = search_result.group(2)
if obj == "деревце":
if group1.split(" ")[0] == "Мёртвый":
text = "Мёртвое деревце ({})".format("".join(search_result.group(0).split(" ")[1:-1]))
else:
text = "Деревце ({})".format(group1)
return text.capitalize()
if " " in group1:
words = group1.split(" ")
first_words = []
gender = get_main_word_gender(obj)
for word in words:
if word in {"Заснеженный", "Неотесанный", "Влажный"}:
if gender is not None:
new_word = inflect_adjective(word, gender)
if not new_word:
new_word = word
else:
new_word = word
first_words.append(new_word)
else:
break
words = words[len(first_words) :]
if words[0] == "из":
words = words[1:]
else:
words = to_genitive_case_list(words)
if not first_words:
text = "{} из {}".format(obj, " ".join(words))
else:
text = "{} {} из {}".format(" ".join(first_words), obj, " ".join(words))
else:
material = group1
text = "{} из {}".format(obj, to_genitive_case(material))
if "иза" in text:
text = text.replace(" иза", "")
return text.capitalize()
def corr_craft_general(text, search_result):
"""
>>> corr_craft_general("Изготовить камень дверь")
'Изготовить каменную дверь'
>>> corr_craft_general("Делать деревянный ловушка для животных")
'Делать деревянную ловушку для животных'
>>> corr_craft_general("Украшать кость")
'Украшать кость'
>>> corr_craft_general("Делать деревянный изделия")
'Делать деревянные изделия'
"""
verb = search_result.group(1)
words = search_result.group(2).split()
product = None
if len(words) > 1:
for i, word in enumerate(words[1:], 1):
if any_in_tag({"NOUN", "nomn"}, custom_parse(word)) and word not in make_adjective:
product = " ".join(words[i:])
words = words[:i]
break
else:
product = words[0]
words = []
product_gender = get_main_word_gender(product)
if " " not in product:
orig_form = {"plur" if product_gender == "plur" else "sing", "inan"}
product = inflect_noun(product, "accs", orig_form=orig_form)
assert product is not None
else:
product = inflect_collocation(product, {"accs"})
if words:
if len(words) == 1 and words[0] not in make_adjective and not is_adjective(words[0]):
material = inflect_noun(words[0], "gent", orig_form={"nomn", "inan"}) # рог -> (из) рога
assert material is not None
result = "{} {} из {}".format(verb, product, material)
else:
adjectives = [
make_adjective[word] if word in make_adjective else word if is_adjective(word) else None
for word in words
]
assert all(adj is not None for adj in adjectives)
adjectives = [inflect_adjective(adj, product_gender, "accs", animated=False) for adj in adjectives]
result = "{} {} {}".format(verb, " ".join(adjectives), product)
else:
result = "{} {}".format(verb, product)
return text.replace(search_result.group(0), result).capitalize()
def corr_wooden_logs(text, search_result):
"""
>>> corr_wooden_logs('древесина дуба брёвна')
'дубовые брёвна'
"""
of_wood = "из " + search_result.group(2)
if of_wood in make_adjective:
adj = inflect_adjective(make_adjective[of_wood], "plur")
text = text.replace(search_result.group(0), adj + " " + search_result.group(3)) # берёзовые брёвна
else:
text = text.replace(
search_result.group(0), search_result.group(1) + " " + search_result.group(2)
) # древесина акации
return text
def corr_clothiers_shop(_, search_result):
"""
>>> corr_clothiers_shop("Делать ткань роба")
'Шить робу из ткани'
>>> corr_clothiers_shop("Делать шёлк роба")
'Шить шёлковую робу'
>>> corr_clothiers_shop("Изготовить ткань мешок")
'Шить мешок из ткани'
>>> corr_clothiers_shop("Вышивать кожа изображение")
'Вышивать изображение на коже'
>>> corr_clothiers_shop("Делать пряжа рубаха")
'Вязать рубаху из пряжи'
>>> corr_clothiers_shop("Делать ткань верёвка")
'Вить верёвку из ткани'
"""
verb = search_result.group(1)
material = search_result.group(2)
product = search_result.group(3).strip()
if not product:
return None # Leave as is eg. 'Ткать шёлк'
elif verb == "Вышивать": # Sew
if material == "пряжа":
# вязать <product> из пряжи
verb, preposition, material = "Вязать", "из", "пряжи"
else:
# вышивать <product> на <material>
preposition = "на"
material = inflect_noun(material, case="loct", orig_form={"nomn"})
return "{} {} {} {}".format(verb, product, preposition, material)
else:
if product in {"щит", "баклер"}:
_, of_material = cloth_subst[material] # Don't change the verb, leave 'Делать'/'Изготовить'
else:
verb, of_material = cloth_subst[material]
if product == "верёвка":
verb = "Вить"
product_accus = inflect_noun(product, case="accs", orig_form={"nomn"})
if material in make_adjective: # "шёлк" -> "шёлковый"
gender = get_gender(product, {"nomn"})
material_adj = inflect_adjective(make_adjective[material], gender, "accs", animated=False)
return "{} {} {}".format(verb, material_adj, product_accus) # {Шить} {шёлковую} {робу}
else:
return "{} {} {}".format(verb, product_accus, of_material) # {Шить} {робу} {из ткани}
def corr_weapon_trap_parts(text, search_result):
"""
>>> corr_weapon_trap_parts('гигантский из меди лезвия топоров')
'гигантские медные лезвия топоров'
>>> corr_weapon_trap_parts('большой зазубренный из берёзы диски')
'большие зазубренные берёзовые диски'
"""
adj = search_result.group(1)
words = search_result.group(2).split()
if " ".join(words[:2]) in make_adjective:
material = " ".join(words[:2])
obj = " ".join(words[2:])
gender = get_main_word_gender(obj)
new_adj = inflect_as_adjective(adj, gender)
new_word_2 = inflect_adjective(make_adjective[material], gender)
text = text.replace(search_result.group(0), "{} {} {}".format(new_adj, new_word_2, obj))
else:
material = " ".join(words[:3])
obj = " ".join(words[3:])
gender = get_main_word_gender(obj)
assert gender is not None
new_adj = inflect_as_adjective(adj, gender)
text = text.replace(search_result.group(0), "{} {} {}".format(new_adj, obj, material))
return text
def corr_container(text, _):
"""
>>> corr_container('(дварфийское пиво бочка (из ольхи))')
'(Бочка дварфийского пива (ольховая))'
>>> corr_container("(дварфийское вино бочка (из клёна) <#8>)")
'(Бочка дварфийского вина (кленовая) <#8>)'
>>> corr_container("(Семя бочка (из лумбанга) <#10>)")
'(Бочка семян (лумбанговая) <#10>)'
"""
search_result = re_container.search(text)
initial_string = search_result.group(0)
containment = search_result.group(1)
if containment in replace_containment:
containment = replace_containment[containment]
if containment.endswith("кровь"):
words = containment.split()
if words[0] in possessive_adjectives:
words[0] = possessive_adjectives[words[0]]
words = to_genitive_case_list(words)
else:
words = [to_genitive_case_single_noun(words[-1])] + list(to_genitive_case_list(words[:-1]))
containment = " ".join(words)
elif containment.startswith("из "):
containment = containment[3:] # Words after 'из' are already in genitive case
elif containment in {"слитков/блоков", "специй"}:
pass # Already in genitive case
elif containment.startswith("семена"):
words = containment.split()
words[0] = to_genitive_case(words[0])
containment = " ".join(words)
else:
containment = to_genitive_case(containment)
container = search_result.group(2)
of_material = search_result.group(3)
if not of_material:
replacement_string = container + " " + containment
elif (
" " not in of_material
and is_adjective(of_material)
or of_material in make_adjective
or of_material[3:] in make_adjective
):
if " " not in of_material and is_adjective(of_material):
adjective = of_material
elif of_material in make_adjective:
adjective = make_adjective[of_material]
elif of_material[3:] in make_adjective:
adjective = make_adjective[of_material[3:]]
else:
adjective = None
gender = get_gender(container, {"nomn"})
adjective = inflect_adjective(adjective, gender)
replacement_string = "{} {} ({})".format(container, containment, adjective)
else:
words = of_material.split()
material = None
if of_material.startswith("из ") or len(of_material) <= 2:
material = of_material
elif (
len(words) >= 2
and words[-2] == "из"
and (words[-1] in materials or any(mat.startswith(words[-1]) for mat in materials))
):
# Try to fix truncated materail names, eg. '(ямный краситель мешок (гигантский пещерный паук из шёл'
if words[-1] not in materials: # Fix partial material name eg. 'шерст', 'шёлк'
candidates = [mat for mat in materials if mat.startswith(words[-1])]
if len(candidates) == 1:
words[-1] = candidates[0]
else:
material = of_material # Partial name is not recognized (too short)
if not material:
material = " ".join(words[-2:] + list(to_genitive_case_list(words[:-2])))
else:
gen_case = list(to_genitive_case_list(of_material.split()))
if None not in gen_case:
material = "из " + " ".join(gen_case)
else:
material = of_material
replacement_string = "{} {} ({}".format(container, containment, material)
if initial_string[-1] == ")":
replacement_string += ")"
text = text.replace(initial_string, replacement_string.capitalize())
return text
def corr_forge(_, search_result):
"""
>>> corr_forge("Ковать из меди болты")
'Ковать медные болты'
>>> corr_forge("Кузница из железа Наконечники стрел баллисты")
'Ковать железные наконечники стрел баллисты'
>>> corr_forge("Делать из адамантина Колчан")
'Делать адамантиновый колчан'
"""
verb = search_result.group(1)
words = search_result.group(2).split()
assert len(words) >= 3
assert words[0] == "из"
# Second word ia adjective in gent case
second_is_adjf_in_gent = any_in_tag({"ADJF", "gent"}, custom_parse(words[1]))
# Third word is noun in gent case
third_is_noun_in_gent = any_in_tag({"NOUN", "gent"}, custom_parse(words[2]))
if second_is_adjf_in_gent and third_is_noun_in_gent:
of_material = words[:3]
obj = words[3:]
else:
# Second word is noun in gent case
assert any_in_tag({"NOUN", "gent"}, custom_parse(words[1]))
of_material = words[:2]
obj = words[2:]
of_material = " ".join(of_material)
noun_index = None
parse = None
gender = None
if len(obj) == 1:
noun_index = 0
parse = custom_parse(obj[noun_index])
noun = filter_noun(parse)
gender = get_gender(obj[noun_index], known_tags={"nomn"})
if not any_in_tag({"accs"}, noun):
obj[0] = noun[0].inflect({"accs"}).word
else:
for i, word in enumerate(obj):
parse = custom_parse(word)
noun = filter_noun(parse)
if noun:
noun_index = i
gender = get_gender(obj[noun_index])
obj[i] = noun[0].inflect({"accs"}).word
break # Words after the 'item' must be left in genitive case
elif not any_in_tag("accs", parse):
obj[i] = parse[0].inflect({"accs"}).word
assert parse is not None
if not any_in_tag("accs", parse):
obj[noun_index] = parse[0].inflect({"accs"}).word
if verb == "Кузница":
verb = "Ковать"
if of_material in make_adjective:
assert gender is not None
material = inflect_adjective(make_adjective[of_material], gender, "accs", animated=False)
text = verb + " " + material + " " + " ".join(obj)
else:
text = verb + " " + " ".join(obj) + " " + of_material
return text.capitalize()
replacement_string = " ".join(adjs) + " " + material
# elif (words[2] not in corr_item_general_except and len(words) > 3 and
elif (
len(words) > 3
and any_in_tag({"gent"}, custom_parse(words[1]))
and any_in_tag({"NOUN", "gent"}, custom_parse(words[2])) # The second word is in genitive
): # The third word is a noun in genitive
# Complex case, eg. "из висмутовой бронзы"
of_material = " ".join(words[:3])
words = words[3:]
if len(words) == 1:
first_part = words[0]
else:
obj = words[-1]
gender = get_gender(obj, "NOUN")
adjs = (inflect_adjective(adj, gender) or adj for adj in words[:-1])
first_part = "{} {}".format(" ".join(adjs), obj)
replacement_string = first_part + " " + of_material
elif any_in_tag({"NOUN", "gent"}, custom_parse(words[1])) and words[1] != "древесины":
# Simple case, eg. "из бронзы"
of_material = " ".join(words[:2])
words = words[2:]
item = words[-1]
for word in words:
if any_in_tag({"NOUN", "nomn"}, custom_parse(word)):
item = word
break
if of_material in make_adjective:
gender = get_gender(item, {"nomn"})