def test_extractnumber_pt(self):
self.assertEqual(extractnumber("isto e o primeiro teste", lang="pt"),
1)
self.assertEqual(extractnumber("isto e o 2 teste", lang="pt"), 2)
self.assertEqual(extractnumber("isto e o segundo teste", lang="pt"),
2)
self.assertEqual(extractnumber(u"isto e um ter�o de teste",
lang="pt"), 1.0 / 3.0)
self.assertEqual(extractnumber("isto e o teste numero quatro",
lang="pt"), 4)
self.assertEqual(extractnumber(u"um ter�o de chavena", lang="pt"),
1.0 / 3.0)
self.assertEqual(extractnumber("3 canecos", lang="pt"), 3)
self.assertEqual(extractnumber("1/3 canecos", lang="pt"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarto de hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("1/4 hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("um quarto hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("2/3 pinga", lang="pt"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 pinga", lang="pt"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 e 3/4 cafe", lang="pt"), 1.75)
self.assertEqual(extractnumber("1 cafe e meio", lang="pt"), 1.5)
self.assertEqual(extractnumber("um cafe e um meio", lang="pt"), 1.5)
self.assertEqual(
extractnumber("tres quartos de chocolate", lang="pt"),
3.0 / 4.0)
self.assertEqual(extractnumber(u"tr�s quarto de chocolate",
lang="pt"), 3.0 / 4.0)
self.assertEqual(extractnumber("sete ponto cinco", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete ponto 5", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete e meio", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete e oitenta", lang="pt"), 7.80)
self.assertEqual(extractnumber("sete e oito", lang="pt"), 7.8)
self.assertEqual(extractnumber("sete e zero oito",
lang="pt"), 7.08)
self.assertEqual(extractnumber("sete e zero zero oito",
lang="pt"), 7.008)
self.assertEqual(extractnumber("vinte treze avos", lang="pt"),
20.0 / 13.0)
self.assertEqual(extractnumber("seis virgula seiscentos e sessenta",
lang="pt"), 6.66)
self.assertEqual(extractnumber("seiscentos e sessenta e seis",
lang="pt"), 666)
self.assertEqual(extractnumber("seiscentos ponto zero seis",
lang="pt"), 600.06)
self.assertEqual(extractnumber("seiscentos ponto zero zero seis",
lang="pt"), 600.006)
self.assertEqual(extractnumber("seiscentos ponto zero zero zero seis",
lang="pt"), 600.0006)
def test_extractnumber(self):
self.assertEqual(extractnumber("this is the first test"), 1)
self.assertEqual(extractnumber("this is 2 test"), 2)
self.assertEqual(extractnumber("this is second test"), 2)
self.assertEqual(extractnumber("this is the third test"), 1.0 / 3.0)
self.assertEqual(extractnumber("this is test number 4"), 4)
self.assertEqual(extractnumber("one third of a cup"), 1.0 / 3.0)
self.assertEqual(extractnumber("three cups"), 3)
self.assertEqual(extractnumber("1/3 cups"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarter cup"), 0.25)
self.assertEqual(extractnumber("1/4 cup"), 0.25)
self.assertEqual(extractnumber("one fourth cup"), 0.25)
self.assertEqual(extractnumber("2/3 cups"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 and 3/4 cups"), 1.75)
self.assertEqual(extractnumber("1 cup and a half"), 1.5)
self.assertEqual(extractnumber("one cup and a half"), 1.5)
self.assertEqual(extractnumber("one and a half cups"), 1.5)
self.assertEqual(extractnumber("one and one half cups"), 1.5)
self.assertEqual(extractnumber("three quarter cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("three quarters cups"), 3.0 / 4.0)
def test_extractnumber_pt(self):
self.assertEqual(extractnumber("isto e o primeiro teste", lang="pt"),
1)
self.assertEqual(extractnumber("isto e o 2 teste", lang="pt"), 2)
self.assertEqual(extractnumber("isto e o segundo teste", lang="pt"), 2)
self.assertEqual(
extractnumber(u"isto e um terço de teste", lang="pt"), 1.0 / 3.0)
self.assertEqual(
extractnumber("isto e o teste numero quatro", lang="pt"), 4)
self.assertEqual(extractnumber(u"um terço de chavena", lang="pt"),
1.0 / 3.0)
self.assertEqual(extractnumber("3 canecos", lang="pt"), 3)
self.assertEqual(extractnumber("1/3 canecos", lang="pt"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarto de hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("1/4 hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("um quarto hora", lang="pt"), 0.25)
self.assertEqual(extractnumber("2/3 pinga", lang="pt"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 pinga", lang="pt"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 e 3/4 cafe", lang="pt"), 1.75)
self.assertEqual(extractnumber("1 cafe e meio", lang="pt"), 1.5)
self.assertEqual(extractnumber("um cafe e um meio", lang="pt"), 1.5)
self.assertEqual(extractnumber("tres quartos de chocolate", lang="pt"),
3.0 / 4.0)
self.assertEqual(
extractnumber(u"tr�s quarto de chocolate", lang="pt"),
3.0 / 4.0)
self.assertEqual(extractnumber("sete ponto cinco", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete ponto 5", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete e meio", lang="pt"), 7.5)
self.assertEqual(extractnumber("sete e oitenta", lang="pt"), 7.80)
self.assertEqual(extractnumber("sete e oito", lang="pt"), 7.8)
self.assertEqual(extractnumber("sete e zero oito", lang="pt"), 7.08)
self.assertEqual(extractnumber("sete e zero zero oito", lang="pt"),
7.008)
self.assertEqual(extractnumber("vinte treze avos", lang="pt"),
20.0 / 13.0)
self.assertEqual(
extractnumber("seis virgula seiscentos e sessenta", lang="pt"),
6.66)
self.assertEqual(
extractnumber("seiscentos e sessenta e seis", lang="pt"), 666)
self.assertEqual(
extractnumber("seiscentos ponto zero seis", lang="pt"), 600.06)
self.assertEqual(
extractnumber("seiscentos ponto zero zero seis", lang="pt"),
600.006)
self.assertEqual(
extractnumber("seiscentos ponto zero zero zero seis", lang="pt"),
600.0006)
def test_extractnumber(self):
self.assertEqual(extractnumber("this is the first test"), 1)
self.assertEqual(extractnumber("this is 2 test"), 2)
self.assertEqual(extractnumber("this is second test"), 2)
self.assertEqual(extractnumber("this is the third test"), 1.0 / 3.0)
self.assertEqual(extractnumber("this is test number 4"), 4)
self.assertEqual(extractnumber("one third of a cup"), 1.0 / 3.0)
self.assertEqual(extractnumber("three cups"), 3)
self.assertEqual(extractnumber("1/3 cups"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarter cup"), 0.25)
self.assertEqual(extractnumber("1/4 cup"), 0.25)
self.assertEqual(extractnumber("one fourth cup"), 0.25)
self.assertEqual(extractnumber("2/3 cups"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 and 3/4 cups"), 1.75)
self.assertEqual(extractnumber("1 cup and a half"), 1.5)
self.assertEqual(extractnumber("one cup and a half"), 1.5)
self.assertEqual(extractnumber("one and a half cups"), 1.5)
self.assertEqual(extractnumber("one and one half cups"), 1.5)
self.assertEqual(extractnumber("three quarter cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("three quarters cups"), 3.0 / 4.0)
def test_extractnumber(self):
self.assertEqual(extractnumber("this is the first test"), 1)
self.assertEqual(extractnumber("this is 2 test"), 2)
self.assertEqual(extractnumber("this is second test"), 2)
self.assertEqual(extractnumber("this is the third test"), 1.0 / 3.0)
self.assertEqual(extractnumber("this is test number 4"), 4)
self.assertEqual(extractnumber("one third of a cup"), 1.0 / 3.0)
self.assertEqual(extractnumber("three cups"), 3)
self.assertEqual(extractnumber("1/3 cups"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarter cup"), 0.25)
self.assertEqual(extractnumber("1/4 cup"), 0.25)
self.assertEqual(extractnumber("one fourth cup"), 0.25)
self.assertEqual(extractnumber("2/3 cups"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 and 3/4 cups"), 1.75)
self.assertEqual(extractnumber("1 cup and a half"), 1.5)
self.assertEqual(extractnumber("one cup and a half"), 1.5)
self.assertEqual(extractnumber("one and a half cups"), 1.5)
self.assertEqual(extractnumber("one and one half cups"), 1.5)
self.assertEqual(extractnumber("three quarter cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("three quarters cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("twenty two"), 22)
self.assertEqual(extractnumber("two hundred"), 200)
self.assertEqual(extractnumber("nine thousand"), 9000)
self.assertEqual(extractnumber("six hundred sixty six"), 666)
self.assertEqual(extractnumber("two million"), 2000000)
self.assertEqual(
extractnumber("two million five hundred thousand "
"tons of spinning metal"), 2500000)
self.assertEqual(extractnumber("six trillion"), 600000000000.0)
self.assertEqual(extractnumber("six trillion", short_scale=False),
6e+19)
self.assertEqual(extractnumber("one point five"), 1.5)
self.assertEqual(extractnumber("three dot fourteen"), 3.14)
self.assertEqual(extractnumber("zero point two"), 0.2)
self.assertEqual(extractnumber("billions of years older"),
10000000000.0)
self.assertEqual(
extractnumber("billions of years older", short_scale=False),
10000000000000.0)
self.assertEqual(extractnumber("one hundred thousand"), 100000)
self.assertEqual(extractnumber("minus 2"), -2)
self.assertEqual(extractnumber("negative seventy"), -70)
self.assertEqual(extractnumber("thousand million"), 1000000000)
def test_extractnumber_sv(self):
self.assertEqual(
extractnumber("1 och en halv deciliter", lang='sv-se'), 1.5)
self.assertEqual(
extractnumber("det här är det första testet", lang='sv-se'), 1)
self.assertEqual(
extractnumber("det här är test nummer 2", lang='sv-se'), 2)
self.assertEqual(
extractnumber("det här är det andra testet", lang='sv-se'), 2)
self.assertEqual(
extractnumber("det här är tredje testet", lang='sv-se'), 3)
self.assertEqual(
extractnumber("det här är test nummer 4", lang='sv-se'), 4)
self.assertEqual(extractnumber("en tredjedels dl", lang='sv-se'),
1.0 / 3.0)
self.assertEqual(extractnumber("tre deciliter", lang='sv-se'), 3)
self.assertEqual(extractnumber("1/3 deciliter", lang='sv-se'),
1.0 / 3.0)
self.assertEqual(extractnumber("en kvarts dl", lang='sv-se'), 0.25)
self.assertEqual(extractnumber("1/4 dl", lang='sv-se'), 0.25)
self.assertEqual(extractnumber("en kvarts dl", lang='sv-se'), 0.25)
self.assertEqual(extractnumber("2/3 dl", lang='sv-se'), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 dl", lang='sv-se'), 3.0 / 4.0)
self.assertEqual(extractnumber("1 och 3/4 dl", lang='sv-se'), 1.75)
self.assertEqual(extractnumber("tre fjärdedels dl", lang='sv-se'),
3.0 / 4.0)
self.assertEqual(extractnumber("trekvarts kopp", lang='sv-se'),
3.0 / 4.0)
def test_extractnumber_fr(self):
self.assertEqual(extractnumber("voici le premier test", lang="fr-fr"),
1)
self.assertEqual(extractnumber("c'est 2 tests", lang="fr-fr"), 2)
self.assertEqual(extractnumber("voici le second test", lang="fr-fr"),
2)
self.assertEqual(extractnumber("voici trois tests", lang="fr-fr"), 3)
self.assertEqual(extractnumber("voici le test numéro 4", lang="fr-fr"),
4)
self.assertEqual(extractnumber("un tiers de litre", lang="fr-fr"),
1.0 / 3.0)
self.assertEqual(extractnumber("3 cuillères", lang="fr-fr"), 3)
self.assertEqual(extractnumber("1/3 de litre", lang="fr-fr"),
1.0 / 3.0)
self.assertEqual(extractnumber("un quart de bol", lang="fr-fr"), 0.25)
self.assertEqual(extractnumber("1/4 de verre", lang="fr-fr"), 0.25)
self.assertEqual(extractnumber("2/3 de bol", lang="fr-fr"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 de bol", lang="fr-fr"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 et 3/4 de bol", lang="fr-fr"), 1.75)
self.assertEqual(extractnumber("1 bol et demi", lang="fr-fr"), 1.5)
self.assertEqual(extractnumber("un bol et demi", lang="fr-fr"), 1.5)
self.assertEqual(extractnumber("un et demi bols", lang="fr-fr"), 1.5)
self.assertEqual(extractnumber("un bol et un demi", lang="fr-fr"), 1.5)
self.assertEqual(extractnumber("trois quarts de bol", lang="fr-fr"),
3.0 / 4.0)
self.assertEqual(extractnumber("32.2 degrés", lang="fr-fr"), 32.2)
self.assertEqual(extractnumber("2 virgule 2 cm", lang="fr-fr"), 2.2)
self.assertEqual(extractnumber("2 virgule 0 2 cm", lang="fr-fr"), 2.02)
self.assertEqual(extractnumber("ça fait virgule 2 cm", lang="fr-fr"),
0.2)
self.assertEqual(extractnumber("point du tout", lang="fr-fr"),
"point tout")
self.assertEqual(extractnumber("32.00 secondes", lang="fr-fr"), 32)
self.assertEqual(
extractnumber("mange trente-et-une bougies", lang="fr-fr"), 31)
self.assertEqual(extractnumber("un trentième", lang="fr-fr"),
1.0 / 30.0)
self.assertEqual(extractnumber("un centième", lang="fr-fr"), 0.01)
self.assertEqual(extractnumber("un millième", lang="fr-fr"), 0.001)
self.assertEqual(extractnumber("un 20e", lang="fr-fr"), 1.0 / 20.0)
def test_extractnumber(self):
self.assertEqual(
extractnumber("dies ist der erste Test", lang="de-de"), 1)
self.assertEqual(extractnumber("dies ist 2 Test", lang="de-de"), 2)
self.assertEqual(extractnumber("dies ist zweiter Test", lang="de-de"),
2)
self.assertEqual(
extractnumber("dies ist der dritte Test", lang="de-de"), 3)
self.assertEqual(
extractnumber("dies ist der Test Nummer 4", lang="de-de"), 4)
self.assertEqual(
extractnumber("ein drittel einer Tasse", lang="de-de"), 1.0 / 3.0)
self.assertEqual(extractnumber("drei Tassen", lang="de-de"), 3)
self.assertEqual(extractnumber("1/3 Tasse", lang="de-de"), 1.0 / 3.0)
self.assertEqual(extractnumber("eine viertel Tasse", lang="de-de"),
0.25)
self.assertEqual(extractnumber("1/4 Tasse", lang="de-de"), 0.25)
self.assertEqual(extractnumber("viertel Tasse", lang="de-de"), 0.25)
self.assertEqual(extractnumber("2/3 Tasse", lang="de-de"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 Tasse", lang="de-de"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 und 3/4 Tassen", lang="de-de"), 1.75)
self.assertEqual(extractnumber("1 Tasse und eine halbe", lang="de-de"),
1.5)
self.assertEqual(
extractnumber("eine Tasse und eine halbe", lang="de-de"), 1.5)
self.assertEqual(
extractnumber("eine und eine halbe Tasse", lang="de-de"), 1.5)
self.assertEqual(
extractnumber("ein und ein halb Tassen", lang="de-de"), 1.5)
self.assertEqual(extractnumber("drei Viertel Tasse", lang="de-de"),
3.0 / 4.0)
self.assertEqual(extractnumber("drei Viertel Tassen", lang="de-de"),
3.0 / 4.0)
def test_extractnumber(self):
self.assertEqual(
extractnumber("this is the first test", ordinals=True), 1)
self.assertEqual(extractnumber("this is 2 test"), 2)
self.assertEqual(extractnumber("this is second test", ordinals=True),
2)
self.assertEqual(extractnumber("this is the third test"), 1.0 / 3.0)
self.assertEqual(
extractnumber("this is the third test", ordinals=True), 3.0)
self.assertEqual(extractnumber("this is test number 4"), 4)
self.assertEqual(extractnumber("one third of a cup"), 1.0 / 3.0)
self.assertEqual(extractnumber("three cups"), 3)
self.assertEqual(extractnumber("1/3 cups"), 1.0 / 3.0)
self.assertEqual(extractnumber("quarter cup"), 0.25)
self.assertEqual(extractnumber("1/4 cup"), 0.25)
self.assertEqual(extractnumber("one fourth cup"), 0.25)
self.assertEqual(extractnumber("2/3 cups"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 and 3/4 cups"), 1.75)
self.assertEqual(extractnumber("1 cup and a half"), 1.5)
self.assertEqual(extractnumber("one cup and a half"), 1.5)
self.assertEqual(extractnumber("one and a half cups"), 1.5)
self.assertEqual(extractnumber("one and one half cups"), 1.5)
self.assertEqual(extractnumber("three quarter cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("three quarters cups"), 3.0 / 4.0)
self.assertEqual(extractnumber("twenty two"), 22)
self.assertEqual(extractnumber("two hundred"), 200)
self.assertEqual(extractnumber("nine thousand"), 9000)
self.assertEqual(extractnumber("six hundred sixty six"), 666)
self.assertEqual(extractnumber("two million"), 2000000)
self.assertEqual(
extractnumber("two million five hundred thousand "
"tons of spinning metal"), 2500000)
self.assertEqual(extractnumber("six trillion"), 60000000000.0)
self.assertEqual(extractnumber("six trillion", short_scale=False),
6e+18)
self.assertEqual(extractnumber("one point five"), 1.5)
self.assertEqual(extractnumber("three dot fourteen"), 3.14)
self.assertEqual(extractnumber("zero point two"), 0.2)
self.assertEqual(extractnumber("billions of years older"),
1000000000.0)
self.assertEqual(
extractnumber("billions of years older", short_scale=False),
1000000000000.0)
self.assertEqual(extractnumber("one hundred thousand"), 100000)
self.assertEqual(extractnumber("minus 2"), -2)
self.assertEqual(extractnumber("negative seventy"), -70)
self.assertEqual(extractnumber("thousand million"), 1000000000)
self.assertEqual(extractnumber("sixth third"), 1 / 6 / 3)
self.assertEqual(extractnumber("sixth third", ordinals=True), 3)
self.assertEqual(extractnumber("thirty second"), 30)
self.assertEqual(extractnumber("thirty second", ordinals=True), 32)
self.assertEqual(
extractnumber("this is the billionth test", ordinals=True), 1e09)
self.assertEqual(extractnumber("this is the billionth test"), 1e-9)
self.assertEqual(
extractnumber("this is the billionth test",
ordinals=True,
short_scale=False), 1e12)
self.assertEqual(
extractnumber("this is the billionth test", short_scale=False),
1e-12)
# TODO handle this case
# self.assertEqual(
# extractnumber("6 dot six six six"),
# 6.666)
self.assertTrue(extractnumber("The tennis player is fast") is False)
self.assertTrue(extractnumber("fraggle") is False)
self.assertTrue(extractnumber("fraggle zero") is not False)
self.assertEqual(extractnumber("fraggle zero"), 0)
self.assertTrue(extractnumber("grobo 0") is not False)
self.assertEqual(extractnumber("grobo 0"), 0)
def test_extractnumber_it(self):
self.assertEqual(extractnumber(u"questo è il primo test",
lang="it"), 1)
self.assertEqual(extractnumber(u"questo è il 2 test",
lang="it"), 2)
self.assertEqual(extractnumber(u"questo è il secondo test",
lang="it"), 2)
self.assertEqual(extractnumber(u"questo è un terzo di test",
lang="it"), 1.0 / 3.0)
self.assertEqual(extractnumber(u"questo è test numero 4",
lang="it"), 4)
self.assertEqual(extractnumber("un terzo di tazza",
lang="it"), 1.0 / 3.0)
self.assertEqual(extractnumber("tre tazze",
lang="it"), 3)
self.assertEqual(extractnumber("1/3 tazze",
lang="it"), 1.0 / 3.0)
self.assertEqual(extractnumber("un quarto di tazza",
lang="it"), 0.25)
self.assertEqual(extractnumber("1/4 tazza",
lang="it"), 0.25)
self.assertEqual(extractnumber("2/3 tazza",
lang="it"), 2.0 / 3.0)
self.assertEqual(extractnumber("3/4 tazza",
lang="it"), 3.0 / 4.0)
self.assertEqual(extractnumber("1 e 1/4 tazza",
lang="it"), 1.25)
self.assertEqual(extractnumber("1 tazza e mezzo",
lang="it"), 1.5)
self.assertEqual(extractnumber("una tazza e mezzo",
lang="it"), 1.5)
self.assertEqual(extractnumber("una e mezza tazza",
lang="it"), 1.5)
self.assertEqual(extractnumber("una e una mezza tazza",
lang="it"), 1.5)
self.assertEqual(extractnumber("tre quarti tazza",
lang="it"), 3.0 / 4.0)
self.assertEqual(extractnumber("tre quarto tazza",
lang="it"), 3.0 / 4.0)
self.assertEqual(extractnumber("sette punto cinque",
lang="it"), 7.5)
self.assertEqual(extractnumber("sette punto 5",
lang="it"), 7.5)
self.assertEqual(extractnumber("sette e mezzo",
lang="it"), 7.5)
self.assertEqual(extractnumber("sette e ottanta",
lang="it"), 7.80)
self.assertEqual(extractnumber("sette e otto",
lang="it"), 7.8)
self.assertEqual(extractnumber("sette e zero otto",
lang="it"), 7.08)
self.assertEqual(extractnumber("sette e zero zero otto",
lang="it"), 7.008)
self.assertEqual(extractnumber("venti tredicesimi",
lang="it"), 20.0 / 13.0)
self.assertEqual(extractnumber("sei virgola sessanta sei",
lang="it"), 6.66)
self.assertEqual(extractnumber("sei virgola sessantasei",
lang="it"), 6.66)
self.assertEqual(extractnumber("seicento sessanta sei",
lang="it"), 666)
self.assertEqual(extractnumber("seicento punto zero sei",
lang="it"), 600.06)
self.assertEqual(extractnumber("seicento punto zero zero sei",
lang="it"), 600.006)
self.assertEqual(extractnumber("seicento punto zero zero zero sei",
lang="it"), 600.0006)
self.assertEqual(extractnumber("tre decimi ",
lang="it"), 0.30000000000000004)
self.assertEqual(extractnumber("dodici centesimi",
lang="it"), 0.12)
self.assertEqual(extractnumber("cinque e quaranta due millesimi",
lang="it"), 5.042)
self.assertEqual(extractnumber("mille e uno",
lang="it"), 1001)
self.assertEqual(extractnumber("due mila venti due dollari ",
lang="it"), 2022)
self.assertEqual(extractnumber(
"cento quattordici mila quattrocento undici dollari ",
lang="it"), 114411)
self.assertEqual(extractnumber("ventitre dollari ", lang="it"), 23)
self.assertEqual(extractnumber("quarantacinque minuti ",
lang="it"), 45)
self.assertEqual(extractnumber("ventuno anni ",
lang="it"), 21)
self.assertEqual(extractnumber("ventotto euro ",
lang="it"), 28)
self.assertEqual(extractnumber("dodici e quarantacinque ",
lang="it"), 12.45)
self.assertEqual(extractnumber("quarantotto euro ",
lang="it"), 48)
self.assertEqual(extractnumber("novantanove euro ",
lang="it"), 99)
self.assertEqual(extractnumber("avvisa se qualcuno arriva ",
lang="it"), False)
def extract_expression_en(string):
expressions = {
"+": ["add", "adding", "plus", "added"],
"-": ["subtract", "subtracting", "minus", "negative", "subtracted"],
"/": ["divide", "dividing", "divided"],
"*": ["multiply", "multiplying", "times", "multiplied"],
"%": ["modulus"],
"!": ["factorial"],
"is": ["set"], # TODO find better keyword for x = value
# "=": ["equals"],
"^": ["**", "^", "pow"
"elevated", "power", "powered", "raised"],
"sqr": ["squared"],
"sqrt": ["square_root"],
"qb": ["cubed", "qubed"],
"exp": ["exponent", "exponentiate", "exponentiated"],
"(": ["open"],
")": ["close"]
}
# clean string
noise_words = [
"by", "and", "the", "in", "at", "a", "for", "an", "to", "with", "off",
"of"
]
# replace natural language expression
for op in expressions:
string = string.replace(op, " " + op + " ")
words = string.replace(",", "").replace("square root", "sqrt").split(" ")
for idx, word in enumerate(words):
if word in noise_words:
words[idx] = ""
else:
for operation in expressions:
if word in expressions[operation]:
words[idx] = operation
words = [word for word in words if word]
exps = []
# convert all numbers
for idx, word in enumerate(words):
if not word:
continue
if extractnumber(word):
words[idx] = str(extractnumber(word))
# join unknown vars nums
if idx + 1 < len(words) and words[idx + 1] not in operations:
# 3 x = 3x
if is_numeric(word) and not is_numeric(words[idx + 1]):
words[idx] = word + words[idx + 1]
words[idx + 1] = ""
if idx - 1 >= 0 and word not in operations:
# 1 2 x = 1 2x
if not is_numeric(word) and is_numeric(words[idx - 1]):
words[idx] = words[idx - 1] + word
words[idx - 1] = ""
words = [word for word in words if word]
# extract operations
for idx, word in enumerate(words):
if not word:
continue
# is an operation
if word in expressions:
operation = word
if idx > 0:
woi = words[idx - 1:idx + 2]
words[idx - 1] = ""
if idx + 1 < len(words):
words[idx + 1] = ""
words[idx] = ""
x = woi[0]
try:
y = woi[2]
except:
y = "next"
if x == "":
x = "prev"
if operation == "sqrt":
x = y
exps.append([x, operation, y])
else:
# operation at first, is a sign
y = words[idx + 1]
words[idx + 1] = ""
words[idx] = ""
if operation == "-":
x = "-" + y
y = 0
operation = "+"
exps.append([x, operation, y])
elif operation == "+":
exps.append(["0", operation, y])
# or square root
elif operation == "sqrt":
x = y
y = "next"
exps.append([x, operation, y])
# TODO exponent, log
if not exps and extractnumber(string):
exps = [["0", "+", str(extractnumber(string))]]
return exps