示例#1
0
 def test_error_on_bad_record(self):
     with self.assertRaises(ValueError):
         NexusMaker([
             Record(Word="leg", Item="", Cognacy="2")  # no language
         ])
     with self.assertRaises(ValueError):
         NexusMaker([
             Record(Language="French", Item="", Cognacy="2")  # no word
         ])
示例#2
0
    def test_is_loan(self):
        assert not Record(Loan="").is_loan
        assert not Record(Loan=None).is_loan

        assert Record(Loan="L").is_loan
        assert Record(Loan="English").is_loan
        assert Record(Loan=True).is_loan
        assert Record(Loan="B").is_loan
        assert Record(Loan="S").is_loan
        assert Record(Loan="X").is_loan
        assert Record(Loan="x").is_loan
示例#3
0
 def test_error_on_cognates_with_loans(self):
     """Test that we generate an error if a loan word enters .cognates()"""
     n = NexusMaker(TESTDATA)
     n.data.append(
         Record(Language="D", Word="eye", Item="", Cognacy="1", Loan=True))
     with self.assertRaises(ValueError):
         n.cognates
示例#4
0
 def test_simple(self):
     r = Record(ID=1,
                WID=2,
                LID=3,
                Language='English',
                Word='Hand',
                Item='hand',
                Annotation='?',
                Cognacy=None,
                Loan="L")
     assert r.ID == 1
     assert r.WID == 2
     assert r.LID == 3
     assert r.Language == 'English'
     assert r.Word == 'Hand'
     assert r.Item == 'hand'
     assert r.Annotation == '?'
     assert r.Loan == "L"
     assert r.Cognacy == None
示例#5
0
 def test_get_token(self):
     r = Record(ID=1,
                WID=2,
                LID=3,
                Language='English',
                Word='Hand',
                Item='hand',
                Annotation='?',
                Cognacy=None,
                Loan="L")
     assert r.get_taxon() == "English_3"
     r = Record(ID=1,
                WID=2,
                LID=None,
                Language='English',
                Word='Hand',
                Item='hand',
                Annotation='?',
                Cognacy=None,
                Loan="L")
     assert r.get_taxon() == "English"
示例#6
0
40167	006	Mopan	00026	Ulrich_and_Ulrich	cloud	muyal	6c	
23136	024	Kiche	00004	Kaufman_1976	cloud	suutZ'	2	
23137	024	Kiche	00009	Campbell_1971b	cloud	suutZ'	2	
23138	024	Kiche	00022	Mayers_ed_1966	cloud	sutZ'	2	
23139	024	Kiche	00061	Stoll_1884	cloud	su'tZ	2	
50386	024	Kiche	00086	Kaufman_2003	cloud	mayuul	6	
50389	024	Kiche	00086	Kaufman_2003	cloud	mayuy	6b	
53557	036	Popti	00086	Kaufman_2003	cloud	xhmoyxi	6d	
53558	036	Popti	00086	Kaufman_2003	cloud	asun	1	
"""

HEADER = [
    'ID', "LID", "Language", "SID", "Source", "Word", "Item", "Cognacy", "Loan"
]
COMPLEX_TESTDATA = [
    Record(**dict(zip(HEADER, r.split('\t')))) for r in RECORDS.split("\n")
    if len(r.strip())
]

EXPECTED_COGNATES = {
    ('cloud', '1'): {
        'Tojolabal_012', 'Chuj_013', 'Mocho_017', 'Jakalteko_014',
        'Qanjobal_016', 'Akateco_015', 'Tuzanteco_037', 'Popti_036'
    },
    ('cloud', '2'): {
        'Achi_025', 'Kaqchikel_022', 'Poqomam_029', 'Sakapulteko_026',
        'Ixil_021', 'Kiche_024', 'Poqomchi_030', 'Tzutujil_023',
        'Uspanteko_028'
    },
    ('cloud', '2b'): {'Qanjobal_016'},
    ('cloud', '3'): {'Awakateko_020'},
示例#7
0
import sys
import unittest

from nexusmaker import Record
from nexusmaker import NexusMaker, NexusMakerAscertained, NexusMakerAscertainedWords

TESTDATA = [
    Record(Language="A", Word="eye", Item="", Cognacy="1"),
    Record(Language="A", Word="leg", Item="", Cognacy="1"),
    Record(Language="A", Word="arm", Item="", Cognacy="1"),
    Record(Language="B", Word="eye", Item="", Cognacy="1"),
    Record(Language="B", Word="leg", Item="", Cognacy="2"),
    Record(Language="B", Word="arm", Item="", Cognacy="2"),
    Record(Language="C", Word="eye", Item="", Cognacy="1"),
    # No Record for C 'leg'
    Record(Language="C", Word="arm", Item="", Cognacy="3"),
    Record(Language="D", Word="eye", Item="", Cognacy="1", Loan=True),
    Record(Language="D", Word="leg", Item="", Cognacy="1"),
    Record(Language="D", Word="leg", Item="", Cognacy="2"),
    Record(Language="D", Word="arm", Item="", Cognacy="2,3"),
]


class TestNexusMakerInternals(unittest.TestCase):
    def test_error_on_non_record(self):
        with self.assertRaises(ValueError):
            NexusMaker(['1'])

    def test_error_on_bad_record(self):
        with self.assertRaises(ValueError):
            NexusMaker([
示例#8
0
import unittest
from nexusmaker import NexusMaker, Record

TESTDATA = [
    Record(Language="A", Word="eye", Item="", Cognacy="1"),
    Record(Language="B", Word="eye", Item="", Cognacy="1"),
    Record(Language="C", Word="eye", Item="", Cognacy="2"),
    Record(Language="D", Word="eye", Item="", Cognacy="2"),
    Record(Language="E", Word="eye", Item="", Cognacy=""),
]


class TestBugUniquesAndAmbig(unittest.TestCase):
    """
    Fix Bug caused by lru_cache on ._is_missing_for_word that made unique
    sites occur as single ones in blocks of missing states.
    """
    def setUp(self):
        self.maker = NexusMaker(data=TESTDATA)
        self.nex = self.maker.make()

    def test_cognate_sets(self):
        assert ('eye', '1') in self.maker.cognates
        assert ('eye', '2') in self.maker.cognates
        assert ('eye', 'u_1') in self.maker.cognates

        assert sorted(self.maker.cognates[('eye', '1')]) == ['A', 'B']
        assert sorted(self.maker.cognates[('eye', '2')]) == ['C', 'D']
        assert sorted(self.maker.cognates[('eye', 'u_1')]) == ['E']

    def test_is_missing_for_word(self):
示例#9
0
Iaai-471	125865	five	thabyŋ	
Lamogai-67	83796	five	elmé	1
Lamogai-67	83881	hand	mulǵu	45
Lamogai-67	83882	hand	melsé	45
Lamogai-67	83883	hand	milpí	45
Lamogai-67	83884	hand	melép	45
Lamogai-67	83885	hand	milpú	45
Lamogai-67	83886	hand	meylá	45
Lamogai-67	83887	hand	melsék	45
Lamogai-67	83942	leg	kaip	1
Lamogai-67	83943	leg	kaŋgú	1
"""

RECORDS = [r.split("\t") for r in RECORDS.split("\n") if len(r)]
COMPLEX_TESTDATA = [
    Record(Language=r[0], Word=r[2], Item=r[3], Cognacy=r[4]) for r in RECORDS
]

EXPECTED_COGNATES = {
    ('five', '1'): {
        'Aiwoo-501', 'Banoni-4', 'Dehu-196', 'Eton-1088', 'Hiw-639',
        'Lamogai-67'
    },
    ('leg', '86'): {'Aiwoo-501'},
    ('hand', '1'): {'Aiwoo-501', 'Banoni-4', 'Hiw-639'},
    ('hand', '64'): {'Banoni-4'},
    ('hand', '66'): {'Aiwoo-501'},
    ('leg', '95'): {'Eton-1088'},
    ('hand', '78'): {'Hiw-639'},
    ('leg', '17'): {'Hiw-639'},
    ('hand', '14'): {'Iaai-471'},