def customize_matrix(path, arch_type, destination=None):
if os.path.isdir(path):
path = os.path.join(path, 'choices')
# if no destination dir is specified, just use the choices file's dir
destination = destination or os.path.dirname(path)
global ch
ch = ChoicesFile(path)
language = ch['language']
grammar_path = get_grammar_path(
ch.get('iso-code', language).lower(), language.lower(), destination)
# delete any existing contents at grammar path
if os.path.exists(grammar_path):
shutil.rmtree(grammar_path)
# the rsync command won't create the target dirs, so do it now
os.makedirs(grammar_path)
# Use the following command when python2.6 is available
#shutil.copytree('matrix-core', grammar_path,
# ignore=shutil.ignore_patterns('.svn'))
IGNORE = open(os.devnull, 'w')
try:
call([
'rsync', '-a', '--exclude=.svn',
get_matrix_core_path() + os.path.sep, grammar_path
],
stdout=IGNORE,
stderr=IGNORE)
except OSError, er:
print "OS Error. Most likely rsync is not installed."
print er.message
sys.exit(1)
def test_word_order_determiners_no_determiner_answer(self):
c = ChoicesFile()
# order but no determiners
c = ChoicesFile()
c['noun-det-order'] = 'noun-det'
self.assertError(c, 'has-dets')
def test_features(self):
choices = ChoicesFile("web/sample-choices/mini-english")
actual = choices.features()
expected = [['number', 'sg|sg;pl|pl', 'LOCAL.CONT.HOOK.INDEX.PNG.NUM', 'noun', 'y'],
['person', '1st|1st;2nd|2nd;3rd|3rd', 'LOCAL.CONT.HOOK.INDEX.PNG.PER', 'noun', 'y'],
['argument structure', 'intrans|intransitive;trans|transitive', '', 'verb', 'n'],
['negation', 'plus|plus;minus|minus', '', 'verb', 'y'],
['information-structure meaning', 'focus|focus;topic|topic;contrast|contrast;semantic-focus|non-contrastive-focus;contrast-focus|contrastive-focus;aboutness-topic|non-contrastive-topic;contrast-topic|contrastive-topic;focus-or-topic|focus-or-topic;contrast-or-focus|contrast-or-focus;contrast-or-topic|contrast-or-topic;non-topic|non-topic;non-focus|non-focus;bg|background', 'LOCAL.CONT.HOOK.ICONS-KEY', 'both', 'n']]
self.assertEqual(actual, expected)
def test_set(self):
c = ChoicesFile() # no file loaded
self.assertEqual(c.choices, {})
c['abc'] = 5
self.assertEqual(c.choices, {'abc':5})
c['def2_ghi'] = 2
self.assertEqual(c.choices, {'abc':5, 'def':[None, {'ghi':2}]})
c['def1_ghi'] = 1
self.assertEqual(c.choices, {'abc':5, 'def':[{'ghi':1}, {'ghi':2}]})
c = ChoicesFile() # no file loaded
c['rst_uvw'] = 2
self.assertEqual(c.choices, {'rst_uvw': 2})
def test_other_features(self):
# a feature without a name, a type, or a name for its value
c = ChoicesFile()
c['feature1_value1_supertype1_name'] = 'dummy'
self.assertErrors(
c, ['feature1_name', 'feature1_type', 'feature1_value1_name'])
# a feature without a name, a type, or a supertype for its value
c = ChoicesFile()
c['feature1_value1_name'] = 'dummy'
self.assertErrors(c, [
'feature1_name', 'feature1_type', 'feature1_value1_supertype1_name'
])
def test_word_order_osv_vp_before_illegal(self):
c = ChoicesFile()
c['word-order'] = 'osv'
c['aux-comp'] = 'vp'
c['aux-comp-order'] = 'before'
self.assertError(c, 'aux-comp')
def test_word_order_determiners_no_determiners_with_determiner_lex(self):
c = ChoicesFile()
# determiners=no but determiner in the lexicon
c['has-dets'] = 'no'
c['det1_name'] = 'dummy'
self.assertError(c, 'has-dets')
def test_del(self):
c = ChoicesFile() # no file loaded
c['abc'] = 5
self.assertEqual(c.choices, {'abc': 5})
del c['abc']
self.assertEqual(c.choices, {})
c['abc1_def'] = 1
c['abc2_def'] = 2
c['abc2_ghi'] = 3
self.assertEqual(c.choices,
{'abc': [{
'def': 1
}, {
'def': 2,
'ghi': 3
}]})
del c['abc2_def']
self.assertEqual(c.choices, {'abc': [{'def': 1}, {'ghi': 3}]})
del c['abc2']
self.assertEqual(c.choices, {'abc': [{'def': 1}, None]})
c['abc2_def'] = 2
c['abc3_def'] = 3
self.assertEqual(c.choices,
{'abc': [{
'def': 1
}, {
'def': 2
}, {
'def': 3
}]})
del c['abc2']
self.assertEqual(c.choices, {'abc': [{'def': 1}, None, {'def': 3}]})
del c['abc']
self.assertEqual(c.choices, {})
def test_lexicon_requires_one_noun_two_verbs(self):
c = ChoicesFile()
# not enough nouns and verbs
self.assertWarnings(c, ('noun1_stem1_orth',
'verb1_valence',
'verb2_valence'))
def test_coordination_missing_affix_answers(self):
# marked by affixes on NP, VP, or S (not supported yet)
for phrase in ('np', 'vp', 's'):
c = ChoicesFile()
c['cs1_mark'] = 'affix'
c['cs1_' + phrase] = 'on'
self.assertError(c, 'cs1_mark')
def test_person_no_first_person_answer(self):
c = ChoicesFile()
# no answer for first-person in a language with a first person
for person in ('1-2-3', '1-2-3-4', '1-non-1'):
c['person'] = person
self.assertError(c, 'first-person')
def test_case_additional_cases_require_other_cases(self):
c = ChoicesFile()
# if there's no case, additional cases can't be specified
c['case-marking'] = 'none'
c['case1_name'] = 'test'
self.assertError(c, 'case1_name')
def test_lexicon_obligatory_determiners_require_determiners(self):
c = ChoicesFile()
# determiners obligatory, but not determiners
c['noun1_det'] = 'obl'
c['has-dets'] = 'no'
self.assertErrors(c, ('has-dets', 'noun1_det'))
def test_general_language_required(self):
c = ChoicesFile()
# no language
vr = validate(c)
self.assertError(c, 'language')
self.assertWarning(c, 'archive')
def test_word_order_vso_vp_after_illegal(self):
c = ChoicesFile()
c['word-order'] = 'vso'
c['aux-comp'] = 'vp'
c['aux-comp-order'] = 'after'
self.assertError(c, 'aux-comp')
def test_sentential_negation_adverb_requires_other_answers(self):
c = ChoicesFile()
# auxiliary selects neg complement,
# but no auxiliaries ('has-aux' != 'yes')
c['comp-neg'] = 'on'
c['comp-neg-head'] = 'aux'
self.assertError(c, 'comp-neg-head')
def test_sentential_negation_adverb_requires_other_answers(self):
c = ChoicesFile()
# negation via adverb and simple negation
# but no answers for other questions
c['adv-neg'] = 'on'
c['neg-exp'] = '1'
self.assertErrors(c, ('neg-mod', 'neg-order', 'neg-adv-orth'))
def test_person_first_person_answer_without_person(self):
c = ChoicesFile()
# an answer for first-person in a language with no first person
c['first-person'] = 'incl-excl'
for person in ('none', '2-non-2', '3-non-3'):
c['person'] = person
self.assertError(c, 'first-person')
def test_person_first_person_answer(self):
c = ChoicesFile()
# answer for first-person in a language with a first person
c['first-person'] = 'incl-excl'
for person in ('1-2-3', '1-2-3-4', '1-non-1'):
c['person'] = person
self.assertNoError(c, 'first-person')
def test_case_answer(self):
c = ChoicesFile()
# no answer for case
c['case-marking'] = 'nom-acc'
c['case-marking-nom-case-name'] = 'nom'
c['case-marking-acc-case-name'] = 'acc'
self.assertNoError(c, 'case-marking')
def test_coordination_no_asyndeton_answers(self):
c = ChoicesFile()
# asyndeton but with mark, order, and orth specified
c['cs1_pat'] = 'a'
c['cs1_mark'] = 'dummy'
c['cs1_order'] = 'dummy'
c['cs1_orth'] = 'dummy'
self.assertErrors(c, ('cs1_mark', 'cs1_order', 'cs1_orth'))
def test_argopt(self):
c = ChoicesFile()
c['subj-drop'] = 'on'
c['obj-drop'] = 'on'
c['context1_feat1_dummy'] = 'dummy'
self.assertErrors(c, ('subj-mark-drop', 'subj-mark-no-drop',
'obj-mark-drop', 'obj-mark-no-drop',
'context1_feat1_head'))
def test_full_key(self):
c = ChoicesFile()
self.assertEqual(c.choices.full_key, None)
c['abc1_def'] = 1
c['abc2_def'] = 2
self.assertEqual(c['abc'].full_key, 'abc')
for i, key in enumerate(c['abc']):
self.assertEqual(key.full_key, 'abc' + str(i+1))
def test_lexicon_verb_requires_valence_orth_pred(self):
c = ChoicesFile()
c['verb1_dummy'] = 'dummy'
c['verb1_stem1_dummy'] = 'dummy'
self.assertErrors(c, ('verb1_valence',
'verb1_stem1_orth',
'verb1_stem1_pred'))
def test_tense_legal_supertype(self):
# subtypes defined for incorrect supertype
for st in self.tenses:
for t in (x for x in self.tenses if x != st):
c = ChoicesFile()
c[st] = 'on'
c[t + '-subtype1_name'] = 'dummy'
self.assertError(c, t + '-subtype1_name')
def test_names(self):
# Pairs of [variable prefix, value suffix]. Appending 'name' to the
# prefix gives the variable name (e.g., 'case1_name'). The suffix
# tells what the system will append to the end of the *value* of
# that variable.
variables = [['nom-acc-nom-case-', ''],
['case1_', ''], ['number1_', ''], ['gender1_', ''],
['feature1_', ''], ['feature1_value1_', ''],
['tense1_', ''], ['aspect1_', ''], ['situation1_', ''],
['form-subtype1_', ''], ['noun1_', '-noun-lex'],
['noun2_', '-noun-lex'], ['det1_', '-determiner-lex'],
['det2_', '-determiner-lex'], ['verb1_', '-verb-lex'],
['verb2_', '-verb-lex'], ['noun-pc1_', '-lex-rule'],
['noun-pc1_lrt1_',
'-lex-rule'], ['noun-pc2_', '-lex-rule'],
['noun-pc2_lrt1_', '-lex-rule']]
# Name for causing collisions, made up of the lower-case letter 'a' in:
# Latin, accented Latin, accented Latin, Greek, Cyrillic, Armenian
value = 'aáāαаա'
for v1 in variables:
# first try colliding with an existing name (head-comp-phrase)
if not v1[1]: # won't collide if there's a suffix
name = v1[0] + 'name'
c = ChoicesFile()
c[name] = 'head-comp-phrase'
self.assertError(c, name)
# next try colliding pairs of names
for v2 in [x for x in variables if x[0] != v1[0]]:
if v1[1] == v2[1]: # won't collide with different suffixes
name1 = v1[0] + 'name'
name2 = v2[0] + 'name'
c = ChoicesFile()
# same case
c[name1] = value
c[name2] = value
# different case (should still collide)
c[name1] = value.lower()
c[name2] = value.upper()
self.assertErrors(c, [name1, name2])
def test_general_illegal_language_characters_anywhere(self):
c = ChoicesFile()
for bad_ch in chr(20) + '?*:<>|/\\"^':
c['language'] = bad_ch + 'x'
self.assertError(c, 'language')
c['language'] = 'x' + bad_ch + 'x'
self.assertError(c, 'language')
def test_lexicon_noun_requires_determiners_orth_pred(self):
c = ChoicesFile()
# noun with no answer about determiners, and a stem with no orth or pred
c['noun1_dummy'] = 'dummy'
c['noun1_stem1_dummy'] = 'dummy'
self.assertErrors(c, ('noun1_det',
'noun1_stem1_orth',
'noun1_stem1_pred'))
def test_other_features_requires_name_type_value(self):
# a feature with a name, a type, and a name for its value
c = ChoicesFile()
c['feature1_name'] = 'dummy_name'
c['feature1_type'] = 'dummy_type'
c['feature1_value1_name'] = 'dummy_value_name'
c['feature1_value1_supertype1_name'] = 'dummy_name'
self.assertNoErrors(c, ('feature1_name', 'feature1_type',
'feature1_value1_name', 'feature1_value1_supertype1_name'))
def test_lexicon_features_illegal_name_value(self):
# try a bad feature and value everywhere
for p in ('scale',
'noun', 'verb', 'adj', 'det', 'aux', #'cop', # TODO: why doesn't copula trigger this?
#'noun-pc1_lrt', 'verb-pc1_lrt', 'det-pc1_lrt', # TODO: Move this to morphotactics?
'context'):
c = ChoicesFile()
c[p + '1_feat1_name'] = 'dummy'
c[p + '1_feat1_value'] = 'dummy'
self.assertErrors(c, (p + '1_feat1_name', p + '1_feat1_value'))
