def test_should_find_lemma_concordance(self):
idx = DictionaryItemConcordanceIndex(self.word_list)
assert_that(idx.offsets(u'something'), equal_to([]))
assert_that(idx.offsets(u"o"), equal_to([0, 1, 2, 3, 4]))
assert_that(idx.offsets(u"o", SyntacticCategory.PRONOUN), equal_to([0, 1, 3, 4]))
assert_that(idx.offsets(u"o", SyntacticCategory.DETERMINER), equal_to([2]))
assert_that(idx.offsets(u"o", SyntacticCategory.PRONOUN, SecondarySyntacticCategory.PERSONAL), equal_to([0, 3]))
assert_that(idx.offsets(u"o", SyntacticCategory.PRONOUN, SecondarySyntacticCategory.DEMONSTRATIVE), equal_to([1, 4]))
assert_that(idx.offsets(u"git"), equal_to([6, 7, 8, 9, 10, 11]))
assert_that(idx.offsets(u"gid"), equal_to([]))
assert_that(idx.offsets(u"git", SyntacticCategory.VERB), equal_to([6, 7, 8, 9, 10, 11]))
def test_should_find_lemma_concordance(self):
idx = DictionaryItemConcordanceIndex(self.word_list)
assert_that(idx.offsets(u'something'), equal_to([]))
assert_that(idx.offsets(u"o"), equal_to([0, 1, 2, 3, 4]))
assert_that(idx.offsets(u"o", SyntacticCategory.PRONOUN),
equal_to([0, 1, 3, 4]))
assert_that(idx.offsets(u"o", SyntacticCategory.DETERMINER),
equal_to([2]))
assert_that(
idx.offsets(u"o", SyntacticCategory.PRONOUN,
SecondarySyntacticCategory.PERSONAL), equal_to([0, 3]))
assert_that(
idx.offsets(u"o", SyntacticCategory.PRONOUN,
SecondarySyntacticCategory.DEMONSTRATIVE),
equal_to([1, 4]))
assert_that(idx.offsets(u"git"), equal_to([6, 7, 8, 9, 10, 11]))
assert_that(idx.offsets(u"gid"), equal_to([]))
assert_that(idx.offsets(u"git", SyntacticCategory.VERB),
equal_to([6, 7, 8, 9, 10, 11]))
def _validate_lemma_concordance_indexes(self, word_list):
idx = DictionaryItemConcordanceIndex(word_list)
for lemma_root in idx._offsets._indices.iterkeys():
offsets = idx.offsets(lemma_root)
words = [word_list[offset] for offset in offsets]
assert_that(
all([word.root.lemma_root == lemma_root for word in words]))
for lemma_root in idx._offsets._indices.iterkeys():
for syntactic_category in idx._offsets._indices[
lemma_root].iterkeys():
offsets = idx.offsets(lemma_root, syntactic_category)
words = [word_list[offset] for offset in offsets]
assert_that(
all([
word.root.lemma_root == lemma_root
and word.root.syntactic_category == syntactic_category
for word in words
]))
for lemma_root in idx._offsets._indices.iterkeys():
for syntactic_category in idx._offsets._indices[
lemma_root].iterkeys():
for secondary_syntactic_category in idx._offsets._indices[
lemma_root][syntactic_category].iterkeys():
offsets = idx.offsets(lemma_root, syntactic_category,
secondary_syntactic_category)
words = [word_list[offset] for offset in offsets]
assert_that(
all([
word.root.lemma_root == lemma_root and
word.root.syntactic_category == syntactic_category
and word.root.secondary_syntactic_category
== secondary_syntactic_category for word in words
]))
def _validate_lemma_concordance_indexes(self, word_list):
idx = DictionaryItemConcordanceIndex(word_list)
for lemma_root in idx._offsets._indices.iterkeys():
offsets = idx.offsets(lemma_root)
words = [word_list[offset] for offset in offsets]
assert_that(all([word.root.lemma_root==lemma_root for word in words]))
for lemma_root in idx._offsets._indices.iterkeys():
for syntactic_category in idx._offsets._indices[lemma_root].iterkeys():
offsets = idx.offsets(lemma_root, syntactic_category)
words = [word_list[offset] for offset in offsets]
assert_that(all([word.root.lemma_root==lemma_root and word.root.syntactic_category==syntactic_category for word in words]))
for lemma_root in idx._offsets._indices.iterkeys():
for syntactic_category in idx._offsets._indices[lemma_root].iterkeys():
for secondary_syntactic_category in idx._offsets._indices[lemma_root][syntactic_category].iterkeys():
offsets = idx.offsets(lemma_root, syntactic_category, secondary_syntactic_category)
words = [word_list[offset] for offset in offsets]
assert_that(all([word.root.lemma_root==lemma_root and word.root.syntactic_category==syntactic_category
and word.root.secondary_syntactic_category==secondary_syntactic_category
for word in words]))