def get_hamlet_term_doc_matrix():
# type: () -> TermDocMatrix
hamlet_docs = get_hamlet_docs()
hamlet_term_doc_matrix = build_from_category_whitespace_delimited_text([
(get_hamlet_snippet_binary_category(text), text)
for i, text in enumerate(hamlet_docs)
])
return hamlet_term_doc_matrix
def get_hamlet_term_doc_matrix():
# type: () -> TermDocMatrix
hamlet_docs = get_hamlet_docs()
hamlet_term_doc_matrix = build_from_category_whitespace_delimited_text(
[(get_hamlet_snippet_binary_category(text), text)
for i, text in enumerate(hamlet_docs)]
)
return hamlet_term_doc_matrix
def test_single_category_term_doc_matrix_should_error(self):
with self.assertRaisesRegex(
expected_exception=
CannotCreateATermDocMatrixWithASignleCategoryException,
expected_regex=
'Documents must be labeled with more than one category. '
'All documents were labeled with category: "a"'):
single_category_tdm = build_from_category_whitespace_delimited_text(
[['a', text]
for category, text in get_test_categories_and_documents()])
def test_single_category_term_doc_matrix_should_error(self):
with self.assertRaisesRegex(
expected_exception=CannotCreateATermDocMatrixWithASignleCategoryException,
expected_regex='Documents must be labeled with more than one category. '
'All documents were labeled with category: "a"'):
single_category_tdm = build_from_category_whitespace_delimited_text(
[['a', text]
for category, text
in get_test_categories_and_documents()]
)
def make_a_test_term_doc_matrix():
# type: () -> TermDocMatrix
return build_from_category_whitespace_delimited_text(
get_test_categories_and_documents())
def make_a_test_term_doc_matrix():
# type: () -> TermDocMatrix
return build_from_category_whitespace_delimited_text(
get_test_categories_and_documents()
)
