def test_to_str(self):
text = AnnotatedText("helo world")
text.annotate(0,
4, ["hello", "hola"],
meta={"key": "value with spaces"})
expected = "{helo=>hello|hola:::key=value with spaces} world"
assert text.get_annotated_text() == expected
def test_parse_no_suggestions(self):
# When parsing text with the special "NO_SUGGESTIONS" value
text = AnnotatedText("{asdasd=>NO_SUGGESTIONS}")
# Then empty suggestions list should be returned
ann = text.get_annotations()[0]
assert ann.suggestions == []
def test_join_by(self):
t1 = AnnotatedText("I {likes=>like:::error_type=SVA} turtles.")
t2 = AnnotatedText("I{,=>:::error_type=Punct} like turtles.")
actual = AnnotatedText.join("\n", [t1, t2])
expected = AnnotatedText("I {likes=>like:::error_type=SVA} turtles.\n"
"I{,=>:::error_type=Punct} like turtles.")
assert expected == actual
def test_wrong_constructor_params():
# Constructing AnnotatedText from anything other than string is forbidden
with pytest.raises(ValueError):
AnnotatedText(42)
annotated = AnnotatedText("hello")
with pytest.raises(ValueError):
AnnotatedText(annotated)
def test_parse(self):
text = AnnotatedText(
"Hello {wold=>World|world:::type=OOV Spell:::status=ok}")
expected_original = "wold"
expected_sugg = ["World", "world"]
expected_meta = {"type": "OOV Spell", "status": "ok"}
ann = text.get_annotations()[0]
assert ann.meta == expected_meta
assert ann.suggestions == expected_sugg
assert ann.source_text == expected_original
def test_get_annotations():
text = AnnotatedText(
"this {was=>is|will be} an {anotated=>annotated} text")
anns = text.get_annotations()
assert len(anns) == 2
assert anns[0].source_text == "was"
assert anns[0].suggestions == ["is", "will be"]
assert anns[0].start == len("this ")
assert anns[0].end == len("this was")
assert anns[1].source_text == "anotated"
assert anns[1].suggestions == ["annotated"]
assert anns[1].start == len("this was an ")
assert anns[1].end == len("this was an anotated")
def iter_documents(self):
for meta in self._get_metadata():
filename = f"{meta.doc_id}.a{self.annotator_id}.ann"
path = self._data_dir / meta.partition / "annotated" / filename
text = AnnotatedText(path.read_text())
doc = Document(text, meta=meta)
yield doc
def test_get_annoation_at():
text = AnnotatedText("Hello {word=>world}")
print(text.get_annotations())
assert text.get_annotation_at(0) is None
assert text.get_annotation_at(0, 100) is None
assert text.get_annotation_at(6, 10).source_text == "word"
assert text.get_annotation_at(6).source_text == "word"
assert text.get_annotation_at(7).source_text == "word"
assert text.get_annotation_at(10) is None
assert text.get_annotation_at(11) is None
def test_parse_colon(self):
text = AnnotatedText("text {.=>::::key=R:PUNCT}")
expected = AnnotatedText("text .")
expected.annotate(5, 6, ":", meta={"key": "R:PUNCT"})
ann = text.get_annotations()[0]
expected_ann = expected.get_annotations()[0]
assert ann.meta == expected_ann.meta
assert ann.suggestions == expected_ann.suggestions
assert ann.source_text == expected_ann.source_text
def test_iter_annotations():
# Given annotated text
text = AnnotatedText("{helo=>Hello}{...=>,} {word=>World}!")
# When changing annotations during annotations iteration
for i, ann in enumerate(text.iter_annotations()):
if i == 1:
text.remove(ann) # {...=>,}
else:
text.apply_correction(ann)
# Then the text should be correctly updated
assert text.get_annotated_text() == "Hello... World!"
def test_remove_non_existing_annotation():
# Given annotated text
text = AnnotatedText("{helo=>Hello} {word=>World}!")
# When removing annotation that doesn't exists in the text
# Then error should be raised
annotation = text.get_annotations()[0]
text.remove(annotation)
with pytest.raises(ValueError):
text.remove(annotation)
def test_apply_correction_insert_word():
text = AnnotatedText("{=>Hello} {word=>World}!")
ann = text.get_annotations()[0] # {=>Hello}
text.apply_correction(ann)
assert text.get_annotated_text() == "Hello {word=>World}!"
def test_remove():
# Given annotated text
text = AnnotatedText("{helo=>Hello} {word=>World}!")
# When removing single annotation
annotation = text.get_annotations()[1]
assert annotation.source_text == "word"
text.remove(annotation)
# Then all other annotations should stay
expected = "{helo=>Hello} word!"
assert text.get_annotated_text() == expected
assert len(text.get_annotations()) == 1
def test_apply_correction_with_no_suggestions():
# Given annotated text
text = AnnotatedText("Hi {sdf=>NO_SUGGESTIONS}!")
# When applyinng a no-suggestions annotation
ann = text.get_annotations()[0]
text.apply_correction(ann)
# Then the annotation should be replaced with the original text
assert text.get_annotated_text() == "Hi sdf!"
def test_apply_correction():
# Given annotated text
text = AnnotatedText("{helo=>Hello}{...=>,} {word=>World}!")
# When applying the corrected text of an annotation
ann = text.get_annotations()[1] # {...=>,}
text.apply_correction(ann)
# Then annotation should be removed, and the correct text used instead
assert text.get_annotated_text() == "{helo=>Hello}, {word=>World}!"
def test_annotate_with_none(self):
# Some checks return None in place of suggestions, meaning
# that there are no good suggestions.
# The convention for those in the light-annotated format is
# {bad=>NO_SUGGESTIONS} annotation.
text = AnnotatedText("asdasd")
text.annotate(0, len("asdasd"), correct_value=None)
assert text.get_annotated_text() == "{asdasd=>NO_SUGGESTIONS}"
# When parsed, special value of "NO_SUGGESTIONS" should be
# represented as empty suggestions list
assert text.get_annotations()[0].suggestions == []
def test_error(self):
text = AnnotatedText("{helllo=>Hello} {world=>World}")
with pytest.raises(OverlapError):
text.annotate(2, 5, "ll")
def test_repr():
text = AnnotatedText("Hello {word=>world}!")
expected = "<AnnotatedText('Hello {word=>world}!')>"
assert repr(text) == expected
def test_not_eq_other_type():
text = "Text {=>tokens} tokens ."
text_1 = AnnotatedText(text)
assert text_1 != text
def test_eq():
text = "Text {=>tokens} tokens ."
text_1 = AnnotatedText(text)
text_2 = AnnotatedText(text)
assert text_1 == text_2
def test_string_representation():
# AnnotatedText should pretend
text = AnnotatedText("Hello {word=>world}!")
expected = "Hello {word=>world}!"
assert str(text) == expected
def test_eq_2():
t1 = AnnotatedText("hello word")
t2 = AnnotatedText("hello word")
t1.annotate(0, 5, "Hello")
t1.annotate(6, 10, "world")
t2.annotate(6, 10, "world")
t2.annotate(0, 5, "Hello")
assert t1 == t2
def test_get_corrected_text_with_highlight():
text = AnnotatedText("{helo=>NO_SUGGESTIONS} world!")
expected = "helo world!"
assert text.get_corrected_text() == expected
def test_get_original_text():
text = AnnotatedText("{helo=>Hello} world!")
expected = "helo world!"
assert text.get_original_text() == expected
def test_annotate_empty_lengthy_intersect(self):
text = AnnotatedText("Did n't know it.")
text.annotate(12, 12, [" about"])
with pytest.raises(OverlapError):
text.annotate(0, len("Did n't know it."), ["Did n't know it"])
def test_curly_braces_in_original_text():
text = AnnotatedText(r"(e.g. \emph{the, {dox=>fox}, jumps})")
expected = r"(e.g. \emph{the, dox, jumps})"
assert text.get_original_text() == expected
def test_get_corrected_text():
text = AnnotatedText("{helo=>Hello} world!")
expected = "Hello world!"
assert text.get_corrected_text() == expected
def test_not_eq():
text_1 = AnnotatedText("Text {=>tokens} tokens .")
text_2 = AnnotatedText("Text {=>TOKENS} tokens .")
assert text_1 != text_2
def test_forbid_joining_annotation_with_multiple_suggestions(self):
text = AnnotatedText("helloworld")
text.annotate(0, len("helloworld"), ["hello", "hola", "hi"])
with pytest.raises(OverlapError):
text.annotate(0, len("helloworld"), "world")
def test_get_annotated_text_same_span():
text = AnnotatedText("{=>b}{=>a}")
assert text.get_annotated_text() == "{=>b}{=>a}"