def test_format(self):
# we show how we can format a snippet (below) when the
# annotations are relative to a larger enclosing text
original_text = (
"Eliot: Time present and time past / "
"Are both perhaps present in time future..."
)
snippet_text = (
"Time present and time past / " "Are both perhaps present in time future..."
)
annotations = [
AnnotatedSpan("SIR_NOT_APPEARING_IN_THIS_SNIPPET", Span(1, 5)),
AnnotatedSpan("LINE", Span(7, 33)),
AnnotatedSpan("BAR", Span(8, 9)),
AnnotatedSpan("LINE", Span(36, 75)),
AnnotatedSpan("PP", Span(61, 75)),
AnnotatedSpan("ELLIPSES", Span(75, 78)),
AnnotatedSpan("FOO", Span(7, 19)),
AnnotatedSpan("SINGLE_CHAR", Span(61, 62)),
]
expected_result = (
"<LINE><FOO>T<BAR>i</BAR>me present</FOO> and time past</LINE> "
"/ <LINE>Are both perhaps present <PP><SINGLE_CHAR>i</SINGLE_CHAR>n "
"time future</PP></LINE><ELLIPSES>...</ELLIPSES>"
)
self.assertEqual(
expected_result,
HTMLStyleAnnotationFormatter().annotated_text(
snippet_text, annotations, text_offsets=Span(7, len(original_text))
),
)
def get_sentence_spans(document_text: str) -> List[Span]:
tokenizer = nltk.tokenize.punkt.PunktSentenceTokenizer()
return [
Span(*s) for s in tokenizer.span_tokenize(document_text,
realign_boundaries=True)
]
def contexts_from_justifications(justifications: ImmutableDict[str, Span],
document) -> ImmutableDict[str, Span]:
document_text = document["fulltext"]
sentence_spans = get_sentence_spans(document_text)
contexts: Dict[str, Span] = {}
for justification_id, justification_span in justifications.items():
for s_span in sentence_spans:
if s_span.contains_span(justification_span):
# the sentence tokenizer doesn't recognize double newline as a potential sentence boundary,
# so we split on double newlines and return the parts of the pre/post context
# that are closest to the mention
precontext_lines = document_text[
s_span.start:justification_span.start].split("\n\n")
precontext_extra = ("\n\n".join(precontext_lines[:-1])
if len(precontext_lines) > 1 else "")
postcontext_lines = document_text[justification_span.
end:s_span.end].split("\n\n")
postcontext_extra = ("\n\n".join(postcontext_lines[1:])
if len(postcontext_lines) > 1 else "")
modified_sentence_start = s_span.start + len(precontext_extra)
modified_sentence_end = s_span.end - len(postcontext_extra)
assert (
justification_id not in contexts
), "justification should not be overlapping with more than one sentence"
contexts[justification_id] = Span(modified_sentence_start,
modified_sentence_end)
return immutabledict(contexts)
def _render_html(corpus: Corpus, output_dir: Path, parent_or_child_id: str,
start: int, end: int) -> Tuple[Path, str]:
"""Outputs either the whole document rendered in HTML or a subspan. `end` is inclusive."""
document = _get_document(corpus, parent_or_child_id)
if not document:
raise ValueError(
f"{document['parent_id']} not present in the document database.")
justification_spans: ImmutableDict[str, Span] = immutabledict(
{f"{start}:{end}": Span(start, end + 1)})
contexts = contexts_from_justifications(justification_spans, document)
to_render, _ = render_document(document["fulltext"], justification_spans,
contexts)
if not to_render:
raise ValueError("Could not find anything to render.")
final_html = _render_template(
document=immutabledict({
"id": document["parent_id"],
"title": document["title"],
"html": to_render,
"span": f"{start}:{end}",
}))
output_file = output_dir / f"{document['parent_id']}_{start}-{end}.html"
output_file.write_text(final_html)
return output_file, document["fulltext"][start:end + 1]
def match_against_tokens(
self,
token_sequence_to_match_against: Tuple[str, ...],
*,
slots_to_filler_spans: Mapping[SyntaxSemanticsVariable, Span],
) -> Optional[Span]:
"""
Gets the token indices, if any, for the first match of this template against
*token_sequence_to_match_against*, assuming any slots are filled by the tokens given by
*slots_to_fillers_spans*.
"""
# First, we turn the template into a token sequence to search for
# by filling in all the slots form the provided token span mapping.
tokens_to_match = []
for element in self.elements:
if isinstance(element, str):
tokens_to_match.append(element)
else:
slot_filler_span = slots_to_filler_spans.get(element)
if slot_filler_span:
# endpoints are exclusive
tokens_to_match.extend(token_sequence_to_match_against[
slot_filler_span.start:slot_filler_span.end])
else:
raise RuntimeError(
f"Template contained variable {element}, "
f"but it was not found in the mapping of slots to spans: "
f"{slots_to_filler_spans}")
# Now we need to check if the tokens to match occur in the given token sequence to
# match against. We don't expect these sequences to be long, so an inefficient solution
# is okay.
if not tokens_to_match:
raise RuntimeError(
"Don't know how to match any empty token sequence")
next_idx_to_search_from = 0
while next_idx_to_search_from < len(token_sequence_to_match_against):
try:
index_of_first_token = token_sequence_to_match_against.index(
tokens_to_match[0], next_idx_to_search_from)
candidate_match_exclusive_end = index_of_first_token + len(
tokens_to_match)
if candidate_match_exclusive_end <= len(
token_sequence_to_match_against):
if tokens_to_match == list(token_sequence_to_match_against[
index_of_first_token:candidate_match_exclusive_end]
):
# span endpoints are exclusive
return Span(index_of_first_token,
candidate_match_exclusive_end)
# False alarm - the first token matched, but not the whole sequence.
next_idx_to_search_from = index_of_first_token + 1
except ValueError:
# If we can't even find the first token of what we are searching for,
# we definitely have no match.
return None
# We got all the way to the end without finding a match
return None
def _sort_mapping_by_token_spans(
pairs) -> ImmutableDict[ObjectSemanticNode, Span]:
# we type: ignore because the proper typing of pairs is huge and mypy is going to screw it up
# anyway.
unsorted = immutabledict(pairs) # type: ignore
return immutabledict(
(matched_node, token_span) for (matched_node, token_span) in sorted(
unsorted.items(),
key=lambda item: Span.earliest_then_longest_first_key(item[1]),
))
def annotated_text(
self,
text: str,
annotations: Collection[AnnotatedSpan],
*,
text_offsets: Optional[Span] = None,
) -> str:
"""
Mark annotations on text in an HTML-like style.
Each annotation will becomes an HTML tag wrapping the text at the corresponding offsets.
Any attributes will become HTML attributes.
This does not add any other HTML annotations (`head`, `body`, etc.), so if desired the
user should add them afterwards.
If `text_offsets` is specified, the annotations are assumed to have offsets with respect
to some larger string, where `text` is a substring of that string with offsets
`text_offsets` relative to it. You might use this, for example, to render a single
paragraph from a document.
"""
if not text_offsets:
text_offsets = Span.from_inclusive_to_exclusive(0, len(text))
check_arg(
len(text_offsets) == len(text),
f"Text offsets length {len(text_offsets)} "
f"does not match text length {len(text)}",
)
# we process the annotations to (a) ensure they all fit within the requested snippet
# and (b) shift their offsets so that all offsets are relative to the text being
# formatted
processed_annotations = self._clip_to_offsets_and_shift(
annotations, text_offsets)
ret = io.StringIO()
last_uncopied_offset = 0
for tag in self._tag_sequence(processed_annotations):
if last_uncopied_offset < tag.offset:
ret.write(text[last_uncopied_offset:tag.offset])
last_uncopied_offset = tag.offset
ret.write(tag.string)
# get any trailing text after last tag
if last_uncopied_offset < text_offsets.end:
ret.write(text[last_uncopied_offset:text_offsets.end])
return ret.getvalue()
def test_intersection(self) -> None:
s1, s2, s3, s4 = (Span(0, 3), Span(2, 25), Span(25, 30), Span(10, 20))
s1_s2_intersection = Span(2, 3)
s2_s4_intersection = Span(10, 20)
self.assertIsNone(s1.intersection(s3))
self.assertIsNone(s2.intersection(s3))
self.assertEqual(s1.intersection(s2), s1_s2_intersection)
self.assertEqual(s2.intersection(s4), s2_s4_intersection)
self.assertEqual(s4.intersection(s2), s2_s4_intersection)
def parse_text_from_source(text_justification_lookup: TextJustificationLookup,
inf_just_pattern,
inf_just_span):
match = re.search(inf_just_pattern, inf_just_span)
if match:
# source = match.group(1)
document = match.group(2)
start = int(match.group(3))
end = int(match.group(4))
text_descriptor = TextDescriptor(doceid=document,
span=Span.from_inclusive_to_exclusive(start, end + 1),
language=None)
try:
lookup = text_justification_lookup.text_for_justification(text_descriptor, 50)
return lookup.spanning_tokens, lookup.original_text
except (RuntimeError, AttributeError):
return 'None', 'None'
else:
return 'None', 'None'
def render_single_justification_document(document: dict,
justification: Justification) -> str:
span_start = justification.span_start
span_end = justification.span_end
if not span_start or not span_end:
raise ValueError(
"Justification to render must have values for span_start and span_end."
)
justification_spans: ImmutableDict[str, Span] = immutabledict(
{f"{span_start}:{span_end}": Span(span_start, span_end + 1)})
contexts = contexts_from_justifications(justification_spans, document)
to_render, _ = render_document(document["fulltext"], justification_spans,
contexts)
if not to_render:
raise ValueError("Could not find anything to render.")
return to_render
def is_legal_template_span(
candidate_token_span: Span, *, invalid_token_spans: ImmutableSet[Span]
) -> bool:
# A template token span can't exceed the bounds of the utterance
if candidate_token_span.start < 0:
return False
if candidate_token_span.end > len(sentence_tokens):
return False
# or be bigger than our maximum template size...
if len(candidate_token_span) > max_length:
return False
for span in invalid_token_spans:
if candidate_token_span.contains_span(span):
return False
# or we have already aligned any of the tokens in between the objects
# to some other meaning.
for token_index in range(candidate_token_span.start, candidate_token_span.end):
if language_concept_alignment.token_index_is_aligned(token_index):
return False
return True
metadata_from_wordpiece = doc.metadata_for(tokens_from_wordpiece)
tokens_from_spacy = metadata_from_wordpiece[
WordPieceTokenizationAnnotator.EXISTING_TOKEN_THEORY_USED_FOR_WORDPIECE
]
map_spacy_to_wordpiece_indices = metadata_from_wordpiece[
WordPieceTokenizationAnnotator.MULTIMAP_FROM_EXISTING_TO_WORDPIECE_TOKENIZATION
]
mentions_from_apf = doc_with_lots_of_mention_algos.mentions(algorithm(ApfIngester))
mentions_from_corenlp = doc_with_lots_of_mention_algos.mentions(
algorithm(CoreNLPNameFinder)
)
mentions_from_spacy = doc_with_lots_of_mention_algos.mentions(
algorithm(SpacyAnnotator)
)
s = Span.from_inclusive_to_exclusive(2778, 2915)
print("== ACE ==")
print(
"\n".join(
str(mention)
for mention in get_items_overlapping_with_this_one(s, mentions_from_apf)
)
)
print("== CoreNLP ==")
print(
"\n".join(
str(mention)
for mention in get_items_overlapping_with_this_one(s, mentions_from_corenlp)
)
)
print("== spaCy ==")
def process_aligned_objects_with_template(
candidate_template: Tuple[AlignmentSlots, ...],
aligned_nodes: Tuple[SemanticNodeWithSpan, ...],
*,
invalid_token_spans: ImmutableSet[Span],
) -> Iterable[Optional[SurfaceTemplateBoundToSemanticNodes]]:
aligned_node_index = 0
template_elements: List[Union[str, SyntaxSemanticsVariable]] = []
slot_to_semantic_node: List[Tuple[SyntaxSemanticsVariable, SemanticNode]] = []
# We need to handle fixed strings that are pre or post fix to the rest of the
# Sentence differently as they don't have a fixed length so we could generate
# multiple options.
prefix_string_end = None
postfix_string_start = None
# In the event we generate a candidate template like:
# A, F, F, A then we want to compute this like A, F, A
# So we keep track if the previous token was a FixedString indicator
previous_node_was_string = False
for token in candidate_template:
# If the token in our template is an argument we need to assign it a
# unique SyntaxSemanticsVariable, and map it to the SemanticNode
if token == AlignmentSlots.Argument:
slot_semantic_variable = STANDARD_SLOT_VARIABLES[aligned_node_index]
template_elements.append(slot_semantic_variable)
aligned_node = aligned_nodes[aligned_node_index].node
if not isinstance(aligned_node, ObjectSemanticNode):
logging.debug(
f"Attempted to make template where an Argument is not an ObjectSemanticNode."
f"Invalid node: {aligned_node}"
)
# Log this failure and then ignore this attempt
yield None
slot_to_semantic_node.append((slot_semantic_variable, aligned_node))
aligned_node_index += 1
previous_node_was_string = False
else:
# We ignore this case to process A, F, F, A like A, F, A
if previous_node_was_string:
continue
# We make a note of where the end of our prefix string can be
# Then continue as we'll handle this case afterwards
elif aligned_node_index == 0:
prefix_string_end = aligned_nodes[aligned_node_index].span.start
# Similiarly to above, we instead mark the start of the postfix string
elif aligned_node_index == len(aligned_nodes):
postfix_string_start = aligned_nodes[aligned_node_index - 1].span.end
else:
# If our FixedString is flanked by two Arguments we just want to acquire all the tokens
# between them
if (
aligned_nodes[aligned_node_index - 1].span.end
!= aligned_nodes[aligned_node_index].span.start
):
candidate_token_span = Span(
aligned_nodes[aligned_node_index - 1].span.end,
aligned_nodes[aligned_node_index].span.start,
)
if not is_legal_template_span(
candidate_token_span, invalid_token_spans=invalid_token_spans
):
# If not a valid span, ignore this attempt
continue
template_elements.extend(
sentence_tokens[
candidate_token_span.start : candidate_token_span.end
]
)
previous_node_was_string = True
# We need to handle searching before or after the aligned token
# And we could generate multiple options of different lengths
# between 1 and _MAXIMUM_ACTION_TEMPLATE_TOKEN_LENGTH
if prefix_string_end and postfix_string_start:
for max_token_length_for_template_prefix in range(1, max_length + 1):
prefix_candidate_token_span = Span(
prefix_string_end - max_token_length_for_template_prefix,
prefix_string_end,
)
if is_legal_template_span(
prefix_candidate_token_span, invalid_token_spans=invalid_token_spans
):
for max_token_length_for_template_postfix in range(1, max_length + 1):
postfix_candidate_token_span = Span(
postfix_string_start,
postfix_string_start + max_token_length_for_template_postfix,
)
if is_legal_template_span(
postfix_candidate_token_span,
invalid_token_spans=invalid_token_spans,
):
final_template_elements: List[
Union[str, SyntaxSemanticsVariable]
] = list(
sentence_tokens[
prefix_candidate_token_span.start : prefix_candidate_token_span.end
]
)
final_template_elements.extend(template_elements)
final_template_elements.extend(
sentence_tokens[
postfix_candidate_token_span.start : postfix_candidate_token_span.end
]
)
yield SurfaceTemplateBoundToSemanticNodes(
surface_template=SurfaceTemplate(
elements=final_template_elements,
determiner_prefix_slots=[
SLOT for (SLOT, _) in slot_to_semantic_node
],
language_mode=language_mode,
),
slot_to_semantic_node=slot_to_semantic_node,
)
elif prefix_string_end:
for max_token_length_for_template_prefix in range(1, max_length + 1):
prefix_candidate_token_span = Span(
prefix_string_end - max_token_length_for_template_prefix,
prefix_string_end,
)
if is_legal_template_span(
prefix_candidate_token_span, invalid_token_spans=invalid_token_spans
):
final_template_elements = list(
sentence_tokens[
prefix_candidate_token_span.start : prefix_candidate_token_span.end
]
)
final_template_elements.extend(template_elements)
yield SurfaceTemplateBoundToSemanticNodes(
surface_template=SurfaceTemplate(
elements=final_template_elements,
determiner_prefix_slots=[
SLOT for (SLOT, _) in slot_to_semantic_node
],
language_mode=language_mode,
),
slot_to_semantic_node=slot_to_semantic_node,
)
elif postfix_string_start:
for max_token_length_for_template_postfix in range(1, max_length + 1):
postfix_candidate_token_span = Span(
postfix_string_start,
postfix_string_start + max_token_length_for_template_postfix,
)
if is_legal_template_span(
postfix_candidate_token_span, invalid_token_spans=invalid_token_spans
):
final_template_elements = list(template_elements)
final_template_elements.extend(
sentence_tokens[
postfix_candidate_token_span.start : postfix_candidate_token_span.end
]
)
yield SurfaceTemplateBoundToSemanticNodes(
surface_template=SurfaceTemplate(
elements=final_template_elements,
determiner_prefix_slots=[
SLOT for (SLOT, _) in slot_to_semantic_node
],
language_mode=language_mode,
),
slot_to_semantic_node=slot_to_semantic_node,
)
else:
yield SurfaceTemplateBoundToSemanticNodes(
surface_template=SurfaceTemplate(
elements=template_elements,
determiner_prefix_slots=[SLOT for (SLOT, _) in slot_to_semantic_node],
language_mode=language_mode,
),
slot_to_semantic_node=slot_to_semantic_node,
)
def _candidate_templates(
self, language_perception_semantic_alignment: LanguagePerceptionSemanticAlignment
) -> AbstractSet[SurfaceTemplateBoundToSemanticNodes]:
ret = []
language_concept_alignment = (
language_perception_semantic_alignment.language_concept_alignment
)
# Find all objects we have recognized...
for (
object_node,
span_for_object,
) in language_concept_alignment.node_to_language_span.items():
if isinstance(object_node, ObjectSemanticNode):
try:
# Any words immediately before them or after them are candidate attributes.
# See https://github.com/isi-vista/adam/issues/791 .
preceding_token_index = span_for_object.start - 1
if (
preceding_token_index >= 0
and not language_concept_alignment.token_index_is_aligned(
preceding_token_index
)
):
ret.append(
SurfaceTemplateBoundToSemanticNodes(
language_concept_alignment.to_surface_template(
{object_node: SLOT1},
restrict_to_span=Span(
preceding_token_index, span_for_object.end
),
language_mode=self._language_mode,
),
{SLOT1: object_node},
)
)
following_token_index = span_for_object.end + 1
if following_token_index < len(
language_concept_alignment.language.as_token_sequence()
) and not language_concept_alignment.token_index_is_aligned(
following_token_index
):
ret.append(
SurfaceTemplateBoundToSemanticNodes(
language_concept_alignment.to_surface_template(
{object_node: SLOT1},
restrict_to_span=Span(
span_for_object.start, following_token_index
),
language_mode=self._language_mode,
),
{SLOT1: object_node},
)
)
# Catches errors in to_surface_template() - we skip this case to prevent the learning from breaking.
except RuntimeError:
continue
return immutableset(
bound_surface_template
for bound_surface_template in ret
# For now, we require templates to account for the entire utterance.
# See https://github.com/isi-vista/adam/issues/789
if covers_entire_utterance(
bound_surface_template,
language_concept_alignment,
# We need to explicitly ignore determiners here for some reason
# See: https://github.com/isi-vista/adam/issues/871
ignore_determiners=True,
)
# this keeps the relation learner from learning things such as "a_slot1" which will pose an issue for
# later learning of attributes since the learner may consider both the attribute and the object to be objects initially,
# leading it to try to match two objects with a template that only has one slot
and not all(
(e in ENGLISH_DETERMINERS or isinstance(e, SyntaxSemanticsVariable))
for e in bound_surface_template.surface_template.elements
)
)
def match_against_tokens(
self,
token_sequence_to_match_against: Tuple[str, ...],
*,
slots_to_filler_spans: Mapping[SyntaxSemanticsVariable, Span],
) -> Optional[Span]:
"""
Gets the token indices, if any, for the first match of this template against
*token_sequence_to_match_against*, assuming any slots are filled by the tokens given by
*slots_to_fillers_spans*.
"""
# First, we turn the template into a token sequence to search for
# by filling in all the slots form the provided token span mapping.
tokens_to_match = []
for element in self.elements:
if isinstance(element, str):
# Hack to handle determiners.
#
# This may not handle Chinese properly; see
# https://github.com/isi-vista/adam/issues/993
try:
index = token_sequence_to_match_against.index(element)
if (index - 1 >= 0
and token_sequence_to_match_against[index - 1]
in ENGLISH_DETERMINERS):
tokens_to_match.append(
token_sequence_to_match_against[index - 1])
except ValueError:
pass
finally:
tokens_to_match.append(element)
else:
slot_filler_span = slots_to_filler_spans.get(element)
if slot_filler_span:
# endpoints are exclusive
start = slot_filler_span.start
# Hack to handle determiners
#
# This may not handle Chinese properly; see
# https://github.com/isi-vista/adam/issues/993
if (slot_filler_span.start - 1 >= 0
and token_sequence_to_match_against[
slot_filler_span.start - 1]
in ENGLISH_DETERMINERS):
start -= 1
tokens_to_match.extend(
token_sequence_to_match_against[start:slot_filler_span.
end])
# If template contains an element not found in the mapping of slots to spans, we can return empty here.
# We don't want to do this now because of generics.
# else:
# return None
# Now we need to check if the tokens to match occur in the given token sequence to
# match against. We don't expect these sequences to be long, so an inefficient solution
# is okay.
if not tokens_to_match:
raise RuntimeError(
"Don't know how to match any empty token sequence")
next_idx_to_search_from = 0
while next_idx_to_search_from < len(token_sequence_to_match_against):
try:
index_of_first_token = token_sequence_to_match_against.index(
tokens_to_match[0], next_idx_to_search_from)
candidate_match_exclusive_end = index_of_first_token + len(
tokens_to_match)
if candidate_match_exclusive_end <= len(
token_sequence_to_match_against):
if tokens_to_match == list(token_sequence_to_match_against[
index_of_first_token:candidate_match_exclusive_end]
):
# span endpoints are exclusive
return Span(index_of_first_token,
candidate_match_exclusive_end)
# False alarm - the first token matched, but not the whole sequence.
next_idx_to_search_from = index_of_first_token + 1
except ValueError:
# If we can't even find the first token of what we are searching for,
# we definitely have no match.
return None
# We got all the way to the end without finding a match
return None
def test_index(self) -> None:
overlapping_items = (
Foo(Span(0, 10)),
Foo(Span(5, 25)),
Foo(Span(20, 30)),
Bar(Span(20, 30)),
)
index = HasSpanIndex.index(overlapping_items)
self.assertEqual(immutableset([overlapping_items[0]]),
index.get_exactly_matching(Span(0, 10)))
self.assertEqual(immutableset([overlapping_items[1]]),
index.get_exactly_matching(Span(5, 25)))
self.assertEqual(
immutableset([overlapping_items[2], overlapping_items[3]]),
index.get_exactly_matching(Span(20, 30)),
)
self.assertEqual(immutableset(),
index.get_exactly_matching(Span(6, 26)))
self.assertEqual(immutableset(), index.get_overlapping(Span(31, 35)))
self.assertEqual(
immutableset([overlapping_items[2], overlapping_items[3]]),
index.get_overlapping(Span(29, 32)),
)
self.assertEqual(immutableset(), index.get_contained(Span(25, 30)))
self.assertEqual(
immutableset([overlapping_items[2], overlapping_items[3]]),
index.get_contained(Span(20, 30)),
)
self.assertEqual(
immutableset([
overlapping_items[0],
overlapping_items[1],
overlapping_items[2],
overlapping_items[3],
]),
index.get_contained(Span(0, 30)),
)
self.assertEqual(immutableset(), index.get_containing(Span(0, 15)))
self.assertEqual(
immutableset([
overlapping_items[1], overlapping_items[2],
overlapping_items[3]
]),
index.get_containing(Span(21, 24)),
)
def span(self, start_index: int, *, end_index_exclusive: int) -> Span:
return Span(start_index, end_index_exclusive)
def test_disjoint_index(self) -> None:
overlapping_items = (
Foo(Span(0, 10)),
Foo(Span(5, 25)),
Foo(Span(20, 30)),
Bar(Span(20, 30)),
)
with self.assertRaisesRegex(
ValueError, "Some range keys are connected or overlapping"):
HasSpanIndex.index_disjoint(overlapping_items)
s1, s2, s3 = (Span(0, 3), Span(5, 25), Span(25, 30))
s2_within = Span(5, 10)
s4_contains = Span(5, 30)
fs1, fs2, fs3 = Foo(s1), Foo(s2), Foo(s3)
index = HasSpanIndex.index_disjoint((fs1, fs2, fs3))
self.assertIsNone(index.get_exactly_matching(s2_within))
self.assertEqual(fs3, index.get_exactly_matching(s3))
self.assertEqual(immutableset(), index.get_overlapping(Span(35, 40)))
self.assertEqual(immutableset([fs3]),
index.get_overlapping(Span(28, 35)))
self.assertEqual(immutableset([fs1, fs2]),
index.get_overlapping(Span(2, 7)))
self.assertEqual(immutableset(), index.get_contained(s2_within))
self.assertEqual(immutableset([fs2, fs3]),
index.get_contained(s4_contains))
self.assertIsNone(index.get_containing(s4_contains))
self.assertEqual(fs2, index.get_containing(s2_within))