def check_matches(tgt_doc, matches, strict=True):
"""
Check that the target document text is indeed a subsequence of
the source document text (the source document is expected to be
"augmented" version of the target with new text interspersed
throughout)
Parameters
----------
tgt_doc :
matches : list of (int, int, int)
List of triples (i, j, n) representing matching subsequences:
a[i:i+n] == b[j:j+n].
See `difflib.SequenceMatcher.get_matching_blocks`.
strict : boolean
If True, raise an exception if there are match gaps in the
target document, otherwise just print the gaps to stderr.
"""
tgt_text = tgt_doc.text()
if not tgt_text:
return
elif not matches:
raise WeaveException('no matches in non-empty target doc')
elif matches[0].b != 0:
oops = ('matches ({}) do not start at beginning of target '
'document <{}>').format(matches[0], tgt_doc.origin)
raise WeaveException(oops)
gaps = tgt_gaps(matches)
if gaps:
# we might want to give some slack because gaps can result from
# manual rewrites that happened here and there in the soclogs
# e.g. a pair of logical not (¬) around _ => ^_^
# in these cases, just print them on stderr for quick checks
for gap in gaps:
gap_txt = tgt_text[gap[0]:gap[0] + gap[1]]
print(u"Match gap in tgt doc ({})\t{}\t{}".format(
tgt_doc.origin, gap, gap_txt),
file=sys.stderr)
print('Matches: ', matches)
tgt_turns = set(
turn_id(x) for x in tgt_doc.units if turn_id(x) is not None)
print('Turns: ', sorted(tgt_turns))
if strict:
oops = 'there are match gaps in the target document {}: {}'
raise WeaveException(oops.format(tgt_doc.origin, gaps))
_, tgt, size = matches[-1]
if tgt + size != len(tgt_text):
raise WeaveException('matches do not cover the full target '
'document')
def check_matches(tgt_doc, matches, strict=True):
"""
Check that the target document text is indeed a subsequence of
the source document text (the source document is expected to be
"augmented" version of the target with new text interspersed
throughout)
Parameters
----------
tgt_doc :
matches : list of (int, int, int)
List of triples (i, j, n) representing matching subsequences:
a[i:i+n] == b[j:j+n].
See `difflib.SequenceMatcher.get_matching_blocks`.
strict : boolean
If True, raise an exception if there are match gaps in the
target document, otherwise just print the gaps to stderr.
"""
tgt_text = tgt_doc.text()
if not tgt_text:
return
elif not matches:
raise WeaveException('no matches in non-empty target doc')
elif matches[0].b != 0:
oops = ('matches ({}) do not start at beginning of target '
'document <{}>').format(matches[0], tgt_doc.origin)
raise WeaveException(oops)
gaps = tgt_gaps(matches)
if gaps:
# we might want to give some slack because gaps can result from
# manual rewrites that happened here and there in the soclogs
# e.g. a pair of logical not (¬) around _ => ^_^
# in these cases, just print them on stderr for quick checks
for gap in gaps:
gap_txt = tgt_text[gap[0]:gap[0] + gap[1]]
print(u"Match gap in tgt doc ({})\t{}\t{}".format(
tgt_doc.origin, gap, gap_txt), file=sys.stderr)
print('Matches: ', matches)
tgt_turns = set(turn_id(x) for x in tgt_doc.units
if turn_id(x) is not None)
print('Turns: ', sorted(tgt_turns))
if strict:
oops = 'there are match gaps in the target document {}: {}'
raise WeaveException(oops.format(tgt_doc.origin, gaps))
_, tgt, size = matches[-1]
if tgt + size != len(tgt_text):
raise WeaveException('matches do not cover the full target '
'document')
def get_eduinfo(config, doc, context, rstuff, edu):
"""
extract the interesting parst of an EDU
:: ... -> EduInfo
"""
turn = context[edu].turn
surface_act = edu.features.get("Surface_Act", "?")
if edu.type == "Segment" or not config.emit_dialogue_acts:
dialogue_act = "?"
else:
dialogue_act = edu.type
addresee_feat = edu.features.get("Addressee", STAC_UNSET)
addressees = set(x.strip() for x in addresee_feat.split(";"))
if not addressees or STAC_UNSET in addressees:
addressees = set("?")
return EduInfo(edu=edu,
turn_id=turn_id(turn),
dialogue_act=dialogue_act,
text=doc.text(edu.text_span()),
surface_act=surface_act,
speaker=turn.features["Emitter"],
addressees=addressees,
rstuff=rstuff)
def get_eduinfo(config, doc, context, rstuff, edu):
"""
extract the interesting parst of an EDU
:: ... -> EduInfo
"""
turn = context[edu].turn
surface_act = edu.features.get("Surface_Act", "?")
if edu.type == "Segment" or not config.emit_dialogue_acts:
dialogue_act = "?"
else:
dialogue_act = edu.type
addresee_feat = edu.features.get("Addressee", STAC_UNSET)
addressees = set(x.strip() for x in addresee_feat.split(";"))
if not addressees or STAC_UNSET in addressees:
addressees = set("?")
return EduInfo(edu=edu,
turn_id=turn_id(turn),
dialogue_act=dialogue_act,
text=doc.text(edu.text_span()),
surface_act=surface_act,
speaker=turn.features["Emitter"],
addressees=addressees,
rstuff=rstuff)
def _get_turn_info(self, anno):
"return the speaker and turn id for a unit-level annotation"
enclosing_turns = [t for t in self.turns if t.span.encloses(anno.span)]
if len(enclosing_turns) > 0:
turn = enclosing_turns[0]
speaker = stac_anno.speaker(turn)
turn_id = stac_anno.turn_id(turn)
return speaker, turn_id
else:
return None, None
def _get_turn_info(self, anno):
"return the speaker and turn id for a unit-level annotation"
enclosing_turns = [t for t in self.turns if t.span.encloses(anno.span)]
if len(enclosing_turns) > 0:
turn = enclosing_turns[0]
speaker = stac_anno.speaker(turn)
turn_id = stac_anno.turn_id(turn)
return speaker, turn_id
else:
return None, None
def subgrouping(self):
"""What abstract subgrouping the EDU is in (here: turn stars)
See also
--------
educe.stac.context.merge_turn_stars
Return
------
subgrouping: string
"""
return self._doc.global_id('t' + str(turn_id(self.tstar)))
def subgrouping(self):
"""What abstract subgrouping the EDU is in (here: turn stars)
See also
--------
educe.stac.context.merge_turn_stars
Return
------
subgrouping: string
"""
return self._doc.global_id('t' + str(turn_id(self.tstar)))
