def transform_utterance(self, utt, override_input_filter=False):
"""
Computes per-utterance attributes of an individual utterance or string. For utterances which do not contain all of the `input_field` attributes as specified in the constructor, or for utterances which return `False` on `input_filter`, this call will not annotate the utterance. For strings, will convert the string to an utterance and return the utterance, annotating it if `input_field` is not set to `None` at initialization.
:param utt: utterance or a string
:param override_input_filter: ignore `input_filter` and compute attribute for all utterances
:return: the utterance
"""
if isinstance(utt, str):
utt = Utterance(text=utt, speaker=Speaker(id="speaker"))
if self.input_field is None:
text_entry = utt.text
else:
if not override_input_filter:
if not self.input_filter(utt, self.aux_input):
return utt
if isinstance(self.input_field, str):
text_entry = utt.retrieve_meta(self.input_field)
elif isinstance(self.input_field, list):
text_entry = {field: utt.retrieve_meta(field) for field in self.input_field}
if sum(x is None for x in text_entry.values()) > 0:
return utt
if text_entry is None:
return utt
if len(self.aux_input) == 0:
result = self.proc_fn(text_entry)
else:
result = self.proc_fn(text_entry, self.aux_input)
if self.multi_outputs:
for res, out in zip(result, self.output_field):
utt.add_meta(out, res)
else:
utt.add_meta(self.output_field, result)
return utt
def transform_utterance(self, utt: Utterance, spacy_nlp: Callable[[str], Doc] = None, markers: bool = False):
"""
Extract politeness strategies for raw string inputs (or individual utterances)
:param utt: the utterance to be annotated with politeness strategies.
:spacy_nlp: if provided, will use this SpaCy object to do parsing; otherwise will initialize an object via `load('en')`.
:return: the utterance with politeness annotations.
"""
if isinstance(utt, str):
utt = Utterance(text=utt, speaker=Speaker(id='speaker'))
if self.parse_attribute_name not in utt.meta:
if spacy_nlp is None:
raise ValueError('spacy object required')
parses = process_text(utt.text, spacy_nlp=spacy_nlp)
utt.add_meta(self.parse_attribute_name, parses)
parsed = utt.retrieve_meta(self.parse_attribute_name)
for i, sent in enumerate(parsed):
for p in sent["toks"]:
p["tok"] = p['tok'].lower()
parses = [x["toks"] for x in parsed]
utt.meta[self.strategy_attribute_name], marks = self._extractor_lookup[self.strategy_collection](parses)
if markers:
utt.meta[self.marker_attribute_name] = marks
return utt