Speaker: "Alice" type: advice Speaker: "Alice" type: birth Speaker: "Alice" type: custom Speaker: "Alice" type: delivered_by Speaker: "Alice" type: emotion Speaker: "Alice" type: location Speaker: "Alice" type: mood Speaker: "Alice" type: name Speaker: "Alice" type: quote Speaker: "Alice" type: shout Speaker: "Mouse" type: advice Speaker: "Mouse" type: birth Speaker: "Mouse" type: custom Speaker: "Mouse" type: delivered_by Speaker: "Mouse" type: emotion Speaker: "Mouse" type: location Speaker: "Mouse" type: mention Speaker: "Mouse" type: opinion Speaker: "Mouse" type: quote Speaker: "Mouse" type: shout Speaker: "Mouse" type: side_character Speaker: "Mouse" type: voice Speaker: "Assistant" type: approval Speaker: "Assistant" type: cohesion Speaker: "Assistant" type: confirming """ def extract_statements(paragraphs): """Given text as argument, extract statements/questions as speak w/ tag.""" marking = ActionNames.Marc.Values.Text # Denote paragraphs as needing to be split into statements paragraph.Annotation(marking, _value=marking(_value=True, count=0)) paragraph.Annotation(marking, _value=marking(_value=False, count=0)) voice_objs = [] for idx, segment in paragraph.segments.get_ut()) -> utterance: # Determine if the index of a statement if (segment.utterance_end() ^ segment.utterance_start()) > 0: speaker = 'ASSISTANT' else: speaker = 'USER' text = segment.utterance().ToText() delim = ':' # quoted if delim in context.lower(): speaker = context.split(delim)[0] else: delims = ['"', "'", '@'] # speaking; assumed to precede body for d in delims: if text.endswith(d): speaker, text = context.split(d)[-1] break assert speaker text = text.strip() speaker = text # nonqualified names are speakers obj_tag = speech (chex).ToText() # Next, rewrite the statement to have all relevant tags out_stmt = f"Speaker: {speaker} {obj_tag} " out_stmt += f"{text}" TAGS_TO_REGEX = { "lik": "----------", "LIK": "----------", "QUOTE": "", ",": ", ", ", ":", ", ", "__": "----", "~": "----", } for k, v in TAGS_TO_REGEX.items(): if k in text: text = text.replace(k, v) scr_out = out_stmt + "\n\n" sentences = format_markup_statements(paragraph=(.)--) scr_out += "\n".join(sentences) first_stat_idx = top[0] top = [] combined_scraps = "".join([scr_out, sentences]) paragraph.Annotation(marking, _value=marking(_value=first_stat_idx, count=0)) obj_type = delimeter.get_ut()). -> utteranc paragraph.Annotation(marking, _value=marking(_value=True, count=0)) return combined_scraps```