Function get_verb_phrases_from_token

data_preprocessing/utils.py:277–355

(token, phrase_config=None)

    return [token.text]

def get_verb_phrases_from_token(token, phrase_config=None):
    if token.pos_ == "VERB" or token.pos_ == "AUX":
        if hasattr(token, "children"):
            dobj = None
            prts = []
            preps = []
            attrs = []
            acomps = []

            for child in token.children:
                if _enabled(phrase_config, "include_dobj") and child.dep_ == "dobj":
                    dobj = child
                elif _enabled(phrase_config, "include_prt") and child.dep_ == "prt":
                    prts.append(child)
                elif _enabled(phrase_config, "include_prep") and child.dep_ == "prep":
                    preps.append(child)
                elif _enabled(phrase_config, "include_attr") and child.dep_ == "attr":
                    attrs.append(child)
                elif _enabled(phrase_config, "include_acomp") and child.dep_ == "acomp":
                    acomps.append(child)
            
            phrases = []
            verb = f"{token.lemma_}"

            if prts:
                verb = f"{verb} {' '.join([prt.text for prt in prts])}"

            dobj_phrases = []

            if dobj:
                dobj_phrases = get_noun_phrases_from_token(dobj, phrase_config=phrase_config)
            
            prep_phrases = []

            for prep in preps:
                prep_phrases.extend(get_prep_phrases_from_token(prep, phrase_config=phrase_config))
            
            attr_phrases = []

            for attr in attrs:
                attr_phrases.extend(get_noun_phrases_from_token(attr, phrase_config=phrase_config))

            acomp_phrases = []

            for acomp in acomps:
                acomp_phrases.extend(get_acomp_phrases_from_token(acomp, phrase_config=phrase_config))
    
            if _enabled(phrase_config, "longest_only"):
                final_phrase = ""

                for ph in dobj_phrases+attr_phrases+acomp_phrases+prep_phrases:
                    final_phrase += f" {ph}"

                if final_phrase:
                    return [f"{verb}{final_phrase}"]
                
                return []