luscenje_struktur/src/match.py

from word import Word

class StructureMatch:
    def __init__(self, match_id, structure):
        self.match_id = str(match_id)
        self.structure = structure

        self.matches = []
        self.representations = {}
    
    @staticmethod
    def from_db(db, colocation_id, structure):
        result = StructureMatch(colocation_id, structure)
        for match_id in db.execute("SELECT mid_match_id FROM ColocationMatches WHERE mid_colocation_id=?", (colocation_id,)):
            to_add = {}

            for component_id, word_lemma, word_text, word_msd, word_id in db.execute("""
                SELECT component_id, word_lemma, word_text, word_msd, word_id 
                FROM Matches WHERE match_id=?""", match_id):

                to_add[str(component_id)] = Word(word_lemma, word_msd, word_id, word_text, False)

            result.matches.append(to_add)
        
        for component_id, text in db.execute("SELECT component_id, text FROM Representations WHERE colocation_id=?", (colocation_id,)):
            result.representations[str(component_id)] = text
        
        return result

    def distinct_forms(self):
        dm = set()
        keys = list(self.matches[0].keys())
        for words in self.matches:
            dm.add(" ".join(words[k].text for k in keys))
        return len(dm)

    def append(self, match):
        self.matches.append(match)

    def __len__(self):
        return len(self.matches)