from model.sense import Sense from model.editable import Editable from model.tags import TAGS class Entry(Editable): def __init__(self, entry_xml): status = entry_xml.querySelector("head status") headword = entry_xml.querySelector("head headword lemma") grammar = entry_xml.querySelector("head grammar category") comment = entry_xml.querySelector("head comment") self.status = status.textContent if status else "" self.headword = headword.textContent if headword else "" self.grammar = grammar.textContent if grammar else "" self.comment = comment.textContent if comment else "" self.variants = [v.textContent for v in entry_xml.querySelectorAll("head variantList variant")] self.lexical_unit = {} lex_unit = entry_xml.querySelector("lexical_unit lexeme,lexicalUnit lexeme") if lex_unit: self.lexical_unit['id'] = lex_unit.getAttribute("lexical_unit_lexeme_id") self.lexical_unit['text'] = lex_unit.textContent self.measure = {} measure = entry_xml.querySelector("measureList measure") if measure: self.measure["source"] = measure.getAttribute("source") self.measure["type"] = measure.getAttribute("type") self.measure["text"] = measure.textContent self.labels = [] for tag_xml in entry_xml.querySelectorAll("head labelList label"): t_type = tag_xml.getAttribute("type") t_value = tag_xml.textContent if t_type not in TAGS: # using some default t_type = TAGS.keys()[0] self.labels.append((t_type, t_value)) self.senses = [Sense(sense_xml) for sense_xml in entry_xml.querySelectorAll("body senseList sense")] def get_measure_text(self): return self.measure["text"] if "text" in self.measure else ""