lexicalUnit from single lexeme to multiple lexeme support

This commit is contained in:
matic_t
2020-10-12 01:28:06 -07:00
parent eea25b9968
commit bd4e64f818
2 changed files with 21 additions and 12 deletions
+12 -5
View File
@@ -42,11 +42,18 @@ class Entry(Data):
self.variants = [v.textContent for v in entry_xml.querySelectorAll("head variantList variant")]
self.homonymy = [{"value": v.textContent, "name": v.getAttribute("name")} for v in entry_xml.querySelectorAll("head headword homonymy homonymyFeature ")]
self.related_entries = [re.textContent for re in entry_xml.querySelectorAll("head relatedEntryList relatedEntry")]
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
lex_units = entry_xml.querySelectorAll("lexical_unit lexeme,lexicalUnit lexeme")
lex_unit_parent = entry_xml.querySelector("lexicalUnit")
self.lexical_unit['lexemes'] = []
self.lexical_unit['id'] = lex_unit_parent.getAttribute('id') if lex_unit_parent and lex_unit_parent.hasAttribute(
"id") else None
self.lexical_unit['type'] = lex_unit_parent.getAttribute("type") if lex_unit_parent and lex_unit_parent.hasAttribute(
"type") else "single"
for unit in lex_units:
lexical_unit = {}
lexical_unit['id'] = unit.getAttribute("lexical_unit_lexeme_id")
lexical_unit['text'] = unit.textContent
self.lexical_unit['lexemes'].append(lexical_unit)
measure = entry_xml.querySelector("measureList measure")
if measure: