X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/9d05775ad3576582aeb43bdd95224c58f17e1673..c7734daf3d6884c358f1644c1f8f7970453fafb1:/src/librarian/document.py diff --git a/src/librarian/document.py b/src/librarian/document.py index fbaf6ca..6cdf191 100644 --- a/src/librarian/document.py +++ b/src/librarian/document.py @@ -36,11 +36,37 @@ class WLDocument: filename=self.provider.by_slug(part_uri.slug), provider=self.provider ) - def build(self, builder, base_url=None, **kwargs): return builder(base_url=base_url).build(self, **kwargs) + def assign_ids(self, existing=None): + # Find all existing IDs. + existing = existing or set() + que = [self.tree.getroot()] + while que: + item = que.pop(0) + try: + item.normalize_insides() + except AttributeError: + pass + existing.add(item.attrib.get('id')) + que.extend(item) + + i = 1 + que = [self.tree.getroot()] + while que: + item = que.pop(0) + que.extend(item) + if item.attrib.get('id'): + continue + if not getattr(item, 'SHOULD_HAVE_ID', False): + continue + while f'e{i}' in existing: + i += 1 + item.attrib['id'] = f'e{i}' + i += 1 + def _compat_assign_ordered_ids(self): """ Compatibility: ids in document order, to be roughly compatible with legacy @@ -75,7 +101,10 @@ class WLDocument: def _compat_assigns_section_ids_in_elem(elem, prefix='sec'): for i, child in enumerate(elem): idfier = '{}{}'.format(prefix, i + 1) - child.attrib['_compat_section_id'] = idfier + try: + child.attrib['_compat_section_id'] = idfier + except: + pass _compat_assigns_section_ids_in_elem(child, idfier + '-') _compat_assigns_section_ids_in_elem(self.tree.getroot().master) @@ -89,3 +118,5 @@ class WLDocument: persons.remove(None) return persons + def references(self): + return self.tree.findall('.//ref')