fixes for edge cases

[librarian.git] / src / librarian / document.py

diff --git a/src/librarian/document.py b/src/librarian/document.py
index a3ee093..6cdf191 100644 (file)
--- a/src/librarian/document.py
+++ b/src/librarian/document.py
@@ -33,14 +33,40 @@ class WLDocument:
     def children(self):
         for part_uri in self.meta.parts or []:
             yield type(self)(
     def children(self):
         for part_uri in self.meta.parts or []:
             yield type(self)(

-                filename=self.provider.by_uri(part_uri),
+                filename=self.provider.by_slug(part_uri.slug),

                 provider=self.provider
             )
                 provider=self.provider
             )

-            

     
     def build(self, builder, base_url=None, **kwargs):
         return builder(base_url=base_url).build(self, **kwargs)
 
     
     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
     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)
         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)
 
                 _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
 
             persons.remove(None)
         return persons
 

+    def references(self):
+        return self.tree.findall('.//ref')