Drop lots of legacy code. Support Python 3.7-3.11.

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

diff --git a/src/librarian/document.py b/src/librarian/document.py
index fbaf6ca..aa6f37f 100644 (file)
--- a/src/librarian/document.py
+++ b/src/librarian/document.py
@@ -1,8 +1,11 @@
+# This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
+# Copyright © Fundacja Wolne Lektury. See NOTICE for more information.
+#
 import gettext
 import os
 import re
+import urllib.request
 from lxml import etree
-import six
 from .parser import parser
 from . import dcparser, DCNS, DirDocProvider
 from .functions import lang_code_3to2
@@ -10,7 +13,7 @@ from .functions import lang_code_3to2
 
 class WLDocument:
     def __init__(self, filename=None, url=None, provider=None):
-        source = filename or six.moves.urllib.request.urlopen(url)
+        source = filename or urllib.request.urlopen(url)
         tree = etree.parse(source, parser=parser)
         self.tree = tree
         tree.getroot().document = self
@@ -32,15 +35,39 @@ class WLDocument:
     @property
     def children(self):
         for part_uri in self.meta.parts or []:
-            yield type(self)(
-                filename=self.provider.by_slug(part_uri.slug),
-                provider=self.provider
-            )
-            
+            with self.provider.by_slug(part_uri.slug) as f:
+                yield type(self)(filename=f, 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 +102,10 @@ 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 +119,5 @@ class WLDocument:
             persons.remove(None)
         return persons
 
+    def references(self):
+        return self.tree.findall('.//ref')