Fix link updating not working

src/calibre/ebooks/oeb/polish/split.py

@@ -372,6 +372,7 @@ def merge_html(container, names, master, insert_page_breaks=False):
     master_body = p(master).findall('h:body', namespaces=XPNSMAP)[-1]
     master_base = os.path.dirname(master)
     anchor_map = {n:{} for n in names if n != master}
+    first_anchor_map = {}
 
     for name in names:
         if name == master:
@@ -419,6 +420,7 @@ def merge_html(container, names, master, insert_page_breaks=False):
         if 'id' not in first_child.attrib:
             first_child.set('id', unique_anchor(seen_anchors, 'top'))
             seen_anchors.add(first_child.get('id'))
+        first_anchor_map[name] = first_child.get('id')
 
         if insert_page_breaks:
             first_child.set('style', first_child.get('style', '') + '; page-break-before: always')
@@ -444,6 +446,8 @@ def merge_html(container, names, master, insert_page_breaks=False):
         repl = MergeLinkReplacer(fname, anchor_map, master, container)
         container.replace_links(fname, repl)
 
+    return first_anchor_map
+
 
 def merge_css(container, names, master):
     p = container.parsed

src/calibre/ebooks/pdf/html_writer.py

@@ -247,13 +247,14 @@ def add_cover(pdf_doc, cover_data, page_layout, opts):
 
 
 # Margin groups {{{
-def create_margin_groups(container):
+def create_margin_groups(container, name_anchor_map):
 
     def merge_group(group):
         if len(group) > 1:
             group_margins = group[0][1]
             names = [name for (name, margins) in group]
-            merge_html(container, names, names[0], insert_page_breaks=True)
+            first_anchor_map = merge_html(container, names, names[0], insert_page_breaks=True)
+            name_anchor_map.update(first_anchor_map)
             group = [(names[0], group_margins)]
         return group
 
@@ -371,8 +372,11 @@ class AnchorLocation(object):
         self.pagenum, self.left, self.top, self.zoom = pagenum, left, top, zoom
 
     def __repr__(self):
-        return 'AnchorLocation(pagenum={}, left={}, top={}, zoom={})'.format(
+        return 'AnchorLocation(pagenum={}, left={}, top={}, zoom={})'.format(self.as_tuple)
-                self.pagenum, self.left, self.top, self.zoom)
+
+    @property
+    def as_tuple(self):
+        return self.pagenum, self.left, self.top, self.zoom
 
 
 def get_anchor_locations(pdf_doc, first_page_num, toc_uuid):
@@ -403,7 +407,7 @@ def fix_links(pdf_doc, anchor_locations, name_anchor_map, mark_links, log):
             loc = anchor_locations.get(name_anchor_map.get(purl.fragment))
             if loc is None:
                 log.warn('Anchor location for link to {} not found'.format(purl.fragment))
-        return loc
+        return None if loc is None else loc.as_tuple
 
     pdf_doc.alter_links(replace_link, mark_links)
 # }}}
@@ -520,8 +524,8 @@ def add_pagenum_toc(root, toc, opts, page_number_display_map):
 def convert(opf_path, opts, metadata=None, output_path=None, log=default_log, cover_data=None, report_progress=lambda x, y: None):
     container = Container(opf_path, log)
     report_progress(0.05, _('Parsed all content for markup transformation'))
-    margin_groups = create_margin_groups(container)
     name_anchor_map = make_anchors_unique(container)
+    margin_groups = create_margin_groups(container, name_anchor_map)
     toc = get_toc(container, verify_destinations=False)
     has_toc = toc and len(toc)
     links_page_uuid = add_all_links(container, margin_groups)

src/calibre/utils/podofo/doc.cpp

@@ -453,17 +453,16 @@ PDFDoc_alter_links(PDFDoc *self, PyObject *args) {
 							PdfObject *uo = A.GetKey("URI");
 							if (uo && uo->IsString()) {
 								const std::string &uri = uo->GetString().GetStringUtf8();
-								PyObject *ret = PyObject_CallObject(alter_callback, Py_BuildValue("(N)", PyUnicode_DecodeUTF8(uri.c_str(), uri.length(), "replace")));
+								pyunique_ptr ret(PyObject_CallObject(alter_callback, Py_BuildValue("(N)", PyUnicode_DecodeUTF8(uri.c_str(), uri.length(), "replace"))));
 								if (!ret) { return NULL; }
-								if (PyTuple_Check(ret) && PyTuple_GET_SIZE(ret) == 4) {
+								if (PyTuple_Check(ret.get()) && PyTuple_GET_SIZE(ret.get()) == 4) {
 									int pagenum; double left, top, zoom;
-									if (PyArg_ParseTuple(ret, "iddd", &pagenum, &left, &top, &zoom)) {
+									if (PyArg_ParseTuple(ret.get(), "iddd", &pagenum, &left, &top, &zoom)) {
 										PdfPage *page = NULL;
 										try {
 											page = self->doc->GetPage(pagenum - 1);
 										} catch(const PdfError &err) {
 											PyErr_Format(PyExc_ValueError, "No page number %d in the PDF file", pagenum);
-											Py_DECREF(ret);
 											return NULL;
 										}
 										if (page) {
@@ -473,7 +472,6 @@ PDFDoc_alter_links(PDFDoc *self, PyObject *args) {
 										}
 									}
 								}
-								Py_DECREF(ret);
 							}
 						}
 					}