From b2c8f6f8ee1c27712bfc8d84cdc1939aab8183e5 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Thu, 3 Mar 2022 15:33:33 +0530
Subject: [PATCH] PDF Output: Fix conversion failing if there are ToC entries
 pointing to removed content. Fixes #1960554 [Private
 bug](https://bugs.launchpad.net/calibre/+bug/1960554)

---
 src/calibre/ebooks/pdf/html_writer.py | 8 +++++++-
 src/calibre/utils/podofo/doc.cpp      | 7 +++++--
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/src/calibre/ebooks/pdf/html_writer.py b/src/calibre/ebooks/pdf/html_writer.py
index ecf8b1304b..986f0153db 100644
--- a/src/calibre/ebooks/pdf/html_writer.py
+++ b/src/calibre/ebooks/pdf/html_writer.py
@@ -569,6 +569,7 @@ def get_anchor_locations(name, pdf_doc, first_page_num, toc_uuid, log):
 
 
 def fix_links(pdf_doc, anchor_locations, name_anchor_map, mark_links, log):
+    pc = pdf_doc.page_count()
 
     def replace_link(url):
         purl = urlparse(url)
@@ -583,7 +584,12 @@ 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(f'Anchor location for link to {purl.fragment} not found')
-        return None if loc is None else loc.as_tuple
+        if loc is None:
+            return None
+        if loc.pagenum > pc:
+            log.warn(f'Anchor location for link to {purl.fragment} is past the end of the document, moving it to last page')
+            loc.pagenum = pc
+        return loc.as_tuple
 
     pdf_doc.alter_links(replace_link, mark_links)
 # }}}
diff --git a/src/calibre/utils/podofo/doc.cpp b/src/calibre/utils/podofo/doc.cpp
index 7cedee1a95..25f814a3c4 100644
--- a/src/calibre/utils/podofo/doc.cpp
+++ b/src/calibre/utils/podofo/doc.cpp
@@ -483,7 +483,7 @@ alter_link(PDFDoc *self, PdfDictionary &link, PyObject *alter_callback, bool mar
                 page = self->doc->GetPage(pagenum - 1);
             } catch(const PdfError &err) {
                 (void)err;
-                PyErr_Format(PyExc_ValueError, "No page number %d in the PDF file", pagenum);
+                PyErr_Format(PyExc_ValueError, "No page number %d in the PDF file of %d pages", pagenum, self->doc->GetPageCount());
                 return ;
             }
             if (page) {
@@ -525,7 +525,10 @@ PDFDoc_alter_links(PDFDoc *self, PyObject *args) {
 		}
         for (auto const & ref: links) {
             PdfObject *lo = self->doc->GetObjects().GetObject(ref);
-            if (lo) alter_link(self, lo->GetDictionary(), alter_callback, mark_links, border, link_color);
+            if (lo) {
+                alter_link(self, lo->GetDictionary(), alter_callback, mark_links, border, link_color);
+                if (PyErr_Occurred()) return NULL;
+            }
         }
     } catch(const PdfError & err) {
         podofo_set_exception(err);