From 78a97a0e37fe66ca870ee9097d91507b67a9c721 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Sun, 27 Oct 2019 12:52:12 +0530
Subject: [PATCH] Handle comments while getting char count

---
 src/calibre/srv/render_book.py   | 12 +++++++++---
 src/calibre/srv/tests/content.py |  2 ++
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/src/calibre/srv/render_book.py b/src/calibre/srv/render_book.py
index 6f7762a894..1ea6166ad6 100644
--- a/src/calibre/srv/render_book.py
+++ b/src/calibre/srv/render_book.py
@@ -202,8 +202,11 @@ def get_length(root):
         strip_space = re.compile(r'\s+')
 
         def count(elem):
+            tag = getattr(elem, 'tag', count)
+            if callable(tag):
+                return len(strip_space.sub('', getattr(elem, 'tail', None) or ''))
             num = 0
-            tname = elem.tag.rpartition('}')[-1].lower()
+            tname = tag.rpartition('}')[-1].lower()
             if elem.text and tname not in ignore_tags:
                 num += len(strip_space.sub('', elem.text))
             if elem.tail:
@@ -213,11 +216,14 @@ def get_length(root):
             return num
     else:
         def count(elem):
-            return fast(elem.tag, elem.text, elem.tail)
+            tag = getattr(elem, 'tag', count)
+            if callable(tag):
+                return fast('', None, getattr(elem, 'tail', None))
+            return fast(tag, elem.text, elem.tail)
 
     for body in root.iterchildren(XHTML('body')):
         ans += count(body)
-        for elem in body.iterdescendants('*'):
+        for elem in body.iterdescendants():
             ans += count(elem)
     return ans
 
diff --git a/src/calibre/srv/tests/content.py b/src/calibre/srv/tests/content.py
index 9cc3122519..ccc51a4045 100644
--- a/src/calibre/srv/tests/content.py
+++ b/src/calibre/srv/tests/content.py
@@ -234,4 +234,6 @@ class ContentTest(LibraryBaseTest):
         self.ae(get_length(root), 5)
         root = html5_parse('<script>xyz</script>a<iMg>b')
         self.ae(get_length(root), 1002)
+        root = html5_parse('<p><!-- abc -->m')
+        self.ae(get_length(root), 1)
     # }}}