[gtk-doc] tools/c10e-html: strip html produced by xslt

[Stefan Sauer <stefkost src gnome org>]

commit 6a9b3c293af818b1a3db8acf39c7758828284abc
Author: Stefan Sauer <ensonic users sf net>
Date:   Fri Apr 13 09:38:15 2018 +0200

    tools/c10e-html: strip html produced by xslt
    
    Remove some content that we won't produce in the new version to ease diffing.

 tools/c10e-html.py |   14 +++++++++++++-
 1 files changed, 13 insertions(+), 1 deletions(-)
---
diff --git a/tools/c10e-html.py b/tools/c10e-html.py
index d90ff7b..81e8a8d 100755
--- a/tools/c10e-html.py
+++ b/tools/c10e-html.py
@@ -6,6 +6,7 @@
 
 import argparse
 import glob
+import re
 import os
 import sys
 
@@ -16,7 +17,18 @@ def prettify(filename):
     with open(filename, 'r') as doc:
         soup = BeautifulSoup(doc.read(), 'lxml')
     with open(filename, 'w') as doc:
-        doc.write(soup.prettify())
+        html = soup.prettify()
+        # strip things that mkhtml2 is not producing to reduce the diff
+        html = html.replace('a class="link" href', 'a href')
+        html = html.replace('summary="Navigation header" ', '')
+        html = html.replace("""  <div class="footer">
+   <hr/>
+   Generated by GTK-Doc V1.28.1
+  </div>
+""", '')
+        html = re.sub(r'\s*<meta content="DocBook[^>]*>', '', html)
+        html = re.sub(r'\s*<meta content="GTK-Doc[^>]*>', '', html)
+        doc.write(html)
 
 
 def main(htmldir):