[gtk-doc] tools/c10e-html: run fixup conditionally

[Stefan Sauer <stefkost src gnome org>]

commit fed55be5678cc8b609e5def1e65c3e985cac8f49
Author: Stefan Sauer <ensonic users sf net>
Date:   Thu May 10 19:23:23 2018 +0200

    tools/c10e-html: run fixup conditionally
    
    We only want to run this on the html files.

 tools/c10e-html.py |   31 ++++++++++++++++---------------
 1 files changed, 16 insertions(+), 15 deletions(-)
---
diff --git a/tools/c10e-html.py b/tools/c10e-html.py
index a568125..f020f67 100755
--- a/tools/c10e-html.py
+++ b/tools/c10e-html.py
@@ -13,36 +13,37 @@ import sys
 from bs4 import BeautifulSoup
 
 
-def prettify(filename):
+def prettify(filename, parser='lxml', fixup=False):
     with open(filename, 'r') as doc:
-        soup = BeautifulSoup(doc.read(), 'lxml')
+        soup = BeautifulSoup(doc.read(), parser)
     with open(filename, 'w') as doc:
         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(' target="_top"', '')
-        html = html.replace('summary="Navigation header" ', '')
-        html = html.replace("""   <a name="idx">
+        if fixup:
+            # strip things that mkhtml2 is not producing to reduce the diff
+            html = html.replace('a class="link" href', 'a href')
+            html = html.replace(' target="_top"', '')
+            html = html.replace('summary="Navigation header" ', '')
+            html = html.replace("""   <a name="idx">
    </a>
 """, '')
-        html = re.sub("""  <div class="footer">
+            html = re.sub("""  <div class="footer">
    <hr/>
    Generated by GTK-Doc V[.0-9]*
   </div>
 """, '', html)
-        html = re.sub(r'\s*<p>\s*</p>', '', html)
-        html = re.sub(r'\s*<a name="id-[.0-9]+">\s*</a>', '', html)
-        html = re.sub(r'\s*<div class="titlepage">\s*</div>', '', html)
-        html = re.sub(r'\s*<meta content="DocBook[^>]*>', '', html)
-        html = re.sub(r'\s*<meta content="GTK-Doc[^>]*>', '', html)
+            html = re.sub(r'\s*<p>\s*</p>', '', html)
+            html = re.sub(r'\s*<a name="id-[.0-9]+">\s*</a>', '', html)
+            html = re.sub(r'\s*<div class="titlepage">\s*</div>', '', html)
+            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):
     for filename in glob.glob(os.path.join(htmldir, '*.devhelp2')):
-        prettify(filename)
+        prettify(filename, parser='lxml-xml')
     for filename in glob.glob(os.path.join(htmldir, '*.html')):
-        prettify(filename)
+        prettify(filename, fixup=True)
 
 
 if __name__ == '__main__':