From 652d069c3509b179790548f03e14cd37b2c6e5dd Mon Sep 17 00:00:00 2001
From: Castro0o <andre@andrecastro>
Date: Tue, 7 Apr 2020 16:33:33 +0200
Subject: [PATCH] removing imgs srcset value

---
 dumpwiki.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/dumpwiki.py b/dumpwiki.py
index a3dc706..80ede08 100644
--- a/dumpwiki.py
+++ b/dumpwiki.py
@@ -67,6 +67,15 @@ def rewritelinks (html):
     html = ET.tostring(t, method="html", encoding="unicode")
     return html
 
+
+def rewriteimgs(html):
+    t = html5lib.parseFragment(html, treebuilder = "etree", namespaceHTMLElements = False)
+    # remove the srcset value as it prevents images from displaying
+    for img in t.findall(".//img[@srcset]"):
+        img.attrib['srcset'] = ""
+    html = ET.tostring(t, method="html", encoding="unicode")
+    return html
+
 publish=site.Categories['Publish']
 for cat in publish.members():
     if cat.namespace != 14:
@@ -83,6 +92,8 @@ for cat in publish.members():
         print(p)
         htmlsrc = site.parse(page=p.name)['text']['*']
         htmlsrc = rewritelinks(htmlsrc)
+        htmlsrc = rewriteimgs(htmlsrc)
+
         if args.local is True:
             html = template.render(page=p, body=htmlsrc, staticpath='..')
         else: