X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_robot_base.py;h=fd7237ae0184033797ef30a6a773fe9b200c10de;hb=ea440b7a93009308e8d87351725b600f1929b181;hp=d8877c6f41c6b6f30c8d8ba4b8dbcec396e6c02b;hpb=8c04e58972d1c58ab82250df093c3d503eed4fe2;p=bookmarks_db.git

diff --git a/Robots/bkmk_robot_base.py b/Robots/bkmk_robot_base.py
index d8877c6..fd7237a 100644
--- a/Robots/bkmk_robot_base.py
+++ b/Robots/bkmk_robot_base.py
@@ -90,7 +90,7 @@ class robot_base(Robot):
                 try:
                     size = headers["Content-Length"]
                 except KeyError:
-                    size = len(content)
+                    pass
 
                 try:
                     last_modified = headers["Last-Modified"]
@@ -99,7 +99,8 @@ class robot_base(Robot):
 
                 if last_modified:
                     last_modified = parse_time(last_modified)
-            else:
+
+            if not size:  # Could be None from headers
                 size = len(content)
 
             if last_modified:
@@ -149,8 +150,14 @@ class robot_base(Robot):
                         if content_type.startswith(ctype):
                             is_html = True
                             break
-                    if content and is_html:
-                        parser = parse_html(content, charset, self.log)
+                    content_stripped = content.strip()
+                    if content_stripped and is_html:
+                        parser = parse_html(
+                            content_stripped, charset, self.log)
+                        if charset:
+                            bookmark.charset = charset
+                        elif parser and parser.meta_charset:
+                            bookmark.charset = parser.meta_charset
                         if parser:
                             bookmark.real_title = parser.title
                             icon = parser.icon
@@ -246,7 +253,7 @@ class robot_base(Robot):
                                                   % (url, timeout)
                                                   )
 
-                    if not content:
+                    if not content_stripped:
                         self.log("   empty response, no content")
                     if not is_html:
                         self.log("   not html")