]> git.phdru.name Git - bookmarks_db.git/commitdiff
Fix(robots): Process response without `Content-Type`
authorOleg Broytman <phd@phdru.name>
Sun, 12 Nov 2023 18:19:58 +0000 (21:19 +0300)
committerOleg Broytman <phd@phdru.name>
Mon, 13 Nov 2023 19:02:26 +0000 (22:02 +0300)
Try to recognize HTML.

Robots/bkmk_robot_base.py

index 9a74cb4ad67e342f96ec087e99186bb2eb69c4d2..b5cac691fe9f2842d12184f30de6c6b2271dbb83 100644 (file)
@@ -126,6 +126,13 @@ class robot_base(Robot):
                 try:
                     content_type = headers["Content-Type"]
                     self.log("   Content-Type: %s" % content_type)
+                    if content_type is None:
+                        if 'html' in content.lower():
+                            content_type = 'text/html'
+                        else:
+                            content_type = 'text/plain'
+                        self.log("   Set Content-Type to: %s"
+                                 % content_type)
                     try:
                         # extract charset from
                         # "text/html; foo; charset=UTF-8, bar; baz;"
@@ -187,9 +194,11 @@ class robot_base(Robot):
                                 icons[icon_url] = None
                             else:
                                 content_type = icon_headers["Content-Type"]
-                                if content_type.startswith("application/") \
-                                   or content_type.startswith("image/") \
-                                   or content_type.startswith("text/plain"):
+                                if content_type and (
+                                    content_type.startswith("application/")
+                                    or content_type.startswith("image/")
+                                    or content_type.startswith("text/plain")
+                                ):
                                     bookmark.icon_href = icon_url
                                     self.log("   got icon       : %s"
                                              % content_type)