Continued:

- "guessing" it didn't really work, let's take <meta name='generator'> as the
  software type, some people even change that and have no /.well-known/ path
  enabled. Then they cannot federate anymore, at least not by auto-discovery

diff --git a/fba.py b/fba.py
index 9427b147b3ed645c13f20ea88020a431f8ef62e1..27ee3bac6f52411477ad86f95ee8f7686c6e3a50 100644 (file)
--- a/fba.py
+++ b/fba.py
@@ -424,20 +424,30 @@ def determine_software(domain: str) -> str:
         update_last_error(domain, json["message"])
         return None
     elif "software" not in json or "name" not in json["software"]:
-        print("WARNING: JSON response does not include [software][name], guessing ...")
-        found = 0
-        for element in {"uri", "title", "short_description", "description", "email", "version", "urls", "stats", "thumbnail", "languages", "contact_account", "registrations", "approval_required"}:
-            # NOISY-DEBUG: print("DEBUG: element:", element)
-            if element in json:
-                found = found + 1
-
-        # NOISY-DEBUG: print("DEBUG: Found elements:", found)
-        if found == len(json):
-            # NOISY-DEBUG: print("DEBUG: Maybe is Mastodon:", domain)
-            return "mastodon"
-
-        print(f"WARNING: Cannot guess software type: domain='{domain}',found={found},json()={len(json)}")
-        return None
+        # NOISY-DEBUG: print(f"DEBUG: JSON response from {domain} does not include [software][name], fetching / ...")
+        try:
+            res = reqto.get(f"https://{domain}/", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
+
+            # NOISY-DEBUG: print("DEBUG: domain,res.ok,res.status_code:", domain, res.ok, res.status_code)
+            if res.ok and res.status_code < 300 and len(res.text) > 0:
+                # NOISY-DEBUG: print("DEBUG: Search for <meta name='generator'>:", domain)
+
+                doc = bs4.BeautifulSoup(res.text, "html.parser")
+                # NOISY-DEBUG: print("DEBUG: doc[]:", type(doc))
+
+                tag = doc.find("meta", {"name": "generator"})
+                # NOISY-DEBUG: print(f"DEBUG: tag[{type(tag)}: {tag}")
+                if isinstance(tag, bs4.element.Tag):
+                    # NOISY-DEBUG: print("DEBUG: Found generator meta tag:", domain)
+                    software = tidyup(tag.get("content"))
+
+        except BaseException as e:
+            print(f"WARNING: Cannot fetch / from '{domain}':", e)
+            update_last_error(domain, e)
+            pass
+
+        # NOISY-DEBUG: print(f"DEBUG: Generator for domain='{domain}' is: {software}, EXIT!")
+        return software
 
     software = tidyup(json["software"]["name"])