From 9a003e49791cffe0fb360a2821086d2bc2dbeea7 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Roland=20H=C3=A4der?= <roland@mxchange.org>
Date: Mon, 22 May 2023 04:52:13 +0200
Subject: [PATCH] Continued: - fetching .well-known/nodeinfo for auto-discovery
 added

---
 fba.py          | 31 +++++++++++++++++++++++++++++++
 fetch_blocks.py |  6 +++---
 2 files changed, 34 insertions(+), 3 deletions(-)

diff --git a/fba.py b/fba.py
index 954318d..4e74fd8 100644
--- a/fba.py
+++ b/fba.py
@@ -156,6 +156,37 @@ def fetch_nodeinfo(domain: str) -> list:
     # NOISY-DEBUG: print("DEBUG: json[]:", type(json))
     if json is None or len(json) == 0:
         print("WARNING: Failed fetching nodeinfo from domain:", domain)
+        json = fetch_wellknown_nodeinfo(domain)
+
+    # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
+    return json
+
+def fetch_wellknown_nodeinfo(domain: str) -> list:
+    # NOISY-DEBUG: print("DEBUG: Fetching .well-known info for domain:", domain)
+    json = {}
+
+    try:
+        res = reqto.get(f"https://{domain}/.well-known/nodeinfo", headers=headers, timeout=(config["connection_timeout"], config["read_timeout"]))
+        # NOISY-DEBUG: print("DEBUG: domain,res.ok:", domain, res.ok)
+        if res.ok and res.json() is not None:
+            nodeinfo = res.json()
+            # NOISY-DEBUG: print("DEBUG: Found entries:", len(nodeinfo), domain)
+            if "links" in nodeinfo:
+                # NOISY-DEBUG: print("DEBUG: Found links in nodeinfo():", len(nodeinfo["links"]))
+                for link in nodeinfo["links"]:
+                    # NOISY-DEBUG: print("DEBUG: rel,href:", link["rel"], link["href"])
+                    if link["rel"] == "http://nodeinfo.diaspora.software/ns/schema/2.0":
+                        # NOISY-DEBUG: print("DEBUG: Fetching nodeinfo from:", link["href"])
+                        res = reqto.get(link["href"])
+                        # NOISY-DEBUG: print("DEBUG: href,res.ok:", link["href"], res.ok)
+                        if res.ok and res.json() is not None:
+                            # NOISY-DEBUG: print("DEBUG: Found JSON nodeinfo():", len(res.json()))
+                            json = res.json()
+                            break
+
+    except:
+        print("WARNING: Failed fetching .well-known info:", domain)
+        pass
 
     # NOISY-DEBUG: print("DEBUG: Returning json[]:", type(json))
     return json
diff --git a/fetch_blocks.py b/fetch_blocks.py
index 4bdf158..0f75e15 100644
--- a/fetch_blocks.py
+++ b/fetch_blocks.py
@@ -226,7 +226,7 @@ for blocker, software in fba.cursor.fetchall():
                     elif blocked.count("*") < 1:
                         # No obsfucation for this instance
                         fba.cursor.execute(
-                            "SELECT hash FROM instances WHERE domain = ? LIMIT 1", [blocked]
+                            "SELECT hash FROM instances WHERE domain = ?", [blocked]
                         )
 
                         if fba.cursor.fetchone() == None:
@@ -369,7 +369,7 @@ for blocker, software in fba.cursor.fetchall():
                     if blocked == "":
                         print("WARNING: blocked is empty:", blocker)
                         continue
-                    if blocked.count("*") > 0:
+                    elif blocked.count("*") > 0:
                         # GTS does not have hashes for obscured domains, so we have to guess it
                         fba.cursor.execute(
                             "SELECT domain FROM instances WHERE domain LIKE ? ORDER BY rowid LIMIT 1", [blocked.replace("*", "_")]
@@ -419,7 +419,7 @@ for blocker, software in fba.cursor.fetchall():
         except Exception as e:
             print("error:", e, blocker, software)
     else:
-        print("WARNING: Unknown software:", software)
+        print("WARNING: Unknown software:", blocker, software)
 
     if fba.config["bot_enabled"] and len(blockdict) > 0:
         send_bot_post(blocker, blockdict)
-- 
2.39.5