From dfa5901bb6dfd97ff000b0808161e4cd1cd4a35c Mon Sep 17 00:00:00 2001
From: =?utf8?q?Roland=20H=C3=A4der?= <roland@mxchange.org>
Date: Mon, 1 Jul 2024 23:58:25 +0200
Subject: [PATCH] Continued: - stop question marks in domain.raise_on()
 function - split off question mark - added space for better readability -
 fixed syntax error

---
 fba/commands.py         | 2 +-
 fba/helpers/domain.py   | 2 ++
 fba/http/federation.py  | 5 +++--
 fba/models/instances.py | 1 -
 4 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/fba/commands.py b/fba/commands.py
index 05cdd7e..a514c79 100644
--- a/fba/commands.py
+++ b/fba/commands.py
@@ -1009,7 +1009,7 @@ def fetch_instances(args: argparse.Namespace) -> int:
         elif row["software"] is not None and software_helper.is_relay(row["software"]) and instances.is_registered(row["domain"]):
             logger.warning("row[domain]='%s' is of software type '%s' which is not supported by this command. Please invoke fetch_relays instead - SKIPPED!", row["domain"], row["software"])
             continue
-        elif not args.force and not args.software in [None, ""]and instances.is_recent(row["domain"]):
+        elif not args.force and not args.software in [None, ""] and instances.is_recent(row["domain"]):
             logger.debug("row[domain]='%s' has recently been crawled - SKIPPED!", row["domain"])
             continue
 
diff --git a/fba/helpers/domain.py b/fba/helpers/domain.py
index df4c703..fcf672a 100644
--- a/fba/helpers/domain.py
+++ b/fba/helpers/domain.py
@@ -38,6 +38,8 @@ def raise_on(domain: str):
         raise ValueError("Parameter 'domain' is empty")
     elif domain.lower() != domain:
         raise ValueError(f"Parameter domain='{domain}' must be all lower-case")
+    elif "?" in domain:
+        raise ValueError(f"Parameter domain='{domain}' contains a question-mark")
     elif not validators.domain(domain.split("/")[0]):
         raise ValueError(f"domain='{domain}' is not a valid domain")
     elif domain.endswith(".onion"):
diff --git a/fba/http/federation.py b/fba/http/federation.py
index eae8f8e..6fea3ea 100644
--- a/fba/http/federation.py
+++ b/fba/http/federation.py
@@ -162,6 +162,7 @@ def fetch_instances(domain: str, origin: str, software: str, command: str, path:
 
         logger.debug("instance='%s' - BEFORE!", instance)
         instance = instance.encode("idna").decode("utf-8")
+        instance = instance.split("?")[0]
         logger.debug("instance='%s' - AFTER!", instance)
 
         if not domain_helper.is_wanted(instance):
@@ -295,8 +296,8 @@ def fetch_generator_from_path(domain: str, path: str = "/") -> str:
     )
 
     logger.debug("response.ok='%s',response.status_code=%d,response.text()=%d", response.ok, response.status_code, len(response.text))
-    response_url = response.url.split("#")[0], response_url)
-    logger.debug("response_url='%s'"
+    response_url = response.url.split("#")[0]
+    logger.debug("response_url='%s'", response_url)
 
     if ((response.ok and response.status_code == 200) or response.status_code == 410) and response.text.find("<html") >= 0 and validators.url(response_url) and domain_helper.is_in_url(domain, response_url):
         logger.debug("Parsing response.text()=%d Bytes ...", len(response.text))
diff --git a/fba/models/instances.py b/fba/models/instances.py
index 325be37..f75bd6b 100644
--- a/fba/models/instances.py
+++ b/fba/models/instances.py
@@ -538,7 +538,6 @@ def set_original_software(domain: str, software: str):
     _set_data("original_software", domain, software)
     logger.debug("EXIT!")
 
-
 def set_software(domain: str, software: str):
     logger.debug("domain='%s',software='%s' - CALLED!", domain, software)
     domain_helper.raise_on(domain)
-- 
2.39.5