Continued:

[fba.git] / fba / helpers / tidyup.py

diff --git a/fba/helpers/tidyup.py b/fba/helpers/tidyup.py
index f09c2824eaeb9032f771398a672ae66ac1533d61..acd8bc655475bfc6cd4b472a0b0b1f53cd6edd54 100644 (file)
--- a/fba/helpers/tidyup.py
+++ b/fba/helpers/tidyup.py
@@ -20,49 +20,63 @@ logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
 def reason(string: str) -> str:
-    logger.debug(f"string='{string}' - CALLED!")
+    logger.debug("string='%s' - CALLED!", string)
+
     if not isinstance(string, str):
-        raise ValueError(f"Parameter string[]='{type(string)}' is not 'str'")
+        raise ValueError(f"Parameter string[]='{type(string)}' is not of type 'str'")
 
     # Strip string
     string = string.strip()
 
-    logger.debug(f"string='{string}' - EXIT!")
+    logger.debug("string='%s' - EXIT!", string)
     return string
 
 def domain(string: str) -> str:
-    logger.debug(f"string='{string}' - CALLED!")
+    logger.debug("string='%s' - CALLED!", string)
+
     if not isinstance(string, str):
-        raise ValueError(f"Parameter string[]='{type(string)}' is not 'str'")
+        raise ValueError(f"Parameter string[]='{type(string)}' is not of type 'str'")
+    elif string == "":
+        raise ValueError("Parameter string is empty")
 
     # All lower-case and strip spaces out + last dot
-    string = string.lower().strip().rstrip(".")
-    logger.debug(f"string='{string}' - #1")
+    string = string.lower().strip().rstrip(".").replace("..", ".")
+    logger.debug("string='%s' - #1", string)
 
     # No port number
-    string = re.sub("\:\d+$", "", string)
-    logger.debug(f"string='{string}' - #2")
+    string = re.sub(r"\:\d+$", "", string)
+    logger.debug("string='%s' - #2", string)
 
     # No protocol, sometimes without the slashes
-    string = re.sub("^https?\:(\/*)", "", string)
-    logger.debug(f"string='{string}' - #3")
+    string = re.sub(r"^https?\:(\/*)", "", string)
+    logger.debug("string='%s' - #3", string)
 
     # No trailing slash
-    string = re.sub("\/$", "", string)
-    logger.debug(f"string='{string}' - #4")
+    string = re.sub(r"\/$", "", string)
+    logger.debug("string='%s' - #4", string)
 
     # No @ or : sign
-    string = re.sub("^\@", "", string)
+    string = re.sub(r"^\@", "", string)
     string = string.split(":")[0]
-    logger.debug(f"string='{string}' - #4")
+    logger.debug("string='%s' - #5", string)
+
+    # Try to "detect" user profiles, not wanted here. Don't block single users
+    # in an instance block list! Everything personal can be solved in a
+    # personal block.
+    string = re.sub(r"(.+)\@", "", string)
+    logger.debug("string='%s' - #6", string)
 
-    # No individual users in block lists
-    string = re.sub("(.+)\@", "", string)
-    logger.debug(f"string='{string}' - #5")
     if string.find("/profile/"):
         string = string.split("/profile/")[0]
     elif string.find("/users/"):
         string = string.split("/users/")[0]
+    elif string.find("/tag/"):
+        string = string.split("/tag/")[0]
+
+    # Some people have TLDs with this word on the end
+    logger.debug("string='%s' - #7", string)
+    if string.endswith("silence"):
+        string = string.split("silence")[0]
 
-    logger.debug(f"string='{string}' - EXIT!")
+    logger.debug("string='%s' - EXIT!", string)
     return string