Continued:

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

diff --git a/fba/helpers/domain.py b/fba/helpers/domain.py
index 4141258665c43c64e6acb2c3c39492961ad3ee51..39fbb7797acee74c7c5207ed0b3b60c507276b2f 100644 (file)
--- a/fba/helpers/domain.py
+++ b/fba/helpers/domain.py
@@ -16,6 +16,7 @@
 
 import logging
 
+from functools import lru_cache
 from urllib.parse import urlparse
 
 import validators
@@ -28,18 +29,6 @@ from fba.models import instances
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-# In-function cache
-_cache = {
-    # Cache for function is_in_url()
-    "is_in_url": {},
-
-    # Cache for function is_wanted()
-    "is_wanted": {},
-
-    # Cache for function raise_on()
-    "raise_on": {},
-}
-
 def raise_on(domain: str):
     logger.debug("domain='%s' - CALLED!", domain)
 
@@ -47,9 +36,6 @@ def raise_on(domain: str):
         raise ValueError(f"Parameter domain[]='{type(domain)}' is not of type 'str'")
     elif domain == "":
         raise ValueError("Parameter 'domain' is empty")
-    elif domain in _cache["raise_on"]:
-        logger.debug("Returning cached raised_on='%s' - EXIT!", _cache["raise_on"][domain])
-        return _cache["raise_on"][domain]
     elif domain.lower() != domain:
         raise ValueError(f"Parameter domain='{domain}' must be all lower-case")
     elif not validators.domain(domain.split("/")[0]):
@@ -63,9 +49,9 @@ def raise_on(domain: str):
     elif domain.endswith(".tld"):
         raise ValueError(f"domain='{domain}' is a fake domain, please don't crawl them!")
 
-    _cache["raise_on"][domain] = True
     logger.debug("EXIT!")
 
+@lru_cache
 def is_in_url(domain: str, url: str) -> bool:
     logger.debug("domain='%s',url='%s' - CALLED!", domain, url)
     raise_on(domain)
@@ -76,9 +62,8 @@ def is_in_url(domain: str, url: str) -> bool:
         raise ValueError(f"Parameter url[]='{type(url)}' is not of type 'str'")
     elif url == "":
         raise ValueError("Parameter 'url' is empty")
-    elif domain + url in _cache["is_in_url"]:
-        logger.debug("Returning cached is_in_url='%s' - EXIT!", _cache["is_in_url"][domain + url])
-        return _cache["is_in_url"][domain + url]
+    elif not validators.url(url):
+        raise ValueError(f"Parameter url='{url}' is not a valid URL")
 
     punycode = domain.encode("idna").decode("utf-8")
 
@@ -87,12 +72,10 @@ def is_in_url(domain: str, url: str) -> bool:
 
     is_found = (punycode in [components.netloc, components.hostname])
 
-    # Set cache
-    _cache["is_in_url"][domain + url] = is_found
-
     logger.debug("is_found='%s' - EXIT!", is_found)
     return is_found
 
+@lru_cache
 def is_wanted(domain: str) -> bool:
     logger.debug("domain='%s' - CALLED!", domain)
 
@@ -100,9 +83,6 @@ def is_wanted(domain: str) -> bool:
         raise ValueError(f"Parameter domain[]='{type(domain)}' is not of type 'str'")
     elif domain == "":
         raise ValueError("Parameter 'domain' is empty")
-    elif domain in _cache["is_wanted"]:
-        logger.debug("Returning cached is_found='%s' - EXIT!", _cache["is_wanted"][domain])
-        return _cache["is_wanted"][domain]
 
     wanted = True
     if domain.lower() != domain:
@@ -133,8 +113,5 @@ def is_wanted(domain: str) -> bool:
         logger.debug("domain='%s' is a tag", domain)
         wanted = False
 
-    # Set cache
-    _cache["is_wanted"][domain] = wanted
-
     logger.debug("wanted='%s' - EXIT!", wanted)
     return wanted