Regex for unwanted extensions (#1907)

* allow regexp for unwanted extensions

* housekeeping

* use rss.convert_filter

* improve function name

* move convert_filter to misc

sabnzbd/filesystem.py

@@ -59,6 +59,25 @@ def get_ext(filename: str) -> str:
         return ""
 
 
+def is_listed_ext(ext: str, ext_list: list) -> bool:
+    """Check if the extension is listed. In case of a regexp the entire extension must be matched;
+    partial matches aren't accepted (e.g. 'r[0-9]{2}' will be treated the same as '^r[0-9]{2}$' and
+    thus return false for extentions such as 'r007' despite the substring match on 'r00').
+    """
+    for item in ext_list:
+        RE_EXT = sabnzbd.misc.convert_filter(item)
+        if RE_EXT:
+            try:
+                if len(RE_EXT.match(ext).group()) == len(ext):
+                    return True
+            except Exception:
+                pass
+        elif item == ext:
+            return True
+    # No match found
+    return False
+
+
 def has_unwanted_extension(filename: str) -> bool:
     """Determine if a filename has an unwanted extension, given the configured mode"""
     extension = get_ext(filename).replace(".", "")
@@ -66,11 +85,11 @@ def has_unwanted_extension(filename: str) -> bool:
         return (
             # Blacklisted
             sabnzbd.cfg.unwanted_extensions_mode() == 0
-            and extension in sabnzbd.cfg.unwanted_extensions()
+            and is_listed_ext(extension, sabnzbd.cfg.unwanted_extensions())
         ) or (
             # Not whitelisted
             sabnzbd.cfg.unwanted_extensions_mode() == 1
-            and extension not in sabnzbd.cfg.unwanted_extensions()
+            and not is_listed_ext(extension, sabnzbd.cfg.unwanted_extensions())
         )
     else:
         # Don't consider missing extensions unwanted to prevent indiscriminate blocking of

sabnzbd/misc.py

@@ -220,6 +220,23 @@ def wildcard_to_re(text):
     return "".join([_wildcard_to_regex.get(ch, ch) for ch in text])
 
 
+def convert_filter(text):
+    """Return compiled regex.
+    If string starts with re: it's a real regex
+    else quote all regex specials, replace '*' by '.*'
+    """
+    text = text.strip().lower()
+    if text.startswith("re:"):
+        txt = text[3:].strip()
+    else:
+        txt = wildcard_to_re(text)
+    try:
+        return re.compile(txt, re.I)
+    except:
+        logging.debug("Could not compile regex: %s", text)
+        return None
+
+
 def cat_convert(cat):
     """Convert indexer's category/group-name to user categories.
     If no match found, but indexer-cat equals user-cat, then return user-cat

sabnzbd/rss.py

@@ -31,7 +31,7 @@ from sabnzbd.constants import RSS_FILE_NAME, DEFAULT_PRIORITY, DUP_PRIORITY
 from sabnzbd.decorators import synchronized
 import sabnzbd.config as config
 import sabnzbd.cfg as cfg
-from sabnzbd.misc import cat_convert, wildcard_to_re, cat_to_opts, match_str, from_units, int_conv, get_base_url
+from sabnzbd.misc import cat_convert, convert_filter, cat_to_opts, match_str, from_units, int_conv, get_base_url
 import sabnzbd.emailer as emailer
 
 import feedparser
@@ -47,23 +47,6 @@ def notdefault(item):
     return bool(item) and str(item).lower() not in ("default", "*", "", str(DEFAULT_PRIORITY))
 
 
-def convert_filter(text):
-    """Return compiled regex.
-    If string starts with re: it's a real regex
-    else quote all regex specials, replace '*' by '.*'
-    """
-    text = text.strip().lower()
-    if text.startswith("re:"):
-        txt = text[3:].strip()
-    else:
-        txt = wildcard_to_re(text)
-    try:
-        return re.compile(txt, re.I)
-    except:
-        logging.debug("Could not compile regex: %s", text)
-        return None
-
-
 def remove_obsolete(jobs, new_jobs):
     """Expire G/B links that are not in new_jobs (mark them 'X')
     Expired links older than 3 days are removed from 'jobs'

tests/test_filesystem.py

@@ -1094,7 +1094,7 @@ class TestRenamer:
 class TestUnwantedExtensions:
     # Only test lowercase extensions without a leading dot: the unwanted_extensions
     # setting is sanitized accordingly in interface.saveSwitches() before saving.
-    test_extensions = "iso, cmd, bat, sh"
+    test_extensions = "iso, cmd, bat, sh, re:r[0-9]{2}, sab*"
     # Test parameters as (filename, result) tuples, with result given for blacklist mode
     test_params = [
         ("ubuntu.iso", True),
@@ -1109,7 +1109,19 @@ class TestUnwantedExtensions:
         ("par2.cmd.notcmd", False),
         ("freedos.tab", False),
         (".SH.hs", False),
+        ("regexp.r0611", False),
+        ("regexp.007", False),
+        ("regexp.A01", False),
+        ("regexp.r9", False),
+        ("regexp.r2d2", False),
+        ("regexp.r2d", False),
+        ("regexp.r00", True),
+        ("regexp.R42", True),
+        ("test.sabnzbd", True),
+        ("pass.sab", True),
+        ("fail.sb", False),
         ("No_Extension", False),
+        ("r42", False),
         (480, False),
         (None, False),
         ("", False),