Locking blacklist while reading, make it consistent amongst multiprocesses

validate_email/domainlist_check.py

@@ -1,9 +1,11 @@
 from logging import getLogger
 from typing import Optional
 
+from filelock import FileLock
+
 from .exceptions import DomainBlacklistedError
 from .updater import (
-    BLACKLIST_FILEPATH_INSTALLED, BLACKLIST_FILEPATH_TMP,
+    BLACKLIST_FILEPATH_INSTALLED, BLACKLIST_FILEPATH_TMP, LOCK_PATH,
     update_builtin_blacklist)
 
 SetOrNone = Optional[set]
@@ -30,8 +32,8 @@ class DomainListValidator(object):
     def _blacklist_path(self) -> str:
         'Return the path of the `blacklist.txt` that should be loaded.'
         try:
-            # Zero size, file is touched to indicate the
-            # preinstalled file is still fresh enough
+            # Zero size: file is touched to indicate the preinstalled
+            # file is still fresh enough
             return BLACKLIST_FILEPATH_INSTALLED \
                 if BLACKLIST_FILEPATH_TMP.stat().st_size == 0 \
                 else BLACKLIST_FILEPATH_TMP
@@ -42,13 +44,14 @@ class DomainListValidator(object):
         '(Re)load our built-in blacklist.'
         if not self._is_builtin_bl_used:
             return
-        bl_path = self._blacklist_path
-        LOGGER.debug(msg=f'(Re)loading blacklist: {bl_path}')
-        try:
-            with open(bl_path) as fd:
-                lines = fd.readlines()
-        except FileNotFoundError:
-            return
+        with FileLock(lock_file=LOCK_PATH):
+            bl_path = self._blacklist_path
+            LOGGER.debug(msg=f'(Re)loading blacklist from {bl_path}')
+            try:
+                with open(bl_path) as fd:
+                    lines = fd.readlines()
+            except FileNotFoundError:
+                return
         self.domain_blacklist = set(
             x.strip().lower() for x in lines if x.strip())
 

validate_email/updater.py

@@ -35,7 +35,6 @@ class BlacklistUpdater(object):
     """
 
     _refresh_when_older_than: int = 5 * 24 * 60 * 60  # 5 days
-    _on_update_callback: Callable = None
     _is_install_time: bool = False
 
     @property
@@ -122,16 +121,18 @@ class BlacklistUpdater(object):
                 BLACKLIST_FILEPATH_TMP.touch()
                 return
             raise
-        if self._on_update_callback:
-            self._on_update_callback()
 
     def process(
             self, force: bool = False, callback: Optional[Callable] = None):
         'Start optionally updating the blacklist.txt file.'
         # Locking to avoid multi-process update on multi-process startup
-        self._on_update_callback = callback
         with FileLock(lock_file=LOCK_PATH):
             self._process(force=force)
+        # Always execute callback because multiple processes can have
+        # different versions of blacklists (one before, one after
+        # updating)
+        if callback:
+            callback()
 
 
 def update_builtin_blacklist(